Overview of the Molecular Biology of Hepatocellular Neoplasms and Hepatoblastomas of the Mouse Liver

Abstract

The molecular pathogenesis of chemically induced hepatocellular neoplasms and hepatoblastomas in the B6C3F1 mouse is unclear but may involve alterations in the β-catenin/Wnt signaling pathway as was recently described for human liver neoplasms. The objectives of this research were to characterize the mutation frequency and spectrum of β-catenin mutations and the intracellular localization of β-catenin protein accumulation in chemically induced hepatoblastomas and hepatocellular neoplasms. In the majority of the hepatoblastomas examined by immunohistochemical methods, both nuclear and cytoplasmic localization of β-catenin protein were detected, whereas in hepatocellular adenomas and carcinomas and normal liver only membrane staining was observed. Genomic DNA was isolated from paraffin sections of each liver tumor. β-catenin exon 2 (corresponds to exon 3 in humans) genetic alterations were identified in the majority of hepatoblastomas from exposed mice. Deletion mutations were identified more frequently than point mutations in hepatoblastomas. Hepatocellular adenomas and carcinomas from treated mice had mutations in exon 2 of the β-catenin gene which ranged from 32–43%, while 10% β-catenin mutations were detected in spontaneous neoplasms. By immunohistochemical methods cyclin D1 was observed in most nuclei of hepatoblastomas and strong expression of cyclin D1 was confirmed by Western analysis regardless of treatment. The cumulative data suggests that β-catenin mutations with upregulation of the β-catenin protein and Wnt signaling most likely increased cyclin D1 expression. Cyclin D1 may provide an advantage during tumor progression of hepatocellular neoplasms and hepatoblastomas. The review will also focus on other genes which are important in mouse and human liver tumors.

Hepatocellular Neoplasms

Human hepatocellular carcinoma (HCC) accounts for more than 90% of all primary liver cancer. High rates of liver cancer incidence are reported in large portions of Asia and Africa (Mathers et al., 2002) and affects men more frequently than women (Parkin et al., 1999). In mice, hepatocellular adenoma is the most common spontaneous liver neoplasm, followed by HCC. Similar to the human tumors, hepatocellular adenomas and HCC in the mouse occur more frequently in males than females (Harada et al., 1999).

The etiology of human HCC is multifactorial and include exposure to naturally occurring carcinogens, viral infection of hepatitis B virus (HBV) and hepatitis C virus (HCV), genetic disease, industrial chemicals, pharmacologic agents, various pollutants, and lifestyle factors (Coleman, 2003). Transgenic mice expressing HBV X protein are more susceptible to carcinogen induced hepatocarcinogenesis (Zhu et al., 2004). Chronic hepatitis and cirrhosis also represent significant risk factors for liver cancer. Regenerative hepatocytes in the inflamed liver give rise to hyperplastic hepatocyte nodules, and these further progress to dysplastic nodules, which may progress to hepatocellular adenomas or carcinomas (Coleman, 2003). Different genes have been implicated in hepatocarcinogenesis including genes regulating DNA damage responses, genes involved in cell cycle control, genes involved in growth inhibition and apoptosis, and genes responsible for cell-cell interaction and signal transduction (Coleman, 2003).

Proliferative preneoplastic hepatocellular lesions have increased expression of growth factors and their receptors such as transforming growth factor α (TGF-α), insulin-like growth factor II (IGF II), hepatocyte growth factor (HGF), and c-met (Coleman, 2003). The cognate receptors and growth factors may be produced in an autocrine manner or through paracrine signaling pathways. Mouse models including TGF-α transgenic mice, c-myc/TGF-α double transgenic mice and c-myc-HGF double transgenic mice demonstrated an increased incidence of liver tumors (Fausto, 1999). One-third of the TGFα transgenic mice had elevated c-myc levels; furthermore, 75% of these tumors had elevated levels of IGF II mRNA, suggesting that TGFα and IGF II genes contribute to the formation of liver tumors. In mice, a genotoxic hepatocarcinogen, diethyl-nitrosamine (DEN), induced re-expression of IGF II during the early stages of liver carcinogenesis (Lahm et al., 2002).

