JGV Papers in Press. Published December 17, 2014 as doi:10.1099/vir.0.070680-0
1
C-terminal truncated hepatitis B virus X protein enhances the development
2
of diethylnitrosamine-induced hepatocellular carcinogenesis
3 4
Ivan Quetier1,2,3 †, Nicolas Brezillon1,2,3 †, Julien Revaud1,2,3, James Ahodantin1,2,3, Lucie
5
DaSilva1,2,3, Patrick Soussan1,2,3,4,5, Dina Kremsdorf1,2,3*
6 7
1
Inserm, U845, team «Pathogenèse des hépatites virales B et immunothérapie», Paris, France.
8
2
Université Paris Descartes, Sorbonne Paris-Cité, Faculté de Médecine Necker, Paris, France.
9
3
Institut Pasteur, Département de Virologie, Paris, France.
10
4
Service de Virologie, Hôpital Tenon, Paris, France.
11
5
Université Pierre et Marie Curie, Paris, France.
12 13
*Corresponding Author: Dina Kremsdorf
14
Inserm U1135 and Université Pierre et Marie Curie.
15
91 Boulevard de l’Hopital 75013 Paris
16
Email:
[email protected] Phone: +33-1-72606487
17 18
†
These authors contributed equally to this work
19 20
Running title: HBx truncation enhances DEN-induced HCC
21
Category: Animal Viruses - Small DNA
22
Word counts: Abstract, 240; Manuscript, 5584
23
1
1
Abstract
2 3
Hepatitis B virus X protein (HBx) is involved in the development of hepatocellular carcinoma
4
(HCC). The HBx sequence is a preferential site of integration into the human genome, leading
5
to the formation of C-terminal truncated HBx proteins (Ct-HBx). We previously reported that
6
Ct-HBx proteins were able to potentiate cell transformation in vitro. Our present goal was to
7
compare the ability of Ct-HBx and full-length HBx (FL-HBx) proteins to develop or enhance
8
HCC in transgenic mice.
9
In the absence of treatment, neither Ct-HBx nor FL-HBx transgenic mice developed
10
HCC. In young mice treated with diethylnitrosamine (DEN), at 8 months of age, a significantly
11
higher incidence and number of liver lesions were observed in Ct-HBx mice than in FL-HBx
12
and control mice. The earlier development of tumours in Ct-HBx transgenic mice was
13
associated with increased liver inflammation. At 10 months, macroscopic and microscopic
14
analyses showed that statistically, FL-HBx mice developed more liver lesions with a larger
15
surface area compared to control mice. Furthermore, during DEN-induced initiation of HCC,
16
Ct-HBx and FL-HBx transgenic mice showed a highest expression of IL-6, TNF-α and IL-
17
1β transcript, an activation of STAT3 ERK and JNK proteins and an increase in cell apoptosis.
18
In conclusion, in DEN-treated transgenic mice, the expression of Ct-HBx protein causes a
19
more rapid onset of HCC than that of FL-HBx protein. HBV genome integration leading to the
20
expression of a truncated form of HBx protein may therefore facilitate HCC development in
21
chronically infected patients.
22 23
Key words: Hepatitis B virus, Hepatocellular carcinogenesis, Truncated HBx protein
2
1
Introduction
2 3
Chronic hepatitis B virus (HBV) infection is a major risk factor for the development of
4
hepatocellular carcinoma (HCC), accounting for more than 50% of cases worldwide (Kew,
5
2011). The direct mechanisms whereby HBV causes malignant transformation remain elusive.
6
Nevertheless, much of the available evidence supports a pathogenetic role of the viral HBx
7
protein. HBx is a small protein of 17 kDa, which is highly conserved among all mammalian
8
hepadnaviruses and is essential to the enhancement of HBV replication (Assrir et al., 2010).
9
HBx protein does not bind directly to DNA but rather acts on cellular promoters via protein-
10
protein interactions and displays pleiotropic effects on different pathways involved in
11
intracellular signalling, transcriptional activation that modulates cell responses to genotoxic
12
stress, protein degradation, apoptosis and cell division which may be responsible for the
13
potential transforming activities of HBx (for review see (Kew, 2011)). HBx is expressed at low
14
levels during acute and chronic hepatitis and induces a humoral and cellular immune response
15
(Malmassari et al., 2007; Wei et al., 2010). The HBx gene is maintained and transcribed in
16
most integrated subviral DNA, and several studies have demonstrated the expression of HBx
17
mRNA and/or protein in human HCC in the absence of any HBV replication (Toh et al., 2013).
