Aflatoxin B1 and/or Hepatitis B Virus Induced Tumor Spectrum in a Genetically Engineered Hepatitis B Virus Expression and Trp53 Haploinsufficient Mouse Model System for Hepatocarcinogenesis

Abstract

The authors investigated the spectrum of tumors and Trp53 mutations in genetically engineered models using the FVB/N mouse that expressed the hepatitis B virus genome and/or carried a Trp53 null and wildtype allele and/or were exposed to aflatoxin B1. Liver tumor incidence was increased when all three risk factors were present. Without aflatoxin B1 exposure, neither Trp53 haploinsufficiency nor HBV expression affected liver tumor development. Liver tumor prevalence increased with aflatoxin B1 exposure (p < .001), as thirteen of fourteen mice with liver tumors were initiated with aflatoxin B1. Liver tumors were more frequent in males (12/190) than females (2/170). Seventy-three mice developed sarcomas. Trp53 haploinsufficiency was associated with increased sarcoma incidence in males and females (p < .001). In Trp53 haploinsufficient mice, the HBV transgene increased the risk of sarcoma in males and females (p < .001). Lymphoma was significantly increased in Trp53 haploinsufficient FVB/N mice. There was no loss of heterozygosity at the wildtype Trp53 locus in twenty-five sarcomas or four hepatocellular tumors examined. No mutations were identified in the mRNA (exons 2–11) of Trp53 in six liver neoplasms or twenty-four sarcomas. In this model system, HBV expression affected only hepatocellular neoplasia in association with both aflatoxin B1 initiation and p53 haploinsufficiency.

Hepatitis B virus (HBV)-related hepatic injury, when combined with aflatoxin B1 (AFB) exposure and p53 mutation, leads to a significant increase in the risk of hepatocellular carcinoma in humans. Mouse models of hepatocarcinogenesis have exhibited similar responses, although the role of HBV is likely to be related to continual necrosis and proliferation of hepatocytes secondary to abnormal viral protein retention in some models (Sell 2003). The mechanisms by which chronic HBV infection leads to liver tumor formation are not clear (Tabor 2007). Chronic hepatic inflammation induced by the host immune system’s response to viral antigens that leads to increased hepatocyte cell proliferation and possible oxidative injury have been proposed as a possible cause (Tornillo et al. 2000). Direct effects of hepadnaviral proteins, such as the X protein or surface proteins, as agents of neoplastic transformation have also been implicated (Block et al. 2003; Brechot 2004; Zheng et al. 2007). AFB exposure, gender, and Trp53 haploinsufficiency (p53+/−) have been previously evaluated in a transgenic mouse model that expressed HBV envelope proteins (Ghebranious and Sell 1998a). In that study, male gender, AFB exposure, and HBV transgene expression were risk factors for liver neoplasia. However, only a portion of the HBV genome, the entire viral envelope coding region, was included leading to a significant but not HBV-specific increase in hepatic necrosis, inflammation, oxidative injury, and hepatocyte turnover (Hagen et al. 1994). Hepatocyte injury and replication can predispose to liver tumor development, and so the relevant role of HBV in the pathogenesis of liver tumor development was unclear. Zheng et al. (2007) found that HBV transgenic mice were more susceptible to hepatic neoplasia if they were treated with a single initiating exposure of diethylnitrosamine (DEN) and that the effect was independent of HBV X gene expression. HBV expression alone was not sufficient in producing liver tumors but increased hepatocellular proliferation caused by DEN exposure combined with HBV expression increased HCC formation (Zheng et al. 2007). As a result, we investigated the effect of expression of the HBV genome on the rate of HCC development in a mouse model that was initiated with AFB and that did not have evidence of hepatic inflammation or increased hepatocyte proliferation associated with HBV expression. In addition, we wanted to evaluate the neoplastic phenotype of this genetically engineered FVB/N mouse model that expressed the HBV genome and had only one functional allele of Trp53 following initiation with AFB₁.