The major molecular features of human HCC include aneuploidy and chromosomal aberrations, activation of protooncogenes, and inactivation of tumor suppressor genes. About 40% of HCC exhibit an aneuploid DNA content and numerical chromosomal abnormalities (Mise et al., 1998). The chromosomal regions most frequently involved in structural rearrangements are on chromosomes 1, 7, and 8 (Parada et al., 1998). DNA hypomethylation on pericentromeric satellite regions has been shown to significantly correlate with the loss of functional DNA methyltransferase activity and is also associated with chromosomal instability in precancerous conditions as well as HCC (Saito et al., 2002). Some HCC exhibit microsatellite instability associated with a loss of heterozygosity in mismatch repair genes (including hMSH2 and hMLH1) (Macdonald et al., 1998).

Increased levels of expression of c-H-ras, c-N-ras, and c-K-ras have been reported in HCC, as well as some pre-neoplastic lesions in humans (Coleman, 2003). The overexpression of the p21 ras protein in nearly all HCC examined suggests that the ras-MAPK pathway is hyperactive in these neoplasms (Huynh, 2004). K-ras mutations were found in 5 of 12 (42%) HCC in workers exposed to vinyl chloride (Weihrauch et al., 2001). H-ras mutations, which are often found in spontaneous mouse liver tumors, were not observed in the diethanolamine (DEA)-induced tumors suggesting that these tumors may develop by pathways which are independent of ras signal transduction (Hayashi et al., 2003). Then, 38 of 50 methylene chloride-induced liver tumors had mutations in H-ras codon 61 (Devereux et al., 1993), indicating that the H-ras signaling pathway plays a major role in development of liver tumors induced by this chemical.

Mutations of the p53 gene occur in conjunction with loss of heterozygosity of the p53 locus in human HCC, with most mutations occurring within the highly conserved DNA binding region of the gene including exons 5–9 (Hainaut et al., 1998). Point mutations in codon 249 of the p53 gene accounted for approximately 30% of p53 mutations in human HCC (Coleman, 2003). G to T transversions at codon 249 (AGG to AGT, arginine to serine) of the p53 gene have been attributed to aflatoxin B₁ exposure (Puisieux et al., 1991). Preneoplastic liver nodules in rats treated with aflatoxin B₁ showed no alterations in p53 expression or mutations in exons 5–8 of the p53 gene, suggesting that p53 mutations are not an early event in AFB₁-induced hepatocarcinogenesis (Liu et al., 1996). A higher percentage of p53 genetic alterations have been detected in advanced stages of HCC (Hsu et al., 1993), suggesting that p53 abnormalities may be associated with tumor progression in humans.

β-catenin mutations in human HCC have been frequently reported (Nhieu et al., 1999; Yamamoto et al., 2003). β-catenin mutations were detected more frequently in HCV-related HCC (45.8%) than in HBV-related HCC (22.2%) (Yamamoto et al., 2003). Some studies suggest there may be a chemical specific involvement of β-catenin activation in mouse liver tumors and that these alterations may be early events in mouse hepatocellular carcinogenesis (Devereux et al., 1999). Eleven of 34 (32%) hepatocellular adenomas and carcinomas from diethanolamine (DEA) treated B6C3F1 mice had mutations in exon 2 of the β-catenin gene and all tumors exhibited mild to moderate membrane staining of β-catenin protein (Hayashi et al., 2003). Similarly, 80% of neoplasms from mice treated with DEN (initiator) and phenobarbital had Catnb (β-catenin) mutations, whereas H-ras mutations were not found (Aydinlik et al., 2001). In contrast, mouse liver tumors induced by diethylnitrosamine (initiator alone), exhibited no Catnb mutations, while 30% of these tumors contained H-ras mutations (Aydinlik et al., 2001). Alterations of the Catnb gene and protein have also been associated with increased expression of cyclin D1 suggesting that Catnb mutation along with β-catenin protein accumulation and the resulting activation of the Wnt signaling pathway lead to an increase in cyclin D1 expression (Anna et al., 2003). Cyclin D1 protein is overexpressed in approximately 40% of human HCC and 34% of mouse hepatocellular neoplasms (Ito et al., 1998). In some cases overexpression is thought to be related to amplification of the cyclin D1 gene and overexpression of mRNA (Ito et al., 1998).

