www.elsevier.com/locate/jhep

Journal of Hepatology 47 (2007) 826–833

Interferon-c exacerbates liver damage, the hepatic progenitor cellresponse and fibrosis in a mouse model of chronic liver injury q,qq

Belinda Knight1,3,�, Rebecca Lim1,2,�, George C. Yeoh2,3, John K. Olynyk1,3,4,*

1School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital Campus, PO Box 480, Fremantle 6959, WA, Australia2School of Biochemical and Chemical Sciences, University of Western Australia, Nedlands, WA, Australia

3Western Australian Institute for Medical Research, Centre for Medical Research, University of Western Australia, Nedlands, WA, Australia4Department of Gastroenterology, Fremantle Hospital, Fremantle, WA, Australia

Background/Aims: Several previous studies have suggested that interferon gamma (IFNc) may play a key role during

hepatic progenitor cell (HPC) mediated liver regeneration. However to date, no studies have directly tested the ability

of IFNc to mediate the HPC response in an in vivo model.

Methods/Results: Administration of IFNc to mice receiving a choline deficient, ethionine (CDE) supplemented diet to

induce chronic injury resulted in an augmented HPC response. This was accompanied by increased inflammation, altered

cytokine expression and hepatic fibrosis. Serum alanine aminotransferase activity, hepatocyte apoptosis and Bak staining

were significantly increased in IFNc-treated, CDE-fed mice, demonstrating that liver damage was exacerbated in theseanimals. Administration of IFNc to control diet fed mice did not induce liver damage, however it did stimulate hepatic

inflammation.

Conclusions: Our results suggest that IFNc increases the HPC response to injury by stimulating hepatic inflammation

and aggravating liver damage. This is accompanied by an increase in hepatic fibrogenesis, supporting previous reports

which suggest that the HPC response may drive fibrogenesis during chronic liver injury.

� 2007 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

Keywords: Hepatocarcinogenesis; Liver regeneration; Progenitor cells; Hepatic fibrosis

0168-8278/$32.00 � 2007 European Association for the Study of the Liver.

doi:10.1016/j.jhep.2007.06.022

Received 4 March 2007; received in revised form 13 June 2007; accepted

26 June 2007; available online 17 September 2007

Associate Editor: C. Trautweinq Rebecca Lim is the recipient of an International Postgraduate

Research Scholarship from the University of Western Australia. DrBelinda Knight is the recipient of a Raine Medical ResearchFellowship from the University of Western Australia. This work wassupported by research grants from the Cancer Council of WesternAustralia.qq The authors declare that they have no conflict of interest todisclose concerning this manuscript.

* Corresponding author. Address: School of Medicine and Pharma-cology, University of Western Australia, Fremantle Hospital Campus,P.O. Box 480, Fremantle 6959, WA, Australia. Tel.: +61 8 94313774;fax: +61 8 9431 2977.

E-mail address: john.olynyk@uwa.edu.au (J.K. Olynyk).� These authors contributed equally to this work.

Abbreviations: MPK, muscle pyruvate kinase; TUNEL, terminaldUTP nick end labelling; CDE, choline deficient ethionine supple-mented; SMA, smooth muscle actin; ALT, alanine aminotransferase.

1. Introduction

The concept that two pathways exist by which adultliver can regenerate is now widely accepted. Followingacute injury, mature hepatocytes divide to replenish losttissue. This occurs in human subjects following severe,usually acute, trauma to the liver, such as followingoverexposure to hepatotoxins (e.g. acetaminophen).This regenerative pathway has been extensively studiedusing animal models and involves stimulation by growthfactors of quiescent hepatocytes to divide, often two orthree times each, until liver mass is restored (for review,see [1,2]). In circumstances of chronic or carcinogenicliver damage, the ability of mature hepatocytes toproliferate is often inhibited, and adult hepatic progeni-tor cells (HPCs; also known as oval cells in rodents[3]) become a major cellular source of parenchymal

Published by Elsevier B.V. All rights reserved.

B. Knight et al. / Journal of Hepatology 47 (2007) 826–833 827

replacement. HPC proliferation is observed in humanliver pathologies, including chronic viral hepatitis, hae-mochromatosis, alcoholic liver disease and non-alco-holic steatohepatitis [4–7]. Studies in our laboratoryand by others have led to the identification of growthfactors that modulate the HPC response to chronicinjury [8–13]. Intriguingly, several of these growth fac-tors also facilitate hepatocyte division [14–16]. However,the pathways governing hepatocyte and oval cell prolif-eration must differ, as HPC expansion is not observedfollowing acute parenchymal damage when hepatocytesare stimulated to divide. Thus, further characterisationis required to determine which cytokines preferentiallystimulate division of each cell type, and thus to clarifyhow the two regenerative pathways are differentiallyregulated.

