Research Article

Oral N-acetylcysteine rescues lethality of hepatocyte-specificGclc-knockout mice, providing a model for hepatic cirrhosis

Ying Chen1,2,⇑, Elisabet Johansson1, Yi Yang1, Marian L. Miller1, Dongxiao Shen1, David J. Orlicky3,Howard G. Shertzer1, Vasilis Vasiliou2, Daniel W. Nebert1, Timothy P. Dalton1

1Department of Environmental Health, University of Cincinnati, Cincinnati, OH 45267, USA; 2Department of Pharmaceutical Sciences, Universityof Colorado Denver, Aurora, CO 80045, USA; 3Department of Pathology, University of Colorado Denver, Aurora, CO 80045, USA

Background & Aims: Certain liver diseases have been associated sis to progress to a chronic stage. Thus, with NAC supplementa-

with depletion of glutathione (GSH), the major antioxidant in theliver. A recent report about Gclch/h mice with a hepatocyte-specific ablation of Gclc (the gene encoding the catalytic subunitof the rate-limiting enzyme in GSH synthesis) has shown anessential role of GSH in hepatic function. Gclch/h mice developsevere steatosis and die of liver failure within one month, dueto �95% depletion of hepatic GSH; mitochondria are the majoraffected organelles, displaying abnormal ultrastructure andimpaired functioning.Methods: Gclch/h mice were fed with L-N-acetylcysteine (NAC;10 g/L) in drinking water, starting at postnatal day 18.Results: Gclch/h mice were rescued by use of NAC supplementation,and survived until adulthood. NAC replenished the mitochondrialGSH pool and attenuated mitochondrial damage, with accompany-ing diminished hepatic steatosis; however, abnormal liver bio-chemical tests, hepatocyte death, and hepatic oxidative stresspersisted in the rescued mice. At 50 days of age, the liver fromrescued Gclch/h mice started to display characteristics of fibrosisand at age 120 days, macronodular cirrhosis was observed. Immu-nohistostaining for liver-specific markers as well as the expressionprofile of hepatic cytokines indicated that the repopulation ofhepatocytes in the cirrhotic nodules involved the expansion of ovalcells.Conclusions: Replenishment of mitochondrial GSH and restora-tion of mitochondrial function by NAC prevents mortality causedby the loss of hepatocyte GSH de novo synthesis, allowing steato-

Journal of Hepatology 20

Keywords: Glutathione; Conditional knockout; Mitochondria; Steatohepatitis;Liver fibrosis and cirrhosis.Received 12 October 2009; received in revised form 11 May 2010; accepted 25 May2010; available online 11 August 2010⇑Corresponding author. Address: Department of Pharmaceutical Sciences, Uni-versity of Colorado Denver, School of Pharmacy, Aurora, CO 80045, USA. Tel.: +1303 724 3521; fax: +1 303 724 7266.E-mail address: ying.chen@ucdenver.edu (Y. Chen).Abbreviations: GSH, glutathione; NAC, L-N-acetylcysteine; ROMs, reactive oxy-genated metabolites; GCLC, glutamate-cysteine ligase catalytic subunit; c-GC,c-glutamylcysteine; PND, postnatal day; GSH-EE, GSH ethylester; DIC, dicarbox-ylate carrier; OGC, oxoglutarate carrier; ALT, alanine aminotransferase; AST, as-partate aminotransferase; MDA, malondialdehyde; HSCs, hepatic stellate cells;a-SMA, a-smooth muscle actin; ER, endoplasmic reticulum; HMOX1, heme ox-ygenase-1; MT1, metallothionein-1; ALB, albumin; AFP, a-fetoprotein; GGT,c-glutamyltranspeptidase; TNF-a, tumor necrosis factor-a; IL-6, interleukin-6;TGF-b1, transforming growth factor-b1; IFN-c, interferon-c; LT-b, lymphotoxin-b.

tion, Gclch/h mice provide a model for the development of liverfibrosis and cirrhosis.� 2010 European Association for the Study of the Liver. Publishedby Elsevier B.V. All rights reserved.

Introduction

Because hepatocytes play a unique role in drug, xenobiotic, andendogenous compound metabolism, they are a target of reactiveoxygenated metabolites (ROMs). To prevent cellular damage,hepatocytes are equipped with robust antioxidant defense sys-tems. Hepatocytes maintain the highest level of, and syntheticcapacity for, glutathione (GSH), a tripeptide thiol-based antioxi-dant of known importance in the elimination of endogenousand xenobiotic ROMs [1]. Numerous studies suggest that over-production of ROMs and/or depletion of hepatic GSH is a commonthread, allowing association of various forms of liver disease,regardless of etiology [2,3].

The function of hepatic GSH has been studied using a hepato-cyte-specific knockout mouse model of glutamate-cysteine ligasecatalytic subunit (Gclc) [4], the gene encoding the essentialenzyme in GSH biosynthesis. The Gclch/h hepatocyte-specificknockout mice display steatosis and necroinflammation in theliver, which is first observable at postnatal day 21 (PND21), andsuccumb to hepatic failure by PND30. Loss of hepatocyte Gclcresults in a dramatic decrease in hepatic GSH, which precedesthe depletion of mitochondrial GSH. Hepatic failure in Gclch/h

mice parallels loss of mitochondrial function, accumulation ofmitochondria with an abnormal ultrastructure, and a dramaticdecline in cellular ATP, suggesting mitochondrial failure as theunderlying cause of hepatic failure.

Hepatic mitochondrial dysfunction has been suggested to par-ticipate in steatohepatitis occurring in liver injuries of various eti-ologies [5]. In patients with steatohepatitis, liver fibrosis andcirrhosis commonly occur as the disease progresses [6]. InGclch/h mice, hepatocyte GSH depletion is perhaps too severe toallow us to observe the cirrhotic phenotype seen in steatohepati-tis. In this current study, we test this hypothesis by providingGclch/h mice with the antioxidant L-N-acetylcysteine (NAC), start-ing at PND18. NAC lessened the mitochondrial damage associatedwith GSH depletion, allowing mice to avoid hepatic failure;

10 vol. 53 j 1085–1094

Table 1. Primers used in Q-PCR analysis.

mRNA Forward Reverse

SLC25A10 5’ -TGGACCTGCTCAAGGTGCATCTAC-3’ 5’ -AGATTGCGAACCGAGTCAGAGAGT-3’SLC25A11 5’ -AGTCTGTCAAGTTCCTGTTTGGG-3’ 5’ -ACCAGCTGACAGCCCAGTGTAAAT-3’ HMOX1 ’3-TACACATGGAGGAACCACCG-’5 5’ -GCTTGTTGCGCTCTATCTCC-3’MT1 5' - CACGACTTCAACGTCC-3’ 5' - CAGCAGGAGCAGCAGCTCTTCTTGCAGTNF-α 5’ -AATTCGAGTGACAAGCCTGTAGCC-3’ 5’ -TGTCTTTGAGATCCATGCCGTTGG-3’