18
Whether HBx directly contributes to the development of HBV-associated HCC or functions as a
19
co-factor in HCC development continues to be debated. Indeed, despite in vivo data showing that
20
HBx transgenic mice may directly develop HCC, it is more likely that HBx expression
21
participates in its development by sensitizing transgenic mice to chemical carcinogens or by
22
cooperating with cellular oncogenes (for review, (Brechot et al., 2010)).
23
Natural mutants of HBx have been described in the liver and serum of patients with
24
different clinical diseases and may play a role in the development of HBV-related HCC
3
1
(Iavarone et al., 2003; Ma et al., 2008; Sirma et al., 1999; Toh et al., 2013; Tu et al., 2001).
2
These include hot-spot mutations in HBx at amino acids 130 and 131, which have been
3
associated with the development of HCC (Iavarone et al., 2003; Kuang et al., 2004; Lee et al.,
4
2011). Furthermore, the random integration of HBV DNA into the host genome frequently
5
leads to 3’ end deleted sequences of the HBx gene and the observation of chimeric transcripts
6
which often carry deletions at the 3 -end of HBx gene, resulting in C-terminal truncated HBx
7
protein (Iavarone et al., 2003; Jiang et al., 2012; Ma et al., 2008; Sirma et al., 1999; Sung et
8
al., 2012; Sze et al., 2013; Toh et al., 2013; Tu et al., 2001). Interestingly, a recent study
9
reports, in human HCC samples, an association between the presence of C-terminal truncation
10
of HBx and venous invasion (Sze et al., 2013). However, most of the data in favour of the
11
contribution of C-truncated HBx proteins in tumorigenesis are from in vitro studies. We
12
previously reported that HBx proteins with deletion of amino acids at the C-terminal region
13
have altered capacity to control cell proliferation and transcriptional activity but were able to
14
potentiate ras and myc cell transformation (Sirma et al., 1999; Tu et al., 2001). Other in vitro
15
data have shown that these HBx mutants retain the ability to bind to p53 and to block p53-
16
mediated apoptosis (Huo et al., 2001). The C-terminal region of HBx is indispensable for HBx
17
specific localization in mitochondria, which is involved in activating several cytosolic signal
18
transduction pathways (Shirakata & Koike, 2003). Although the extent of deletion of the 3 -end
19
of HBx gene varies, the integrity of the DDB1 binding domain (aa 76-125) probably protects
20
HBx truncated protein from proteasome-mediated degradation (Bergametti et al., 2002; Lin-
21
Marq et al., 2001). By contrast, C-terminal truncated HBx proteins showed specific
22
mechanisms involved in regulation of cell cycle and apoptosis (Bock et al., 2008; Jiang et al.,
23
2010; Liu et al., 2008b).
′
′
24
These in vitro results suggest that the HBx protein with C-terminal deletions may be
25
implicated in the development of HCC through distinct biological functions than full-length
4
1
HBx protein. Therefore, in order to better define the involvement of naturally-occurring HBx
2
protein mutants in the evolution of liver disease to HCC, we decided to perform an in vivo
3
comparison of the carcinogenic or co-carcinogenic capacities of C-terminal truncated HBx (Ct-
4
HBx) and non-truncated HBx (FL-HBx) proteins obtained from tumour and non-tumour
5
biopsies from the same patient.
5
1
Results
2 3
Characterization of HBx transgenic strains
4
We previously reported that, in vitro, that HBx proteins with natural deletion at the C-
5
terminal region have altered capacity to control cell proliferation and transcriptional activity
6
but were able to potentiate ras and myc cell transformation and that a deletion of 16 aa was
7
enough to observe these effects (Sirma et al., 1999; Tu et al., 2001). Thus, to generate HBx
8
transgenic mice we used a couple of tumor /non tumoral sequences (patient 1) differing only
9
by a deletion of 18 aa of HBx and the addition of three amino acids from the cellular genome
10
in the tumoral region (Fig.1b) (Sirma et al., 1999; Tu et al., 2001).
11
Transgenic mice expressing full-length HBx (FL-HBx) (pX-HBx-1336 and pX-HBx-
12
1368) or C-terminal truncated HBx (Ct-HBx) (pX-HBx-1280), under HBV regulatory
13
elements, were generated (Figure 1a). HBx transcripts were quantified using quantitative RT-
14
PCR and the highest expression of transcripts was observed in pX-HBx-1336 transgenic mice
15
(Fig.1c). Assuming an identical level of expression in each hepatoctye, the HBx RNA copy
16
number was estimated to be 8 to 25 copies/hepatocyte for the 1336 lineage; 0.2 to 0.5
17
copies/hepatocyte for the 1368 lineage and 0.07 to 0.2 copies/hepatocyte for the 1280 lineage.
