www.elsevier.com/locate/jhep

Journal of Hepatology 46 (2007) 134–141

Attenuated liver progenitor (oval) cell and fibrogenic responsesto the choline deficient, ethionine supplemented diet

in the BALB/c inbred strain of mice

Belinda Knight1,3,*, Barbara Akhurst2,3, Vance B. Matthews2,3, Richard G. Ruddell4,Grant A. Ramm4, Lawrence J. Abraham3, John K. Olynyk1,3, George C. Yeoh2,3

1School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Alma Rd., Fremantle, WA 6101, Australia2School of Biomedical and Chemical Sciences, University of Western Australia, Australia

3Western Australian Institute for Medical Research, Centre for Medical Research, University of Western Australia, Australia4Queensland Institute for Medical Research, Australia

Background/Aims: Liver regeneration following chronic injury is associated with inflammation, the proliferation of liver

progenitor (oval) cells and fibrosis. Previous studies identified interferon-gamma as a key mediator of oval cell prolifera-

tion. Interferon-c is known to regulate Th1 cell activities during immune challenge. Therefore, we hypothesised that pro-

genitor cell-mediated regeneration is associated with a Th1 immune response.

Methods: C57Bl /6 (normal Th1 response) and BALB/c mice (deficient in Th1 signalling) were placed on a carcinogenic

diet to induce liver injury, progenitor cell proliferation and fibrosis.

Results: Serum transaminases and mortality were elevated in BALB /c mice fed the diet. Proliferation of liver progenitor

cells was significantly attenuated in BALB/c animals. The pattern of cytokine expression and inflammation differedbetween strains. Liver fibrosis and hepatic stellate cell activation were significantly inhibited in BALB/c mice compared

to C57Bl/6. In addition, interferon-c knockout mice also showed reduced fibrosis compared to wild type. These findings

are in contrast to published results, in which interferon-gamma is shown to be anti-fibrogenic.

Conclusions: Our data demonstrate that the hepatic progenitor cell response to a CDE diet is inhibited in mice lacking

Th1 immune signalling and further show that this inhibition is associated with reduced liver fibrosis.

� 2006 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.

Keywords: Liver regeneration; Progenitor cell; Fibrosis; Hepatic stellate cell; Interferon c

1. Introduction

Liver regeneration is characterised by a variety of cel-lular responses. Following acute injury, repair of lost tis-

0168-8278/$32.00 � 2006 European Association for the Study of the Liver.

doi:10.1016/j.jhep.2006.08.015

Received 22 March 2006; received in revised form 24 July 2006; accepted

21 August 2006; available online 9 October 2006* Corresponding author. Tel.: +61 8 9431 3971; fax: +61 8 9431

2977.E-mail address: belindat@cyllene.uwa.edu.au (B. Knight).Abbreviations: IFN, interferon; IL, interleukin; Th1/(2), T helper

type 1 (or 2); CDE, choline deficient, ethionine supplemented; AST,aspartate aminotransferase; LT, lymphotoxin; MPK, muscle isoformof pyruvate kinase; SMA, smooth muscle actin; TUNEL,Terminaldeoxynucleotidyl Transferase Biotin-dUTP Nick End Labeling.

sue mass is typically accomplished by proliferation ofhepatocytes. In contrast, following chronic liver injury,regeneration mainly relies on the adult liver’s progenitorcell compartment [1,2]. Both of these processes areaccompanied by liver remodelling, which is an essentialpart of the wound-healing process, however can lead tofibrosis if continued in an uncontrolled fashion.

During chronic liver injury, hepatic progenitor cells(HPCs; oval cells) appear in periportal liver regions,then undergo proliferation, migration and differentia-tion, to replenish hepatocytes and biliary cells. Thisoccurs in numerous rodent models, and human diseasessuch as Hepatitis B, C, hemochromatosis and alcoholic

Published by Elsevier B.V. All rights reserved.

B. Knight et al. / Journal of Hepatology 46 (2007) 134–141 135

liver disease [1,3,4]. Interestingly, these conditions areassociated with liver inflammation [4–6]. We observedan infiltration of inflammatory cells immediately priorto HPC proliferation during experimental chronic liverinjury, suggesting that immunomodulatory cytokinespromote the initial expansion of oval cells [7]. In accor-dance, anti-inflammatory agents reduce the HPCresponse [8,9]. Using knockout mice, we tested therequirement for specific cytokines in mediating HPCproliferation, leading to the identification of factors con-tributing to the injury-induced HPC response [6,10,11].However, in each knockout model, a residual oval cellresponse remained, suggesting that the full responserequires multiple factors.