IL-6 5’ -TGTCTTTGAGATCCATGCCGTTGG-3’ 5’ -TCTGCAAGTGCATCATCGTTGTTC-3’TGF-β1 5’ -CAACTTCTGTCTGGGACCCT-3’ 5’ -TAGTAGACGATGGGCAGTGG-3’

IFN-γ 5’ -TCTTCCTCATGGCTGTTTCTGGCT-3’ 5’ -TGCCAGTTCCTCCAGATATCCAAG-3’LT-β 5’ -TACTTCCCTGGTGACCCTGTTGTT-3’ 5’ -GACGGTTTGCTGTCATCCAGTCTT-3’B2M 5’ -CATGGCTCGCTCGGTGACC-3’ ’3-GTCAAGGTGGGCGGAGTGTAA-’5

SLC25A10, dicarboxylate carrier; SLC25A11, oxoglutarate carrier; HMOX1, heme oxygenase-1; MT1, metallothionein-1; TNF-a, tumor necrosis factor-a; IL-6, interleukin-6;TGF-b1, transforming growth factor-b1; IFN-c, interferon-c; LT-b, lymphotoxin-b; B2M, b2-microglobulin.

Research Article

however, the rescued mice developed macronodular hepatic cir-rhosis by PND120.

Body

wei

ght (

g)

0

20

10

40

30

% o

f sur

viva

l

0

20

40

60

80

100

120

Gclcf/f + NACGclch/h + NACGclch/h

0 20 40 60 80 100 120 140Postnatal day

A

B

Fig. 1. Restoration of survival and growth of Gclch/h mice by NAC supplemen-tation. (A) Survival rates and (B) growth curves of Gclcf/f and Gclch/h mice. NAC(10 g/L) was supplemented in the drinking water, starting at PND18. Data arereported as means ± SEM (N = 7–10 mice for each group). NAC supplementationoffered no growth advantage to Gclcf/f or Gclc+/+ mice (not shown).

Materials and methods

Mice and treatment

The Gclch/h knockout mouse line was generated as reported earlier [4]. All studieswere conducted on littermates and approved by the Institutional Animal Care andUse Committee (IACUC). NAC (Sigma, 10 g/L) was dissolved in regular tap waterand the solution was adjusted to pH 7.0 using NaOH. Freshly made NAC-contain-ing water was supplied to mice every 2 days.

Measurement of hepatic ATP and GSH levels

Fresh liver pieces were processed for ATP measurement using the ATP lumines-cence kit (Sigma) in accordance with the manufacturer’s protocols. Whole liverhomogenates and liver cytosolic fractions were prepared from frozen liver pieces.GSH levels were determined spectrophotofluorometrically using o-phthalalde-hyde (OPA) as described [7]. Although cysteine, including NAC, does not reactappreciably with OPA, in mice supplemented with NAC, the authenticity ofGSH-OPA conjugates was verified chromatographically [8]. In addition, chromato-graphic separation of the c-glutamylcysteine (c-GC) peak from the GSH peakrevealed that c-GC was always at low concentrations (<15%), compared to GSH.

Mitochondrial isolation and in vitro mitochondrial analysis

Liver mitochondrial suspension was prepared from freshly excised liver as previ-ously described [9]. Aliquots were subject to measurements of oxygen consump-tion [9], as well as membrane potential using 5,50 ,6,60 ,-tetracholoro-1,1,3,30-tetraethylbenzimidazolylcarbocyanine iodide (JC-1, Sigma) [7], and GSH levelsusing OPA [10], as described. Mitochondrial GSH as the percent of total hepaticGSH was calculated as: Percent (%) = (GSHmit � Vmit � 100)/(GSHliv �Wliv),where: GSHmit is the concentration of mitochondrial GSH (lmol/ll), Vmit is thetotal volume of mitochondrial suspension (ll), GSHliv is the concentration ofhepatic GSH (lmol/g liver), and Wliv is the wet weight of liver (g) used for themitochondrial isolation.

Histopathological examination and histochemical staining of the liver

Liver paraffin sections (5-lm) and thin sections (1-lm) were prepared for lightmicroscopy and electron microscopy, respectively, as described [4]. HE and Mas-son’s trichrome staining on paraffin sections was performed by the Department ofPathology at the University of Cincinnati, using standard procedures. Liver histo-pathology was examined and blindly scored by a pathologist (DJO) using theBrunt scoring system [11]. TUNEL stain was performed on paraffin sections using

1086 Journal of Hepatology 2010

the In-Situ Cell Death Detection kit (Roche; Indianapolis, IN) in accordance withthe manufacturer’s protocol. Thin sections were stained with uranyl acetateand lead citrate. Images were obtained using a Nikon Eclips TE-300 microscope.

Immunohistochemistry (IHC) and immunofluorescence (IHF)

Deparaffinized and rehydrated liver sections (5-lm) were prepared in the samemanner as for light microscopy. IHC was conducted using TSA™ Biotin SystemKit in accordance with the manufacturer’s protocol (NEN Life Science Products,Boston, MA). IHF was performed in a similar manner, except that sections were

vol. 53 j 1085–1094

Table 2. Liver weight and plasma clinical pathology tests of liver function atPND120.

Parameters clcG clcGh/h

Body weight (g) 28 ± 2.3

(5.8 ± 0.5*)

Liver weight (% of body weight)

8.5 ± 0.4(5.5 ± 0.8)

ALT (IU/L)428 ± 41(1720 ± 124*)

AST (IU/L)853 ± 31(3400 ± 261*)

Unconjugated bilirubin(mg/dl)

0.6 ± 0.2(0.2 ± 0.0*)

Conjugated bilirubin (mg/dl)

11.3 ± 1.0(9.2 ± 0.6*)

f/f

30 ± 2.5

(14.3 ± 0.7)

4.7 ± 0.7 (5.7 ± 0.6) 28 ± 6 (37 ± 2) 107 ± 8 (122 ± 8) <0.01 (0.0 ± 0.0) <0.01 (0.0 ± 0.0)

Gclc

Results are reported as means ± SEM of 4–5 mice. Numbers in parentheses aredata from mice without NAC supplementation at PND28 (4). � p <0.01 whencomparing NAC-treated Gclcf/f and Gclch/h mice. * p <0.01 when comparinguntreated Gclcf/f and Gclch/h mice.