Materials and Methods

Animal Care

Mice were maintained and treated in accordance with the guidelines of the North Carolina State University Institutional Animal Care and Use Committee. Mice were housed in filter-top cages, fed commercial chow (Picolab Rodent Diet 5058, Granville Milling, Creedmoor, NC), and offered water ad libitum. The light cycle was twelve hours of light and dark. Female mice were group housed, up to four mice in a cage, and males were housed individually or in groups of two. Sentinel mice were tested to ensure that there was no evidence of infectious pathogens. For retro-orbital blood collections, mice were anesthetized by isoflurane inhalation. Mice were euthanized by CO₂ inhalation.

Study Mice

Normal FVB/NCrIBR mice were obtained commercially (Charles River Laboratories, Raleigh, NC). A lineage of FVB/N-TgN (HBV) mice (HBVTg) that expressed the entire HBV genome was derived from FVB/N mice and developed at Stanford University (Marion 2003). Briefly, this lineage was developed by inserting 4.1 kb of HBV genome, including the entire genome plus a redundancy for the sequences between 1,067 and 1,996 of the subtype of HBV designated ayw, into a plasmid construct pTHBVG2. These mice express HBV particles and HBV surface antigen (HBsAg) in serum and urine. HBV DNA levels peak at 10⁶ to 10⁸ viral genome equivalents/ml. For these studies, groups were generated by crossing homozygous female FVB/N-Tg(HBV), designated STC, mice with normal FVB/N male mice or with homozygous null FVB/N-Trp53^tm1Brd male mice.

Male FVB/N-Trp53^tm1Brd (−/−) mice, designated p53−/−, were obtained as a generous gift from Drs. Aya and Philip Leder at Harvard University to Ray Tennant at the National Institute of Environmental Health Sciences. These FVB/N p53(−/−) mice were the product of more than ten backcrosses of wildtype Trp53 FVB/N female starting with a homozygous null B6.129-Trp53^tm1Brd male. The homologous recombination targeting construct used to create these mice had a neoinsert between exon 4 and exon 5 of the Trp53 allele that completely disrupted p53 mRNA and protein expression in the 129 embryonic stem cell of origin. For this study, male FVB/N-Trp53^tm1Brd (−/−) mice were intercrossed to either wildtype FVB/N or homozygous female FVB/N-Tg (HBV) mice to generate mice haploinsufficient for the wildtype Trp53 allele [p53 (+/−)] in addition to other risk factors.

HBV Testing

Offspring were tested for the presence of HBsAg in serum using the Auszyme Assay (Abbott Laboratories, Abbott Park, IL) before being included in the study.

AFB Exposure

AFB₁ (Aldrich, Milwaukee, WI) was dissolved in trioctanoin (Sigma-Aldrich, St. Louis, MO) at a final concentration of 1.0 mg/ml. Final concentrations of AFB in were confirmed by HPLC analysis conducted by the North Carolina Agricultural Research Center, Raleigh, NC. Seven-day-old mice were given a single intraperitoneal injection with a total dose of 10 mg/kg.

Study Groups

The design of this experiment is a 2 × 2 × 2 full factorial in which the three factors are AFB treated (Yes, No), HBVTg hemizygous (Yes, No), and p53 haploinsufficient (Yes, No) for both males and females. Except for the p53 haploinsufficient and AFB-treated group (group 6 below), sample sizes were nearly equal across the combinations. Eight groups of FVB/NCrIBR mice were established: (1) untreated FVB/N (n = 19 males, 21 females); (2) AFB, AFB-treated (n = 21 males, 20 females); (3) HBVTg, hemizygous HBVTg (n = 32 males, 23 females); (4) HBVTg/AFB, hemizygous HBVTg and AFB-treated (n = 20 males, 19 females); (5) p53(+/−), p53 haploinsufficient (n = 30 males, 19 females); (6) p53(+/−)/AFB, p53 haploinsufficient and AFB-treated (n = 15 males, 17 females); (7) HBVTg/p53(+/−), hemizygous HBVTg and p53 haploinsufficient (n = 29 males, 22 females); and (8) HBVTg/p53(+/−)/AFB, hemizygous HBVTg and p53 haploin-sufficient and AFB-treated (n = 24 males, 29 females).