One cytokine identified as regulating HPC prolifera-tion is interferon gamma (IFNc). Expression of IFNcincreases during HPC proliferation [7,10], and knockoutattenuates the progenitor cell response to injury [10].Bisgaard et al. identified several IFNc-induced genesas being regulated during oval cell-mediated liver regen-eration [12]. A key function of IFNc is its ability to reg-ulate the innate immune response, in particular thatinvolving type 1 T-helper (Th1) cells. This responsepolarises towards either a Th1 or Th2 phenotype,depending on the type of immune challenge invoked.The difference in the biological activity of Th1 versusTh2 cells is achieved through the cytokines they express:Th1 cells produce interleukin-12 (IL-12) and IFNc whileTh2 cells secrete interleukin-4. Interestingly, severalhuman chronic liver diseases invoke a Th1 response[13–15]. Collectively these findings led us to hypothesisethat HPC proliferation may be regulated by the Th1innate immune response.

To address this, we compared the oval cell responsein inbred strains of mice known to differ in their Th1/Th2 phenotype: the C57Bl/6 strain, which has normalimmune signalling, producing IFNc-secreting Th1 cellsupon host challenge; and the BALB/c strain, which isdeficient in its ability to produce IL-12, and thus tomount a Th1 response [16]. This approach enabled usto directly test the association between the injury-in-duced HPC response and the presence of Th1 cytokines.

2. Methods

2.1. Animals

Four-week-old C57Bl/6 and BALB/c mice were obtained from theAnimal Resources Centre (Western Australia). IFNc knockout(IFNc�/�) mice were from The Jackson Laboratory (Bar Harbour,USA). Mice were housed under pathogen-free conditions under 12 hlight–dark cycles. Animal care was provided in accordance with theAustralian Code of Practice for the Care and Use of Animals for Sci-entific Purposes, as specified by the National Health and MedicalResearch Council of Australia.

Control diet mice received normal chow and drinking water. CDEdiet mice received choline deficient chow (ICN, USA) and drinkingwater supplemented with 0.165% (w/v) ethionine (Sigma) [17].

Nine animals of each strain were analysed per time point, with theexception of the 3-day groups, which comprised 6 animals each.

Serum transaminase levels were assayed by a commercial system(Sigma).

2.2. Histology and immunohistochemistry

Sirius red staining was by a standard histological protocol [18].Immunohistochemistry for m-pyruvate kinase (MPK), A6 and alpha-smooth muscle actin (aSMA) was performed as previously [10,19].Staining for pan-cytokeratin was performed exactly as in Kofmanet al. [20]. Inflammatory cells were detected by standard two-step indi-rect immunohistochemistry using primary antibodies directed againstCD45 (BD Biosciences), B220 (BD Biosciences) or CD4 (Santa Cruz).For double-staining, dewaxed sections were simmered in 0.1 M Tris–EDTA buffer (pH 8.0) for 20 min and allowed to cool. Blocking wasperformed with 3% H2O2, followed by Serum Free Block (DAKO)for 10 min each. Sections were incubated with anti-pan-cytokeratin(DAKO) was overnight at room temperature. Detection was per-formed using a peroxidase-coupled secondary antibody (Santa Cruz)and visualised by DAB (DAKO). Sections were re-blocked in H2O2

and incubated with either anti-Ki67 (DAKO) or anti-aSMA for 1 hat 37 �C. Detection was performed using either anti-rat or anti-mouseIgG-peroxidase and visualised by Vector VIP (Vector Labs).

Terminal deoxynucleotidyl Transferase Biotin-dUTP Nick EndLabeling (TUNEL) assay was performed using a commercially avail-able kit (Promega) exactly as the manufacturer recommended.

2.3. Validation of specificity of HPC and stellate cell

markers

Primary hepatic stellate cells from adult rat liver were isolated andcultured as previously described [21]. RNA was isolated following either24 h (quiescent stellate cells) or 10 days (activated stellate cells) in cul-ture. Concurrently, RNA was isolated from two HPC cell lines [22].Expression of hepatic (albumin, alpha fetoprotein), biliary (cytokeratin19) and myofibroblast (aSMA) markers was examined by RT-PCR.Primer sequences for HPC markers were previously described [23].Primer sequences for aSMA were fwd: 5 0GGCCACTGCTGCTTCCTCTTCTT30, rev: 5 0TGCCCGCCGACTCCATTC3 0.

