Hepatocarcinogenesis in transgenic mice carrying hepatitis B virus pre-S/S gene with the sW172* mutation

Hepatitis B virus (HBV) carrying the rtA181T/sW172* mutation conferred cross-resistance to adefovir and lamivudine. Cell-based and clinical studies indicated that HBV carrying this mutation had an increased oncogenic potential. Herein, we created transgenic mouse models to study the oncogenicity of the HBV pre-S/S gene containing this mutation. Transgenic mice were generated by transfer of the HBV pre-S/S gene together with its own promoter into C57B6 mice. Four lines of mice were created. Two of them carried wild-type gene and produced high and low levels of HBV surface antigen (HBsAg) (TgWT-H and L). The other two carried the sW172* mutation with high and low intrahepatic expression levels (TgSW172*-H and L). When sacrificed 18 months after birth, none of the TgWT mice developed hepatocellular carcinoma (HCC), whereas 6/26 (23.1%) TgSW172*-H and 2/24 (8.3%) TgSW172*-L mice developed HCC (TgWT vs TgSW172* P=0.0021). Molecular analysis of liver tissues revealed significantly increased expression of glucose-regulated protein 78 and phosphorylated extracellular signal-regulated kinases 1 in TgSW172* mice, and decreased expression of B-cell lymphoma-extra large in TgSW172*-H mice. Higher proportion of apoptotic cells was found in TgSW172*-H mice, accompanied by increased cyclin E levels, suggesting increased hepatocyte turnover. Combined analysis of complimentary DNA microarray and microRNA array identified microRNA-873-mediated reduced expression of the CUB and Sushi multiple domains 3 (CSMD3) protein, a putative tumor suppressor, in TgSW172* mice. Our transgenic mice experiments confirmed that HBV pre-S/S gene carrying the sW172* mutation had an increased oncogenic potential. Increased endoplasmic reticulum stress response, more rapid hepatocyte turnover and decreased CSMD3 expression contributed to the hepatocarcinogenesis.


INTRODUCTION
Worldwide, hepatocellular carcinoma (HCC) is the sixth most commonly diagnosed solid cancer and the second most common cause of cancer-related death. 1 In Taiwan, it ranked the first and second respectively in male and female cancer mortality. 2 HCC is multifactorial in origin and the three most important causes of liver cancer are chronic hepatitis B virus (HBV) infection, chronic hepatitis C virus infection and alcoholic liver disease. 3,4 Other risk factors include old age, male gender, underlying chronic liver diseases, liver cirrhosis, aflatoxin exposure and diabetes. 5,6 In Southeast Asia, chronic hepatitis B is highly prevalent, leading to a correspondingly high incidence of HCC. Despite a successful vaccination program, HBV remains the major etiological factor of HCC in Taiwan to date.
Because of a multifactorial etiology, the key molecular pathways leading to hepatocarcinogenesis are not well understood. It is believed that hepatocarcinogenesis involved not only multiple steps of molecular events but also heterogeneous cellular pathways. [7][8][9] In HBV-associated HCC, several virological factors have been demonstrated to manifest important prognostic correlation. For example, high serum or intrahepatic HBV-DNA levels and the presence of basal core promoter A1762T/G1764A mutation can independently predict outcomes of long-term chronic HBV infection as well as postoperative survival in HCC. [10][11][12] In addition, antiviral therapy using nucleos(t)ide analogs to suppress HBV-DNA levels significantly reduced long-term HCC occurrence and postoperative HCC relapse. [13][14][15] Despite the success of antiviral strategy, several studies indicated that in cirrhosis patients, anti-HBV therapy with nucleos(t)ide analogs cannot completely eliminate HCC occurrence. 16,17 One of the explanations for such observation is the failure to completely suppress HBV-DNA replication. Our groups have shown that potent anti-HBV treatment frequently leads to selection of the S gene mutants. 18,19 Such mutants could be generated due to emergence of the overlapping polymerase resistant mutations or alternatively, by an enhanced humoral immunity. Of them, the S gene truncation mutants, such as the rtA181T/sW172* mutant, were associated with hepatocarcinogenesis. 20,21 It has been suggested that intrahepatic accumulation of mutant pre-S/S proteins due to secretion defect could be the reason why such mutants were selected during antiviral therapy and why they were associated with HCC. [20][21][22][23][24] Other supporting evidence included previous studies showing that S gene stop codon mutations found in the integrated HBV genomes in HCC tissues harbored transactivation and oncogenic activities. 25,26 However, hepatic expression levels of the S truncation proteins derived from HBV integration was usually low and the sW172* mutation was not located on the characterized 'transactivation-on' region. On the other hand, HBV carrying the rtA181T/sW172* mutation could accumulate to a high serum level, insuring the generation of significant amount of truncated pre-S/S proteins. 27 Clinical analysis showed that the rtA181T/sW172* mutation was associated with an increased risk of HCC in cirrhotic patients. 28 As such, this truncated pre-S/S protein is considered an oncogenic hepatitis B viral protein in review articles. 29,30 In the present study, we aimed to provide transgenic mouse evidence for the increased oncogenicity of pre-S/S proteins carrying the rtA181T/sW172* mutation. Such experiment could illustrate that the truncated pre-S/S proteins alone contribute to an increase of oncogenicity, excluding the interference effect of HBV-DNA level and other HBV viral proteins.