JOURNAL OF HEPATOLOGY

incubated with fluorescence-labeled secondary antibodies and DAPI to stain thenuclei. Images were obtained using a Nikon Eclips TE-300 microscope. Specificprimary antibodies were used at dilutions of 1:100 for anti-rat a-fetoprotein(AFP; Santa Cruz Biotechnology, Santa Cruz, CA), 1:10 for rat monoclonal A6 anti-body (generous gift from Dr. Valentina Factor, National Institutes of Health), 1:50for anti-CK19 (ICN/Cappel, Aurora, OH), 1:400 for anti-mouse a-smooth muscle

Gclch/h

** * *† †

†

††

Cyt

osol

ic G

SH (n

mol

/mg)

0

4

8

40

60

Gclcf/f

PNDNAC

28-

28+

50+

120+

**

†

Mito

chon

dria

l GSH

(nm

ol/m

g)

0

4

8

12

16

PNDNAC

28-

28+

* *

*†

†

*†

†

PNDNAC

28-

28+

50+

120+

0.00.5

1.0

1.5

2.0

2.53.0

Rel

ativ

e m

RN

A le

vels

SLC25A10

A

B

Fig. 2. Liver cytosolic and mitochondrial GSH levels. (A) Cytosolic and mitochondrialMitochondrial GSH, as the percent of total hepatic GSH, was calculated as described in M(SLC25A10) and oxoglutarate carrier (SLC25A11). Relative mRNA levels were reportedmicroglobulin (B2M). Data are expressed as means ± SEM (N = 4–6 mice for each group).when compared with untreated mice of the same genotype.

Journal of Hepatology 2010

actin (a-SMA; Sigma, St. Louis, MO), and at 25 lg/ml for anti-bovine TGF-b1,2

(R&D Systems, Minneapolis, MN. Corresponding secondary antibodies were pur-chased from Southern Biotech (Birmingham, AL) and used at dilutions of 1:200–500.

Plasma liver biochemical tests and lipid peroxidation assay

Mouse blood was collected by cardiac puncture using heparin-coated syringes.Plasma was collected and immediately assayed in the Clinical Laboratory for ala-nine aminotransferase (ALT), aspartate aminotransferase (AST), and bilirubin lev-els, following standard procedures. A whole liver homogenate was prepared fromfrozen liver pieces to measure hepatic lipid peroxidation, using the malondialde-hyde (MDA) assay kit in accordance with the manufacturer’s protocol (OxfordBiomedical Research, MI).

Reverse transcription and real-time quantitative PCR (Q-PCR)

Mouse livers were harvested, frozen in liquid nitrogen, and stored at �80 �C untiluse. Total RNA was isolated using Tri-Reagent (Molecular Research Center, Inc.).Synthesis of first strand cDNA and Q-PCR reactions using gene-specific primersets was performed as described [4]. Primers used for Q-PCR are summarizedin Table 1. For comparison, the mRNA levels in untreated Gclcf/f mice at PND28were set as control (=1) and relative mRNA levels were reported as fold valueincreases of the control, after normalization with housekeeping gene b-2 micro-globulin (B2M).

Statistical analyses

Statistics were performed using SigmaStat Statistical Analysis software (SPSS Inc.,Chicago, IL). Group means were compared by one-way ANOVA, followed by thestudent t-test for pair-wise comparison-of-means. Normality testing (Kolmogo-rov–Smirnov test with Lilliefors’ correction) was performed using residuals ofthe dependent variable from one-way ANOVA, and the data did not deviate from

* *† †

††

50+

120+

**†

†

††

Mito

chon

dria

l GSH

as

% o

f tot

al G

SH

0

15

30

60

75

PNDNAC

28-

28+

50+

120+

45

*

*†

*†

*†

PNDNAC

28-

28+

50+

120+

0.0

0.5

1.0

1.5

2.0

2.5SLC25A11

*†

Gclcf/f Gclch/h

GSH were determined using o-phthalaldehyde and reported as nmol/mg protein.aterials and methods. (B) Q-PCR analysis of liver mRNA for dicarboxylate carrieras fold values of control, after normalization to the housekeeping gene b-2

* p <0.05, when compared with age- and treatment-matched Gclcf/f mice. � p <0.05

vol. 53 j 1085–1094 1087

Research Article

the normality assumption. Results are reported as the mean value ± SEM of 7–10mice for growth curve, survival rate, and Brunt score, or results values from 4 to 6mice for other measurements. The p-values of <0.05 were considered as statisti-cally significant.

Gclcf/f, PND120 Gclch/h, PND120

Gclcf/f, PND120

Gclch/h,PND30

Gclch/h, PND50

Gclch/h, PND120

Steatosis

PND: 30 50 1200.00.51.01.52.02.53.0

Brun

t sco

re

Lobular inflammation

30 50 1200

1

2

3

Hepatocellularballonning

30 50 1200.0

0.5

1.0

2.0

1.5

Fibrosis

30 50 1200

23

54

1

p <0.05

p <0.05

p <0.05 p <0.05

p <0.05

p <0.05

Fig. 3. Liver histopathology. (A) Representative gross morphology at PND120 oflivers from Gclcf/f (left) and Gclch/h (right) mice. Magnification: 6�. (B) Represen-tative images of histological and histochemical analysis of liver sections fromNAC-treated mice. H&E-stained liver sections (top row) from NAC-treated Gclch/h

mice displayed macrovesicular and microvesicular steatosis at all ages (rightinsets); areas of focal hepatocyte necrosis surrounded by inflammatory infiltratewere most abundant at PND30 (lower right inset). Inflammatory infiltrateextended pericellularly at PND50. At PND120, the liver showed nodular structures(left inset, *). At all time points, apoptosis of hepatocytes (red) was noted in NAC-treated Gclch/h mice by TUNEL stain (second row). Mason’s trichrome stain forcollagen (third row) shows regular staining (blue) in the vascular walls of Gclcf/f

mice and NAC-treated Gclch/h mice at PND30; perisinusoidal fibrosis was visible inNAC-supplemented Gclch/h mice at PND50 and intensified at PND120, whenfibrous septa became clearly visible (left inset, arrows). Activation of hepaticstellate cells identified by immunopositivity of a-smooth muscle actin a-SMA;red, bottom row) was detected at PND50 and increased at PND120 in NAC-treatedGclch/h mice, whereas regular positive staining of a-SMA was observed in thevascular wall in Gclcf/f mice and NAC-treated Gclch/h mice at PND30. Magnifica-tions: 100� (left insets) and 400� (right insets). (C) Brunt score of liverhistopathology of NAC-treated Gclch/h mice. Data are reported as means ± SEM(N = 7–8 mice for each group). Individual p-values (<0.05) for pair-wisecomparisons are indicated.

Results

Rescue of Gclch/h mice with NAC-supplemented drinking water

We previously reported the generation of a conditional null Gclcallele, flanked by loxP sites (i.e. ‘‘floxed”), and termed this alleleGclc(f) [4]. Cre-mediated recombination of Gclc(f), using a hepato-cyte-specific ALB-Cre transgene, was complete and detected onlyin hepatocytes by PND14. The resultant hepatocyte-specific Gclcnull allele was termed Gclc(h). Progressive depletion of hepaticGSH, up to 95% depletion, caused growth retardation and earlydeath in Gclch/h mice.