2.4. Cell counts

HPC proliferation was examined by counting cells double-positivefor cytokeratin and Ki67, relative to numbers positive for cytokeratinalone. Hepatocyte proliferation was determined by staining for Ki67and counting positive hepatocyte nuclei, relative to total hepatocytenumbers. Activated hepatic stellate cells were counted as cells positivefor aSMA possessing a clear nucleus (visualised by hematoxylin count-erstaining). All other cell counts were performed as previously described[10,11,19].

2.5. Analysis of hepatic cytokine expression

RNA was isolated from whole liver tissue using TRIzol (Invitro-gen). Hepatic cytokine expression was examined either by ribonucleaseprotection assay (RPA) (Pharmingen) or using quantitative real-timePCR. RPA results were analysed as previously described to give finalresults as normalised expression relative to control [7,10]. Primers,annealing temperatures and extension times for RT-PCR were exactlyas described in Liu et al. [24]. Quantitated cytokine mRNA levels werenormalised to a housekeeping gene (b-actin).

2.6. Sirius red quantification

Sirius red stained photomicrographs were quantified by digital imageanalysis using the Discern software package for Windows (Pawel Jew-stafjew, Poland). For each sample, three images were captured at 100·magnification and the percentage of area positive for Sirius red deter-

7 14 210

100

200

300

400

500

600

*

*

AST

(IU/L

)

136 B. Knight et al. / Journal of Hepatology 46 (2007) 134–141

mined, using background and active threshold settings of 0.82 and 0.33,respectively. The active area score (representing percentage of area Siriusred stained) for each image was then averaged to give a mean score foreach animal. These averages were pooled within each experimentalgroup and the overall mean and standard error calculated.

2.7. Statistical analysis

All data are presented as means ± standard error (SEM). Statisticalcomparisons were performed by one-way analysis of variance or Stu-dent’s t-test. Correlation was assessed by linear regression. Statisticaltests were performed using PRISM (GraphPad).

Days on CDE diet

7 14 210.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Days on CDE diet

TUN

EL+

hepa

tocy

tes

(%)

7 14 210

1

2

3

4

5

6 C57Bl/6BALB/c

Days on CDE diet

Ki6

7+H

epat

ocyt

es(%

)

A

B

C

Fig. 1. Liver injury in C57Bl/6 and BALB/c mice fed the CDE diet.

Serum AST was comparable in both strains following 1 week on the CDE

diet (A). Following 2 weeks however, levels remained high in BALB/c

mice, while falling in C57Bl/6 animals (A). The percentage of hepato-

cytes undergoing apoptosis as judged by TUNEL staining was not

different between strains, at any time point examined (B). Similarly, Ki67

labelling of hepatocytes to illustrate proliferation was equivalent between

strains throughout the experiment (C). Data represent means ± SEM,

n = 6. Asterisk denotes a significant difference as judged by one way

ANOVA, *p < 0.05.

3. Results

3.1. Mortality and liver injury, but not hepatocyte

apoptosis, are increased in BALB/c mice

Unexpected mortality is a feature seen in mice fed theCDE diet. This typically occurs within the first 10 daysand is associated with attenuated weight gain [17]. In theseexperiments, mortality amongst C57Bl/6 mice was 7% (3/43). In contrast, a significantly higher rate of mortality(p < 0.05) representing 37% (13/35) of BALB/c micewas observed in the same period. This was associated witha significant increase in serum AST levels at 2 and 3 weeksof the diet (Fig. 1A). Control-diet fed mice of both strainsshowed normal AST levels (<50 IU/L) throughout theexperiments (data not shown). Hepatocyte apoptosiswas not different between C57Bl/6 and BALB/c mice atany time point examined (Fig. 1B). The mitotic index ofhepatocytes was also equivalent between strains through-out the experiment (Fig. 1C).

3.2. Progenitor cell proliferation is attenuated in BALB/c

mice

HPCs were identified in liver sections using three mark-ers: A6, MPK and cytokeratin. C57Bl/6 mice showed atypical oval cell response to the CDE diet (Figs. 2A andC). In contrast, BALB/c mice showed few HPCs at anytime point examined (Figs. 2B and D). Quantification ofcells positive for A6 demonstrated a significant attenua-tion of the oval cell response in BALB/c mice comparedto C57Bl/6 (Fig. 2E). Similarly, numbers of MPK positiveoval cells were reduced by 75–80% in BALB/c mice at the2- and 3-week time points (Fig. 2F).

The percentage of oval cells proliferating in BALB/clivers was significantly reduced compared to C57Bl/6, atall time points examined (Fig. 3).