Deletion of Rb1 (loss of heterozygosity) is a frequent event in HCC and is typically accompanied by loss of expression of pRb (Zhang et al., 1994). It has been suggested that deletion of Rb1 is an early event in hepatocarcinogenesis, occurring in preneoplastic cirrhotic livers (Ashida et al., 1997). However, other studies suggest that deletion of Rb1 may be a late genetic alteration in HCC, associated with tumor progression (Zhang et al., 1994).

Epigenetic events including altered methylation also are involved in hepatocarcinogenesis. DLC-1 (frequently deleted in liver cancer), a putative tumor suppressor gene, was hypermethylated in some HCC cell lines and primary HCC of humans (Wong et al., 2003). Metallothionein (MT) is down-regulated in HCC and this may be related to hypermethylation of the MT-promoter (Cherian et al., 2003). HCC induced by treatment with DEN and other chemicals in F344 rats revealed increased expression of p53 and hypermethylation of p16 INK4A exon 1 in the later stages of carcinogenesis (Lim, 2002).

Neoplastic transformation of hepatocytes to HCC results from the accumulation of a significant number of genetic and epigenetic alterations and is accompanied by increased genetic instability. Currently, more than 20 cellular genes have been found to be either down- or up-regulated or mutated in HCC. However, studies suggest that multiple molecular mechanisms are involved in hepatocarcinogenesis and many of the specific target genes remain to be determined. Comprehensive elucidation of the specific genes and molecular pathways involved in neoplastic transformation of the liver will facilitate development of new strategies for diagnosis, prevention, and therapy.

Hepatoblastomas

Hepatoblastoma is the most common malignant hepatic neoplasm in children with an incidence of 1.5 cases per million children (Koesters and von Knebel Doeberitz, 2003). Most cases are sporadic and occur with a predominance in males, however some are associated with familial syndromes such as Beckwith-Wiedemann syndrome or familial adenomatous polyposis coli (FAP) (Ishak and Glunz, 1967; Steenman et al., 2000). Individual case reports have described this tumor in a number of different animal species including dogs, horses, alpacas and llamas, however, most animal cases have been described in mice (Neu, 1993; Shiga et al., 1997; Watt et al., 2001). In the mouse, spontaneous hepatoblastomas are rare but certain chemicals can induce a high incidence of this tumor type and, like its human counterpart, these often develop with a higher incidence in males (Turusov et al., 2002). Unlike their human equivalents, however, chemically induced mouse hepatoblastomas generally occur in aged rather than young animals. Despite this difference a number of similarities have been noted between mouse and human hepatoblastomas.

Alterations in the Wnt signaling pathway are involved in a high percentage of both mouse and human hepatoblastomas (Blaker et al., 1999; Koch et al., 1999; Anna et al., 2000; Jeng et al., 2000; Takayasu et al., 2001; Udatsu et al., 2001; Hayashi et al., 2003; Koesters and von Knebel Doeberitz, 2003). β-catenin is an essential component of the Wnt signaling pathway (Cadigan and Nusse, 1997) playing roles in intercellular adhesion, signal transduction and gene transcription (Nelson and Nusse, 2004). Mutations of the β-catenin gene in human hepatoblastomas have been reported to be as high as 89% (Jeng et al., 2000) while some chemically induced hepatoblastomas in mice reveal incidences of up to 100% (Anna et al., 2000; Takayasu et al., 2001). Both mouse and human hepatoblastomas have shown a high prevalence of deletion mutations affecting the GSK-3β binding region of the β-catenin gene (Takayasu et al., 2001; Taniguchi et al., 2002; Hayashi et al., 2003). Mutations in this region prevent degradation of the β-catenin protein allowing for enhanced signaling. Immunohistochemical studies in both mouse and human hepatoblastomas have confirmed these findings by showing a loss of β-catenin membrane staining along with increased cytoplasmic and nuclear staining (Blaker et al., 1999; Anna et al., 2000; Park et al., 2001). Studies looking at the expression of downstream targets, such as Cyclin D1, also suggest that Wnt-mediated signaling is indeed activated in both human (Kuniyasu et al., 1996; Takayasu et al., 2001) and mouse (Anna et al., 2003) hepatoblastomas. Interestingly, some of these same studies show that c-myc, another potential downstream target of the Wnt signaling pathway, is not elevated in either mouse or human hepatoblastomas. Additional support for involvement of the Wnt signaling pathway come from several human studies showing alterations in other key members of this pathway including both APC and AXIN (Oda et al., 1996; Taniguchi et al., 2002; Miao et al., 2003). Furthermore, people with the familial syndrome FAP (which carry a germline mutation of APC) are known to have up to a thousand times greater risk for development of hepatoblastomas as compared to the general population (Kingston et al., 1982; Koesters and von Knebel Doeberitz, 2003).