Tissue Collection

Surviving animals were euthanized (C0₂ narcosis) at twelve to thirteen months of age. Tissues were collected from moribund mice and mice found dead if tissue was adequate for diagnostic purposes. All animals had thorough postmortem examination and tissues, including liver, kidney, lung, and any grossly observed lesions, were collected for histologic examination. Liver and all tumor samples were placed in cryogenic vials and snap frozen in liquid nitrogen for nucleic acid extraction or fixed in 10% zinc buffered formalin for histology and immunohistochemistry.

Histology

Fixed tissues were embedded in paraffin, sectioned at 6 microns, stained with hematoxylin and eosin, and reviewed. Hepatic neoplasms and sarcomas were diagnosed using published criteria for mouse neoplasia (Maronpot 1999; Mohr 2001).

Immunohistochemistry

Normal and neoplastic liver as well as sarcomas were stained for HBV core antigen (HBcAg), p53, vimentin, cytokeratin, and Ki67. Detailed methods of staining methods for all immunohistochemistry (IHC) are available at http://dir.niehs.hin.gov/dirlep/immuno/protocols.htm.

Hepatocyte Proliferation Index

Normal liver that was collected at the conclusion of the study from 9 HBVTg mice and 8 FVB mice and stained for Ki67 was examined using Image Pro-Plus software Version 5.0 (Media Cybernetics. Inc., Bethesda, MD) using methods described by Maronpot et al. (2000). A t-test including the Satterthwaite correction for unequal variance was used and performed using SAS software version 9.2 (Cary, NC) to compare hepatocyte proliferation in the two groups.

Loss of Heterozygosity (LOH) Determination

DNA was extracted from flash-frozen tumor and normal liver tissue for twenty-five of the sarcomas and six of the hepa-tocellular neoplasms using Qiagen’s DNeasy Tissue Kit (Valencia, CA). DNA was also isolated from liver of three p53 +/+ mice, as positive controls. To selectively amplify the wildtype p53 allele, priming was targeted to the region of exon 5 deleted from the null allele by primers described previously (Hulla, French, and Dunnick 2001). Primers (Integrated DNA Technologies, Inc., Coralville, IA) amplified a 260bp fragment of the wildtype allele, an 180bp pseudogene, and a 386bp fragment of the null p53 allele.

PCR was performed in duplicate for normal and neoplastic liver and sarcomas using the Perkin Elmer GeneAmp 9600 Thermocycler (Waltham, MA) using previously published methods (Hulla, French, and Dunnick 2001). Each sample was run on an agarose gel on three separate occasions, and an average LOH was calculated for each sample.

LOH of p53 was determined through a fluorescence measurement calculated by Kodak Electrophoresis Documentation and Analysis System (Rochester, NY). The fluorescence of the wildtype and pseudogene amplicons were compared to each other and tumor and nontumor values from the same mouse were compared. Percentage LOH was calculated for each tumor as [1 − (Trp53/(Trp53+ pseudogene))_tumor/(Trp53/(Trp53+pseudogene))_nontumor] × 100. The wildtype/pseudogene ratio was also measured for each tumor and corresponding normal tissue, as was done previously (Hulla, French, and Dunnick 2001).