3.3. Hepatic inflammation and cytokine expression

induced by the CDE diet differ between BALB/c and

C57Bl/6 mice

Hepatic IFNc mRNA levels were consistently lowerin control-diet fed BALB/c mice than in C57Bl/6 (Figs.

4A and B). CDE-feeding invoked an increase in IFNcexpression in the livers of C57Bl/6 mice, which peakedat 2 weeks, as in previous studies (Fig. 4A; [10,19]). InBALB/c liver samples, this increase was not evident(Fig. 4A). Expression of both lymphotoxin alpha(LTa) and beta (LTb) was increased in BALB/c liversfrom control-diet mice, as well as in 1- and 3-weekCDE-diet fed animals (Figs. 4B and C). Hepatic TNFmRNA levels were not significantly different betweenanimal strains at any time point on either CDE or con-trol diets (data not shown).

Fig. 2. Oval cell numbers are reduced in BALB/c mice fed the CDE diet.

Identification of oval cells by immunohistochemical staining for cyto-

keratin showed a typical pattern of induction in C57Bl/6 mice fed the

CDE diet for 1 (A) to 3 weeks (C). In contrast, BALB/c animals showed

reduced numbers of cytokeratin positive oval cells at both 1 (B) and 3

weeks (D) of the CDE diet. Bar = 50 lm. Quantification of oval cell

numbers using two markers: A6 (E) and MPK (F) showed a significant

attenuation in the oval cell response in BALB/c mice compared to

C57BL/6. This was most evident following 14 and 21 days of the CDE

diet. Data represent means ± SEM, n = 6 � 9. Asterisk denotes a

significant difference as judged by one way ANOVA, **p < 0.01,

***p < 0.001.

Fig. 4. Cytokine expression in hepatic tissue from control and CDE diet

fed mice. Ribonuclease protection assay was used to quantify cytokine

mRNA levels in extracts of liver total RNA. A sample autoradiogram is

illustrated (A). Quantified data illustrates that levels of hepatic IFNcmRNA rose to a peak at 2 weeks of the CDE diet, but remained constant

in BALB/c mice (D). Both LTa (B) and LTb (C) expression was

consistently higher in BALB/c samples, compared to C57Bl/6, with a

significant difference detected at 7 and 21 days CDE, as well as in control

diet mice. Data are presented as means ± SEM, n = 6. Asterisk denotes

a significant difference as judged by one way ANOVA, ***p < 0.001.

1 C57Bl/6A

B. Knight et al. / Journal of Hepatology 46 (2007) 134–141 137

To confirm that the injury response in BALB/c micewas Th2 polarised, expression of interleukins 4, 10 and12 was examined in 2-week CDE liver samples. The

7 14 210

10

20C57Bl/6BALB/c

**

***

Days on CDE diet

Ki6

7+ oval

cells

(%)

Fig. 3. Oval cell proliferation is attenuated in BALB/c mice. Double

immunohistochemistry for cytokeratin and Ki67 (a marker of proliferating

cells) facilitated quantification of proliferating oval cells, relative to total oval

cell numbers. The results show a significant reduction in the percentage of

proliferating oval cells at all time points in BALB/c mice, compared to C57Bl/

6. Data represent means ± SEM, n = 5. Asterisk denotes a significant

difference as judged by one way ANOVA, *p < 0.05, **p < 0.01.

results showed significantly higher expression of IL-4and IL-10 in BALB/c mice than in C57Bl/6, and signif-icantly lower IL-12 mRNA levels, consistent with a Th2polarised response (Fig. 5).

Staining of liver sections for markers of totalinflammatory cells (CD45), B lymphocytes (B220)and T lymphocytes (CD4) revealed a significantincrease in the number of total leukocytes present inBALB/c livers at 2 weeks, but no difference at othertime points. Numbers of B220 positive cells were

IL-4 IL-10 IL-12p400.001

0.01

0.1

BALB/c**

***

*

No

rma

lis

ed

cy

tok

ine

mR

N

lev

el

(arb

itra

ry u

nit

s)

Fig. 5. BALB/c mice display a Th2 hepatic cytokine profile. Total RNA

from 2 week CDE fed BALB/c and C57Bl/6 mice was subjected to real

time RT-PCR analysis for IL-4, IL-10 and IL-12. Results show that

levels of hepatic IL-4 and IL-10 were relatively higher in BALB/c liver

samples than C57Bl/6. Conversely, IL-12 was barely detected in BALB/c

but was highly expressed in C57Bl/6 livers. This pattern of cytokine

expression is consistent with a Th2 polarisation of BALB/c animals.