18
In accordance with the level of viral transcript expression, the HBx protein was detected in the
19
two FL-HBx transgenic mouse liver tissues but barely detected in Ct-HBx transgenic mice
20
(Fig.1d). Finally, different organs were investigated for the expression of HBx mRNA, in
21
addition to the liver, a significant expression of HBx transcript was only observed in kidney by
22
RT-PCR (Quetier et al., 2013).
23 24
Susceptibility of HBx transgenic mice to DEN-induced hepatocarcinogenesis
6
1 2
At 18 months of age not direct carcinogenesis was observed in Ct-HBx or FL-HBx transgenic mice (data not shown).
3
Thus, DEN-induced liver tumour model was used to explore whether the Ct-HBx
4
protein was able as FL-HBx protein to participate to hepatocarcinogenesis development (Lee et
5
al., 2004; Madden et al., 2001). To exert its carcinogenic effect DEN needs to be bioactivated
6
in the liver by several P450 isozymes, including CYP2E1, resulting in DNA-adducts form
7
through an alkylation mechanism (Verna et al., 1996). The resulting O6-alkylguanine leads to
8
GC–AT transitions, which are believed to be largely responsible for DEN-induced
9
carcinogenesis. Exposure to DEN is also associated with hepatocellular accumulation of
10
reactive oxygen species, which may enhance DEN-induced hepatocarcinogenesis (Santos et
11
al., 2012). Livers of Ct-HBx and FL-HBx transgenic mice, and genetically matched WT
12
treated mice were removed from post mortem mice at 8, 10 or 12 months. Macroscopically
13
visible lesions on the liver surface, representing as well as preneoplastic focal lesions and
14
benign or malignant neoplasmic lesions were counted and microscopic analyses were
15
performed (Fig.2 and 3). At 8 months, the incidence and the number of macroscopic liver
16
lesions were significantly higher in Ct-HBx transgenic mice compared to FL-HBx transgenic
17
and control mice (Fig.2b and e). Similarly, the microscopic evaluation of hematoxylin and
18
eosin stained slides showed a higher incidence of tumours in Ct-HBx transgenic mice
19
compared to FL-HBx transgenic and control mice (Fig.2b). However, probably because of the
20
small number and size of observed lesions per slides, no statistical difference in tumour area
21
was observed between the three groups of mice (Fig.2e). The microscopic evaluation of
22
hematoxylin and eosin stained slides showed essentially preneoplastic foci and area with
23
inflammatory cells (Fig.3a-c). At 10 months, the incidence of macroscopic liver lesions was a
24
similar in the WT and FL-HBx transgenic mice (Fig.2c). However, the number of macroscopic
25
lesions was statistically higher in FL-HBx transgenic mice compared to WT mice (Fig.2f).
7
1
Similarly, histological assessment of the liver tumours showed a significantly higher tumour
2
area in FL-HBx transgenic mice than in WT mice (Fig.2f). Histopathological examination
3
revealed that most of the tumour nodules seen in the livers were of a neoplastic type (Fig.3d-f).
4
In 12-month-old FL-HBx and Ct-HBx transgenic animals, macroscopic and microscopic
5
examinations do not evidenced differences (Fig.2d and g). Histopathological examination of
6
the livers revealed enlarged cancerous tissue in a high proportion of animals, corresponding to
7
well-differentiated HCC or multifocal nodules of poorly-differentiated HCC (Fig.3g-i).
8
In order to evaluate the liver damage in DEN-treated mice, blind scoring for steatosis,
9
necrosis and inflammatory cell infiltration were determined (Fig.4). In 8 month-old Ct-HBx
10
transgenic mice, but not in FL-HBx transgenic mice and control mice, significant increase in
11
inflammation was observed (Fig.4d). Indeed, in Ct-HBx transgenic mice, scattered small
12
inflammatory cells in the lobule, associated or not with portal inflammation, were observed in
13
all the mice (data not shown). In 10 month-old FL-HBx transgenic and control mice, no
14
differences were seen with respect to hepatic damage (Fig.4e).
15
Overall, in this model of DEN-induced carcinogenesis, our findings are in favour of a
16
co-carcinogenic role for FL-HBx and Ct-HBx proteins, the effect of the latter occurring earlier.