P53 mRNA Sequencing

RNA was extracted using Qiagen RNeasy Kit (Valencia, CA) and reverse transcribed using previously described methods (Hulla, French, and Dunnick 2001). The following primers were used in the PCR reactions:

• 579: 5’-CTC CGT CAT GTG CTG TGA CTT C-3’

• 1045: 5’-cag gaa aca gct atg acc TCT CCA TCA AGT GGT TTT TT-3’

• 1135: 5’-tgt aaa acg acg gcc agt CTA GCA TTC AGG CCC TCA TC-3’

• 1137: 5’-CCT GGC TGT AGG TAG CGA CTA C-3’

• 1150: 5’-tgt aaa acg acg gcc agt CGT GGT GGT ACC TTA TGA GCC A-3’

• 1156: 5’-TTC GCC ACA GCG TGG TGG TAC C-3’

• 1157: 5’-AGA AGG GAC CGG GAG GAT TGT G-3’

• 1158: 5’-cag gaa aca gct atg acc ACG GGA TGC AGA GGC AGT CA-3’

• 1312: 5’-cag gaa aca gct atg acc CAT GTG CTG TGA CTT CTT GTA GA-3’

• 1314: 5’-tgt aaa acg acg gcc agt GAG CGC AAA GAG AGC GCT GCC-3’

• 1334: 5’-CCA AGT CTG TTA TGT GCA CGT AC-3’

• 1335: 5’-tgt aaa acg acg gcc agt CTG TTA TGT GCA CGT ACT CTC CT-3’

• 1336: 5’-TTC TTC TGT ACG GCG GTC TCT C-3’

• 1337: 5’-cag gaa aca gct atg acc TGT ACG GCG GTC TCT CCC AGG-3’

• tgt aaa acg acg gcc agt = M13-fw

• cag gaa aca gct atg acc = M13-rev

The entire coding region of the p53 RNA transcript was amplified with nested PCR using the above primers. Templates for the first set were generated for exons 2–5, 5–6, 7–11 using primer sets 1137–579, 1334–1336, and 1156–1157, respectively. PCR products of 600, 500, and 639bp, respectively, were gel purified using a kit (QIAquick, Qiagen, Valencia, CA). Secondary nested PCR templates were generated for exons 2–5, 5–6, 7–8, and 9–11 using primers 1135–1312, 1335–1337, 1150–1045, and 1314–1158, respectively. PCR products of 500bp for exons 2–5 and 5–6 and PCR product of 300bp for exons 7–8 and 9–11 were gel purified.

Sequencing was performed on p53 for exons 2–5, 5–6, 7–8, and 9–11 using the Big Dye Terminator Sequence kit (Applied Biosystems Inc., Foster City, CA) and the 3700 DNA sequencer (Applied Biosystems Inc.). Exons for p53 were sequenced in the forward and reverse directions using secondary template and corresponding primer sets. Each forward and reverse reaction was analyzed to create a consensus sequence using Vector NTI’s Contig (Invitrogen, Carlsbad, CA). The sample identification number was inserted for each sample, and the exon sequences of the samples were aligned as follows using Vector NTI AlignX: alignment of all normal liver samples (4 total), alignment of all liver tumor samples (6 total), and alignment of all sarcoma samples (24 total). GenBank sequence and NT consensus sequences were included in the sarcoma and liver tumor alignments. Consensus sequences were taken from sarcomas, normal liver and HCCs, based on the appropriate alignment. The exon consensus sequences were then combined to create a single sequence from exon 2 through 11. The primer sequences were then removed for nonhomologous tail regions (tgt aaa acg acg gcc agt, cag gaa aca gct atg acc). Exon 2–11 sequences for the three tissue types were then aligned with Genbank sequence GI\6755880\REF\NM_011640.1.

Statistical Methods

Kaplan-Meier survival curve estimation was used to estimate the probability of survival for each of the eight groups; survival differences between groups were tested using the log-rank statistic (Kaplan and Meier 1958; Hosmer and Lemeshow 1999). Because survival differed among the groups (Table 1) and because the mice that died early were not at risk of tumor development for as long as those surviving for the entire study, two approaches to survival adjustment of tumor rates were used. The first approach was the Poly-3 test, which gives less weight to tumor-free animals that die early (Bailer and Portier 1988; Portier and Bailer 1989). The second approach was logistic regression with survival time as a covariate (Gart 1979). Both approaches were used to compare survival-adjusted tumor rates of each group to the normal FVB control group, as well as between combinations of groups such as AFB-treated versus untreated. The two survival-adjustment approaches yielded nearly identical results, so only the Poly-3 analyses are presented.