In 1999, Bisgaard and colleagues suggested a centralrole for IFNc in mediating the differential regenerativeresponses of oval cells versus hepatocytes [17].Recently, we confirmed that IFNc is required to mounta full HPC response to chronic injury [9] and that micein which the Th1 innate immune response (which iscentrally mediated by IFNc) is lacking are subject tomortality due to their lack of HPC-mediated liverrepair [18]. Additionally, Sun and Gao have shown thatabsence of IFNc leads to increased hepatocyte replica-tion following partial hepatectomy, suggesting thatIFNc is anti-proliferative to hepatocytes [19]. Broolingand colleagues confirmed this, demonstrating thatIFNc is growth inhibitory to hepatocytes both in vitroand in vivo [20].

These studies suggest that IFNc mediates HPC-dri-ven liver regeneration by inhibiting the division of hepa-tocytes, and facilitating the proliferation of HPCs.However, this is yet to be directly tested in an in vivo

model. Further, the data of Brooling et al. showed onlymodest growth stimulation of HPC lines followingIFNc-treatment, and this required the presence of co-factors [20]. Thus, in the present study we sought toclarify the effect of IFNc supplementation on theregenerative response to chronic liver injury in vivo.

2. Materials and methods

2.1. Materials

Recombinant murine IFNc (rmIFNc) was purchased fromR&D systems (Minneapolis, MN, USA) and resuspended at100 lg/mL in sterile PBS containing 0.1% w/v bovine serum albu-min. Aliquots were stored at �20 �C and defrosted only immedi-ately prior to use.

2.2. Animals

Four-week-old male mice were fed either a control diet (normalchow and drinking water) or a CDE diet (choline deficient chow(ICN, USA) and drinking water supplemented with 0.165% ethionine

(Sigma, St. Louis, MO, USA)) [21]. During this time, animals werealso administered either placebo (PBS containing 0.1% w/v bovineserum albumin) or rmIFNc (2 lg dissolved in PBS/0.1% BSA) viaintraperitoneal injection, once daily. This dose was selected basedon the previous study by Brooling et al., who showed anti-prolifera-tive effects of IFNc on hepatocytes in vivo at this dose [20]. At sacri-fice, blood was collected via cardiac puncture. Livers were salineperfused via the portal vein and portions collected for histologicaland gene expression studies. All animal experiments were performedin a temperature and light controlled room in a pathogen-free animalholding facility and were conducted in strict accordance with guide-lines of the National Health and Medical Research Council of Aus-tralia, and the University of Western Australia Animal EthicsCommittee.

2.3. Histology and immunohistochemistry

Liver portions were either formalin fixed or immersed in cryoma-trix (Tissue-Tek OCT, Sakura Finetek, Japan) and snap-frozen. Fivemicrometer sections were stained with haematoxylin and eosin or Sir-ius red, or for the presence of specific antigens using immunohisto-chemistry. Immunostaining for M-pyruvate kinase (MPK), A6 andsmooth muscle actin (SMA) was performed as previously described[9,11,22]. CD45 staining was performed on sections pre-treated withRetreivagen-A (BD Biosciences) using two-step indirect immunohisto-chemistry with anti-mouse CD45 (1/10 dilution; BD). Staining forphospho-p38, phospho-Akt and Bak was performed on citrate-bufferantigen-retrieved sections using two-step indirect immunoperoxidasedetection with antibodies from Cell Signalling Technologies (p38and Akt, diluted 1/20; Danvers, MA) or Invitrogen (Bak diluted 1/200).

For analysis of in situ oval cell proliferation, sections were double-stained for Ki67 (a marker of cell proliferation) and cytokeratin (anoval cell marker) using a previously described method [18].

Apoptotic cells were detected in situ using TUNEL assay accordingto the manufacturer’s directions (Promega Corp, USA).

Sirius red staining was quantified by digital image analysis usingthe Discern software package for Windows PC (Pawel Jewstafjew,Poland). The percentage area of Sirius red stained (minus background)for each image was averaged and results then pooled within eachexperimental group to give the overall mean and standard error.

2.4. Cell counts

Cell counts were performed at 400· magnification on 10 adjacent,non-overlapping periportal fields. HPCs were identified as cellsapproximately 10 lm in diameter with ovoid nucleus and scant cyto-plasm, staining positive for c-kit, A6, cytokeratin or MPK. Hepato-cytes were identified as cells 20 lm or more in diameter with roundnucleus and abundant cytoplasm. Proliferating HPCs were identifiedas cells staining double-positive for cytokeratin and Ki67 and whichwere not bile duct cells. Ki67/cytokeratin cell counts were normalisedto total cytokeratin cell counts to give the mitotic index as apercentage.

2.5. Analysis of hepatic gene expression

RNA was isolated from whole liver tissue using TRIzol (Invitro-gen, USA), DNAse treated (DNA-free, Ambion, USA) and cDNAsynthesized using the Thermoscript Reverse Transcription system(Invitrogen). Quantitative real time PCR was performed using aRotorgene 2000 thermal cycler (Fisher Biotech, Subiaco, Western Aus-tralia) with a final concentration of 2.5 mM MgCl2, 0.3· SYBR green(Molecular Probes, USA) and 1 U Platinum Taq DNA polymerase(Invitrogen). Collagen-1 gene expression was analysed as describedby Ellmers et al. [23]. Gene expression was determined in arbitraryunits by comparison with plasmid DNA standards specific for the geneof interest, as described previously [24]. b-Actin expression was deter-mined in the same samples and used for normalisation purposes. Cyto-kine primer sequences, annealing temperatures and expected productsizes are detailed in Table 1.

