Embed Size (px)
Citation preview
Research Article
Effect of prolyl hydroxylase domain-2 haplodeficiency on thehepatocarcinogenesis in mice
Femke Heindryckx1, Anna Kuchnio2,3, Christophe Casteleyn4, Stephanie Coulon1, Kim Olievier1,Isabelle Colle1, Anja Geerts1, Louis Libbrecht5, Peter Carmeliet2,3, Hans Van Vlierberghe1,⇑
1Department of Gastroenterology and Hepatology, Ghent University Hospital, Ghent, Belgium; 2Laboratory of Angiogenesis and NeurovascularLink, Vesalius Research Center, VIB, Leuven, Belgium; 3Laboratory of Angiogenesis and Neurovascular Link, Vesalius Research Center, KU Leuven,Leuven, Belgium; 4Applied Veterinary Morphology, Department of Veterinary Sciences, University of Antwerp, Wilrijk, Belgium; 5Department
of Pathology, Ghent University Hospital, Ghent, Belgium
Background & Aims: The two major primary liver cancers in Introduction
adults are hepatocellular carcinoma and cholangiocarcinoma.These tumors rapidly outgrow their vascular supply and becomehypoxic, resulting in the production of hypoxia inducible factors.Recently, interest has grown in the regulators of these factors.Several reports have been published describing the role of prolylhydroxylase domains – the key oxygen sensor responsible for thedegradation of hypoxia inducible factors – in tumor progressionand vascularisation. The effect of prolyl hydroxylase domain 2on the pathogenesis of liver cancer has never been studied.Methods: A diethylnitrosamine-induced mouse model was used inthis study, allowing primary hepatic tumors to occur as a result ofchronic liver damage. Several parameters of prolyl hydroxylasedomain 2-haplodeficient mice were compared to those of wild typemice, thereby focussing on the expression of angiogenic factors andon the hepatic progenitor cell activation and differentiation.Results: This study shows that inhibiting prolyl hydroxylasedomain 2 increases the hepatocarcinogenesis and stimulatesthe development of cholangiocarcinoma. Furthermore, PHD2deficiency and the accompanying continuous HIF activation,selected for a more metastatic tumor phenotype.Conclusions: The effect of prolyl hydroxylase domain 2 defi-ciency on hepatocarcinogenesis hold a great potential for thera-peutic intervention, since hypoxia and the selection for a moreaggressive cholangiocarcinoma phenotype might also have arepercussion on patients receiving long-term treatment withanti-angiogenic compounds.� 2012 European Association for the Study of the Liver. Publishedby Elsevier B.V. All rights reserved.Journal of Hepatology 20
Keywords: Hepatocellular carcinoma; Cholangiocarcinoma; Angiogenesis;Hypoxia; Notch signaling; Hepatic progenitor cells.Received 16 December 2011; received in revised form 21 February 2012; accepted 26February 2012; available online 13 March 2012⇑ Corresponding author. Address: De Pintelaan 185, 9000 Gent, Belgium. Tel.: +329 332 23 70; fax: +32 9 332 26 74.E-mail address: [email protected] (H. Van Vlierberghe).Abbreviations: HCC, hepatocellular carcinoma; CC, cholangiocarcinoma; HIF, hy-poxia inducible factors; PHD, prolyl hydroxylase domains; HPC, hepatic progen-itor cells; VEGF, vascular endothelial growth factor; PlGF, placental growth factor;DEN, diethylnitrosamine; CK19, cytokeratin 19.
Hepatocellular carcinoma (HCC) is a primary malignancy thatmostly emerges from a background of chronic liver diseases. Thisprocess is characterized by increased hypoxia due to the forma-tion of fibrotic septa, as well as sinusoidal capillarisation, result-ing in an increased resistance to blood flow and decreased oxygendelivery [1]. Furthermore, HCC is a fast growing tumor that rap-idly exceeds its blood supply and becomes hypoxic. Through theexpression of the hypoxia inducible factor (HIF), several angio-genic factors, such as vascular endothelial growth factor (VEGF)and placental growth factor (PlGF), are regulated. HIF is tightlycontrolled by prolyl hydroxylase domain (PHD) molecules, andparticularly, PHD2 plays an essential role as key oxygen sensor[2]. Proline hydroxylation regulates the stability of the a-sub-units of HIF. In the event of hypoxia, PHD2 is unable to hydroxyl-ate HIF, preventing its degradation and inducing an angiogenicswitch. This leads to an uncoordinated production of angiogenicfactors, inducing a poorly functional vasculature incapable ofmeeting tumoral oxygen demands.