Results

Hepatocellular Neoplasms

Fourteen of 360 mice developed hepatocellular neoplasms (Table 2A). Male mice had a higher rate of hepatocellular neo-plasia than did females. All of the hepatic neoplasms arose in mice treated with AFB with the single exception of a hepatic adenoma in a male mouse from the p53 +/− group. Male mice, but not female mice, treated with AFB had significantly more liver tumors than those not treated (p < .001; Table 2B). Male mice with all three risk factors were more likely to have hepatocellular tumors than controls. Significantly more HBVTg/p53(+/−)/AFB males developed hepatocellular tumors than FVB/N males (p < .05), and while not statistically significant, the two females that developed hepatocellular tumors were HBVTg/p53(+/−)/AFB females (Table 2A). However, neither p53 status nor expression of the HBV transgene affected the incidence of hepatocellular neoplasia in either gender in any of the groups of mice with two risk factors. There were four mice, all males, with HCC (3 in HBVTg/p53/AFB and 1 in HBVTg/AFB) and ten mice with hepatic adenomas. Multiple tumors were found in five male mice (3 HBVTg/AFB, 1 p53/AFB, and 1 HBVTg/p53/AFB).

Histologically, adenomas formed two- to four-cell-thick trabeculae and circumferentially compressed adjacent hepatic parenchyma. HCC were composed of pleomorphic hepatocytes that formed irregular trabeculae up to six cells thick, often containing irregular invasive borders that projected into adjacent hepatic parenchyma. In scattered sites, acinar formations of neoplastic hepatocytes around an open lumen were evident in HCC (Figures 1A and 1B).

Nonneoplastic liver from all of the mice was unremarkable. There were no differences in the mean proliferation indices for nonneoplastic liver from transgenic mice (x = 0.34% +/− 0.27) or normal FVB/N mice (x = 0.33% +/− 0.20) killed at the conclusion of the study.

Immunohistochemistry of hepatocellular neoplasms revealed an increased proportion of nuclear HBcAg staining in five (3 adenomas and 2 carcinomas) of the six tumors examined compared to surrounding nonneoplastic hepatocytes. One hepatocellular adenoma had reduced numbers of stained nuclei, suggesting deregulated expression of the transgene in the neoplasms. HBcAg staining was evident in the centrilobular hepatocytes in all of the fifteen sections of normal liver examined. There was no evidence of p53 staining in any of the fourteen adenomas and two HCC examined.

Sarcomas

Seventy-three mice developed sarcomas, most of which were soft-tissue sarcomas found mainly in the subcutaneous regions, although abdominal and bone neoplasms were also detected.

Sarcomas occurred only in p53(+/−) mice (p < .0001; Table 3A). Interestingly, in both male and female p53 (+/−) mice, the presence of the HBVTg transgene was also associated with a significantly increased risk of sarcomas (p < .0001; Table 3B). AFB exposure did not affect the incidence of sarcoma in either gender of mice.

Sarcomas were poorly differentiated with regions of differentiation toward fat, endothelium, bone, or smooth muscle. Findings suggest origin from a primitive mesenchymal cell with multipotent differentiation capacity. The majority of neoplasms had a component that included spindle-shaped cells, but there was a great deal of individual cell variation in size and morphology (Figure 1C). The sarcomas were often characterized by an aggressive growth pattern with frequent invasion into adjacent soft tissue and markedly variable cellular morphology. Mitotic figures were common and often abnormal, and there were scattered foci of hemorrhage and necrosis. Many had features of histiocytic sarcoma (Figure 1D). Tumors were divided into categories based on the predominant histologic characteristics (Table 4).