Table 1

Primers for cytokine profile analysed by quantitative PCR

Name Sequence 50 fi 30 Size(bp)

Annealingtemperature(� C)

LTb F TCGGGTTGAGAAGATCATTGG 640 52LTb R GCTCGTGTACCATAACGACCTNF F CGGAGTCCGGGCAGGT 74 55TNF R GCTGGGTAGAGAATGGATGAACAIL-6 F GAGGATACCACTCCCAACAGACC 141 55IL-6 R AAGTGCATCATCGTTGTTCATACA

828 B. Knight et al. / Journal of Hepatology 47 (2007) 826–833

2.6. Serum alanine aminotransferase (ALT) assay

Serum ALT activity was analysed by the Fremantle Hospitalpathology service (Fremantle, WA, Australia).

2.7. Statistical methods

Animal experiments were performed with between 3 and 12 ani-mals per experimental group. Comparisons between groups were per-formed using the non-parametric Mann–Whitney U-test, or one-wayanalysis of variance (GraphPad Prism version 4.03 for Windows;San Diego, CA). A P-value of <0.05 was considered to be indicativeof a significant difference.

Placebo IFNγ

A B

G H

D E

Fig. 1. IFNc treatment exacerbates the HPC response to the CDE diet. Daily

resulted in an increased HPC response as judged by staining for the HPC mark

(A). Cell count data revealed a significant increase in the number of A6 positive

(D–F) or c-kit (G–I) were used as markers. Data represent means ± SEM, n =

between the placebo and IFNc-treated groups (Mann–Whitney U-test ***P < 0

3. Results

3.1. IFNc treatment augments the HPC response to the

CDE diet

Three markers were used to quantify the HPCresponse: A6, MPK and c-kit. These markers differ intheir selectivity for HPC sub-populations, with A6staining biliary-lineage HPCs; MPK identifying hepaticlineage HPCs; and c-kit marking the most immatureHPC compartment. Analysis of liver sections from 14-day CDE-fed mice revealed a 3-fold increase (comparedto placebo) in the number of A6 (Fig. 1A–C) and MPK(Fig. 1D–F) positive cells present in IFNc-treated miceand a 4-fold increase in c-kit-positive HPCs (Fig.1G–I, P < 0.001). IFNc treatment did not invoke anHPC response in control diet mice (data not shown).

To determine whether IFNc increased the rate of pro-liferation of HPCs, sections were double-stained forcytokeratin (an oval cell marker) and Ki67 (a markerof proliferating cells). Resultant cell counts showed a2.5-fold increase in the mitotic index of oval cells in

C

Placebo0

5

10

15

20 ***

MPK

-pos

itive

HPC

(per

100

hep

atoc

ytes

)

PlaceboI0.0

0.6

1.2

1.8 ***

c-ki

t pos

itive

cel

ls(p

er 1

00 h

epat

ocyt

es)

Placebo IFNγ

IFNγ

0

5

10

15

20 ***

A6-

posi

tive

cells

(per

100

hep

atoc

ytes

)

I

IFNγ

F

administration of rmIFNc (2 lg/day) to mice fed a CDE diet for 14 days

er A6 (B) compared to animals which received vehicle alone as a placebo

HPCs in the IFNc cohort (C). Similar findings were observed when MPK

10 (IFNc); n = 12 (placebo). Asterisks represent a significant difference

.001).

3 5 70

3

6

9Ctrl PlaceboCtrl IFNγCDE PlaceboCDE IFNγ

* *

Days

Prol

ifera

ting

HPC

(% o

f tot

al)E

A B

DC

Fig. 2. IFNc increases the mitotic index of HPCs in CDE-fed mice. The

mitotic index of HPCs was evaluated as the percentage of cytokeratin

positive cells also positive for Ki67. Representative photomicrographs

illustrating double-staining results in control diet placebo (A), control

diet IFNc-treated (B), CDE diet placebo (C) and CDE diet IFNc-treated

animals (D) are presented. Arrows depict proliferating HPCs. Arrow-

heads illustrate proliferating non-HPC non-parenchymal cells. Cell count

data showed increased proliferation of HPCs in IFNc-treated livers at 3

and 5 days on the CDE diet (E). Data represent means ± SEM, n = 3.

Asterisks represent a significant difference between the placebo and

IFNc-treated groups (Mann–Whitney U-test *P < 0.05).

3 5 7 140

5

10

15

20 **

*

N.A.

Days

% li

ver a

rea

Siriu

s R

ed p

ositi

ve

*

3 5 7 1405

1015202530354045

Ctrl PlaceboCtrl IFNγCDE PlaceboCDE IFNγ

**

*

N.A.