Data represent means ± SEM, n = 4. Asterisk denotes a significant

difference as judged by Student’s t-test, *p < 0.05, **p < 0.01,

***p < 0.001.

138 B. Knight et al. / Journal of Hepatology 46 (2007) 134–141

reduced in BALB/c livers at 2 and 3 weeks comparedto wild type, while numbers of CD4 positive lympho-cytes were increased relative to C57Bl/6 animals(Fig. 6).

3.4. BALB/c and IFNc�/� mice show a reduced fibroticresponse to the CDE diet

Sirius red staining of liver sections revealed a dramat-ic reduction in the extent of collagen deposition in the

1 2 30

1

2

3

4

5

6 C57Bl/6BALB/c

*

*

Weeks on CDE diet

CD

4po

sitiv

ece

lls(p

erfie

ld)

A

B

C

1 2 30123456789 *

Weeks on CDE diet

CD

45po

sitiv

ece

lls(p

erfie

ld)

1 2 30

1

2

3

4

5

6*

*

Weeks on CDE diet

B22

0po

sitiv

ece

lls(p

erfie

ld)

Fig. 6. Hepatic inflammation in BALB/c and C57Bl/6 mice fed the CDE

diet. Numbers of inflammatory cells positive for CD45 (A), B220 (B) and

CD4 (C) were determined in liver sections by immunohistochemistry and

cell counting. Data illustrate that BALB/c mice have significantly more

total inflammatory cells (CD45 positive) at the 2 week CDE time point

(A). Numbers of B220 positive inflammatory cells were lower in BALB/c

than C57Bl/6 mice at 2 and 3 weeks (B), while numbers of CD4 positive

cells showed the opposite pattern (C). Data represent means ± SEM,

n = 4. Asterisk denotes a significant difference as judged by Student’s t-

test, *p < 0.05.

livers of BALB/c mice (Fig. 7C) following 2 weeks ofCDE diet feeding, compared to C57Bl/6 (Fig. 7A).Numbers of aSMA positive cells were reduced between5- and 10-fold in BALB/c mice (Fig. 7D), compared toC57Bl/6 (Fig. 7B).

A similar reduction in both Sirius red and aSMAstaining was also detected in 2-week CDE-fed IFNc�/�

mice (Figs. 7E and F).Quantification of Sirius red staining by digital image

analysis revealed a significant difference in the extent offibrosis present in wild type C57Bl/6 livers comparedwith either BALB/c or IFNc�/� (Fig. 7G).

Fig. 7. Liver fibrosis is reduced in BALB/c and IFNc�/� mice. Sirius red

(A, C and E) or aSMA (B, D and F) staining of 2 week CDE treated

livers revealed a reduction in both collagen deposition and hepatic stellate

cell activation in both BALB/c (C and D) and IFNc (E and F) knockout

mice, compared to wild type C57Bl/6 (A and B) animals. Bar = 50 lm.

Digital image analysis of liver sections from replicate 2 week CDE

animals showed a significant reduction in the percentage of liver area

positive for Sirius red in both BALB/c and IFNc�/� samples. Data

represent means ± SEM, n = 4. Asterisk denotes a significant difference

as judged by one way ANOVA, *p < 0.05, **p < 0.01.

B. Knight et al. / Journal of Hepatology 46 (2007) 134–141 139

3.5. Activated hepatic stellate cells and oval cells correlate

in number and co-localise in CDE liver

A close correlation was observed between numbers ofactivated hepatic stellate cells (aSMA positive) andHPCs (positive for cytokeratin) present in CDE liversfrom both C57Bl/6 and BALB/c mice (Fig. 8A). Thiscorrelation was statistically significant by linear regres-sion analysis (p < 0.005). Double immunostaining foraSMA and cytokeratin revealed a close spatial associa-tion between the two cell types in CDE liver (Fig. 8).This was most evident in C57Bl/6 mice (B), howeverwas also observed in both BALB/c (C) and IFNc�/�

(D) livers.To confirm the specificity of HPC and stellate cell

markers, isolates of primary quiescent and activated

Fig. 8. Correlation and co-localisation between oval cells and stellate

cells in CDE liver. Quantification of numbers of cells positive for

cytokeratin or aSMA was performed in serial liver sections from C57Bl/6

and BALB/c mice fed the CDE diet for 1–3 weeks. Cytokeratin counts for

individual animals were correlated against those for aSMA (A). Linear

regression analysis showed a significant positive correlation between the

two parameters in both C57Bl/6 and BALB/c samples (p < 0.005). Double

immunohistochemical staining of 2 week CDE liver sections from C57Bl/

6 mice (B) showed strong co-localisation between cytokeratin positive

(purple) and aSMA positive (brown) cells. In both BALB/c (C) and

IFNc�/� (D) livers fewer HPC and aSMA positive cells were present,

however co-localisation between the two cell types was still evident.