All twenty-five sampled sarcomas had areas that were vimentin positive, consistent with mesenchymal differentiation, but virtually none of the tumors stained uniformly, with up to 30% of the surface area of some sarcomas failing to stain. Within some of the sarcomas, distinct cytokeratin positive acinar structures could be discerned, interpreted as mammary glands or other adnexal glands entrapped by invading tumor. However, in four cases, there were areas of cytokeratin positive cells that formed irregular skeins of oval to spindle-shaped cells, and the morphology of both the epithelial and the mesenchymal component warranted a diagnosis of malignant. These neoplasms were diagnosed as malignant mixed tumors.

There was only one sarcoma with detectable p53 of sixteen sampled sarcomas. It was a metastatic mass of an undifferentiated sarcoma in the liver of a mouse from the HBVTg/p53/AFB group. IHC of ninteen sarcomas revealed a lack of detectable HBcAg staining.

Lymphomas

The incidence of lymphomas in p53(+/−) mice was significantly higher than in p53 (+/+) mice for females (p = .006) and marginally higher for males (p = .051; Tables 5A and 5B). Lymphomas were found in four male and six female mice, all of which were p53(+/−). Among female p53(+/−) mice, the rate of lymphomas was unrelated to the presence or absence of HBVTg or of AFB. HBV expression in lymphoid tissue was not detected by IHC.

No tumors were detected in the kidneys or the pancreas from any of the HBVTg mice, although HBcAg is expressed in these tissues (Marion et al. 2003).

LOH Determination

All of the tumor and normal livers from p53(+/−) mice yielded a 260bp amplicon from the wildtype p53 allele and an 180bp p53 pseudogene, as expected, as well as a 386bp fragment of the null allele in p53(+/−) mice (Figures 2A and 2B). The 3 hepatocellular neoplasms sampled from mice with normal p53 alleles did not generate a product from the null allele, as expected.

The estimates of percentage LOH of p53 for the four hepatocellular neoplasms sampled from p53(+/−) mice were 0%, 0.45%, 0.73%, and 60% (Figure 3). Results were interpreted to indicate that there was no LOH in three of the four HCC, and the hepatocellular neoplasm with 60% LOH of p53 was interpreted to have an equivocal LOH that may have been confounded by the presence of vascular cells with a wildtype p53 allele. The average LOH of p53 allele for the twenty-five sarcomas examined was 11.2% and ranged from 0% to 22.5% (Figure 4). This was interpreted as indicative of retention of heterozygosity based on previous studies (Hulla, French, and Dunnick 2001). Similar observations have been made of p53 LOH using identical methods in carcinogen-treated p53(+/−) mice that developed lymphomas, sarcomas, and urinary bladder tumors (French et al. 2001; Hulla, French, and Dunnick 2001; French 2004).

P53 cDNA (mRNA) Sequencing

No mutations were evident in any of the twenty-four sarcomas or six sampled hepatocellular neoplasms in exons 2–11 of the coding sequence of p53.

Discussion

In this study, development of hepatocellular neoplasia was associated with AFB exposure and male gender. The effect of the HBV transgene or p53 haploinsufficiency on liver tumor development was only evident when all three factors were present in combination. The overall incidence of liver tumors was relatively low during the thirteen-month, in-life observation period with only fourteen hepatic tumors observed in 360 mice. These results differ from those of Ghebranious and Sell (1998a), which demonstrated a high incidence of hepatocellular tumor development in association with HBsAg expression, male gender, and aflatoxin exposure in C57BL/6 mice. In the previous study, the HBV transgenic mice expressed an excess of the large envelope polypeptide, and because these viral proteins cannot be efficiently secreted from the hepatocytes, the polypeptides are retained and lead to hepatocellular necrosis and associated chronic inflammation and regeneration (Chisari and Ferrari 1995). Thus, the most significant impact of HBsAg expression on hepatocellular tumor development is likely tied to the injury, including oxidative injury and replication, rather than the effects of HBV proteins (Zheng et al. 2007). The incidence of hepatic tumors may have been influenced by genetic differences, as the FVB/N mice used in this study have a lower rate of spontaneous hepatocellular neoplasms (Mahler et al. 1998) than C57BL/6 mice (Ghebranious and Sell 1998a). An increased risk of hepatocellular neoplasia in aflatoxin-treated male mice compared to female mice has been previously demonstrated (Ghebranious and Sell 1998a).