The main rationale behind the intensive interest in prolylhydroxylases is their potential to serve as therapeutic targets.Recent studies have highlighted the importance of PHD2 intumorigenesis [3–5] and have shown a differential role of PHD2in tumoral and stromal compartments. Some studies suggest thatPHD2 haplodeficiency normalizes vessel morphology, thereforeimproving tumor perfusion and oxygenation, resulting in a lessaggressive tumor phenotype [5]. However, this is in contrast toother findings describing an increased capillary density, vesselbranching and an increased tumor growth as a result of PHD2silencing in the tumor environment [3]. Furthermore, studies inpancreatic cancer have shown that PHD2 serves as a potentialtumor inhibitor [6] and results in glioma tumor cells supportthe positive effect of PHD2 expression on mice survival [4]. Mostof these studies use different mouse models to assess the effect ofPHD2 inhibition on tumor development, which might explainpart of the discrepancies [7] and also the distinct role of PHD2in specific tumor compartments will influence the outcome.The effect of PHD2 deficiency on the development and progres-sion of primary liver tumors has never been studied. Therefore,we used a DEN-induced mouse model and compared PHD2
12 vol. 57 j 61–68
Research Article
heterozygous knock-out mice with their wild type (WT) counter-parts, thereby focusing on angiogenesis, tumor development andtumor phenotype.Material and methods
HCC induction
Five-week-old male mice received intraperitoneal injections of DEN (35 mg/kgbodyweight) diluted in saline, once per week, as previously described [8].PHD2+/� mice were obtained from the Vesalius Research Centre (KU Leuven,Belgium). A heterogenic couple was used for breeding and the offspring weregenotyped. The Ethical Committee of experimental animals at the Faculty ofMedicine and Health Sciences, Ghent University, Belgium, approved the protocols.
Sampling and histology
Eight mice per group were sacrificed by isoflurane (Forene�, Abott, Wavre, Bel-gium) anesthesia while blood was obtained from the ophthalmic venous sinus.After macroscopic evaluation and quantification of the number of tumors, allorgans were sampled as previously described [8,9]. Tumor-lesions and non-tumor-tissue were separately collected. Hematoxylin-eosin staining was per-formed to evaluate the morphological changes and to identify HCC and CC lesions.Sirius Red was used to stain CC lesions and was carried out to score CC burden,which was also staged by distinguishing normal bile ducts (stage 0), pre-malignant cholangioma (stage 1), mixed cholangioma-cholangiocarcinoma(stage 2) and cholangiocarcinoma (stage 3), on hematoxylin-eosin-stained slidesas previously described [10]. An experienced pathologist used histological charac-teristics of hematoxylin-eosin-stained slides and immunohistochemistry to dis-tinguish HCC from CC. Pre-malignant hepatocellular lesions were separatedfrom HCC using the absence of reticulin and micropathological characteristicsas diagnostic criteria [11]. Fibrosis was quantified using Metavir score on SiriusRed-stained slides [12,13], and analysing five ROI’s per slide. Stainings were doneusing standard histology protocols and blindly evaluated by an experiencedpathologist.
Protein levels
Immunohistochemistry was used to quantify protein levels in hepatic tumors andin the surrounding non-tumor tissue. Immunohistochemical staining protocolsfor HPC markers (CK19, LIF and Oct4) [14], angiogenesis (CD105 and VEGF), mac-rophage recruitment (F4/80) and hypoxia (HIF1a) are provided in SupplementaryTable 1 and were quantified as previously described [13]. Intercapillary distance(ICD) was used as a marker for microvessel density, by measuring the average dis-tance between vessels in HCC nodules on CD105-stained slides. Protein levels ofPlGF (R&D Biosystems, Mouse PlGF-2 Quantikine ELISA Kit, Abingdon, UK) andVEGF (R&D Biosystems, Mouse VEGF Quantikine ELISA Kit, Abingdon, UK) weremeasured in liver tissue.
Quantitative real time PCR (qPCR)
RNA was extracted from 20 mg frozen liver tissue (macroscopic tumor or non-tumor) preserved in RNAlater, following manufacturer’s guidelines (Qiagen,RNeasy Mini Kit, Venlo, The Netherlands) and diluted to a concentration of100 ng/ll. RNA quality was evaluated using spectrophotometry. cDNA wasobtained from 10 ll RNA with the iScript cDNA Synthesis Kit using random hexa-mere primers, according to manufacturer’s protocol (Bio-Rad, Eke, Belgium).Expression was measured using qRT-PCR, with 3 ll cDNA and the LightCycler480 Green I Master Mix (Roche, Vilvoorde, Belgium). Primers are shown in Sup-plementary Table 2.
Human gene expression
Gene expression data of HCC and CC lesions were obtained from the GSE15765dataset [15]. Seventy HCC patients and 13 CC patients were included andexpression of the Notch pathway was assessed using robust multichip averages(RMA).
62 Journal of Hepatology 20
Vascular corrosion casting
Vascular corrosion casts were created by perfusing Batson n�17 (PolysciencesRef.: 07349, Eppelheim, Germany) through the aorta abdominalis and vena ileoco-lica; followed by dissolving soft tissue in 25% KOH. Casts were evaluated withscanning electron microscopy (Jeol JSM5600LV, Zaventem, Belgium).
Statistics
Data were analyzed with SPSS16. After testing for normality and homoscedastic-ity, data were either subjected to a Student’s t test or a Mann–Whitney U test.p <0.05 was considered statistically significant.
Results
Macroscopic evaluation
No significant difference was observed in mortality betweenPHD2-inhibited mice and WT (Fig. 1). Nevertheless, while WTmice displayed a steady increase in mortality that starts earlyin the experiment, in the PHD2+/� group, 80% of the mice diedin the last 5 weeks of the experiment. PHD2+/� mice sufferedfrom a significantly increased weight loss compared to theirWT counterparts at 20 weeks (p <0.01) and 25 weeks (p <0.001)(Fig. 1). Furthermore, there was an increased number of macro-scopic liver tumors after 25 weeks (p <0.05) (Fig. 1); in PHD2-inhibited mice and at 30 weeks this resulted in a higher liverweight (p <0.05) (Fig. 1). No ascites was present and no extrahe-patic metastases were observed.