We, therefore, attempted to rescue these mutant mice byadministering GSH ethylester (GSH-EE) or NAC at PND18. Deliv-ery of the GSH-EE by intraperitoneal injection (5 mmol/kg) wasnot successful in rescue, possibly due to the difficulty of usingthis methodology over a lengthy duration in pre-weanling mice.When supplied in drinking water, GSH-EE (10 g/L) did notincrease hepatic GSH levels, nor rescue the mice (data notshown), suggesting that this route of delivery for GSH-EE mightbe inappropriate. By contrast, we were successful in avertingdeath in �80% of the Gclch/h mice by supplying NAC (10 g/L) inthe drinking water (Fig. 1A). Rescued Gclch/h mice showed a muchslower weight gain, similar to untreated Gclch/h mice earlier intheir development, but the rescued mice began to gain weightsignificantly around PND50 attaining a body weight comparableto that of the control group by the age of 3 months (Fig. 1B). Inthis study, mice were fed with standard rodent diet containing0.4% methionine and 0.34% cystine (Harlan Teklad).

NAC supplementation preferentially replenishes mitochondrial GSHpool

Oral NAC is absorbed in the small intestine and rapidly deacety-lated by the small intestine and liver, providing cysteine for GSHsynthesis [12]. We compared the GSH levels of treated animalswith those of untreated animals (Fig. 2A). Hepatic cytosolic andmitochondrial GSH in untreated Gclcf/f mice that representedthe control levels were at 44 ± 4 and 7.9 ± 0.9 nmol/mg protein,respectively. NAC supplementation increased cytosolic GSH inGclcf/f mice by 25% at PND120, whereas it doubled (untreated =6% and treated = 13% of control) the cytosolic GSH in Gclch/h mice.Compared to control levels, this increase in cytosolic GSH is mod-est. On the other hand, the mitochondrial GSH from NAC-treatedGclch/h mice increased to >40% of control levels. We calculated theamount of mitochondrial GSH as the percent of total hepatic GSH.In untreated Gclch/h mice, mitochondrial GSH accounted for �30%of the GSH at PND28 (compared to �10% in untreated controlmice); NAC supplementation increased mitochondrial GSH inGclch/h mice to 56%, �52%, and �58% of total hepatic GSH atPND28, PND50, and PND120, respectively.

Mitochondrial GSH derives from transport of cytoplasmic GSHby the transporter functions present in the mitochondrial innermembrane [13]. The dicarboxylate carrier (DIC, Slc25a10) andthe oxoglutarate carrier (OGC, Slc25a11) have been identified aspartially accounting for GSH transport across the mitochondrialmembrane in the liver [13]. We therefore assessed the mRNA lev-

1088 Journal of Hepatology 2010

els of these transporters (Fig. 2B). Upregulation of Slc25a10 by1.9- and 2.5-fold was observed in untreated and treated Gclch/h

vol. 53 j 1085–1094

JOURNAL OF HEPATOLOGY

mice, respectively. NAC also caused a mild but significantincrease (1.3-fold) of SLC25A10 mRNA in Gclcf/f mice. However,upregulation of SLC25A11 mRNA (�1.5-fold) was only observedin NAC-treated Gclch/h mice. Taken together, our data suggest thatinduction of these transporters could contribute to the preferen-tial accumulation of mitochondrial GSH in Gclch/h mice followingNAC supplementation.

Progression from steatosis to fibrosis and cirrhosis in rescued Gclch/h

mice

NAC-rescued Gclch/h mice appeared healthy until studies wereconcluded at PND120. To our surprise, livers from these mice atPND120 were nodular in appearance (Fig. 3A, right). At this age,

A

BGclcf/f, PND120 Gclch/h, PND28

* *

††

PND 21 28 50 1200

200

400

600

800

1000

O2/m

in/m

g pr

otei

n

†

†

State 3 respiration

p <0.05

PND 210

2

4

6

Stat

e 3/

Stat

e 4

ratio

Respiratory control ratio

* †

28

* †

50

†

120

†

p <0.05

PND 210

2

4

6

FU ra

tio

Membra

*

28

*

†

p <p <0.05

8

Gclcf/f Gclch/h Gclcf/f +Gclch/h +

Fig. 4. Ultrastructure of hepatocytes and liver mitochondrial function in vitro. (AHepatocytes from NAC-treated Gclch/h mice at PND28 and PND50 revealed mitochondpreserved; vesiculated rough endoplasmic reticulum was still present at these time poiPND120. RER, rough endoplasmic reticulum; V, cytoplasmic vesicles; M, mitochondripotential, and hepatic ATP levels were measured as described in the Materials and methwhen compared with age-matched NAC-treated Gclcf/f mice. Individual p-values for pair

Journal of Hepatology 2010

the liver weight, as a percentage of body weight for Gclch/h mice,almost doubled, compared with liver weight for Gclcf/f mice(Table 2). In addition, plasma ALT, AST, and bilirubin levelsremained elevated in the rescued Gclch/h mice, indicating the per-sistence of hepatic functional abnormalities (Table 2).

Examination of liver sections from these mice revealed multi-ple age-dependent pathological changes (Fig. 3B, top row). AtPND30, hepatic steatosis of both macrovesicular and microvesic-ular forms was focally present (right upper inset) and localizedinflammatory infiltrate surrounding necrotic foci was noted(right lower inset). At PND50, steatosis became enhanced andpericellular inflammation became more noticeable. At PND120,steatotic hepatocytes and lobular inflammation diminished;however, regular hepatic acinar architecture was lost and was

Gclch/h, PND50 Gclch/h, PND120

PND 21 28 50 1200

50

100

150

O2/m

in/m

g pr

otei

n

State 4 respiration

ne potential

50

†

120

p <0.010.05

NAC NAC

PND 210

2

4

6

μmol

/g li

ver

Hepatic ATP

28

*

†

50

†

120

p <0.01p <0.05

p <0.018

10

†

) Electron microscopy of hepatocytes from NAC-treated Gclcf/f and Gclch/h mice.ria with rarefaction of matrix, but lamellar cristae and proportional size were

nts. Both mitochondria and rough endoplasmic reticulum recovered to normal aton; F, fat droplet. Bar = 1 lm. (B) Mitochondrial respiration indices, membraneods. * p <0.05, when compared with age-matched untreated Gclcf/f mice. � p <0.05-wise comparisons are indicated.

vol. 53 j 1085–1094 1089

*

**

MD

A (n

mol

/mg

prot

ein)

0.3

0.4

0.5

0.6

0.7 Gclcf/f Gclch/h

*

*

**

0

10

20

30

HMOX1

Rel

ativ

e m

RN

A le

vels

(fol

d)

*

*

**

MT1

0

10

20

30

PNDNAC

28-

28+

120+

A

B

Fig. 5. Persistent oxidative stress and stress response in the liver from NAC-rescued Gclch/h mice. (A) Hepatic lipid peroxidation was measured by malondi-aldehyde (MDA) assay. Results are expressed as nmol MDA/mg liver protein,according to a MDA-diacetal standard curve. * p <0.05, when compared with age-matched Gclcf/f mice. (B) Q-PCR analysis of liver mRNA for heme oxygenase-1(HMOX1) and metallothionein-1 (MT1). Relative mRNA levels were reported asfold values of control, after normalization with the housekeeping gene b-2microglobulin (B2M). * p <0.05, when compared with age-matched Gclcf/f mice.** p <0.05, when compared with age-matched Gclcf/f mice and with untreatedGclch/h mice.