Although specific mutations in p53 are associated with an increased risk for HCC in humans, liver tumors in rats or other rodent models of AFB and hepadnavirus-related hepatic carcinogenesis do not appear to require mutations in p53 (Rivkina et al. 1994; Rivkina et al. 1996; McGlynn et al. 2003). P53(+/−) mice with a nonfunctional null and a functioning wildtype allele that are p53 haploinsufficient are not at a greater risk for liver tumor development following AFB exposure (McGlynn et al. 2003), and p53 mutations are not detected in liver tumors of AFB-treated mice (Tong et al. 2006), although transgenic mice that express the equivalent of the human p53 mutation in codon 249 have increased susceptibility to AFB-induced hepatic carcinogenesis (Ghebranious and Sell 1998b).

A recent study by Zheng et al. (2007), which used a design similar to this study, found that HBV transgenic mice were more susceptible to hepatic neoplasia when C57BL6 (B6) or CD1/B6 mice were treated with a single initiating exposure to diethylnitrosamine (DEN) and that the effect was independent of HBV X gene expression. Similar to our findings, HBV expression alone was not sufficient in producing liver tumors. Zheng et al. found evidence of increased hepatocellular proliferation in response to increased apoptosis caused by DEN exposure and interpreted HBV as promoting liver tumor formation by sensitizing hepatocytes to apoptosis following DEN initiation. Hepatic inflammation was not evident in response to HBV expression, as would be expected for mice exposed to viral antigens in utero, and there was no increase in hepatocyte proliferation in HBV transgenic mice compared to control mice at the end of our study. We did not evaluate the acute effects of AFB injection on hepatocyte proliferation and injury, however, and cannot compare the effect of AFB with that of DEN in the immediate posttreatment period. Explanations for the different responses observed in this study include the diverse metabolic and mutational spectra involved in AFB and DEN carcinogenesis and disparity in the response of different mouse genetic backgrounds to the carcinogen exposure. Also, the number of integrations and the sites of integration of HBV into the recipient mouse genome may influence tumor formation by disrupting oncogene or tumor suppressor gene functions or increasing susceptibility to aneuploidy. Alternatively, although our study had more animals than the Zheng et al. study, we may not have had sufficient power to detect these responses in a strain less susceptible to hepatic neoplasia.

The loss of both alleles of p53 has a clear effect on tumor risk in mice, particularly for lymphoma and other sarcomas (Donehower et al. 1992). The loss of a single allele can also increase the risk for tumor development (Donehower et al. 1992; Attardi and Donehower 2005), although this may vary depending on the strain of mouse, which stimulated us to assess the responses of the FVB/N strain (Harvey et al. 1993a). In addition, p53(+/−) mice have a shorter latency for tumor development for some chemically induced tumors (Eastin et al. 1998). This may explain the differences in the tumor spectrum in p53 haploinsufficient mice. However, in this study, the loss of one functional allele of p53 did not increase liver tumor risk, as has been shown previously in a different mouse strain (McGlynn et al. 2003). Mesenchymal neoplasia is, in general, more common than epithelial neoplasia in p53 haploinsufficient mice, and sarcomas were significantly increased in the FVB/N mice in our study by thirteen months of age. The most common sarcomas in this study were primarily poorly differentiated soft-tissue sarcomas that were classified as histiocytic sarcomas, hemangiosarcomas, fibrosarcomas, and rhabdomyosarcomas and occasional osteosarcomas, as seen in other strains of mice that were p53(+/−) (Harvey et al. 1993b; Jacks et al. 1994; Attardi and Donehower 2005). Surprisingly, p53(+/−) FVB/N mice from the same source as our mice were shown to have a low background of spontaneous tumor incidence up to twelve months of age (Mahler et al. 1998).