Histological evaluation
Microscopic analyses revealed that tumors in WT mice are pre-dominantly HCC, while PHD2+/� mice are characterized by bothHCC and CC (Fig. 2). Starting from 20 week DEN, all PHD2+/� micehad developed pre-malignant dysplastic nodules and HCC burdenwas significantly increased at 20 and 25 week DEN (Fig. 2). Inter-estingly, dysplastic foci were predominantly of the small-celltype, yet large dysplastic hepatocytes were also found in bothgroups. No vascular invasion was seen in tumor lesions. SiriusRed showed there was a significant increase of CC lesions at 25and 30 weeks (Fig. 2). Evaluation of CC stage revealed that WTmice occasionally displayed pre-malignant cholangioma lesionsat 25 and 30 week DEN (Fig. 2). At 20 week DEN, PHD2+/� liversshowed several biliary hyperplastic and cholangioma lesions,lined by flattened epithelium (stage 1) (Fig. 2), suggesting apre-malignant state (Table 1). These progressed further at25 week DEN to a mixed occurrence of cholangioma and CC, char-acterized by the appearance of some goblet-like cells as well asflattened epithelium and biliary dysplasia (stage 2) (Fig. 2). After30 week DEN, several CCs were present, tumor cells remainedheterogeneous even within the same gland but often displayedan intestinal goblet cell-like phenotype (stage 3) (Fig. 2). Percent-ages of malignant and pre-malignant lesions from all groups areprovided in Table 1.
Inflammation and angiogenesis
HIF1a staining was significantly increased in PHD2+/� livers andexpression data of Pfk and Glut1 confirmed that PHD2 inhibitionincreased the activation of the HIF pathway (Fig. 3). PHD2 inhibi-
12 vol. 57 j 61–68
)%(lavivrus
evitalumu
C
100
80
60
40
20
00 5 10 15 20 25 30
PHD2+/-WT
Weeks
A B
C D
E F
***
***
p <0.05p <0.01p <0.001
0)g(
ssolthgieW
121086
2
Healthy 20 25 30 wk
4
DEN
** ***PHD2+/-WT
0
sromut
cipocsorcaM
)#( 30
20
Healthy 20 25 30 wk
10
DEN
*
*PHD2+/-WT
PHD2+/-
(after 30 wk DEN)WT
0.00
oitarthgiew
ydob/reviL
0.06
0.04
Healthy 20 25 30 wk
0.02
DEN
*
PHD2+/-WT
Fig. 1. Macroscopic parameters. Mortality of PHD2+/� and WT mice injectedwith DEN did not yield a significant difference. (A) However, a sudden increase inmortality of PHD2+/� mice is observed in the last 5 weeks. (B) PHD2+/� micesuffered from more weight loss compared to WT mice, suggesting that PHD2deficiency negatively influenced the mice’s wellbeing. (C) The number ofmacroscopic liver tumors was significantly increased in PHD2+/� mice after 25and 30 week DEN. (D) The liver to bodyweight ratio was significantly increasedafter 30 week DEN in PHD2+/� mice, correlating with tumor burden. Livers of (E)WT and (F) PHD2+/� mice after 30 week DEN.
G H I J
FE
A
C
B
D
0
)%(
dersuiriS
8
6
Healthy 20 25 30 wk
2
4
DEN
*
**
001
x(nedrubro
mutC
CH
7 )
8
6
20 25 30 wk
2
4
DEN
*
* PHD2+/-WT
***
***
p <0.05p <0.01p <0.001
100 µm 100 µm
100 µm 100 µm
100 µm 100 µm 100 µm 100 µm
PHD2+/-WT
Fig. 2. Histological evaluation. Hematoxylin-eosin (H&E) staining of (A) WT and(B) PHD2+/� livers respectively, showing HCC lesions and CC lesions. Sirius Redstaining showing some fibrotic septa in (C) WT mice and (D) a CC lesion inPHD2+/� mice. (E) HCC burden was significantly increased in PHD2+/� mice after20 and 25 week of DEN. (F) Percentage of Sirius Red showed that PHD2 deficiencysignificantly increased CC burden. Staging of CC happened as followed: (G) stage0 = healthy biliary ducts; (H) stage 1 = ductular reaction and cholangioma withflattened epithelium; (I) stage 2 = lesions characterized some goblet-like cells butalso flattened epithelium and (J) stage 3 = cholangiocarcinoma.
JOURNAL OF HEPATOLOGY
tion did not significantly alter VEGF levels in liver tissue, butdecreased the production of PlGF in liver tissue. PlGF levels weresignificantly lower (p <0.05) in the surrounding non-tumoroustissue of healthy and 25 week DEN PHD2+/� compared to WT,yet at 20 and 30 weeks a trend towards higher PlGF levels wasobserved (Table 2). No PlGF was measured in PHD2+/� tumors,while a time-dependent increase was noted in WT tumors, differ-ing significantly at 30 week DEN (p <0.05). aSMA stainingshowed that PHD2+/� mice had less activated HSC in HCC lesionsafter 25 weeks of DEN. The CC lesions were characterized by ahigh activation of HSCs (not shown). No significant differencewas seen in Metavir score.