RESULTS
Establishment of transgenic mice carrying HBV pre-S/S gene Before generation of transgenic mice, the plasmids carrying the wild type and mutant pre-S/S gene were transiently transfected into Hepa1-6 cells to insure the endogenous HBV pre-S/S promoter sufficiently driven the expression of pre-S/S proteins ( Figure 1b). The method to generate transgenic mice carrying HBV pre-S/S gene was described in the Materials and methods section. Four lines of mice were generated (Table 1). TgWT-L and TgWT-H carried wild-type pre-S/S gene and expressed low and high levels (419 ± 162 IU/ml versus 599 ± 159 IU/ml) of serum HBsAg, respectively. TgSW172*-L and TgSW172*-H carried HBV pre-S/S gene containing the sW172* mutation and expressed low and high levels of intrahepatic HBsAg respectively, determined by western blot and immunohistochemistry (IHC) analysis (Figures 1c-e). Low level of serum HBsAg (10.2 ± 19.1 IU/ml) was detected in TgSW172*-H mice, whereas serum HBsAg was undetectable in TgSW172*-L mice. Western blot analysis using monoclonal antibody against HBsAg detected small surface (S) proteins, gp24 and gp27, only in TgWT-L and H mice but not in TgSW172*-L and H mice (Figure 1c, left panel). When the film was overexposed, the middle surface protein (M), gp33, was visualized ( Figure 1c, right panel). The gp36 was only barely seen (not marked). However, large surface proteins could not be clearly demonstrated in this study. To detect truncated surface proteins, polyclonal antibody against HBsAg was used. Small amounts of truncated S proteins (St), gp18 and gp22, were detected ( Figure 1d, left panel) in TgSW172*-L and H mice. When pre-S2 antibody was used, truncated M proteins, gp27 and gp30, were visualized ( Figure 1d, right panel). A larger protein corresponding to the truncated large surface protein was seen in some but not all mice ( Figure 1d, right panel, under the non-specific bands). The polyclonal anti-Pre-S2 was raised against the pre-S2 region, which was absent in the major S protein. Detection using the polyclonal anti-HBs, unfortunately, displayed heavy nonspecific (NS) bands over the predicted wild type HBsAg positions (Figure 1d). IHC was performed to cross-examine the surface protein expression (Figure 1e). The results were consistent with the western blot data.
Copy number of the transgene was determined by real-time PCR method. It was found that each of the four lines carried only one copy of transgene.
Increased ER stress response and hepatocyte turnover in HBV pre-S/S transgenic mice To explore mechanisms of hepatocarcinogenesis in the transgenic mice, several assays were carried out. TUNEL assay to estimate the numbers of apoptotic cells in liver showed a significantly higher percentage of apoptotic cells in TgsW172*-H mice compared with that in TgWT-H mice (Figures 2a and b). IHC analysis for active caspase 3 expression was also performed for confirmation ( Figure 2c). Consistently, a higher percentage of positive cells was found for all three Tg mice compared with that of the null mice and a higher percentage was observed for TgsW172*-H mice compared with that of TgWT-H mice. Semi-quantitative assessment using western blot analysis showed an increase of GRP78 expression suggesting an increase of endoplasmic reticulum (ER) stress (Figures 2d and e). Decrease of B-cell lymphoma-extra large expression was also found in these transgenic mice (Figures 2e and f). On the other hand, an increase of cyclin E and phosphorylated extracellular signal-regulated kinases 1 expression levels was found, suggesting activation of proliferation signal and increased cell proliferation (Figures 2e and f). IHC analysis for Ki-67 expression also confirmed the results (Figure 2g).