Days

Nor

mal

ised

col

lage

n-1

mR

NA

leve

l

G H

C D

A B

E F

ig. 3. Fibrosis and hepatic stellate cell activation are increased in IFNc-

reated, CDE-fed mice. Analysis of fibrosis in control and CDE livers by

irius red staining and digital quantification revealed a significant increase

the amount of fibrosis in CDE diet IFNc-treated mice compared to

lacebo at 5-, 7- and 14-day time points (A). Collagen-1 mRNA levels were

lso increased in CDE diet IFNc-treated livers at 7 and 14 days (B). Data

epresent means ± SEM, n = 3 for all time points excepting 14-day where

= 10 for IFNc and n = 12 for placebo groups. Asterisks represent a

ignificant difference (Mann–Whitney U-test *P < 0.05; **P < 0.01).

.A., not analysed. Representative photomicrographs illustrate typical

irius red staining in 14-day CDE placebo (C) and IFNc-treated (D) livers.

epatic stellate cell activation, as judged by SMA staining, was notably

creased in IFNc-treated CDE animals by as early as day 3 (F) compared

o the corresponding placebo animals (E). By 14 days this effect was further

ronounced [(G) placebo; (H) IFNc].

B. Knight et al. / Journal of Hepatology 47 (2007) 826–833 829

IFNc-treated livers at 3 days of CDE treatment and a0.8-fold increase at 5 days (Fig. 2E). No stimulation ofHPC proliferation was observed in control diet fed miceat any time point. Representative photomicrographsillustrate typical Ki67/CK double-staining in 3-daysamples from control placebo (Fig. 2A), control IFNc(Fig. 2B), CDE placebo (Fig. 2C) and CDE IFNc(Fig. 2D) treated animals.

3.1.1. IFNc treatment accelerates and worsens fibrosis

induced by the CDE diet

Sirius red staining and digital quantification illus-trated a significant increase in fibrosis in 5-, 7- and14-day IFNc-treated, CDE-fed animals compared tothe paired placebo-treated samples (Fig. 3A). This wasaccompanied by an increase in collagen-1 mRNA levelsas judged by quantitative RT-PCR (Fig. 3B). Represen-tative photomicrographs illustrating Sirius red stainingin 14-day placebo (Fig. 3C) and IFNc-treated(Fig. 3D) CDE samples are presented.

F

t

S

in

p

a

r

n

s

N

S

H

in

t

p

In placebo-treated CDE mice, the earliest noticeableincrease in SMA staining occurred between days 5 and7 of the diet (data not shown). IFNc-treatment resultedin early activation of hepatic stellate cells; by day 3 ofthe CDE diet SMA positive cells were apparent in peri-portal liver regions of IFNc treated animals (Fig. 3E)but not in corresponding placebo mice (Fig. 3F). By14 days, this difference was even more noticeable, withplacebo animals exhibiting only mild periportal SMAstaining (Fig. 3G) while in IFNc-treated livers SMAcells were observed throughout the liver parenchyma(Fig. 3H).

830 B. Knight et al. / Journal of Hepatology 47 (2007) 826–833

3.2. IFNc invokes inflammatory cell infiltration and alters

CDE-diet induced cytokine expression in the liver

Analysis of hepatic cytokine expression revealednumerous changes in the livers of mice administeredIFNc. Expression of TNF and IL-6 was significantlyincreased compared to placebo samples (Fig. 4A andC. This was observed both in control and CDE-dietgroups, although not uniformly across all time points.In contrast, levels of LTb were significantly reduced inIFNc-treated CDE liver samples at 14 days, but werenot different at other time points (Fig. 4B).

Inflammation, as judged by CD45 staining and cellcounts, was significantly increased in IFNc-treated miceat all time points of the CDE diet time course (Fig. 4D).Further, numbers of CD45 positive cells were also sig-nificantly elevated in control diet samples followingIFNc treatment at 3 and 5 days. Representative photo-micrographs illustrating CD45 staining in 3-day samplesfrom control and CDE diet livers are presented inFig. 5A–D. Haematoxylin and eosin staining showedthat in 3- and 5-day control diet-fed, IFNc-treated mice,pockets of inflammatory cells were present in periportalliver regions (Fig. 5F and inlay). These clusters wereapproximately 20–100 cells in size. Placebo-treated con-

3 5 7 140

5

10

15

20 *

**

***

N.A.

Days

Nor

mal

ised

IL-6

mRN

Ale

vel

3 5 7 140

5

10

15 *

*

*

*

**

N.A.

Days

Nor

mal

ised

TN

F m

RN

A le

vel

0

2

4

6

8

10

Nor

mal

ised

LTβ

mR

NA

leve

l

0

5

10

15

20

CD

45+

cells

(per

fiel

d)

A B

C D

Fig. 4. Hepatic inflammation and cytokine expression are altered in IFNc-tre

samples from control and CDE diet mice revealed several changes in hepatic cy

treated with IFNc had significantly more hepatic TNF and IL-6 mRNA at 3 a

differences in hepatic LTb levels observed between placebo or IFNc-treated con

and IL-6 with IFNc-treatment at all time points examined (A,C). In contrast, L

compared to placebo (B). Evaluation of the number of CD45 positive cells show

receiving IFNc, both on the control diet as well as on the CDE diet (D). Th

means ± SEM, n = 3 for all time points excepting 14-day where n = 10 for

difference between the placebo and IFNc-treated groups (Mann–Whitney U-te

trol diet animals did not show this pathology (Fig. 5E).In 3-day CDE mice IFNc treatment leads to an exacer-bated inflammatory response and pathological featuresof parenchymal damage, such as steatosis and dysplasia(Fig. 5H). These were not as evident in placebo mice(Fig. 5G).