F4/80 staining showed that macrophage recruitment was sig-nificantly increased in the non-tumor tissue of the PHD2+/�
Journal of Hepatology 20
groups after 20 and 25 week DEN (p <0.05) (Fig. 3). HCC lesionsoccurring after 30 week DEN in PHD2+/� mice were characterizedby higher macrophage infiltration compared to similar WTlesions (p <0.05). In CC lesions, there was also a high F4/80expression (not shown). PHD2+/� tumors were characterized bya higher microvessel density (20 week, p <0.01; 30 weekp <0.05) determined by the ICD (Fig. 3) and confirmed byCD105 staining in tumor tissue (Fig. 3). Scanning electron micros-copy images of the vascular corrosion casts and histological eval-uation of the number of abnormal vessels on CD105-stainedslides revealed that PHD2 inhibition did not improve the vesselmorphology after 30 week DEN (Fig. 3). On the contrary, PHD2+/
� livers were characterized by more vascular buds in the micro-vasculature and a more irregular organization (SupplementaryFig. 1), supporting the increased vascularisation in CD105-stainedslides (Supplementary Fig. 2).
12 vol. 57 j 61–68 63
Table 1. Percentages of pre-malignant HCC lesions (dysplastic nodules), HCC lesions, pre-malignant CC lesions (cholangioma) and CC lesions in mice treated withDEN.
Dysplastic nodules HCC Cholangioma Cholangiocarcinoma20 wk DEN
WT 86% 14% 0% 0%PHD2+/- 100% 43% 43% 14%
25 wk DENWT 100% 88% 13% 0%PHD2+/- 100% 100% 47% 50%
30 wk DENWT 100% 100% 20% 0%PHD2+/- 100% 100% 100% 75%
Research Article
Hepatic progenitor cells (HPC)
Occurrence of both HCC and CC could imply involvement of HPCs,thus, the HPC niche was immunohistochemically assessed usingHPC markers CK19, LIF and Oct4 (Supplementary Fig. 3). CK19staining showed that HCC and CC were characterized by CK19+cells. In WT mice, activation of HPCs mostly occurred in smallHCC lesions and in the surrounding non-tumor tissue. While CClesions of PHD2+/� mice remained CK19+ at all time, larger HCClesions tended to have a lower CK19 expression compared tosmall HCC lesions. The majority of WT HCC lesions were Oct4+,while little expression was observed in PHD2+/� CC lesions,although their surroundings showed several Oct4+ cells espe-cially in highly inflamed regions. LIF was expressed both in HCCand CC in respectively WT and PHD2+/� livers, however, CClesions often showed a stronger LIF+ staining than HCC lesions.
PHD2+/�mice showed a significant higher expression of recep-tors and ligands of the Notch pathway, especially during earlystages, Jagged2 was significantly increased (p <0.05) in non-tumorand early tumor lesions at 20 week DEN compared to WT (Fig. 4).Notch2 was significantly increased in non-tumor tissue at 20 and30 weeks of DEN administration (p <0.05) (Fig. 4).
Furthermore, a comparison of human gene expression databetween HCC and CC patients, confirmed that CC have a higherexpression of NOTCH1 and JAGGED1 (p <0.001) and levels of PFKshowed an increased activation of the HIF pathway (Fig. 4).
Metastasis
The occurrence of extrahepatic tumors in the DEN-inducedmouse model is very rare. Therefore, we used pro-metastaticmarkers to assess the influence of PHD2 deficiency on the meta-static potential. Indeed, after 30 weeks of DEN, PHD2+/�
expressed significantly higher levels of the pro-metastatic mark-ers Icam (p <0.05) and Mmp9 (p <0.05) in tumor tissue, whereasItgav expression was not significantly altered (Fig. 4). No signifi-cant difference in expression of these metastastic markers wasobserved in the surrounding tissue (Fig. 4).
Discussion
Two main primary liver cancers in adults are HCC and CC. Thesetumors rapidly outgrow their vascular supply and become hyp-
64 Journal of Hepatology 20
oxic, initiating the production of several angiogenic factors[8,9,16]. The identity and regulation of factors responsible forthese angiogenic processes are areas of active investigation asthese pathways could provide valuable targets for therapeuticintervention, especially in HCC where sorafenib (Nexavar�) iscurrently the standard of care for advanced HCC patients. HIFand VEGF have gathered extensive attention for their roles inmediating tumor angiogenesis. Recently, interest has grown inregulators of HIF and in alternative mechanisms of pathologicalangiogenesis. Thus, several reports have been published describ-ing the role of PHD2 in tumor progression and vascularisation,leading to contradictory findings. In our study, we used an ortho-topic model for liver cancer in which PHD2 is silenced both in thetumor and in the stroma, while the study of Giacca et al. usedPHD2-silenced tumor cells and in the study of Mazzone et al.PHD2 was silenced only in the tumor environment. Macroscopicevaluation showed that PHD2+/� mice developed more tumorscompared to their WT counterparts. This was also found in otherstudies [3,17]. Microscopic analysis confirmed an increased HCCburden at 20 and 25 weeks, yet HCC progression stagnates after30 weeks, perhaps due to the increased presence of CC, possiblyresulting in a competition for nutrients between tumors [18].