Research Article

replaced by nodules (left inset, *). At all time points, hepatocellu-lar ballooning degeneration and hepatocyte apoptosis detectedby TUNEL staining (Fig. 3B, second row) were observed. We per-formed Mason’s trichrome staining for collagen accumulationand IHC detection of a-smooth muscle actin (a-SMA) for acti-vated hepatic stellate cells (HSCs), the key player in the patho-genesis of liver fibrosis. As shown in Fig. 3B, periportal fibrosiswas minimally present at PND30, whereas both perisinusoidaland periportal fibrosis were noted at PND50, and further pro-gressed at PND120, when extensive portal fibrosis was observed(left inset, arrows showing fibrous septa). Consistent with thetiming and intensity of hepatic fibrosis, immunopositivity of a-SMA was identified (Fig. 3B, bottom row). The overall liverhistopathology for each age group of NAC-rescued Gclch/h mice(N = 7–8) was scored by the Brunt scoring system (Fig. 3C). Thisresult showed the presence of steatosis, inflammation, and hepa-tocyte injury in these mice at all ages, all of which were mostsevere at PND50; in contrast, hepatic fibrosis progressively devel-oped with age and advanced to cirrhosis by PND120.

Restoration of mitochondrial function by NAC

In our previous study with Gclch/h mice, hepatic failure paralleleddeclines in mitochondrial structure and function [4]. Moreover,several studies have suggested a pivotal role of mitochondrialdysfunction in determining the severity of steatohepatitis [5].In Gclch/h mice, NAC supplementation replenished mitochondrialGSH to >40% of control levels; however, steatosis still developed.We therefore examined the mitochondrial morphology and func-tion (Fig. 4). In untreated Gclch/h mice at PND28, �20% of hepato-cyte mitochondria appeared swollen, with tubular cristaeuncharacteristic of hepatic mitochondria [4]. In NAC-treatedGclch/h mice from PND28 through PND50 (Fig. 4A), the mitochon-dria appeared much healthier than those of untreated Gclch/h

mice, showing lamellar cristae and proportional size; however,in our judgment, mitochondria from NAC-treated mice stillappeared more damaged than mitochondria from untreated ortreated Gclcf/f mice. At PND120, mitochondria appeared normal(Fig. 4A). Overall, all the parameters related to mitochondrialin vitro function in NAC-treated Gclch/h mice were improved, com-pared to those in untreated Gclch/h mice (Fig. 4B), which are con-sistent with the improved mitochondrial ultrastructure.Nevertheless, hepatic mitochondria from NAC-rescued Gclch/h

mice remained significantly uncoupled (having a respiratory con-trol ratio less than 3.0) by PND50; similarly, mitochondrial mem-branes appeared somewhat depolarized relative to those fromGclcf/f mice. At PND120, mitochondrial coupling and membranepotential were within the normal ranges. Mitochondrial energet-ics, reflected by hepatic ATP levels, were drastically improved byNAC (untreated = 1.3 versus NAC-treated = 4.5 lmol/g at PND28),although they remained significantly lower than the controls.Thus, we show above that NAC treatment specifically restoresmitochondrial GSH; consequently, mitochondrial ultrastructureand function are largely restored. It is worth noting that NACwas not effective at restoring ER structure in Gclch/h mice byPND50, with ER structure still appearing vesicular rather thanlamellar; however, by PND120, the ER appeared normal (Fig. 4A).

Persistent oxidative stress in livers of rescued Gclch/h mice

Ablation of the Gclc gene, specifically in the hepatocyte, resultingin depletion of GSH, leads to hepatic oxidative stress [4]. In the

1090 Journal of Hepatology 2010

current study, livers from NAC-treated Gclch/h mice still accumu-lated lipid peroxides (Fig. 5A), as measured by malondialdehyde(MDA) levels, not less than levels seen in untreated Gclch/h mice.Furthermore, increases in mRNA levels of oxidant-stress-respon-sive genes, heme oxygenase-1 (Hmox1), and metallothionein-1(Mt1), were observed (Fig. 5B); however, the inductions of bothgenes decreased at PND120. Thus, NAC treatment, surprisingly,did not diminish hepatic oxidant stress.

vol. 53 j 1085–1094

JOURNAL OF HEPATOLOGY

Oval cell expansion in cirrhotic livers of rescued Gclch/h mice

Following liver damage, hepatocyte regeneration occurs to com-pensate for hepatocyte loss. Hepatocyte replication is a commonmeans of hepatic regeneration following acute liver injury. Alter-natively, studies on animal models with chronic liver injury of var-ious forms suggest the involvement of intra-hepatic precursorcells, called oval cells, in this process [14,15]. Oval cells are bipolarprogenitor cells capable of generating both hepatocytes and bili-ary cells. To test whether oval cells contribute to hepatocyteregeneration in Gclch/h mice, we performed co-localization studies(Fig. 6), using a panel of antibodies for liver-specific markers,which included A6 (for oval cells), a-fetoprotein (AFP, for imma-ture hepatocytes), CK19 (for biliary cells), and albumin (ALB, formature hepatocytes). All hepatic sections showed CK19 stainingin the biliary epithelial architecture. In livers from NAC-treatedGclch/h mice, at PND30 the A6-positive cells were rarely observedand noted only in bile ducts; a similar staining pattern of A6 hasbeen observed in control mice in prior studies [16]. AFP-positivecells were not detected at PND30. At PND50, clusters of A6-posi-tive cells, not displaying the co-staining for either CK19 or AFPindicative of cells that were not lineage-committed, were detectedoutside the biliary epithelium. At PND120, there was a dramaticincrease in A6-positive cells, among which �80% were co-stainedwith AFP, indicating a commitment to hepatocyte lineage,whereas only �10% were co-stained with CK19. The above datasuggest that the expansion of hepatic oval cells likely serves as acritical source of repopulated hepatocytes in the cirrhotic liver.