3.3. IFNc-treatment increases mortality, serum

transaminase activity and hepatocyte apoptosis induced by

the CDE diet

During CDE experiments, 7 of the 23 (30%) animalsin the IFNc-treated cohort were subject to unexpectedmorbidity or mortality between days 4 and 6 of theexperiment. In contrast, no mortality (0%; 0/21) wasobserved in the placebo-treated CDE group, or in con-trol diet animals administered either placebo (0%; 0/9)or IFNc (0%; 0/9).

Serum ALT activity was 9-fold higher in CDE-IFNcmice (1439 ± 387 U/mL) than CDE-Placebo animals(164 ± 51 U/mL) at 3 days and 3.4-fold higher at 5 days(971 ± 30 U/mL versus 282 ± 68 U/mL). By 7 days lev-els were not significantly different between the groups.All animals fed the control diet had normal serum trans-aminase levels (20–30 U/mL).

3 5 7 14

*

N.A.

Days

3 5 7 14

Ctrl PlaceboCtrl IFNγ

CDE PlaceboCDE IFNγ

N.A.

*

* **

* ***

Days

ated, mice. Quantitative RT-PCR for three cytokines of interest in liver

tokine expression during the time course examined. Control diet animals

nd 5 days than their placebo-treated counterparts (A,C). There were no

trol diet animals. CDE-fed animals showed up-regulation of hepatic TNF

Tb expression was down-regulated in 14-day CDE-fed IFNc-treated mice

ed an increase in the inflammatory response invoked in the liver in mice

is was most pronounced at 3 days in the CDE cohort. Data represent

IFNc and n = 12 for placebo groups. Asterisks represent a significant

st *P < 0.05; ***P < 0.001). N.A., not analysed.

Fig. 5. IFNc stimulates inflammation and parenchymal damage in

control and CDE diet mice. Control diet fed mice administered IFNc for

3 days showed pockets of CD45-positive inflammatory cells present in the

liver (arrow; B) which were not apparent in placebo-treated animals (A).

Three-day CDE-fed mice treated with IFNc (D) also showed increased

inflammation compared to paired placebo-treated CDE animals (C).

Haematoxylin and eosin staining confirmed that pockets of inflammatory

cells were present in the periportal region of 3-day control diet IFNc-

treated mice (F) but not placebo (E). Three-day CDE-fed animals showed

little morphological evidence of damage following placebo treatments (G)

while IFNc-treated samples from the same time point showed disrupted

morphology, steatosis and hepatocyte dysplasia (H).

3 5 70.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5 PlaceboIFNγ

Days on the CDE diet

**

TUN

EL p

ositi

ve h

epat

ocyt

es(p

er fi

eld)

*

*

3 5 70

10

20

30 PlaceboIFNγ

**

**

Days on CDE diet

Bak

pos

itive

hep

atoc

ytes

(per

fiel

d)

C D

A B

E F

Fig. 6. Hepatocyte apoptosis and Bak staining are increased in IFNc-

treated, CDE-fed mice. TUNEL staining of liver tissue from mice fed the

CDE diet for 3–7 days showed an increase in the number of hepatocytes

undergoing apoptosis in the IFNc-treated group, compared to placebo, at

all time points examined (A). Representative TUNEL staining of 3-day

CDE livers from placebo (C) and IFNc-treated (D) mice showed

numerous hepatocytes positive in the IFNc group while far fewer were

evident in the placebo-treated animals. Hepatocytes positive for the pro-

apoptotic signalling intermediate Bak were more numerate in IFNc-

treated livers (B and F) than in placebo (E), with a similar pattern of

induction to TUNEL staining. Data represent means ± SEM, n = 3

(IFNc); n = 3 (placebo). Asterisks represent a significant difference

between the placebo and IFNc-treated groups (Mann–Whitney U-test*P < 0.05 **P < 0.01).

B. Knight et al. / Journal of Hepatology 47 (2007) 826–833 831

TUNEL staining of liver sections revealed a signifi-cant elevation in numbers of hepatocytes undergoingapoptosis in the CDE-fed, IFNc-treated cohort atall time points examined compared to placebo(Fig. 6A–C). Hepatocyte apoptosis was minimal in bothgroups of control diet animals (data not shown). No dif-ference was apparent in the number of oval cells under-going apoptosis, at any time point examined (data notshown).