We found that PHD2 inhibition did not significantly alterVEGF levels. This corresponds to in vitro studies showing thatPHD2 does not influence VEGF expression, but exerts its pro-angiogenic potential through other factors [3]. In addition, stud-ies have also shown that CC lesions are characterized by lowerexpression of angiogenic factors compared to HCC [19]. Ourresearch suggests that this is also independent of PlGF in tumorlesions, which can be partially explained by the decrease ofHSC activation in HCC lesions, the main source of PlGF in chronicliver diseases. Studies have shown that PlGF exerts its mainangiogenic effects in the non-tumorous surrounding [20], andindeed, here we see an upregulation in DEN-treated PHD2+/�
mice compared to healthy livers.Previous studies have shown that tumors implanted in PHD2
haplodeficient mice are characterized by a morphologically nor-malized vasculature, enhancing tumor oxygenation and promot-ing a less aggressive tumor phenotype [5,21]. This effect was notobserved in our study, however, only advanced tumors wereassessed using vascular corrosion casting and in certain cancermodels vascular normalization only occurs at an early stage. Ear-lier tumor stages were histologically evaluated using CD105-stained slides and no improvement of the vascular morphologywas seen. Furthermore, the PHD2+/� tumor lesions were charac-
12 vol. 57 j 61–68
E F G H
I J K ML
B C D
* p <0.05** p <0.01PHD2+/-
WT
A
0
kfP noisserpxe
eussit laromut-non ni
50
40
Healthy 20 25 30 wk
20
10
30
DEN
*
PHD2+/-
0
gniniats 08F/4F )
%( eussit laromut-non ni
6
8
4
Healthy 20 25 30 wk
2
DEN
*
*
0
nilgodnE noisserpxe )501
DC(
200
300
100
Healthy 20 25 30 wk
DEN
*
**
0
ecnatsid yrallipacretni naeM
30
40
20
20 25 30 wk
10
DEN
***
edisni gniniats 08F/4F )
%( eussit laromut
0
6
8
20 25 30 wk
2
4
DEN
*
0
1-tulG
noisserpxe eussit laro
mut-non ni
60
40
Healthy 20 25 30 wk
20
DEN
PHD2+/-
0
kfP noisserpxe eussit laro
mut ni
40
20 25 30 wk
20
10
30
DEN
*
PHD2+/-
0
1-tulG
noisserpxe eussit laro
mut ni
40
20 25 30 wk
20
10
30
DEN
*
*PHD2+/-
PHD2+/- PHD2+/- PHD2+/- PHD2+/-
100 µm100 µm 50 µm 50 µm 50 µm
Fig. 3. Hypoxia, inflammation and vascularisation. Pfk expression was significantly increased in (A) non-tumoral and (B) tumor tissue of PHD2+/�. A trend towards higherGlut1 expression was seen in (C) non-tumoral tissue and a significant upregulation was observed in (D) tumors of PHD2+/� mice. F4/80 showed increased macrophagerecruitment in (F) HCC and (E) non-tumoral tissue of PHD2+/� mice. (G) ICD showed that HCC lesions of PHD2+/� were more vascularised, (H) and this was confirmed by theexpression of CD105 in tumor tissue. (I) PHD2+/� and (J) WT livers stained for HIF1a. SEM images of 30 week DEN (K) PHD2+/�, (L) WT and (M) healthy livers showed novascular normalization.
JOURNAL OF HEPATOLOGY
terized by a higher expression of pro-metastatic markers, thusproviding evidence that PHD2 deficiency did not select for a lessaggressive hepatic tumor phenotype. This is in line with multiplereports showing a positive correlation between HIF activationand metastasis [22]. Histologically, we were not able to identifythe loss of epithelial characteristics and gain of mesenchymalattributes, as typically seen during epithelial-mesenchymal tran-sition. However, expression of Mmp9 and Icam is known to beinvolved in this process, thus PHD2 might influence this event.
An interesting finding was that PHD2 inhibition influencedhepatic carcinogenesis, resulting in a shift from HCC to CC.Indeed, the prolonged administration of DEN induced inflamma-tion and fibrosis around cancer foci or in non-cancer liver tissuesand could thereby influence ductular reaction. PHD2 deficiencyincreased the infiltration of macrophages, however, no significantdifference in Metavir score, a fibrosis-staging system, was
Journal of Hepatology 20
observed between WT and PHD2+/� mice, suggesting the ductularreaction is not merely a result of increased inflammation. SinceCC and HCC could be derived from HPCs [23–25], PHD2 mighthave altered their differentiation. HPC proliferation is one of theearliest responses to DEN, however, contradictory results havebeen found on whether DEN induced tumors HPC-derived[25,26]. Staining of CK19, Oct4 and LEF in the surrounding non-tumoral tissue and tumor lesions of DEN-injected WT andPHD2+/� could suggest an HPC origin of these tumors.