17
Furthermore, the earlier development of tumours in Ct-HBx transgenic mice was associated
18
with increased liver inflammation.
19 20
Effect of HBx expression on DEN-induced initiation of carcinogenesis
21
The next step was to determine whether FL-HBx and Ct-HBx transgenic mice were more
22
sensitive to DEN-induced acute hepatic injury. In aged mice, DEN act only as an initiator of
23
liver injury. This is possibly associated to the lesser capacities of hepatocyte to profiler in adult
24
liver. Indeed, hepatocyte proliferation is needed for DEN-induced complete carcinogenesis
8
1
(Maeda et al., 2005). Thus, two-month old mice we injected with a high dose of DEN (100
2
mg/kg body weight) and cell death and compensatory proliferation were investigated.
3
To quantify cell death, TUNEL assays were performed on liver sections 48h after the
4
DEN injection. Ct-HBx and FL-HBx transgenic mice displayed a significantly larger labelled
5
area than WT mice (Fig.5a). The highest rate of hepatocytes apoptosis (with a 2-fold increase)
6
was observed in Ct-HBx transgenic mice (Fig.5b). At the same time, compensatory cell
7
proliferation was measured by Ki-67 nucleus labelling (Fig.5a and c). At this short time of
8
DEN treatment, in agreement with previous data (Schneider et al., 2012), proliferation was
9
essentially observed in non parenchymal cells. Despite a higher degree of cell proliferation in
10
Ct-HBx transgenic mice when compared to both FL-HBx transgenic and WT mice, no
11
statistical difference was observed (Fig.5c). As, in vitro, this specific Ct-HBx protein has lost
12
its cytotoxic activity (Tu et al., 2001); we wanted to investigate if the observed higher increase
13
of apoptosis in Ct-HBx transgenic mice might be associated to a sensitisation of cells to TNF-
14
induced apoptosis. For that, we used the model of fulminant hepatitis induced by TNF-α and
15
D-galactosamine treatment and evaluated the survival of FL-HBx and Ct-HBx transgenic mice
16
(Fig.5d). Interestingly, Ct-HBx transgenic mice were significantly more sensitive to TNF-α
17
and D-galactosamine induced-fulminant hepatitis.
18
The inflammatory process that occurred soon early after the DEN injection is a factor
19
involved in hepatocarcinogenesis. We therefore decided to analyse the expression of IL-6,
20
TNF- , IL-1 and TGF transcripts using RT-qPCR, 4 hours after the DEN treatment. An
21
increased induction of IL-6, TNF- and IL-1β in FL-HBx and Ct-HBx mice was observed,
22
when compared to the control mice (Fig.5e). The expression of TGFβ (which is mostly
23
involved in the development of fibrosis) was not affected by DEN treatment (Fig.5e).
24
Activation of signalling pathways under IL-6 and TNF- , namely STAT3, JNK and ERK1/2,
25
were also analysed (Fig.5f). Four hours after the DEN treatment, we observed an increase in
α
β
β
α
α
9
1
the phosphorylation of JNK, ERK1/2 and STAT3 in both Ct-HBx and FL-HBx transgenic
2
mice livers when compared to the livers of WT mice (Fig.5f). In untreated animals, expression
3
of transcripts and proteins were similar. Interestingly, activation of ERK1/2 was higher in Ct-
4
HBx mice as compare to FL-HBx mice. Finally, 48h post treatment, we did not observe major
5
modifications in expression of mRNAs and proteins in liver of HBx transgenic mice as
6
compared to control (data not shown).
7
These observations suggest that the inflammatory and apoptosis responses related to
8
DEN administration is enhanced in HBx transgenic mice and confirmed the role of FL-HBx
9
and Ct-HBx proteins in the modulation of hepatocyte metabolism, which could contribute to
10
the induction of tumours following treatment with DEN.
11 12
Activation of signalling pathways in the livers of DEN-treated mice
13
In order to investigate whether Ct-HBx or FL-HBx proteins could modified
14
inflammatory or oncogenic pathways in DEN-treated transgenic mice 8 and 12 months liver
15
biopsies we analyzed by western blot (Fig.6a). It should be noted, that at 8 months, lesions
16
were very small (Fig.2a) render difficult to have biopsies with only preneoplasic tumours. In
17
contrary, for biopsies obtained at the 12 months, most of the liver tissues corresponded to HCC
18
tumours (Fig.2a). In addition, HBx mRNA expression was investigated in DEN-treated or not
19
FL-HBx and Ct-HBx transgenic mice at 8 and 12 months. DEN treatment did not modified
20
HBx mRNA expression (Fig.6c).