The 4 hepatocellular neoplasms examined in this study did not possess a mutation of the wildtype p53 allele or a LOH in three of four tumors examined. There was only equivocal LOH in one HCC. Thus, given the limited data available, LOH does not appear to be a frequent development in the HCC from p53 deficient FVB/N mice, suggesting that if p53 loss was involved, it is a very low penetrance event in the FVB/N strain, as also seen in other strains (Ghebranious and Sell 1998a).

There was no evidence of LOH in any of the sarcomas. This conclusion was drawn by making comparisons with a previous study using identical methods. In that study, the percentage of p53 loss ranged from 73% to 100% for lymphomas, 41% to 79% for sarcomas, and 10% to 63% in bladder tumors (Hulla, French, and Dunnick 2001). The average percentage LOH of p53 for the twenty-five sarcomas in the current study was 11.2% and ranged from 0% to 22.5%. This was interpreted as indicative of retention of heterozygosity as supported by the rtPCR expression of p53 mRNA for sequencing. In chimeric or incompletely inbred C57BL/6 x 129/Sv p53(+/−) mice, LOH was found in about half of the sarcomas initially, but about 85% of sarcomas from mice over eighteen months of age retained one intact p53 allele (Venkatachalam et al. 1998). Our results are discordant with earlier studies in which LOH occurred in about one-quarter to one-third of p53(+/−) tumors (Donehower et al. 1992; Jacks et al. 1994; Purdie et al. 1994). Overall, loss of one functioning allele of p53 appears to be sufficient for cancer predisposition, given that many lymphomas and sarcomas in p53(+/−) mice retain an intact functional allele (Ghebranious and Sell 1998a; Venkatachalam et al. 1998; Hulla, French, and Dunnick 2001; Attardi and Donehower 2005). In this study, sarcomas occurred only in p53 haploinsufficient mice, indicating that p53 haploinsufficiency was critical to sarcoma development. However, the absence of LOH at the Trp53 wildtype allele or evidence of mutation (in the Trp53 gene coding region determined by sequencing analysis of cDNA for the isolated total RNA) in the sarcomas suggests that although loss of the first p53 allele likely initiated tumorigenesis, an acquired mutation in another tumor suppressor gene or activation of an oncogene may have promoted sarcoma development. Therefore, p53 mutation or LOH may not be required in some mouse strains that because of different susceptibility alleles that have been conserved during strain development, are susceptible to tumorigenesis by other pathways.

In this study with FVB/N mice, the risk for sarcomas and lymphoma development was increased in mice that were transgenic for HBV as well as Trp53 haploinsufficient. The explanation for this observation is not clear. It is possible that HBx protein or other viral proteins could deregulate cellular homeostatic controls in mesenchymal cells based on strain-tissue-specific genetic susceptibility, but there was no evidence, based on immunohistochemical detection of HBcAg expression in the sarcomas or direct proof, that HBx protein is expressed in this model. HBV transgene integration, depending on the site(s) of integration, such as within or near oncogenes, altering their expression or predisposing cells to develop aneuploidy, could disrupt normal cellular proliferative controls.

In conclusion, in this model, and under the conditions of this study, expression of a complete HBV transgene has little influence on liver tumor development in mice with intact p53 alleles or one mutant copy following exposure to AFB. AFB exposure is clearly a risk factor for liver tumor development in male FVB/N mice. Trp53 heterozygosity for a null and wildtype allele was also a risk factor for a variety of soft-tissue sarcomas, and interestingly, the presence of the HBV transgene was associated with increased sarcoma formation among male and female p53(+/−) mice, although the role of the virus in this process remains unclear.