Bar = 20 lm. The specificity of HPC and stellate cell markers was

confirmed by RT-PCR (E). HPC isolates were positive for the HPC

markers albumin and CK19 but negative for aSMA. In contrast, isolates

of primary stellate cells were negative for HPC markers but positive for

aSMA. SMA expression was stronger in activated (HSC(a)) than

quiescent (HSC(q)) stellate cell preparations.

hepatic stellate cells, as well as two HPC lines, weresubjected to RT-PCR analysis. Results showed astrong transcript for cytokeratin 19 and albumin inHPC but not stellate cells (Fig. 8E) and conversely,for aSMA in activated hepatic stellate cells but notHPC (E). Other HPC markers including glucose-6phosphatase, tyrosine aminotransferase and cytokera-tin 7 were also not detected in stellate cell preparations(data not shown).

4. Discussion

Based on previous observations suggesting a linkbetween inflammation and the proliferation of oval cellsduring chronic liver injury, we investigated the effects ofan HPC inducing regime on two strains of mice knownto differ in their immune response to host-challenge. Theresults show a significantly blunted progenitor cellresponse to injury in BALB/c mice, which exert a pre-dominantly Th2 response, compared to the Th1 polar-ised C57Bl/6 strain. We were surprised to find aconcurrent reduction in fibrosis in BALB/c mice, whichis in contrast to published findings [25].

Previous studies showed either no difference, or adecrease in transaminase activity in BALB/c mice fol-lowing liver injury [25,26]. In our study, serum AST roserapidly to an equivalent level in both strains followingthe commencement of the CDE diet. However as report-ed previously, in C57Bl/6 animals transaminase levelsfell after the 1 week time point [7], whereas in BALB/cmice AST levels remained elevated. Additionally,BALB/c mice were slow to recover from the initialweight loss induced by the CDE regime (data notshown). These findings suggest that BALB/c mice wereless able to compensate for the effects of the CDE diet,consistent with an impairment in their ability to achieveHPC-mediated liver regeneration. Accordingly, mortali-ty was also eleveated in the BALB/c cohort. Interesting-ly, the proportion of hepatocytes undergoing eitherapoptosis or proliferation was not different betweenstrains. The consistently low (<5%) Ki67 labelling indexof hepatocytes observed in CDE-fed mice is indicative ofthe inhibitory effects of the regime on their ability todivide.

Quantification of HPC numbers using three distinctmarkers showed a clear reduction in the CDE-inducedoval cell response in BALB/c mice. Accordingly, dou-ble-labelling of cells for cytokeratin (an oval cell marker)and Ki67 (a proliferating cell marker) showed a lowerpercentage of proliferating oval cells in BALB/c animalsat all time points. Importantly, the absence of oval cellproliferation in BALB/c mice was not due to a protec-tion from liver injury in this strain, as serum transami-nases were comparable or elevated compared toC57Bl/6 animals, as previously discussed.

140 B. Knight et al. / Journal of Hepatology 46 (2007) 134–141

The significant reduction in peak hepatic IFNc expres-sion observed in BALB/c mice fed the CDE diet is consis-tent with an impaired ability to produce IFNc during Th1challenge-response in this strain. Accordingly, liver sam-ples from 2-week CDE-fed C57Bl/6 animals showed sig-nificantly higher levels of the Th1 cytokine IL-12 andsignificantly lower levels of the Th2 cytokines IL-4 andIL-10. This suggests that oval cell proliferation followingchronic injury may be dependent on cytokines elaboratedby infiltrating lymphocytes and immune cells during theinitial inflammatory burst seen in the CDE model. Wepreviously documented the requirement for IFNc inmediating the CDE-induced oval cell response in mice[10]. However, the attenuation in oval cell numbersobserved in BALB/c mice was greater than that seen inIFNc knockout animals, which showed a 50–80% reduc-tion in MPK positive oval cell numbers, but no change inA6 expressing cells [10]. This suggests that in BALB/cmice, attenuated progenitor cell proliferation is mediatedby more than just impaired IFNc signalling. To furtherdissect this, we examined the expression of other candi-date oval cell growth factors in BALB/c livers, includingTNF and LTb, two cytokines shown previously to medi-ate the oval cell response [6,10,11]. However, we did notdetect any difference in hepatic TNF expression inBALB/c mice, and LTb expression was upregulated inBALB/c livers. This indicates that the blockade in ovalcell proliferation in BALB/c mice occurs independentlyof TNF and LTb expression, suggesting that the presenceof these cytokines is not sufficient to sustain HPCproliferation in this model.