21
The activation of STAT3, JNK and ERK, which is frequently observed in a context of
22
HCC was investigated. Surprisingly, we did not observe any STAT3 activation at either 8 or 12
23
months, in any of the mice. Activation of the stress-related JNK pathway was observed
24
between 8 and 12 months in WT mice, and was more pronounced between 8 and 12 months in
25
Ct-HBx and FL-HBx transgenic mice. Finally, the ERK pro-proliferative pathway was
10
1
markedly activated in the livers of Ct-HBx and FL-HBx transgenic mice at 12 months of age
2
when compared to the WT mice livers. We also investigated the level of expression of the
3
alpha foetal protein (AFP), re-expressed in the majority of HCC. AFP was highly expressed in
4
12 months treated mice and more specifically in HBx transgenic mice. In addition, in
5
agreement with a compensatory liver cell proliferation that drives carcinogenesis in liver,
6
proliferation cell nuclear antigen (PCNA), a marker of cell proliferation, was highly expressed
7
in 12 months treated mice, independently to the mice genotype (Fig.6a). To evaluate the
8
impact of HBx expression in the absence of DEN treatment, we have investigated the
9
expression of ERK and PCNA in biopsies of 12 months from untreated transgenic and control
10
mice (Fig.6b). In contrast to ERK activation in 12 months Ct-HBx and FL-HBx DEN-treated
11
transgenic mice, no difference was observed in 12 months untreated mice. Interestingly, in 12
12
months Ct-HBx transgenic mice, PCNA expression seems to be increased. This difference was
13
no more observed in treated animals.
14 15
Thus, after DEN treatment, pathways involved in cell proliferation and HCC development were up-regulated in Ct-HBx and FL-HBx transgenic mice livers.
16
11
1
Discussion
2 3
Recent studies using new generation sequencing technologies have highlighted the fact
4
that following the integration of HBV into the cellular genome, the 3’end of the HBx gene is
5
often deleted and this event is observed more frequently in tumour tissues (Jiang et al., 2012;
6
Sung et al., 2012; Toh et al., 2013). Furthermore, truncated HBx-human chimeric transcripts
7
may be seen and expressed as chimeric proteins (Sirma et al., 1999; Toh et al., 2013; Tu et al.,
8
2001). This has raised the question of a putative role for Ct-HBx proteins in the development
9
of HCC. Our study was carried out in order to enable in vivo investigation of the oncogenic
10
potential of the C-terminal truncated form of HBx by comparison with full-length HBx protein.
11
First of all, we observed that in the absence of any treatment, neither FL-HBx nor Ct-
12
HBx transgenic mice developed HCC, as evaluated by macroscopic and microscopic analyses
13
performed during a 18 months follow-up. The data available on the ability of FL-HBx protein
14
to directly induce carcinogenesis is contradictory. Indeed, although few FL-HBx transgenic
15
mice strains might develop spontaneous HCC, there is more evidence that FL-HBx protein
16
participates in HCC development by sensitising the animals to carcinogen inducers. Genetic
17
background of the mice and the level of expression or genotype of HBx protein may explain
18
these discrepancies (Kremsdorf et al., 2006).
19
To evaluate in our model, where FL-HBx and Ct-HBx were expressed under their own
20
promoter, whether FL-HBx or Ct-HBx proteins were involved in liver carcinogenesis, DEN-
21
induced HCC treatment was chosen because of its relevance to human HCC genes induction.
22
Indeed, when injected into young animals, the gene expression patterns of HCC induced by
23
DEN in mice were generally similar to those seen in the poorer survival group of human HCC
24
(Lee et al., 2004). Our findings were in favour of an accelerated development of liver tumours
25
in Ct-HBx and FL-HBx transgenic mice when compared to WT mice. Indeed, in Ct-HBx
12
1
transgenic mice, 8 months after DEN treatment, livers displayed a higher macroscopic and
2
microscopic incidence of tumours, associated with an increase in liver inflammation. Whereas,
3
in FL-HBx transgenic mice, increases in the number of tumours and in tumour area versus WT
4
mice were only observed 10 months after DEN treatment. To our knowledge, this study
5
provides the first indication under in vivo conditions of the ability of the Ct-HBx mutant, not
6
only to induce HCC but to do it more rapidly than FL-HBx. This emphasises that the HBx
7
deletion mutant may develop distinct biological properties which contribute to liver
8
carcinogenesis. Initial in vitro data demonstrated the ability of the Ct-HBx protein to abrogate
9
the growth-suppressive and apoptotic effects of FL-HBx protein (Huo et al., 2001; Sirma et al.,
10
1999; Tu et al., 2001) and that Ct-HBx might enhance cell proliferation (Jiang et al., 2010; Liu
11
et al., 2008b; Ma et al., 2008). The Ct-HBx protein has also been reported to modulate
12
microRNA expression (Fu et al., 2012; Yip et al., 2011), to regulate Wnt-5a expression (Liu et
13
al., 2008a) and to play a role in cell invasiveness by enhancing the transcription of
14
metalloproteinase 10 (MMP10) through C-Jun (Sze et al., 2013).
