Abstract
Overexpression of c-myc and transforming growth factor-alpha (TGF-α) has been frequently observed in human hepatocellular carcinoma (HCC), suggesting a pivotal role played by these protooncogenes in liver oncogenesis. In order to investigate the molecular events underlying human hepatic malignant transformation, we have generated c-myc and c-myc/TGF-α transgenic mice that are prone to liver cancer. These transgenic mice develop HCCs with different incidence, kinetics and histopathological features. Indeed, co-expression of c-myc and TGF-α transgenes results in a dramatic synergistic effect on liver tumor development when compared with respective single transgenic lines, including a shorter latency period and a more aggressive phenotype. The more malignant histopathological features characteristic of c-myc/TGF-α HCCs are the result of the increased proliferation and reduced apoptosis in this model of liver cancer when compared with single parental lines. Accordingly, c-myc and c-myc/TGF-α transgenic mice display a different molecular pathogenesis of HCC. Importantly, the genetic and molecular mechanisms that are involved in c-myc and c-myc/TGF-α liver cancer development are major oncogenic events in human hepatocarcinogenesis, indicating that these mouse models represent a useful tool to dissect and elucidate the molecular basis of human HCC.
Introduction
Co-expression of transforming growth factor (TGF)- α and c-myc protooncogenes has been frequently detected in human hepatocellular carcinoma (HCC), suggesting a crucial role for these genes in the malignant growth of the liver (Thorgeirsson and Grisham, 2002). To investigate the functional relevance of c-myc and TGF-α cooperation in human hepatocarcinogenesis, we have previously generated c-myc and c-myc/TGF-α transgenic mice that develop liver cancer (Murakami et al., 1993; Santoni-Rugiu et al., 1996). The work in our laboratory has shown that hepatic expression of c-myc alone results in chronic hepatic proliferation and increased incidence of liver cancer (Murakami et al., 1993), whereas co-expression of c-myc and TGF-α transgenes in the liver accelerates HCC development in c-myc/TGF-α double transgenic mice when compared with both parental lines (Murakami et al., 1993; Santoni-Rugiu et al., 1996). In particular, combined up-regulation of c-myc and TGF-α resulted in rapid progression from early preneoplastic focal lesions to HCC in 4 months in the transgenic mice, with 100% frequency of HCC before 8 months and survival reduced to 1 year. In striking contrast, both single transgenic mouse lines exhibited longer tumor latency as well as decreased incidence of HCC (Murakami et al., 1993; Santoni-Rugiu et al., 1996). These data indicate the presence of different molecular mechanisms of hepatocarcinogenesis in c-myc and c-myc/TGF-α transgenic mice. Since the transgenic system offers a rare opportunity to examine the molecular and morphological changes associated with the sequential steps of malignant transformation, considerable efforts have been devoted to define these events.
Results and Discussion
The earliest effect of c-myc and/or TGF-α overexpression in the liver is the induction of persistent proliferation of the hepatocytes, which disrupts the normal mitogenic silencing of the liver occurring during the first weeks of life (Murakami et al., 1993). This continuous replication, associated with elevated levels of the urokinase-type plasminogen activator, leads to the appearance of perivascular dysplastic hepatocytes expanding into the hepatic lobules and the central vein by the second month of age, accompanying the development of neoplastic lesions. To determine whether the generation of initiated cells would take place in the early dysplastic stage, dysplastic liver pieces were transplanted onto nude mice. The results of the experiment showed that livers containing the combination of dysplasia and apparent vascular invasion yielded more HCCs than those having only dysplastic lesions, indicating that the initiated cell population is generated during the early stage of the neoplastic process (Santoni-Rugiu et al., 1996).
Furthermore, the latter is able to progress to HCC without going through focal and nodular stages. Since the large dysplastic hepatocytes are extremely vulnerable to apoptosis and HCCs are composed of small diploid cells, it seems more likely that the small dysplastic cells are the putative tumor precursors. In accordance with this hypothesis, large dysplastic cells displayed the autocrine up-regulation of transforming growth factor-beta 1(TGF-β1), a potent growth inhibitor and inducer of apoptosis in hepatocytes (Roberts et al., 1988). Further experiments have shown that overexpression of mature TGF-β1 may also provide a selective environment in which (pre)neoplastic cells with reduced sensitivity to this cytokine will progress more rapidly toward a malignant phenotype (Factor et al., 1997). Yet, loss of transforming growth factor receptor (TβR)-II was detected in c-myc/TGF-α pre-neoplastic lesions with a small/clear cell phenotype, indicating this cell population as the potential tumor precursors, due to being refractory to antiproliferative and pro-apoptotic effects of TGF-β1(Santoni-Rugiu et al., 1999). In contrast, TβRII negative cell clusters were not observed in c-myc livers, whereas only 47% of c-myc HCCs exhibited TβRII down-regulation.