Search for microRNA-mediated differentially expressed genes between transgenic mice carrying wild type and mutant pre-S/S gene. Despite a significantly increased number of apoptotic cells in TgSW172*-H mice compared with that of TgWT-H mice (Figure 2a), both lines shared similar profiles of ER stress response, increased apoptosis and increased hepatocyte proliferation. It was difficult to explain the 23.1% versus 0% of difference in HCC development. To search for other differential molecular pathways related to hepatocarcinogenesis, we performed two microarray assays: microRNA array and complimentary DNA (cDNA) microarray. This experimental strategy was aimed to identify a microRNA-mediated regulatory pathway shared by both human and mice, and to minimize the number of false positive targets. After eliminating the microRNAs that did not have human equivalents, the distribution of fold-change and statistical significance of the remaining candidates were depicted in Figure 3a. Only four microRNAs (microRNA-211, 290, 138-1 and 873) had the highest fold-change and the difference was statistically significant (P o0.05). Subsequently, the data derived from these two arrays were matched to identify two sets of targets: down-regulated microRNAs with matched up-regulated target genes ( Figure 3b, left panel) and up-regulated microRNA with matched down-regulated target genes ( Figure 3b, right panel). Only microRNA-211 and micriRNA-873 (miR-873) satisfied these criteria. Real-time PCR (RT-qPCR) was carried out to verify the up-and downregulation of the identified supposedly dysregulated genes using an independent cohort of transgenic mice. The results showed that only upregulation of miR-873 and downregulation of its putative target genes, CSMD3 (CUB and Sushi multiple domains 3) and ATP8B2 (ATPase, aminophospholipid transporter, class I, type 8b, member 2), were consistently verified. Because CSMD3 had been implicated as a cancer-related gene in previous reports, we selected this gene for further study. On the other hand, we could not obtain suitable antibodies against ATP8B2 for clear western blot and IHC analysis (data not shown). Western blot analysis confirmed CSMD3 downregulation in TgSW172*-H and L mice, but not in TgWT-H mice ( Figure 3c). IHC analysis showed consistent results ( Figure 3d). Finally, inhibition of miR-873 expression was achieved using a lentivirus-based vector in various hepatoma cell lines. Before inhibition experiments, the baseline CSMD3 expression was undetectable or barely detected in J7, SK-Hep-1 and HepG2 cells. Following miR-873 inhibition, the CSMD3 expression remained undetectable in these three cell lines. However, marked upregulation (5-10 folds) of CSMD3 was found in Mahlavu and Hepa1-6 cells and mild increase (1.5 to 2 folds) was found in BNL and Hep-Y2 cells. No significant change was found in Huh7 cells (Figures 3e and f). The efficiency of anti-miR-873 expression was also assessed for comparison ( Figure 3g).