Expression profile of hepatic inflammatory cytokines

Increasing lines of evidence suggest that several key cytokines[such as tumor necrosis factor-a (TNF-a), interleukin-6 (IL-6),

α-CK19 α-A6 CK19/A6 overla

Gclcf/f

PND120

Gclch/h

PND30

Gclch/h

PND50

Gclch/h

PND120

Fig. 6. Detection of oval cells in the liver of NAC-rescued Gclch/h mice. Immunofluidentifying oval cells (red), AFP as a hepatocyte marker (green), and CK19 as bile duct epitoverlay in which nuclei are stained with DAPI (blue). Magnification: 400�.

Journal of Hepatology 2010

transforming growth factor-b1 (TGF-b1), interferon-c (INF-c),and lymphotoxin-b (LT-b)] act in a paracrine manner to stimulateoval cell expansion and differentiation to hepatocytes; these fac-tors have been found upregulated during oval cell-mediated liverregeneration [17]. To begin to understand whether the hepaticmicroenvironment in our model is in accordance with other mod-els, in which oval cells are known to play a major role in hepato-cyte repopulation, we measured the liver mRNA levels of thesecytokines. As shown in Fig. 7A, increased mRNA levels of TNF-aand IL-6 were observed in livers from NAC-treated Gclch/h mice,with peak levels occurring at PND50. Conversely, the TGF-b1 mes-senger decreased by 2-fold in untreated Gclch/h mice, whereas itshowed control levels in NAC-rescued Gclch/h mice. Given thatTGF-b is primarily regulated at the post-translational level [18],we examined TGF-b expression in liver sections from NAC-trea-ted Gclch/h mice and control mice (Fig. 7B). Under the conditionsemployed, TGF-b immunoreactivity was not detected in liversfrom Gclcf/f mice. By contrast, TGF-b immunoreactivity wasrobust in livers from NAC-treated Gclch/h mice. At PND30, TGF-bstaining was associated with necrotic foci and thereafter wasrobust and diffuse. These results are consistent with this cyto-kine’s role in hepatic fibrosis. Elevated mRNA expression of bothIFN-c and LT-b, two cytokines that have been implicated in play-ing specific roles in oval cell-mediated liver regeneration [19],was observed at PND120 when oval cells expanded in the cir-rhotic liver of NAC-rescued Gclch/h mice.

Discussion

Gclch/h mice succumb to hepatic failure around PND30 due to up to95% depletion of hepatic GSH [4]. In the present study, we showthat Gclch/h mice survive to adulthood when orally supplied with

y α-AFP α-A6 AFP/A6 overlay

orescent co-localization analysis was performed using specific markers: A6 forhelium marker (green); co-localization of two markers (orange) is presented in the

vol. 53 j 1085–1094 1091

*

0

20

40

60 TNF-α *

*

Gclcf/f

Gclch/h

*

0

9

80

100 IL-6 *

* 63

0.0

1.0

1.5 TGF-β1

0.5 *

* *

*

PNDNAC

28-

28+

50+

120+

PNDNAC

28-

28+

50+

120+

PNDNAC

28-

28+

50+

120+

Gclcf/f PND120 Gclch/h PND30 Gclch/h PND50 Gclch/h PND120

A

B

0

6

10 LT-β

4

8

2

0

2

3 IFN-γ

1

Rel

ativ

e m

RN

A le

vels

Fig. 7. Upregulation of inflammatory cytokines in livers from NAC-rescued Gclch/h mice. (A) Liver mRNAs for inflammatory cytokines were determined by Q-PCRanalysis. Relative mRNA levels were reported as values of control, after normalization with the housekeeping gene b-2 microglobulin (B2M). * p <0.05, when compared withage-matched Gclcf/f mice. (B) Immunohistochemical staining for TGF-b on liver sections revealed negative staining in Gclcf/f mice; in contrast, livers from NAC-rescued Gclch/h

mice revealed TGF-b-immunopositive staining in a focal pattern at PND30 and diffusely at PND50 and PND120. Magnification: 400�.

Research Article

NAC, starting at PND18. Comparison of untreated, versus NAC-treated, hepatocytes from Gclch/h mice demonstrates that a majoreffect of NAC is the dramatic improvement of mitochondrial ultra-structure and in vitro functioning. Although mitochondrial energycoupling in NAC-treated Gclch/h mice is still somewhat impaired,membrane potential is remarkably improved, and importantly,hepatic ATP levels are largely maintained. The improvement ofmitochondrial morphology and function afforded by NAC isreflected in a drastic increase in mitochondrial GSH. NAC is knownto replenish GSH by providing cysteine for GSH biosynthesis [12].In the hepatocytes from Gclch/h mice, loss of the Gclc gene abol-ishes de novo GSH biosynthesis because c-GC cannot be produced.c-GC can, however, be synthesized by the salvage pathway inwhich c-glutamyltransferase (GGT) utilizes cystine as theacceptor receiving the c-glutamyl moiety from GSH exportedfrom non-hepatocyte cell types [20]. This reaction producesc-glutamylcystine, which can be reduced intracellularly to c-GC.Interestingly, compared to controls, hepatic GGT activity inNAC-rescued Gclch/h mice was elevated 21- and 27-fold atPND28 and PND120, respectively (not shown). Thus, formationof c-GC by the action of GGT represents a likely mechanism forGSH synthesis in Gclch/h hepatocytes and remains a testable idea

1092 Journal of Hepatology 2010

for future studies. Compared with a 2-fold increase in cytoplasmicGSH, NAC-mediated restoration of mitochondrial GSH (>3-fold) inGclch/h mice is much greater, accounting for >50% of the total hepa-tic GSH pool. Such a preferential increase in mitochondrial GSH,following NAC treatment, has been observed before [21,22] andis supported by considerable evidence demonstrating that mito-chondria represents a separate and high-affinity pool for GSH[23]. This effect of NAC is seemingly explained, at least in part,as consequent to increased active transport of cytoplasmic GSHinto mitochondria; this is supported by the induction of Slc25a10and Slc25a11 transporter genes in the current study. NAC, bymechanisms yet to be elucidated, favors mitochondrial GSH accu-mulation and may be important in the protective effects of NAC ina number of experimental systems [21,22].