In this study, we propose that the switch from HCC towardsCC might be the result of the continuous HIF activation inPHD2+/� mice. Current literature showed that HPC differentiationand embryonic stem cells towards cholangiocytes are regulatedthrough the Notch pathway [14,27]. This supports our findingsthat PHD2+/� livers have a higher expression of Jagged2 andNotch2 in the DEN model, especially in the pre-malignant lesions
12 vol. 57 j 61–68 65
A B D
0
2deggaJnoisserpxeeussitlaro
mut-nonni
40
Healthy 20 25 30 wk
20
30
10
DEN
**
*
*
PHD2+/-
C
E GF
NOTCH2
NOTCH1
JAGGED2
JAGGED1
PFK
0
2hctoN
noisserpxeeussitlaro
mut-nonni
500
Healthy 20 25 30 wk
200
300
400
100
DEN
* *PHD2+/-
0
AM
R 4
6
8
2
*** *** ***
***
0
2deggaJnoisserpxe
eussitlaromut
ni
40
20 25 30 wk
20
10
30
DEN
PHD2+/-500
200
300
400
100
0
2hctoN
noisserpxeeussitlaro
mutni
20 25 30 wk
DEN
PHD2+/-
600
200
400
0
srekram
citatsatemfo
noisserpxEeussitro
mut-nonni
Mmp9 Itgav Icam
PHD2+/-
800
200
400
600
0
srekram
citatsatemfo
noisserpxEeussitro
mutni
Mmp9 Itgav Icam
PHD2+/-
PHD2+/-WT
PHD2+/-WT
PHD2+/-WT
CCHCC
PHD2+/-WT
PHD2+/-WT
PHD2+/-WT
***
***
p <0.05p <0.01p <0.001
Fig. 4. Notch and metastasis. Expression levels of ligands of the Notch pathway were increased in (A and C) non-tumor and (B and D) tumor tissue of PHD2+/�mice. Humanexpression levels showed as the robust multichip average (RMA), demonstrated that NOTCH1, JAGGED1, JAGGED2 and PFK were also significantly upregulated in CC lesionsof patients compared to HCC. (E) However, NOTCH2 was downregulated in human CC. Expression of metastatic markers Mmp9 and Icam was significantly increased inPHD2+/� tumors. No significant difference was observed in (F) Itgav levels or (G) non-tumor lesions. No extrahepatic metastasis was observed.
Table 2. Protein levels of PlGF and VEGF in tumoral and non-tumoral tissues of PHD2+/� and WT mice injected with DEN.
Concentration of PlGF in non-tumoral tissue Concentration of PlGF in tumoral tissueWT PHD2+/- p value WT PHD2+/- p value
Mean SEM Mean SEM Mean SEM Mean SEMHealthy 6.67 1.83 1.33 0.65 * - - - -20 wk DEN 10.39 5.31 29.08 11.46 11.82 2.68 12.22 4.6725 wk DEN 66.35 31.59 11.85 4.93 * 101.27 46.92 25.19 9.1730 wk DEN 73.33 22.02 106.29 50.63 1497.04 264.28 20.37 9.25 *
Concentration of VEGF in non-tumoral tissue Concentration of VEGF in tumoral tisue WT PHD2+/- p value WT PHD2+/- p value
Mean SEM Mean SEM Mean SEM Mean SEMHealthy 110.39 10.49 106.77 8.1320 wk DEN 116.25 6.30 68.24 9.86 * 144.86 23.80 106.79 20.7825 wk DEN 143.79 18.32 118.15 19.16 145.96 31.00 139.61 14.6530 wk DEN 106.87 20.52 105.33 10.23 202.31 16.39 173.88 15.50
⁄p<0.05.
Research Article
and surrounding non-tumoral tissue. PHD2+/� mice have a higherlevel of HIFs stimulating the Notch pathway, possibly causingHPCs to differentiate towards a cholangiocytic lineage. Additionalstudies using transgenic mice models will be needed to supportthis hypothesis, but human expression data also show a signifi-cant upregulation of the Notch pathway in CC lesions. However,
66 Journal of Hepatology 20
the difference is not as distinct as in the mouse model, perhapsbecause only established HCC and CC were assessed in the humanstudy, while a follow-up study was conducted in the mousemodel. It also needs to be clarified that the pre-malignant ‘‘chol-angioma-lesions’’ are not often found in patients, while they areabundant in this and other rodent models [10].
12 vol. 57 j 61–68
JOURNAL OF HEPATOLOGY
Alternatively, the occurrence of CC lesions after continuousactivation of HIF could also be a consequence of the resistanceof cholangiocytes to ischemic conditions and their potential toproliferate under these conditions [28,29]. Liver ischemia isknown to stimulate ductular reaction, by inducing proliferationof the bile duct epithelium, providing a functional response to bil-iary dysfunction [28]. Further studies will be needed to clarify theexact influence of hypoxia on different types of liver cells. Condi-tional knock-out models that fully silence PHD2 expression inspecific hepatic cell types could provide additional insights intohow PHD2 influences the hepatocarcinogenesis.