On the other hand, when miR-873 was overexpressed, suppression of CSMD3 expression was observed in Mahlavu, BNL and Hep-Y2 cells. Mild suppression (0.8-fold only) was observed for Hepa1-6 cells ( Figure 4a). CSMD3 was undetectable in J7 cells with or without miR-873 overexpression. To understand whether the increased ER stress in transgenic mice was caused by pre-S/S-sW172* alone or also by miR-873 upregulation, the GRP78 levels were also compared between hepatoma cells with or without miR-873 overexpression. The results showed significant increase of GRP78 expression in J7, Mahlavu, Hep-Y2 and Hepa1-6 cells (Figure 4a). (a) Genomic structure of the HBV pre-S/S transgenes. PC, precore region (yellow); (c) core region (orange); POL, polymerase (pink); X, X region (blue); PS1/PS2/S, pre-S1/pre-S2/surface region (light, medium, dark green); Horizontal solid bar, BglII to BglII HBV DNA fragments as the transgenes. Triangle, the point mutation leading to sW172* mutation. (b) Transient transfection using the plasmids carrying the transgenes to transfect Hepa1-6 mouse hepatoma cells before generation of transgenic mice. (c) Western blot analysis to detect wild type pre-S/S protein expression using monoclonal anti-HBs antibodies. Right panel, over-exposure to visualize middle surface protein. S, small surface proteins (gp24 and gp27); M, middle surface proteins (only gp33 was clearly visible, gp36 was barely seen (not marked)). (d) Western blot analysis to detect mutant pre-S/S protein using polyclonal anti-HBs antibodies (left panel) or anti-pre-S2 antibodies (right panel). St, truncated surface proteins (sW172* truncation, gp18 and gp22); NS, non-specific bands; Mt, truncated middle surface proteins (gp27 and gp30); Empty arrowhead, a non-specific band also found in null mice. Molecular weights of all protein are obtained through electrophoresis of a molecular weight marker in parallel. (e) IHC analysis for pre-S/S protein expression. Left, anti-pre-S2 antibody is used for detection; right, polyclonal anti-S antibody is used. NonTg, non-transgenic control mice.
Activation of miR-873 to target CSMD3 by pre-S/S-sW172* expression in hepatoma cell lines To further clarify the regulation cascade, pTg-sW172* was transfected into BNL, Hepa1-6, HepY2, Huh7, J7, and Mahlavu cells and miR-873 was assayed 3 days after transfection. Up-regulation of miR-873 was found in Hepa1-6 and HepY2 cells (Figure 4b). Independent transfection experiments were preformed for these two cell lines and the cells were harvested 5 and 7 days after transfection. Increasing levels of miR-873 upregulation were noted (Figure 4c).
To understand whether CSMD3 was an authentic target of miR-873, the 3ʹ-UTR of CSMD3 gene was inserted into the 3ʹ-UTR of a luciferase gene for miR-873 co-transfection experiments. It was found that in HepY2, J7, Hepa1-6 and BLN cells, CSMD3 3ʹ-UTR-associated luciferase activities were significantly downregulated ( Figure 4d). All downregulations were disrupted when the presumed recognition seed sites were mutated.
These results indicated that pre-S/S-sW172* could upregulate miR-873 expression in at least two hepatoma cell lines (Hepa1-6 and HepY2). Subsequently, miR-873 could target at CSMD3 by binding to the 3ʹ-UTR of its transcripts. This occurred in all hepatoma cell lines tested, except for Huh7 cells. These results were consistent with the actual changes of CSMD3 protein levels upon miR-873 inhibition ( Figure 3f) and overexpression (Figure 4a).

DISCUSSION
The rtA181T/sW172* mutant exerts cross-resistance to both lamivudine and adeforvir. 31,32 In lamivudine-resistant patients receiving adefovir rescue therapy, about 8-20% of patients will eventually develop cross-resistance. 32 In Taiwan and Hong Kong, this problem is gradually overcome since lamivudine is rarely used as the first-line antiviral drug at this time. Instead, nucleot(s)ide analogs with high resistance barrier, such as entecavir, are being widely used. The situation, however, remains critical in China, where low-cost, generic adefovir is generally used as the first-line rescue drug for lamivudine, resulting in emergence of crossresistance in a significant proportion of patients. The actual prevalence of rtA181T/sW172* mutant in China is unknown, 33 nor do we know whether it is capable of transmission among people.
S gene truncation can develop not only in lamivudine and adefovir-treated patients but also in peginterferon-treated patients, vaccine failure patients and antiviral treatment naïve patients. 19,20 Recent studies using HBcAg/HBsAg-postive HCC tissues uncovered another frequent S gene truncation mutation, sW182*. 34 It has been reported to be highly prevalent in Korean HCC patients and it has been found in HBV and HIV coinfected patients. 35,36 Tumorigenicity assessment in nude mice showed an even higher oncongenicity. 21 Taken together, our transgenic mice model can serve as a representative tool to study molecular pathways of hepatocarcinogenesis originated from the S gene truncation mutation.