With NAC supplementation, Gclch/h mice continue to showdepleted hepatic GSH and develop steatosis, suggesting thatmitochondrial function in vivo could still be defective, contribut-ing to increased accumulation of lipids in hepatocytes. Moreover,with age, these mice exhibit hepatic fibrosis and cirrhosis, andshow abnormalities in plasma liver biochemical functions. Thisfinding might reflect chronic oxidative stress in NAC-treatedGclch/h mice. Indeed, despite of having >40% of normal mitochon-

vol. 53 j 1085–1094

GSH

Hepatocyte

GSH

GSH

Hepatocyte

GSH

GSHGSH

GSHGSH

Hepatocyte-specific loss of Gclc

Hepatic GSH depletion by 95 %

NAC

Mitochondrial GSH depletion

Impaired mitochondrialultrastructure and function

Partially preserved mitochondrialultrastructure and function

Inhibition of liver repair Induction of hepatic cytokines(TNF-α, IL-6, TGF-β)

Hepatic failure Liver fibrosis(activation of HSCs)

Liver cirrhosis(Oval cell expansion

Induction of IFN- and LT- )

Inflammatory infiltration

Hepatocyte death

Oxidativestress

Inflammatory infiltration

Depleted hepatic ATP

Hepatocyte death

ATP Sustained hepatic ATP

(>75%) (<60%)

Fig. 8. Proposed mechanisms for the protective effects by NAC and for hepaticcirrhotogenesis in rescued Gclch/h mice. Loss of hepocyte Gclc results in adepletion of hepatic GSH by 95% in Gclch/h mice, which precedes the depletion ofmitochondrial GSH (<25% of control). Gclch/h mice reveal necroinflammation in theliver, loss of mitochondrial function, and a dramatic decline in hepatic ATP, anddie of liver failure within a month. Oral NAC supplementation rescues Gclch/h miceto survive to adulthood. NAC specifically replenishes mitochondrial GSH pool(>40% of control), leading to partially restored mitochondrial function andmorphology and sustained ATP levels. However, rescued Gclch/h mice still showongoing hepatocyte death and persistent oxidative stress which, when suppliedwith preserved mitochondrial function, signals the induction of key cytokinemediators for liver fibrosis and oval cell-mediated liver regeneration.

JOURNAL OF HEPATOLOGY

drial GSH and greatly improved mitochondria, livers from NAC-rescued Gclch/h mice exhibit a persistent oxidant stress measuredboth as oxidized lipids and gene expressions. This observation ispuzzling and could suggest that the bulk of oxidant stress, mea-sured as a consequence of GSH depletion in NAC-treated hepato-cytes, is not of mitochondrial origin. This idea could be supportedby following findings: (i) H2O2 liberated from NAC-treated mito-chondria was at the control level (data not shown); and (ii) hepa-tic CYPs 2E1 and 4A, which are reported inducers of lipidperoxidation in models of steatohepatitis [24,25], were upregu-lated by 3- to 14-fold in NAC-treated Gclch/h mice (data notshown). With regard to this report, the most important aspectis the demonstration of a persistent oxidative stress existing inour mouse model and that this stress could contribute to thedevelopment of observed chronic liver pathologies. Future exper-iments, applying other antioxidants to correct hepatic oxidativestress in this model, will be valuable in dissecting the role ofGSH deficit in the pathogenesis of chronic liver diseases.

Liver fibrosis is a reparative response to prolonged liver injuryand can advance into cirrhosis, following sustained insult. Keymediators involved in the process of hepatic fibrogenesis includeinflammatory cytokines, ROMs and growth factors [26]. In addi-tion, hepatic inflammatory milieu created by specific insults hasgreat impact on proliferative response by either hepatocytes oroval cells [27]. In this study, NAC-rescued Gclch/h mice exhibitongoing hepatocyte death and persistent hepatic oxidative stress,accompanied by (starting at PND28) elevated levels of hepaticcytokines, including TNF-a (mRNA), IL-6 (mRNA), and TGF-bimmunostaining). Consistent with these observations, hepaticfibrosis occurs in these mice by PND50 and progresses to cirrho-sis by PND120. Furthermore, elevated expression of IFN-c and LT-b accompanies proliferation of oval cells in the cirrhotic liver,which is in line with other studies that suggest a specific roleof these cytokines in oval cell-mediated liver regeneration [27].This observation also agrees with our mouse model, in whichmature hepatocytes experience persistent GSH depletion due toalbumin-driven-Cre-mediated deletion of the Gclc gene and canbe compromised in their replicative capacity, although the possi-bility of hepatocytes replication cannot be excluded. The precisesignals that direct the expansion of oval cells are not clearlyunderstood, but it is suggested that the signals could be derivedfrom inflammatory cells or activated HSCs [28,29]. Furthermore,oxidant stress is believed to be a signal activating HSCs [26]. Con-sistent with this observation, both activated HSCs and oval cellsare rare in NAC-treated Gclch/h livers at PND30; by PND50, how-ever, both are readily observed in all hepatic sections and theirappearance is further increased in cirrhotic nodules at PND120.It should be mentioned that oval cell proliferation has been sug-gested as being associated with an increased risk of developinghepatocellular carcinoma [30]. At PND120, even with hepatic cir-rhosis, NAC-treated Gclch/h mice are seemingly healthy. It will beinteresting to determine whether these mice develop hepatocel-lular carcinoma, as they grow older.

Past studies using GCLC inhibitors [31,32], as well as Gclch/h

mice, indicate that the pathological consequences of GSH deficitare correlated most closely with mitochondrial GSH depletion.This notion is supported herein by the demonstration that NAC-bolstered GSH repletion, which is specific to mitochondria, avertshepatic failure by preserving mitochondrial function and ultra-structure. NAC-mediated restoration of mitochondrial functionlikely provides necessities (such as ATP) for critical events

Journal of Hepatology 2010

involved in repairing liver damage: for instance, the inductionof key cytokine mediators of fibrogenesis. NAC-rescued Gclch/h

mice continue to display a chronic oxidant stress response. They

vol. 53 j 1085–1094 1093

Research Article

develop steatosis, which progresses to nodular cirrhosis, a condi-tion that is similar to the human disease. The proposed mecha-nisms for the protective effects by NAC, and for hepaticcirrhotogenesis, in our mouse model are depicted in Fig. 8. Futurestudies are required to test these hypotheses and to identify theprincipal signaling pathways involved. Importantly, hepatocyte-specific ablation of GSH synthesis, accompanied with partialrepletion of GSH using NAC, offers many possibilities for under-standing the cellular function of GSH and the disease states thatpresent upon its chronic depletion.

Conflict of interest

The authors who have taken part in this study declared that theydo not have anything to disclose regarding funding or conflict ofinterest with respect to this manuscript. The authors do not havea relationship with the manufacturers of the drugs involvedeither in the past or present and did not receive funding fromthe manufacturers to carry out their research.

Acknowledgements

We thank our colleagues for a careful reading of this manuscript.Portions of this work were presented at the 44th Annual Meetingof the Society of Toxicology, New Orleans, LA (March, 2005). Thiswork was funded, in part, by NIH Grants R01 ES012463 (T.P.D.and H.G.S), P30 ES06096 (T.P.D., D.W.N., H.G.S. and M.L.M), R21AA017754 (V.K.V) and R01 EY011490 (V.K.V. and Y.C.).