15
DEN-induced carcinogenesis promotion requires the proliferation of hepatocytes, thus
16
in adults, where mature hepatocytes display an extremely low turnover rate, DEN
17
administration induce only initiation stage without tumour development. Our experiments
18
revealed that during the DEN-induced initiation stage, there was an increase in the
19
sensitization of both FL-HBx and Ct-HBx transgenic mice to liver injury. Interestingly, a more
20
pronounced apoptosis was observed in Ct-HBx transgenic mice. We have evidenced that this
21
might be linked to a highest sensitivity of Ct-HBx transgenic mice to TNF-α induced
22
apoptosis. These observations are in accordance with previous data demonstrating that the
23
severity of acute hepatic injuries correlates to the level of cell proliferation and the
24
development of liver tumours during the promotion stages (Lee et al., 2004; Maeda et al.,
25
2005). For this reason, the increased inflammatory response and apoptosis observed in FL-
13
1
HBx and Ct-HBx transgenic mice during the initiation phase of DEN-induced carcinogenesis
2
might have been related to the more rapid onset and development of DEN-induced HCC,
3
particularly in Ct-HBx transgenic mice. We also observed at early stages of DEN-induction an
4
increase in cell proliferation in the livers of Ct-HBx transgenic mice, but not in those of FL-
5
HBx transgenic mice, as compared to WT mice. In a previous study, increased hepatocellular
6
proliferation in transgenic mice expressing a full-length HBx gene was observed 48 hours after
7
the injection of DEN in 12-day-old mice (Madden et al., 2001). The use of different
8
experimental protocols (newborn vs. adult animals) probably reflects the differences in the
9
results obtained. However, in line with our data, in the same study, at 8 months of age, no
10
measurable effect of HBx protein expression on hepatocytes proliferation was evidenced
11
(Madden et al., 2001). Taken together, this indicates that the involvement of HBx protein in
12
the induction of cell proliferation probably occurs during the early stages of DEN-induced
13
carcinogenesis. Furthermore, the greater capacity of Ct-HBx protein to induce cell proliferation
14
may account for the more rapid tumour development observed during the promotion stage of
15
DEN-induced carcinogenesis.
16
During DEN-induced initiation stage of carcinogenesis we have observed an increase
17
activation of mitogen-activated protein kinases (JNK and ERK) and STAT3 pathways in both
18
transgenic mice compare to WT mice. DEN treatment leads to an accumulation of reactive
19
oxygen species inducing the activation of JNK and STAT3, which are both involved in liver
20
carcinogenesis (He et al., 2010). Two in vitro studies demonstrated the ability of full-length–
21
but not C-terminal truncated HBx–to induce the generation of reactive oxygen species, leading
22
to the activation of STAT3 (Jung & Kim, 2013; Waris et al., 2001). This suggests that, in our
23
model, the increased activation of JNK and STAT3 by HBx proteins was probably independent
24
of the accumulation of reactive oxygen species. Alternatively, JNK may be activated by
25
diverse stimuli, including cytokines such as IL-1 and TNF-α (Seki et al., 2012). It could
14
1
therefore be hypothesised that the observed increase in IL-1β and TNF-α transcripts might
2
account for JNK activation. Furthermore, in a hepatoma cell line, it has been reported that HBx
3
protein potentiated the phosphorylation of JNK by interacting with Jab1 (Tanaka et al., 2006).
4
Along the same line of evidence, we have observed an increased expression of IL-6 in both
5
transgenic mice. Interestingly, in addition to DEN-induction of IL-6 production in Kupffer
6
cells (Naugler et al., 2007), HBx, stimulates the production of IL-6 in a MyD88-dependent
7
manner in hepatoma cells (Xiang et al., 2011).