Interestingly, a difference was observed both in thepattern of liver inflammation and the type of inflamma-tory cells infiltrating the livers of BALB/c and C57Bl/6mice fed the CDE diet. BALB/c livers possessed signifi-cantly higher numbers of total inflammatory cells (CD45positive) following 2 weeks of CDE feeding. Comparedto C57Bl/6 animals, their total pool of B220 positivecells was reduced, while their CD4-positive cell countswere elevated. This suggests that BALB/c mice respondto chronic hepatic injury by inducing a more robust, T-cell skewed inflammatory response. The significance ofthis difference in terms of cytokines elaborated will bethe subject of future studies.

An unexpected finding from this study was that thefibrogenic response was impaired in both BALB/c andIFNc knockout mice fed the CDE diet. Previous publica-tions have reported that IFNc is anti-fibrotic in vitro andin vivo [27–30]. Further, Shi et al. documented a signifi-cantly greater fibrotic response in BALB/c mice com-pared to C57Bl/6 following CCl4 administration [25]. Inaddition, IFNc�/� mice were shown to have increasedCCl4-stimulated fibrosis compared to wild type controls[25]. These findings are in direct contrast to our results,in which both the BALB/c and IFNc�/� strains were pro-tected from fibrosis induced by the CDE diet. This dis-

crepancy suggests that the fibrogenic response elicitedduring CDE-induced liver remodelling differs from thatinduced by more classic models of hepatic fibrosis, suchas CCl4. We suggest that in the CDE model, the fibrogenicprocess is controlled by oval cells, whereas in CCl4-in-duced liver injury, the fibrogenic stimulus is elicited bydamaged hepatocytes. A recent study by Clouston et al.supports this view, documenting a correlation betweenHPC numbers and stage of fibrosis in human HepatitisC liver samples [31]. Our data are consistent with this find-ing, showing that independent of animal strain, oval cellsand activated hepatic stellate cells expand in close pro-ximity, and correlate in terms of numbers. The spatialassociation observed between the two cell types in CDEliver suggests there may be direct communication betweenthem. Thus, the proposal of Clouston that in Hepatitis C,progenitor cell expansion triggers fibrosis at the portalinterface [31], is directly supported by our findings in theCDE model. Future studies, including co-culture experi-ments, will address this hypothesis in greater detail.

In summary, this study demonstrates that the regen-erative HPC response to chronic liver injury inducedby the CDE diet is impaired in BALB/c mice, support-ing an association between progenitor cell proliferationand the Th immune phenotype. This strengthens previ-ous observations documenting the role of IFNc in medi-ating oval cell-mediated liver regeneration, however alsoillustrates that additional co-factors may be required toinitiate a full response. We also report that in the CDEmodel, a lack of IFNc signalling is associated with areduced fibrogenic response and suggest that this maybe indicative of the lack of a requirement for liverremodelling in the absence of oval cell proliferation.

Acknowledgement

Dr Belinda Knight is the recipient of a fellowshipfrom the Raine Medical Research. Grant Ramm andRichard Ruddell were supported by a grant from theNational Health and Medical Research Council ofAustralia, Program Grant #339400.

References

[1] 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.

[2] 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.

[3] Roskams TA, Libbrecht L, Desmet VJ. Progenitor cells indiseased human liver. Semin Liver Dis 2003;23:385–396.

[4] 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.

[5] Libbrecht L, De Vos R, Cassiman D, Desmet V, Aerts R,Roskams T. Hepatic progenitor cells in hepatocellular adenomas.Am J Surg Pathol 2001;25:1388–1396.

B. Knight et al. / Journal of Hepatology 46 (2007) 134–141 141

[6] 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.

[7] Knight B, Matthews VB, Akhurst B, Croager EJ, Klinken E,Abraham LJ, Olynyk JK, et al. Liver inflammation and cytokineproduction,but not acute phase protein synthesis, accompany theadult liver progenitor (oval) cell response to chronic liver injury.Immunol Cell Biol 2005;83:364–374.

[8] Davies R, Knight B, Tian Y-W, Yeoh G, Olynyk JK. Hepaticprogenitor cell response is attenuated by to the choline deficient,ethionine supplemented diet model of murine liver injury isattenuated by the administration of a selective cyclooxygenase 2inhibitor. Carcinogenesis 2006.