These results provide interesting insights into the effect oflong-term HIF activation on hepatic tumors, and could hold agreat potential for therapeutic intervention. Recent studies haveshown that inducing ischemic injury by transarterial chemoemb-olization (TACE) leads to an increased occurrence of a mixedhepatocholangiocellular phenotype [30,31]. This biliary pheno-type of HCC is more aggressive and could possibly be derivedfrom HPCs, activated through the TACE-induced hypoxia to dif-ferentiate towards a cholangiocytic lineage. A possible linkbetween hypoxia and the selection for a more aggressive tumorphenotype might also have a repercussion on patients receivinglong-term treatment with anti-angiogenic compounds, such assorafenib, which create a more hypoxic environment [32]. Sincerecent studies have shown an association between the epidermalgrowth factor pathway and CK19 expression in liver tumors, thiscould open a possible role for therapeutic agents targeting thispathway, such as erlotinib and gefitinib, in patients with thisaggressive class of HCC or as an adjuvant therapy for thosereceiving long-term administration of anti-angiogenic treatments[33].
To conclude, this study provides new insights into the role ofPHD2, in the pathogenesis of primary liver cancer. We haveshown that PHD2 inhibition aggravates HCC growth and inducesthe presence of CC, possibly through the Notch pathway. Futureresearch is necessary to clarify the exact role of PHD2 or HIF stim-uli on the differentiation of liver progenitor cells, especially forpatients receiving long-term administration of anti-angiogenictherapy.
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.
Financial support
F. Heindryckx received a scholarship (FWO 09/ASP/161) from theFund for Scientific Research (FWO Flanders). S. Coulon received ascholarship (BOF 09/24J/012) from the University Ghent ResearchFund (BOF). H. Van Vlierberghe and I. Colle received a fundamen-tally clinical mandate of the Fund for Scientific Research (FWOFlanders). P. Carmeliet received Long term Structural fundingMethusalem – by the Flemish Government, InteruniversityAttraction Poles Program – Federal Government – P06/30, fundfor Scientific Research (FWO G.0652.08) and the Belgian Federa-tion against Cancer. A. Kuchnio received a fellowship (FWO
Journal of Hepatology 20
1.1.B68.11.N.00) from Fund for Scientific Research (FWOFlanders).
Acknowledgements
The TROMA-III antibody developed by Rolf Kemler was obtainedfrom the Developmental Studies Hybridoma Bank developed un-der the auspices of NICHD and maintained by The University ofIowa, Department of Biological Sciences.
Supplementary data
Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jhep.2012.02.021.
References
[1] Coulon S, Heindryckx F, Geerts A, Van Steenkiste C, Colle I, Van VlierbergheH. Angiogenesis in chronic liver disease and its complications. Liver Int2011;31:146–162.
[2] Takeda K, Cowan A, Fong GH. Essential role for prolyl hydroxylase domainprotein 2 in oxygen homeostasis of the adult vascular system. Circulation2007;116:774–781.
[3] Chan DA, Kawahara TLA, Sutphin PD, Chang HY, Chi J-T, Giaccia AJ. Tumorvasculature is regulated by PHD2-mediated angiogenesis and bone marrow-derived cell recruitment. Cancer Cell 2009;15:527–538.
[4] Henze AT, Riedel J, Diem T, Wenner J, Flamme I, Pouyseggur J, et al. Prolylhydroxylases 2 and 3 act in gliomas as protective negative feedbackregulators of hypoxia-inducible factors. Cancer Res 2010;70:357–366.
[5] Mazzone M, Dettori D, Dettori D, Leite de Oliveira R, Loges S, Schmidt T, et al.Heterozygous deficiency of PHD2 restores tumor oxygenation and inhibitsmetastasis via endothelial normalization. Cell 2009;136:839–851.
[6] Su Y, Loos M, Giese N, Metzen E, Buchler MW, Friess H, et al. Prolylhydroxylase-2 (PHD2) exerts tumor-suppressive activity in pancreaticcancer. Cancer 2011 (ePub).
[7] Chan DA, Giaccia AJ. PHD2 in tumor angiogenesis. Br J Cancer 2010;103:1–5.[8] Heindryckx F, Mertens K, Charette N, Vandeghinste B, Casteleyn C, Van
Steenkiste C, et al. Kinetics of angiogenic changes in a new mouse model forhepatocellular carcinoma. Mol Cancer 2010;9:219.
[9] Van de Veire S, Stalmans I, Heindryckx F, Oura H, Tijeras-Raballand A,Schmidt T, et al. Further pharmacological and genetic evidence for theefficacy of PlGF inhibition in cancer and eye disease. Cell 2010;141:178–190.
[10] Yang H, Li TW, Peng J, Tang X, Ko KS, Xia M, et al. A mouse model ofcholestasis-associated cholangiocarcinoma and transcription factorsinvolved in progression. Gastroenterology 2011;141:378–388, 388e1–4.
[11] Rogers AB, Theve EJ, Feng Y, Fry RC, Taghizadeh K, Clapp KM, et al.Hepatocellular carcinoma associated with liver-gender disruption in malemice. Cancer Res 2007;67:11536–11546.
[12] Pilette C, Rousselet MC, Bedossa P, Chappard D, Oberti F, Rifflet H, et al.Histopathological evaluation of liver fibrosis: quantitative image analysis vssemi-quantitative scores. Comparison with serum markers. J Hepatol1998;28:439–446.
[13] Van Steenkiste C, Ribera J, Geerts A, Pauta M, Tugues S, Casteleyn C, et al.Inhibition of placental growth factor activity reduces the severity of fibrosis,inflammation, and portal hypertension in cirrhotic mice. Hepatology2011;53:1629–1640.