8
Interestingly, it has been reported that HBx protein may up-regulate FoxM1 expression
9
through the ERK/CREB pathway, and that FoxM1 over-expression indicated a poor prognosis
10
in HBV related-HCC patients (Xia et al., 2012). Surprisingly, insofar as the IL-6/STAT3
11
pathway is clearly up-regulated during the DEN-induced initiation phase of HCC in HBx
12
transgenic mice, STAT3 activation was barely detectable 8 or 12 months after DEN-induced
13
carcinogenesis in our animals.
14
Considering activation of mitogen-activated protein kinases it was reported, in vivo,
15
that the prolonged activation of ERK by HBx expression is also accompanied by the activation
16
of JNKs (Nijhara et al., 2001). Furthermore, it was identified that the internal region (amino
17
acids 58–140) of HBx was as effective as the full-length HBx in activating MAPKs (Nijhara et
18
al., 2001). Our observations that the ERK and JNK pathways are simultaneously activated by
19
HBx demonstrate that activation of both signalling pathways is required to trigger a full
20
spectrum of phenotypic traits associated with oncogenesis in HBx expressing livers. This
21
reinforces the idea that HBx may be involved in HBV-mediated liver carcinogenesis, through
22
pro-proliferation mechanisms. Consistent with these findings, in our study the ERK and JNK
23
pathways in both transgenic mice livers, 8 and 12 months after the DEN treatment, were up-
24
regulated when compared to WT mice. In addition, we showed that AFP, an important
25
biomarker of HCC, was highly expressed in 12 months treated transgenic mice. AFP is. It was
15
1
demonstrated that HBx derepress AFP expression through its interaction with p53 and that the
2
expression of AFP receptor was a pivotal event for HBx inducing malignant transformation of
3
liver cells (Li et al., 2013; Ogden et al., 2000).
4
The mechanisms underlying liver carcinogenesis involve numerous steps during which
5
both the immune system and viral proteins may be responsible for initiating environmental and
6
biological modifications that lead to the development of HCC. Among them, it is important to
7
mention that abnormal activation of AFP, ERK and JNK is often observed in human HCC
8
(Min et al., 2011; Seki et al., 2012). The HBx protein by modulating these pathways might
9
contribute to the development of HCC. Our study showed that the expression of Ct-HBx
10
protein led to a more rapid onset of HCC than that of FL-HBx protein. It could be assumed that
11
the expression of truncated forms of HBx, by accelerate the initiation step of carcinogenesis
12
may facilitate cell transformation. This suggests that the integration of HBV into the host
13
genome in addition to cis-acting effects may exert trans-acting effects via the expression of
14
truncated HBx proteins and thus contribute to tumorigenesis.
15
16
1
Materials and methods
2 3
Transgenic mice
4
Two sequences of the HBx gene isolated from either a non-tumour (full-length HBx;
5
FL-HBx) or tumour (3’ end truncated HBx; Ct-HBx) region in the same patient were used to
6
generate transgenic mice (Tu et al., 2001). HBx sequences containing the myc-his tag were
7
cloned downstream of the HBx promoter and enhancer I region (832-1371) and β-globin
8
intronic sequences (Fig.1). Three independent lines were derived from the founders (pX-HBx-
9
1336 and pX-HBx-1368 expressing FL-HBx, and pX-HBx-1280 expressing Ct-HBx), and
10
expanded by back-cross with C57BL/6J strain (Institut Clinique de la Souris, Strasbourg,
11
France). All the experiments were performed on heterozygous transgenic mice. The animals
12
were treated in accordance with European Union regulations on animal care (Directive
13
86/609/EEC).
14 15
Animal experiments
16
For the direct carcinogenic study, untreated FL-HBx, Ct-HBx and WT mice from the
17
same litter were followed for 8, 12 and 18 months. For the co-carcinogenesis study, diethyl
18
nitrosamine (DEN) was injected intraperitoneally at 2µg/g body weight into 10 day-old male
19
mice. The mice were euthanized 8, 10 and 12 months after the DEN injection. At the time of
20
sacrifice, macroscopic lesions were counted on the surface of the liver. In order to induce acute
21
liver damage, DEN was injected intraperitoneally at 100µg/g body weight into 2 month-old
22
male FL-HBx, Ct-HBx (pX-HBx-1336 strain) and WT mice. The mice were euthanized 4 or
23
48 hours post-injection. The left lobe was fixed in 4% paraformaldehyde and embedded in
24
paraffin for histological analysis. The remnant liver was frozen for protein or RNA analysis.