To determine the mechanism underlying IFNc-induced toxicity in CDE livers, we stained 3-, 5- and7-day CDE samples for three signalling intermediatesknown to be associated with cytokine-induced apopto-sis: Akt, p38 and Bak. Cell counts were performed to

assess the number of hepatocytes positive for each mar-ker, at each time point. The results showed no differ-ences in the numbers of phospho-Akt or phospho-p38positive hepatocytes between IFNc and placebo-treatedgroups at any time point (data not shown). HoweverIFNc-treated 3- and 5-day CDE livers showed signifi-cantly more Bak positive hepatocytes than placebo(P < 0.01; Fig. 6B). Representative immunohistologyfrom Bak stained liver sections illustrates this difference(Fig. 6E and F).

4. Discussion

Several previous reports suggested a role for IFNc inmediating liver regeneration following chronic injury[9,10,17]. However, the direct effects of IFNc treatmenton the HPC response in vivo had not been previouslyexamined. In this report, we demonstrate that adminis-tration of IFNc augments the HPC response to CDEdiet-induced liver injury. Numbers of HPCs positivefor three distinct markers were 3- to 4-fold elevated in

832 B. Knight et al. / Journal of Hepatology 47 (2007) 826–833

IFNc-treated mice fed the CDE diet for 14 days, com-pared with animals receiving placebo. Analysis of themitotic index of HPCs demonstrated a pro-proliferativeadvantage for HPCs in IFNc-treated CDE animals. How-ever no increase in HPC proliferation was observed inIFNc-treated control diet animals, demonstrating thatIFNc alone is not sufficient to activate the HPC response.

A striking observation arising from these studies wasthe increase in the incidence of mortality seen in IFNc-treated mice fed the CDE diet. This was accompanied bymassive (>1000 U/mL) serum transaminase activity andhepatocyte apoptosis, suggesting that the animals devel-oped acute liver failure. IFNc has been shown to primeseveral cell types, including hepatocytes, for apoptosisvia a Bcl-2 mediated pathway [25,26]. In accordance,we found increased expression of Bak (a pro-apoptoticBcl-2 family member) in hepatocytes in IFNc-treatedlivers. Thus, we hypothesise that treatment with IFNcincreased the susceptibility of hepatocytes to CDE-dietinduced apoptosis, by a Bcl-2 mediated pathway.

It is well known that HPC expansion occurs in cir-cumstances of liver injury in which damage is sustainedin the setting of inhibited hepatocyte replication (forreview, see [27–29]). Further, recent findings demon-strate that the greater the extent of hepatocyte damage,the more HPCs are induced [7]. Thus the increase inHPC numbers observed in IFNc-treated CDE liverswas likely due to indirect effects mediated by increasedparenchymal damage, rather than a direct effect of IFNcon HPC proliferation. Supporting this view, in a parallelin vitro study we did not find any pro-proliferative effectsof recombinant IFNc on 3 murine oval cell lines or onprimary oval cell cultures (Knight B, et al., manuscriptin preparation).

Another interesting observation from these studieswas the stimulation of hepatic inflammation and cyto-kine expression by IFNc. The CDE diet is known toinduce inflammatory cell infiltration and increasedexpression of several pro-inflammatory cytokines inthe liver [10]. In the current work we found that IFNcsignificantly increased the numbers of hepatic leukocytesinvoked by the diet. Given the exacerbated damagecaused by IFNc in CDE mice, this increase in inflamma-tion is not surprising. However, pockets of inflamma-tory cells were also observed in control diet fedanimals treated with IFNc, demonstrating that IFNcstimulates hepatic inflammation even in the absence ofinjury. This inflammation was accompanied by anincrease in hepatic levels of TNF and IL-6, in bothCDE and control diet samples. In contrast, LT-b levelswere either unchanged or down-regulated in IFNc sam-ples. This suggests that IFNc-treatment ‘‘primes’’ theliver for its response to damage, as the cytokine-enrichedhepatic microenvironment afforded by IFNc-inducedinflammation likely contributes to the enhancedgrowth-potential of HPC in CDE samples. However,

evidently inflammation and cytokine production aloneare not sufficient to induce an HPC response, as controldiet animals did not show HPC expansion despite ele-vated inflammatory signalling in these livers.

An unexpected finding from these studies was anincrease in the severity of CDE-induced hepatic fibrosisin IFNc-treated mice. We observed increased liver colla-gen content and an accelerated and augmented hepaticstellate cell response with IFNc-treatment of CDE-dietmice. Interestingly, IFNc has been shown in numerousprevious reports to be anti-fibrogenic, both inhibitingthe activation of hepatic stellate cells in culture, as wellas attenuating fibrosis in models such as carbon tetra-chloride (CCl4) and dimethylnitrosamine [30,31]. Theextensive damage and inflammation seen in IFNc-trea-ted CDE-fed mice could underlie the enhanced fibrosisobserved. However, the failure of IFNc to prevent stel-late cell activation in our model (as it had in previousstudies) leads us to an alternate hypothesis: that thepathway(s) mediating stellate cell function in the CDEmodel differ from those invoked in other models. Thisis consistent with results recently obtained in IFNc-defi-cient mice, in which CDE-induced fibrosis was reducedcompared to wild type, despite elevated serum transam-inase levels and sustained inflammatory response [18].Again, this observation was in direct contrast to previ-ous reports, in which IFNc deficiency was associatedwith increased fibrosis [32]. Thus it appears that theCDE diet may induce a distinct fibrotic response in com-parison to that invoked in more traditional fibrosismodels, such as CCl4. Interestingly, a recent report byClouston et al. suggests that in chronic hepatitis C, por-tal fibrosis is controlled by expanding progenitor cells[33]. Our data support this proposal. Further, we sug-gest that activated HPCs may directly communicatewith stellate cells to facilitate matrix remodelling andcollagen deposition, and that this pathway may not besubject to the anti-fibrogenic effects elicited by adminis-tration of systemic IFNc. However, further studies usingthese and other models are required to definitively testthese hypotheses.