Beyond the Wnt signaling pathway few molecular analyses on mouse hepatoblastomas have been performed. A couple of mouse studies report that ras mutations are uncommon as are alterations in p53 (Devereux et al., 1994; Hayashi et al., 2003). Similarly, most studies on human hepatoblastomas have shown a lack of p53 involvement (Chen et al., 1995) although only a few studies, involving a small number of cases, have looked at RAS mutations in human hepatoblastomas. One study found a single K-RAS mutation in codon 12 from 1 human hepatoblastoma (Stork et al., 1991) and another found no mutations in 2 cases that were examined (Tada et al., 1990).

TGF-α is a potent stimulator of cell proliferation in the liver and its overexpression is thought to be important in hepatocarcinogenesis. In the mouse, TGF-α expression was shown to be elevated in chemically induced hepatoblastomas and the levels were even higher than those found in hepatocellular carcinomas (Sakairi et al., 2001). Similar findings have been described in human hepatoblastomas where TGF-α expression was shown to be inversely correlated with the level of cell proliferation and positively correlated with the degree of differentiation (Kiss et al., 1998). An additional growth factor frequently implicated in human hepatoblastomas is IGF II. Up-regulation of IGF II has been described in a high percentage of human hepatoblastomas (Li et al., 1995; Hartmann et al., 2000) and although the exact mechanism for this alteration is not entirely certain it appears that loss of imprinting and/or promoter demethylation are at least partly involved (Eriksson et al., 2001; Zatkova et al., 2004). Other studies have shown that the level of IGF II expression in these tumors is inversely correlated with the degree of tumor differentiation (Akmal et al., 1995). Histologically, chemically-induced mouse hepatoblastomas frequently appear similar to the less differentiated human hepatoblastoma subtypes (embryonal or small cell) (Turusov et al., 1973; Diwan et al., 1995), however, to our knowledge, IGF II’s involvement in chemically induced mouse hepatoblastomas has not been thoroughly explored. Interestingly, activation of the IGF II receptor has been shown to result in redistribution of β-catenin from the cell membrane to the nucleus and cause intracellular sequestration and degradation of E-cadherin (Morali et al., 2001). Other studies have shown a decrease in E-cadherin membrane expression in both mouse (Anna et al., 2003) and human hepatoblastomas (von Schweinitz et al., 1996).

A few studies have looked at the role of growth factor receptors in hepatoblastomas including c-MET (hepatocyte growth factor receptor), which was reported as being overexpressed in human hepatoblastomas (von Schweinitz et al., 2002), and epidermal growth factor receptor (EGF-R), which was reported as showing reduced expression in some chemically induced mouse hepatoblastomas (Anna et al., 2003). Interestingly, both EGF-R (Hoschuetzky et al., 1994; Takahashi et al., 1997) and c-MET (Danilkovitch-Miagkova et al., 2001; Nelson and Nusse, 2004) are also involved in regulation of β-catenin.

Although a number of different chromosomal alterations have been described in human hepatoblastomas these are not as common as those found in hepatocellular carcinomas (Buendia, 2002). In addition, it appears as though the presence of β-catenin mutations in either of these tumor types is associated with chromosomal stability (Buendia, 2002). Cell cycle inhibitors have been studied to a limited extent in human but not mouse hepatoblastomas. p16 was shown to be down-regulated due to hypermethylation (Shim et al., 2003) and p27 exhibited reduced expression in undifferentiated hepatoblastomas (Brotto and Finegold, 2002).

Conclusions

Mouse liver carcinogenesis is a complex process which involves growth factors, oncogenes, tumor suppressor genes, cell cycle genes and other mechanisms such as receptor based signal transduction pathways (Figure 1). In addition, factors such as altered methylation and chromosomal alterations may contribute to liver cancer. As more information is obtained on the multiple factors which contribute to the progression of liver cancer in mice and humans, it will provide opportunities for successful intervention and treatment strategies.