25
For survival experiments, animals were fasted for 14 hours and fulminant hepatitis was
17
1
induced by intraperitoneal injection of murin TNF-α (20µg/kg, AbCys) and D-galactosamine
2
(700mg-kg, Sigma). The experimental protocols were approved by the French Committee on
3
the Ethics of Animal Experiments (MESR N°0079303).
4 5
Histological analysis
6
From each mouse, three liver sections of 4µm, each separated by 200µm, were stained
7
with hematoxylin and eosin and used for histological analysis. Liver sections were digitized
8
(Hammamatsu Nanozoomer) and analysed with NDP view software. For each mouse, the three
9
liver sections were used to determine mean tumour number and area per slides and to blind-
10
scored (0 to 4) (two investigators) for steatosis, necrosis, and inflammation using a modified
11
scoring scale (Brandon-Warner et al., 2012). This score was representative of the mean of the
12
three sections from each mouse.
13 14
Cell proliferation and apoptosis
15
Cell proliferation analysis was performed by immunohistochemistry using Ki-67
16
antibody (Novocastra, clone NCL-Ki67p). The number of positive nuclei was evaluated on 15
17
defined areas representing about 10% of the whole liver section and reported as the number of
18
positive cells per mm² of liver.
19
To study apoptosis, TUNEL staining was performed using the In Situ Cell Death
20
Detection Kit (Roche). After digitizing the slides, TUNEL labelling was quantified using
21
Image-J software. TUNEL brown labelling and the counter-staining were separated using
22
threshold tools and the percentage labelled area in the whole liver section was determined. For
23
each experiment, the mean TUNEL-positive area found in WT mice was set at one and the
24
TUNEL area was expressed in relative units.
25
18
1
Western blotting
2
Total proteins were extracted from frozen liver tissues as previously described (Quetier
3
et al., 2013). For HBx protein detection, a total of 200 µg liver protein and 10 µg of HBx-myc
4
transfected HuH7 protein (positive control) was loaded onto 12% SDS-PAGE and transferred
5
onto a nitrocellulose membrane. The membranes were incubated with c-myc antibody (1:100)
6
(Santacruz: sc-40). For cellular protein detection, 20 µg of protein extracts were resolved on
7
SDS-PAGE (10%) and transferred onto a nitrocellulose membrane which was then incubated
8
with the following primary antibodies (1/1000) in PBS-BSA (5%): GAPDH (Santacruz),
9
PCNA (M0879, Dako), phospho-STAT3 (#9131), STAT3 (#9132), phospho-ERK1/2 (#9101),
10
ERK (#9102), phospho-JNK (#9251), JNK (#9258) and AFP (#3903) (Cell Signaling). The
11
secondary antibody was anti-rabbit antibody coupled to horseradish peroxidase, and the bands
12
were revealed using the ECL plus (Amersham) or ECL Prime (Invitrogen) revelation systems.
13 14
RNA quantification
15
Total RNA was extracted from frozen liver tissues (Qiagen kit) and quantified using
16
Nanodrop. 1.5 µg of RNA was used for reverse transcription with random primers and
17
SuperScript II (Invitrogen). cDNAs quantification was performed on a Taqman 7500 (Applied
18
Biosystems) using either the TaqMan Master Mix (Applied Biosystems) and primers/probes
19
specific to IL-6, TNF-α, IL-1β and TGF-β (Integrated DNA Technology), or the SybrGreen
20
PCR Master Mix for GAPDH (forward: 5’-AGACGGCCGCATCTTCTTGTGCA, reverse: 5’-
21
GCCCAATACGGCCAAATCCGTTC). To analyse real-time PCR data, the comparative CT
22
method (2-ΔΔCT) was used (Schmittgen & Livak, 2008). Data were normalized using GAPDH
23
and expressed as the relative mRNA level compared to untreated control animals. HBx mRNA
24
expression was quantified using Lightcycler DNA MasterMix SybrGreen I (Roche) and
25
specific HBx primers (forward: 5’-GCGTCCTTTGTCTACGTCCCGTC; reverse: 5’-
19
1
GCGGAAGTATGCCTCAAGGTCGG). A standard curve was included to quantify HBx
2
mRNA in copy numbers per µg of total RNA.
3 4
Statistical analysis
5
Statistical analyses were performed the non-parametric Mann–Whitney U-test,
6
Student’s t test, or chi-square test to compare the different groups. A p value