In conclusion, this study demonstrates that IFNcadministration augments the HPC response to CDE-dietinduced liver injury, via indirect effects involving stimu-lation of hepatic inflammation and increased parenchy-mal damage. Accompanying the increased HPCresponse, IFNc treatment leads to exacerbated fibrosisin this model, supporting the existence of a distinct path-way mediating fibrogenesis that is not subject to theinhibitory effects of IFNc on myofibroblast activity.Although at the present time we cannot extrapolateour findings beyond the scope of the CDE model, webelieve these results have potentially important implica-tions therapeutically, as IFNc is currently being trialledas an anti-fibrogenic therapeutic agent in patients withchronic liver disease [34,35].

B. Knight et al. / Journal of Hepatology 47 (2007) 826–833 833

References

[1] Fausto N, Campbell JS, Riehle KJ. Liver regeneration. Hepatol-ogy 2006;43:S45–S53.

[2] Fausto N. Liver regeneration. J Hepatol 2000;32:19–31.[3] Roskams TA, Theise ND, Balabaud C, Bhagat G, Bhathal PS,

Bioulac-Sage P, et al. Nomenclature of the finer branches of thebiliary tree: canals, ductules, and ductular reactions in humanlivers. Hepatology 2004;39:1739–1745.

[4] Hsia CC, Evarts RP, Nakatsukasa H, Marsden ER, Thorge-irsson SS. Occurrence of oval-type cells in hepatitis B virus-associated human hepatocarcinogenesis. Hepatology1992;16:1327–1333.

[5] Libbrecht L, Desmet V, Van Damme B, Roskams T. Deepintralobular extension of human hepatic ‘progenitor cells’ corre-lates with parenchymal inflammation in chronic viral hepatitis:can ‘progenitor cells’ migrate? J Pathol 2000;192:373–378.

[6] Lowes KN, Brennan BA, Yeoh GC, Olynyk JK. Oval cellnumbers in human chronic liver diseases are directly related todisease severity. Am J Pathol 1999;154:537–541.

[7] Yang S, Koteish A, Lin H, Huang J, Roskams T, Dawson V,et al. Oval cells compensate for damage and replicative senescenceof mature hepatocytes in mice with fatty liver disease. Hepatology2004;39:403–411.

[8] Knight B, Yeoh GC, Husk KL, Ly T, Abraham LJ, Yu C, et al.Impaired preneoplastic changes and liver tumor formation intumor necrosis factor receptor type 1 knockout mice. J Exp Med2000;192:1809–1818.

[9] Akhurst B, Matthews V, Husk K, Smyth MJ, Abraham LJ, YeohGC. Differential lymphotoxin-beta and interferon gamma signal-ing during mouse liver regeneration induced by chronic and acuteinjury. Hepatology 2005;41:327–335.

[10] Knight B, Matthews VB, Akhurst B, Croager EJ, Klinken E,Abraham LJ, et al. Liver inflammation and cytokine production,but not acute phase protein synthesis, accompany the adult liverprogenitor (oval) cell response to chronic liver injury. ImmunolCell Biol 2005;83:364–374.

[11] Knight B, Yeoh GC. TNF/LTalpha double knockout micedisplay abnormal inflammatory and regenerative responses toacute and chronic liver injury. Cell Tissue Res 2005;319:61–70.

[12] Lim R, Knight B, Patel K, McHutchison JG, Yeoh GC, OlynykJK. Antiproliferative effects of interferon alpha on hepaticprogenitor cells in vitro and in vivo. Hepatology2006;43:1074–1083.

[13] Lowes KN, Croager EJ, Abraham LJ, Olynyk JK, Yeoh GC.Upregulation of lymphotoxin beta expression in liver progenitor(oval) cells in chronic hepatitis C. Gut 2003;52:1327–1332.

[14] Cressman DE, Greenbaum LE, DeAngelis RA, Ciliberto G, FurthEE, Poli V, et al. Liver failure and defective hepatocyte regener-ation in interleukin-6-deficient mice. Science 1996;274:1379–1383.

[15] Yamada Y, Kirillova I, Peschon JJ, Fausto N. Initiation of livergrowth by tumor necrosis factor: deficient liver regeneration inmice lacking type I tumor necrosis factor receptor. Proc Natl AcadSci USA 1997;94:1441–1446.