References

[1] Kretzschmar M. Regulation of hepatic glutathione metabolism and its role inhepatotoxicity. Exp Toxicol Pathol 1996;48:439–446.

[2] Cederbaum AI. CYP2E1 – biochemical and toxicological aspects and role inalcohol-induced liver injury. Mt Sinai J Med 2006;73:657–672.

[3] Okanoue T, Mitsuyoshi H. Non-alcoholic steatohepatitis. 3. Oxidative stressand NASH. Nippon Naika Gakkai Zasshi 2006;95:51–56.

[4] Chen Y, Yang Y, Miller ML, Shen D, Shertzer HG, Stringer KF, et al.Hepatocyte-specific Gclc deletion leads to rapid onset of steatosis withmitochondrial injury and liver failure. Hepatology 2007;45:1118–1128.

[5] Pessayre D, Fromenty B, Mansouri A. Mitochondrial injury in steatohepatitis.Eur J Gastroenterol Hepatol 2004;16:1095–1105.

[6] Day CP. Pathogenesis of steatohepatitis. Best Pract Res Clin Gastroenterol2002;16:663–678.

[7] Shen D, Dalton TP, Nebert DW, Shertzer HG. Glutathione redox stateregulates mitochondrial reactive oxygen production. J Biol Chem2005;280:25305–25312.

[8] Chen Y, Shertzer HG, Schneider SN, Nebert DW, Dalton TP. Glutamatecysteine ligase catalysis: dependence on ATP and modifier subunit forregulation of tissue glutathione levels. J Biol Chem 2005;280:33766–33774.

[9] Senft AP, Dalton TP, Nebert DW, Genter MB, Hutchinson RJ, Shertzer HG.Dioxin increases reactive oxygen production in mouse liver mitochondria.Toxicol Appl Pharmacol 2002;178:15–21.

[10] Senft AP, Dalton TP, Shertzer HG. Determining glutathione and glutathionedisulfide using the fluorescence probe o-phthalaldehyde. Anal Biochem2000;280:80–86.

1094 Journal of Hepatology 2010

[11] Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ, Cummings OW,et al. Design and validation of a histological scoring system for nonalcoholicfatty liver disease. Hepatology 2005;41:1313–1321.

[12] Sjodin K, Nilsson E, Hallberg A, Tunek A. Metabolism of N-acetyl-L-cysteine.Some structural requirements for the deacetylation and consequences forthe oral bioavailability. Biochem Pharmacol 1989;38:3981–3985.

[13] Lash LH. Mitochondrial glutathione transport: physiological, pathologicaland toxicological implications. Chem Biol Interact 2006;163:54–67.

[14] Davies RA, Knight B, Tian YW, Yeoh GC, Olynyk JK. Hepatic oval cell responseto the choline-deficient, ethionine supplemented model of murine liverinjury is attenuated by the administration of a cyclo-oxygenase 2 inhibitor.Carcinogenesis 2006;27:1607–1616.

[15] Evarts RP, Nagy P, Nakatsukasa H, Marsden E, Thorgeirsson SS. In vivodifferentiation of rat liver oval cells into hepatocytes. Cancer Res1989;49:1541–1547.

[16] Engelhardt NV, Factor VM, Yasova AK, Poltoranina VS, Baranov VN, LasarevaMN. Common antigens of mouse oval and biliary epithelial cells. Expressionon newly formed hepatocytes. Differentiation 1990;45:29–37.

[17] Lowes KN, Croager EJ, Olynyk JK, Abraham LJ, Yeoh GC. Oval cell-mediatedliver regeneration: role of cytokines and growth factors. J GastroenterolHepatol 2003;18:4–12.

[18] Gressner AM, Dooley S, Weiskirchen R. TGF-ß and the Smad pathway in liverfibrogenesis. Signaling pathways in liver diseases. Berlin Heidel-berg: Springer; 2005, p. 139–150.

[19] Akhurst B, Matthews V, Husk K, Smyth MJ, Abraham LJ, Yeoh GC. Differentiallymphotoxin-beta and interferon gamma signaling during mouse liverregeneration induced by chronic and acute injury. Hepatology2005;41:327–335.

[20] Njalsson R, Norgren S. Physiological and pathological aspects of GSHmetabolism. Acta Paediatr 2005;94:132–137.

[21] Circu ML, Rodriguez C, Maloney R, Moyer MP, Aw TY. Contribution ofmitochondrial GSH transport to matrix GSH status and colonic epithelial cellapoptosis. Free Radic Biol Med 2008;44:768–778.

[22] Xiong Y, Peterson PL, Lee CP. Effect of N-acetylcysteine on mitochondrialfunction following traumatic brain injury in rats. J Neurotrauma1999;16:1067–1082.

[23] Fernandez-Checa JC, Kaplowitz N. Hepatic mitochondrial glutathione:transport and role in disease and toxicity. Toxicol Appl Pharmacol2005;204:263–273.

[24] Leclercq IA, Farrell GC, Field J, Bell DR, Gonzalez FJ, Robertson GR. CYP2E1and CYP4A as microsomal catalysts of lipid peroxides in murine nonalco-holic steatohepatitis. J Clin Invest 2000;105:1067–1075.

[25] Ip E, Farrell GC, Robertson G, Hall P, Kirsch R, Leclercq I. Central role ofPPARalpha-dependent hepatic lipid turnover in dietary steatohepatitis inmice. Hepatology 2003;38:123–132.

[26] Bataller R, Brenner DA. Liver fibrosis. J Clin Invest 2005;115:209–218.[27] Santoni-Rugiu E, Jelnes P, Thorgeirsson SS, Bisgaard HC. Progenitor cells in

liver regeneration: molecular responses controlling their activation andexpansion. APMIS 2005;113:876–902.

[28] Knight B, Matthews VB, Akhurst B, Croager EJ, Klinken E, Abraham LJ, et al.Liver inflammation and cytokine production, but not acute phase proteinsynthesis, accompany the adult liver progenitor (oval) cell response tochronic liver injury. Immunol Cell Biol 2005;83:364–374.

[29] Lowes KN, Brennan BA, Yeoh GC, Olynyk JK. Oval cell numbers in humanchronic liver diseases are directly related to disease severity. Am J Pathol1999;154:537–541.

[30] Knight B, Yeoh GC, Husk KL, Ly T, Abraham LJ, Yu C, et al. Impairedpreneoplastic changes and liver tumor formation in tumor necrosis factorreceptor type 1 knockout mice. J Exp Med 2000;192:1809–1818.

[31] Martensson J, Meister A. Mitochondrial damage in muscle occurs aftermarked depletion of glutathione and is prevented by giving glutathionemonoester. Proc Natl Acad Sci USA 1989;86:471–475.

[32] Meister A. Mitochondrial changes associated with glutathione deficiency.Biochim Biophys Acta 1995;1271:35–42.

vol. 53 j 1085–1094