[14] Spee B, Carpino G, Schotanus BA, Katoonizadeh A, Vander Borght S, Gaudio E,et al. Characterisation of the liver progenitor cell niche in liver diseases:potential involvement of Wnt and Notch signalling. Gut 2010;59:247–257.
[15] Woo HG, Lee JH, Yoon JH, Kim CY, Lee HS, Jang JJ, et al. Identification of acholangiocarcinoma-like gene expression trait in hepatocellular carcinoma.Cancer Res 2010;70:3034–3041.
[16] El-Khoueiry AB, Rankin CJ, Ben-Josef E, Lenz H-J, Gold PJ, Hamilton RD, et al.SWOG 0514: a phase II study of sorafenib in patients with unresectable ormetastatic gallbladder carcinoma and cholangiocarcinoma. Invest NewDrugs 2011. http://dx.doi.org/10.1007/s10637-011-9719-0, [Epub ahead ofprint].
12 vol. 57 j 61–68 67
Research Article
[17] Lee KA, Lynd JD, O’Reilly S, Kiupel M, McCormick JJ, LaPres JJ. The biphasicrole of the hypoxia-inducible factor prolyl-4-hydroxylase, PHD2, in modu-lating tumor-forming potential. Mol Cancer Res 2008;6:829–842.
[18] Elser JJ, Nagy JD, Kuang Y. Biological stoichiometry: an ecological perspectiveon tumor dynamics. Bioscience 2003;53:1112–1120.
[19] Kawahara N, Ono M, Taguchi K, Okamoto M, Shimada M, Takenaka K, et al.Enhanced expression of thrombospondin-1 and hypovascularity in humancholangiocarcinoma. Hepatology 1998;28:1512–1517.
[20] Schmidt T, Kharabi Masouleh B, Loges S, Cauwenberghs S, Fraisl P, Maes C,Schmidt T, et al. Loss or inhibition of stromal-derived PlGF prolongs survivalof mice with imatinib-resistant Bcr-Abl1(+) leukemia. Cancer Cell2011;19:740–753.
[21] Ameln AK, Muschter A, Mamlouk S, Kalucka J, Prade I, Franke K, et al.Inhibition of HIF prolyl hydroxylase-2 blocks tumor growth in mice throughthe antiproliferative activity of TGFbeta. Cancer Res 2011;71:3306–3316.
[22] Lu X, Kang Y. Hypoxia and hypoxia-inducible factors: master regulators ofmetastasis. Clin Cancer Res 2010;16:5928–5935.
[23] Roncalli M, Park YN, Di Tommaso L. Histopathological classification ofhepatocellular carcinoma. Digest Liver Dis 2010;42:S228–S234.
[24] Roskams T. Liver stem cells and their implication in hepatocellular andcholangiocarcinoma. Oncogene 2006;25:3818–3822.
[25] Anilkumar TV, Golding M, Edwards RJ, Lalani EN, Sarraf CE, Alison MR. Theresistant hepatocyte model of carcinogenesis in the rat: the apparentindependent development of oval cell proliferation and early nodules.Carcinogenesis 1995;16:845–853.
68 Journal of Hepatology 20
[26] Bralet M-P, Pichard V, Ferry N. Demonstration of direct lineage betweenhepatocytes and hepatocellular carcinoma in diethylnitrosamine-treatedrats. Hepatology 2002;36:623–630.
[27] Jin LF, Ji SH, Yang JF, Ji WZ. Notch signaling dependent differentiation ofcholangiocyte-like cells from rhesus monkey embryonic stem cells. Zoo Res2011;32:391–395.
[28] Beaussier M, Wendum D, Fouassier L, Rey C, Barbu V, Lasnier E, et al.Adaptative bile duct proliferative response in experimental bile ductischemia. J Hepatol 2005;42:257–265.
[29] Fouassier L, Beaussier M, Schiffer E, Rey C, Barbu V, Mergey M, et al.Hypoxia-induced changes in the expression of rat hepatobiliary transportergenes. AMJ Phys 2007;293:G25–G35.
[30] Sapisochin G, Fidelman N, Roberts JP, Yao FY. Mixed hepatocellularcholangiocarcinoma and intrahepatic cholangiocarcinoma in patients under-going transplantation for hepatocellular carcinoma. Liver Transplant2011;17:934–942.
[31] Zen C, Zen Y, Mitry RR, Corbeil D, Karbanova J, O’Grady J, et al. Mixedphenotype hepatocellular carcinoma after transarterial chemoembolizationand liver transplantation. Liver Transplant 2011;17:943–954.
[32] Paez-Ribes M, Allen E, Hudock J, Takeda T, Okuyama H, Vinals F, et al.Antiangiogenic therapy elicits malignant progression of tumors to increasedlocal invasion and distant metastasis. Cancer Cell 2009;15:220–231.
[33] Yoneda N, Sato Y, Kitao A, Ikeda H, Sawada-Kitamura S, Miyakoshi M, et al.Epidermal growth factor induces cytokeratin 19 expression accompanied byincreased growth abilities in human hepatocellular carcinoma. Lab Invest2011;91:262–272.
12 vol. 57 j 61–68