[16] Anders RA, Subudhi SK, Wang J, Pfeffer K, Fu YX. Contribu-tion of the lymphotoxin beta receptor to liver regeneration. JImmunol 2005;175:1295–1300.

[17] Bisgaard HC, Muller S, Nagy P, Rasmussen LJ, Thorgeirsson SS.Modulation of the gene network connected to interferon-gammain liver regeneration from oval cells. Am J Pathol1999;155:1075–1085.

[18] Knight B, Akhurst B, Matthews V, Ruddell R, Ramm GA,Abraham LJ, et al. Attenuated liver progenitor (oval) cell andfibrogenic responses to the choline deficient, ethionine supple-

mented diet in the BALB/c inbred strain of mice. J Hepatol2007;46:134–141.

[19] Sun R, Gao B. Negative regulation of liver regeneration by innateimmunity (natural killer cells/interferon-gamma). Gastroenterol-ogy 2004;127:1525–1539.

[20] Brooling JT, Campbell JS, Mitchell C, Yeoh GC, Fausto N.Differential regulation of rodent hepatocyte and oval cell prolif-eration by interferon gamma. Hepatology 2005;41:906–915.

[21] Akhurst B, Croager EJ, Farley-Roche CA, Ong JK, Dumble ML,Knight B, et al. A modified choline-deficient, ethionine-supple-mented diet protocol effectively induces oval cells in mouse liver.Hepatology 2001;34:519–522.

[22] Knight B, Yeap BB, Yeoh GC, Olynyk JK. Inhibition of adultliver progenitor (oval) cell growth and viability by an agonist ofthe peroxisome proliferator activated receptor (PPAR) familymember gamma, but not alpha or delta. Carcinogenesis2005;26:1782–1792.

[23] Ellmers LJ, Scott NJ, Piuhola J, Maeda N, Smithies O, FramptonCM, et al. Npr1-regulated gene pathways contributing to cardiachypertrophy and fibrosis. J Mol Endocrinol 2007;38:245–257.

[24] Matthews VB, Knight B, Tirnitz-Parker JE, Boon J, Olynyk JK,Yeoh GC. Oncostatin M induces an acute phase response but doesnot modulate the growth or maturation-status of liver progenitor(oval) cells in culture. Exp Cell Res 2005;306:252–263.

[25] Ahn EY, Pan G, Vickers SM, McDonald JM. IFN-gammaupregulates apoptosis-related molecules and enhances Fas-medi-ated apoptosis in human cholangiocarcinoma. Int J Cancer2002;100:445–451.

[26] Schoemaker MH, Ros JE, Homan M, Trautwein C, Liston P,Poelstra K, et al. Cytokine regulation of pro- and anti-apoptoticgenes in rat hepatocytes: NF-kappaB-regulated inhibitor ofapoptosis protein 2 (cIAP2) prevents apoptosis. J Hepatol2002;36:742–750.

[27] Alison M, Golding M, Lalani el N, Sarraf C. Wound healing inthe liver with particular reference to stem cells. Philos Trans R SocLond B Biol Sci 1998;353:877–894.

[28] Lowes KN, Croager EJ, Olynyk JK, Abraham LJ, Yeoh GC.Oval cell-mediated liver regeneration: role of cytokines andgrowth factors. J Gastroenterol Hepatol 2003;18:4–12.

[29] Knight B, Matthews VB, Olynyk JK, Yeoh GC. Jekyll and Hyde:evolving perspectives on the function and potential of the adultliver progenitor (oval) cell. Bioessays 2005;27:1192–1202.

[30] Rockey DC, Chung JJ. Interferon gamma inhibits lipocyteactivation and extracellular matrix mRNA expression duringexperimental liver injury: implications for treatment of hepaticfibrosis. J Invest Med 1994;42:660–670.

[31] Baroni GS, D’Ambrosio L, Curto P, Casini A, Mancini R,Jezequel AM, et al. Interferon gamma decreases hepatic stellatecell activation and extracellular matrix deposition in rat liverfibrosis. Hepatology 1996;23:1189–1199.

[32] Shi Z, Wakil AE, Rockey DC. Strain-specific differences in mousehepatic wound healing are mediated by divergent T helpercytokine responses. Proc Natl Acad Sci USA1997;94:10663–10668.

[33] Clouston AD, Powell EE, Walsh MJ, Richardson MM, DemetrisAJ, Jonsson JR. Fibrosis correlates with a ductular reaction inhepatitis C: roles of impaired replication, progenitor cells andsteatosis. Hepatology 2005;41:809–818.

[34] Weng HL, Cai WM, Liu RH. Animal experiment and clinicalstudy of effect of gamma-interferon on hepatic fibrosis. World JGastroenterol 2001;7:42–48.

[35] Muir AJ, Sylvestre PB, Rockey DC. Interferon gamma-1b for thetreatment of fibrosis in chronic hepatitis C infection. J Viral Hepat2006;13:322–328.