From the results in our search for oncogenic mechanisms, we speculated increase of ER stress response accompanied by increased apoptotic hepatocytes was followed by increased compensatory liver cell proliferation and regeneration. This scenario has been observed in other HBV-related transgenic models developing HCC. [37][38][39] In the present study, however, we found that these events also occurred in TgWT-H mice, in which no HCC developed. Such discrepancy urged us to search for other differentially regulated pathways between TgWT and TgSW172*. To avoid misidentification of insignificant background genes and to focus on microRNA-mediated dysregulation, we used two-array assays for cross reference. After verification process, only miR-873mediated downregulation of CSMD3 expression was consistently confirmed in an independent cohort of TgSW172* mice. Inhibition of miR-873 expression resulted in upregulation of CSMD3 in several hepatoma cell lines, indicating that CSMD3 is indeed a target of miR-873. According to the Cancer Atlas of the Human Protein, CSMD3 are expressed in low to medium levels in the cytoplasm/membrane of 20 different cancer tissues. In 12 HCC tissues, CSMD3 is undetectable in five and weakly positive in four. 40,41 CSMD3 has been found to have a high mutation rate and a low expression level in lung cancer, ovarian cancer and several other cancers and thus possibly has a tumor suppressor function. [42][43][44][45][46] CSMD3 somatic mutations have been reported in HBV-associated HCC with portal vein tumor thrombosis and intrahepatic metastasis. 47 To date, the physiological function of The two concentrations were significantly different, P = 0.0001. b P = 0.0021, when the two mutant type TgSW172*-L and TgSW172*-H mice were combined and compared with the two wild type TgWT-L and TgWT-H mice combined. c P = 0.0008, when the TgSW172*-H mice were compared with the two wild type TgWT-L and TgWT-H mice combined.
HBV sW172* transgenic mice M-W Lai et al CSMD3 has not been extensively studied and its role in hepatocarcinogenesis has not been fully investigated. Further studies are needed to clarify its role in liver cancer. When examining the regulatory cascade from pre-S/S-sW172* to CSMD3 suppression, it was found that activation of miR-873 only occurred in Hepa1-6 and HepY2 cells, suggesting that other cellular factors were involved in this process. Interestingly, following miR-873 upregulation, GRP78 expression was enhanced in most of the hepatoma cell lines. As such, the increased ER stress in transgenic mice not only caused by pre-S/S-sW172* accumulation but also by miR-873 activation. Notably, in almost all cell lines (except for Huh7 and HepY2), miR-873 activation led to enhanced cell proliferation, consistent with the view that its target, CSMD3, was a tumor suppressor gene. At this time, the reason for the paradoxical growth regulation effect in HepY2 cells was unclear.
In summary, we provided transgenic mouse evidence showing an increased oncogenicity of truncated HBV pre-S/S protein with sW172* mutation. Besides increased ER stress response and apoptosis, resulting in enhanced hepatocyte proliferation and turnover, a miR-873-mediated downregulation of CSMD3 contributed partly to hepatocarcinogenesis.

Construction of transgenic mice
The BglII to BglII HBV-DNA fragment (nt. 2434-1990) was isolated from pCMV-HBV 48,49 by restriction enzyme digestion (BglII (nt. 2434 in HBV genome) to XbaI (multiple cloning site of pRc/CMV, Invitrogen)) and was used to replace the BglII to XbaI region of pRc/CMV. This replacement also deleted the region containing CMV promoter. As a result, the final product, pTg-WT, encoded the whole pre-S/S gene driven by its native promoters but not CMV promoter. To construct the plasmid encoding the rtA181T/sW172* mutation, site-directed mutagenesis was performed to generate the desired point mutation, G673A, as described previously. 31 Following the same procedure, the resulting plasmid, pTg-sW172*, was the same as pTg-WT except for the presence of target mutation (Figure 1a).