[9] Nagy P, Kiss A, Schnur J, Thorgeirsson SS. Dexamethasoneinhibits the proliferation of hepatocytes and oval cells but not bileduct cells in rat liver. Hepatology 1998;28:423–429.

[10] 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.

[11] Knight B, Yeoh GC, Husk KL, Ly T, Abraham LJ, Yu C, RhimJA, et al. Impaired preneoplastic changes and liver tumorformation in tumor necrosis factor receptor type 1 knockoutmice. J Exp Med 2000;192:1809–1818.

[12] 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.

[13] Sobue S, Nomura T, Ishikawa T, Ito S, Saso K, Ohara H, Joh T,et al. Th1/Th2 cytokine profiles and their relationship to clinicalfeatures in patients with chronic hepatitis C virus infection.J Gastroenterol 2001;36:544–551.

[14] Song K, Coleman RA, Alber C, Ballas ZK, Waldschmidt TJ,Mortari F, LaBrecque DR, et al. TH1 cytokine response ofCD57+ T-cell subsets in healthy controls and patients withalcoholic liver disease. Alcohol 2001;24:155–167.

[15] Jiang R, Feng X, Guo Y, Lu Q, Hou J, Luo K, Fu N. T helpercells in patients with chronic hepatitis B virus infection. Chin MedJ (Engl) 2002;115:422–424.

[16] Karupiah G. Type 1 and type 2 cytokines in antiviral defense. VetImmunol Immunopathol 1998;63:105–109.

[17] Akhurst B, Croager EJ, Farley-Roche CA, Ong JK, Dumble ML,Knight B, Yeoh GC. A modified choline-deficient, ethionine-supplemented diet protocol effectively induces oval cells in mouseliver. Hepatology 2001;34:519–522.

[18] Junqueira LC, Bignolas G, Brentani RR. Picrosirius staining pluspolarization microscopy, a specific method for collagen detectionin tissue sections. Histochem J 1979;11:447–455.

[19] 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. Carcinogenesis 2005.

[20] Kofman AV, Morgan G, Kirschenbaum A, Osbeck J, Hussain M,Swenson S, Theise ND. Dose- and time-dependent oval cellreaction in acetaminophen-induced murine liver injury. Hepatol-ogy 2005;41:1252–1261.

[21] Bridle KR, Crawford DH, Ramm GA. Identification and char-acterization of the hepatic stellate cell transferrin receptor. Am JPathol 2003;162:1661–1667.

[22] Dumble ML, Croager EJ, Yeoh GC, Quail EA. Generation andcharacterization of p53 null transformed hepatic progenitor cells:oval cells give rise to hepatocellular carcinoma. Carcinogenesis2002;23:435–445.

[23] 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.

[24] Liu L, Zhou X, Liu H, Xiang L, Yuan Z. CpG motif acts as a’danger signal’ and provides a T helper type 1-biased microenvi-ronment for DNA vaccination. Immunology 2005;115:223–230.

[25] Shi Z, Wakil AE, Rockey DC. Strain-specific differences in mousehepatic wound healing are mediated by divergent T helper cytokineresponses. Proc Natl Acad Sci USA 1997;94:10663–10668.

[26] Mizuhara H, Kuno M, Seki N, Yu WG, Yamaoka M, YamashitaM, Ogawa T, et al. Strain difference in the induction of T-cellactivation-associated, interferon gamma-dependent hepatic injuryin mice. Hepatology 1998;27:513–519.

[27] Baroni GS, D’Ambrosio L, Curto P, Casini A, Mancini R,Jezequel AM, Benedetti A. Interferon gamma decreases hepaticstellate cell activation and extracellular matrix deposition in ratliver fibrosis. Hepatology 1996;23:1189–1199.

[28] Rockey DC, Maher JJ, Jarnagin WR, Gabbiani G, Friedman SL.Inhibition of rat hepatic lipocyte activation in culture by interfer-on-gamma. Hepatology 1992;16:776–784.

[29] Sakaida I, Uchida K, Matsumura Y, Okita K. Interferon gammatreatment prevents procollagen gene expression without affectingtransforming growth factor-beta1 expression in pig serum-inducedrat liver fibrosis in vivo. J Hepatol 1998;28:471–479.

[30] Czaja MJ, Weiner FR, Takahashi S, Giambrone MA, van derMeide PH, Schellekens H, Biempica L, et al. Gamma-interferontreatment inhibits collagen deposition in murine schistosomiasis.Hepatology 1989;10:795–800.

[31] 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.