The transgenic mice were generated by pronucleus microinjection of C57BL/6 fertilized egg. 50 All mice were maintained in a specific pathogen free facility in Chang Gung Medical Center, Linkou, Taiwan. The tails of mice were cut at 3 weeks of age and genomic DNA was isolated by use of the proteinase K digestion followed by phenol/chloroform extraction method as described previously. 31 The success of gene transfer was determined by PCR using primers franking the target mutation: 5ʹ-TGTGTCTGCGGCGTTTTATC-3ʹ (nt. 381-400; sense) and 5ʹ-GTTTAA ATGTATACCCAGAGAC-3ʹ (nt. 840-819; antisense). The resulting PCR product was sent for nucleotide sequencing to confirm the genotype. RT-PCR was also performed using these primers to calculate the copy number of transgene in each line.  The sample size used to calculate the incidence of HCC for each line of transgenic mice was selected to ensure that development of HCC in 415% of TgSW172* mice could reach statistical significance. All mice were examined by liver biopsy for tissue HBsAg expression by IHC method. Transgenic mice that did not express HBsAg were excluded for calculation. No randomization process was involved in this study. No blinding process was performed in this study. Cell lines Hepa1-6, BNL, SK-Hep-1, Huh7, J7 and Mahlavu cells were maintained in Gibco Dulbecco's Modified Eagle Medium supplemented with 5% fetal bovine serum. HepG2 cells were maintained in Minimal Essential Medium supplemented with 5% fetal bovine serum. Hep-Y2 cells were maintained in RPMI medium 1640 with 5% fetal bovine serum. All culture medium was purchased from Life Technologies, Carlsbad, CA, USA. All cell lines were purchased from American Type Culture Collection except for Hep-Y2 cells, which were developed in our laboratory. 51 All cell lines were tested negative for mycoplasma contamination. cDNA microarray Three liver tissues from the noncancerous parts of negative control (Null; non-transgenic), TgWT-H, TgSW172*-H and TgSW172*-L mice were collected for cDNA microarray experiments. Total RNA was isolated from liver tissues using Trizol (Invitrogen Corporation, Carlsbad, CA, USA). The cDNA microarray system used in this study was GeneChip Gene 1.0 ST Array System for Mouse by Affymetrix which interrogated 28,853 well-annotated genes with 770,317 distinct probes. microRNA array The microRNA array system used in this study was GeneChip miRNA 2.0 Array by Affymetrix which contained 100 percent miRBase v15 coverage, 15,644 probe sets, 2334 snoRNAs and scaRNAs and 2202 probe sets unique to pre-miRNA hairpin sequences. Total RNA was isolated from liver tissues using Trizol (Invitrogen Corporation). All microarray data were submitted to the GEO web site of NCBI (https://submit.ncbi.nlm.gov/geo/submission).

Gene expression microarray data analysis
The fluorescent signals of the microarrays were scanned as digital images and then converted and summarized to intensity readings per probe set using Affymetrix expression console. The intensity readings were normalized using robust multiarray average algorithm. Gene expression levels were compared across groups using t-test, ANOVA, Mann-Whitney tests, fold-change comparison or regression-based methods. The genes with significant changes were visualized as heatmaps using Cluster 3.0 and TreeView. The annotation of the genes was analyzed using Affymetrix NetAffx, Metacore and/or NIH/DAVID.

RT-qPCR validation of gene expression
The highest or critical up-and down-regulated genes from microarray experiments were validated by quantitative reverse transcription PCR (RT-qPCR). Briefly, the first-strand cDNA was synthesized from 0.5 μg of total RNA using SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA), followed by PCR using the SYBR Green PCR Master Mix (Life Technologies Co., Carlsbad, CA). Primers used for RT-qPCR are listed in Table 2. For microRNA quantification, the NCode VILO miRNA cDNA Synthesis Kit and NCode EXPRESS SYBRR GreenER miRNA qRT PCR Kit (Life Technologies Co., Carlsbad, CA) were used according to the manufacturer's protocol.

Inhibition of microRNA-873 expression
To inhibit miR-873 expression in human and murine hepatoma cell lines, miR-873 inhibition plasmid was purchased. The plasmid was generated based on a lentivirus expression vector, TOOLSilent shRNA Vector (TOOLS Biotechnology, New Taipei City, Taiwan), with insertion of a short stretch DNA sequence complementary to a target area of the miR-873 stem-loop sequence. The targeted sequence of miR-873 was 5'-AGGAGACTCACAA GTTCCTG C-3'. This vector also contained a green fluorescent protein gene under control of cytomegalovirus promoter for transfection efficiency monitoring.

Statistical analysis
Except for the cDNA microarray and microRNA array, all other statistical comparison was performed using student's t-test with 45 samples in each group. All the data met the criteria of normal distribution. Variance in each group was estimated to ensure there was no significant difference.