Thus, selection and expansion of TGF-β1-insensitive cells is a late event during c-myc-induced hepatocarcinogenesis. These studies are in accordance with previous findings showing that human HCC can become resistant to TGF-β 1-mediated apoptosis through reduction of TβRII expression and or function (Bedossa et al., 1995; Kiss et al., 1997). Interestingly, eosinophilic preneoplastic lesions (characteristic of c-myc overexpressing livers) with an intact TβRII-signaling exhibited nuclear focal positivity for β-catenin (Calvisi et al., 2001), a multifunctional member of the Wingless/Wnt cascade involved in cell–cell adhesion and embryogenesis, as well as in the malignant transformation of many cell types (Polakis, 2000). Although the meaning of this relationship remains to be elucidated, it is tempting to speculate that activation of the Wnt/β-catenin pathway may provide a selective proliferative advantage for preneoplastic cells exposed to TGF-β 1 (Calvisi et al., 2001).
Dysplastic hepatocytes showed nuclear pleomorphism, multiple nucleoli, and the presence of abnormal mitotic figures, which are signs of an abnormal cell cycle progression (Santoni-Rugiu et al., 1996). We have hypothesized the presence of chronic oxidative stress in c-myc/TGF-α proliferating cells and a concomitant deficiency of the DNA repair system as a possible molecular pathogenesis of dysplastic hepatocytes. In agreement with this hypothesis, we have recently found that c-myc transgenic mice displayed a more efficient induction of genes involved in DNA repair than c-myc/TGF-α mice after treatment with the peroxisome proliferator Wy-14,643 (Hironaka et al., 2003). On the other hand, Factor et al. (1998) demonstrated that the production of reactive oxygen species (ROS) was significantly elevated in c-myc/TGF-α mice at 10 weeks of age when compared with wild-type or c-myc lesions and occurred in parallel with increase in lipid peroxidation. In addition, in vitro studies have shown that the appearance of dysplastic changes in c-myc/TGF-α livers is associated with a dramatic increase in aneuploidy, chromosome breakage and translocations, whereas a relatively stable genome was detected in both parental lines (Sargent et al., 1996).
Accordingly, a high rate of genomic instability and loss of heterozygosity was observed in vivo in c-myc/TGF-α dysplastic livers as early as 10 weeks of age, when no genomic alterations were detected in single transgenic lines (Calvisi et al., 2004a). Indeed, genomic instability began at 28 weeks of age in TGF-α mice and was less pronounced than in c-myc/TGF-α double transgenic mice. Furthermore, the degree of genetic instability progressively increased in c-myc/TGF-α but not TGF-α HCCs (Calvisi et al., 2004a). Together with the fact that accumulation of genetic alterations was detected only at the later stage of tumor development in c-myc trans-genic mice, these data identify TGF-α and c-myc as driving forces of genomic instability at the preneoplastic and neoplastic stage, respectively.
As a result, co-expression of c-myc and TGF-α accelerates both the onset and extent of genomic instability in the liver. The specific sites of breakages observed during early stages of c-myc/TGF-α hepatocarcinogenesis are later involved in stable rearrangements in the tumors (Sargent et al., 1999). The breakage of chromosomes 1, 5, 6, 7, and 12 may represent an early event, while the deletions of chromosomes 4, 12, 14, and X appear during the tumor progression (Sargent et al., 1999). Importantly, the breakpoints described on c-myc/TGF-α chromosomes 1, 4, 7, and 12 correspond to human 1q, 1p, 11p, and 14q that are also rearranged in human liver tumors (Marchio et al., 1997; Zimonjic et al., 1999). The alteration of the same genetic linkage groups in mouse and human liver tumors indicates that these regions of the genome are critical in the etiology of hepatic malignant transformation. Furthermore, Vitamin E, a potent free-radical scavenging antioxidant, reduces liver dysplasia, prevents malignant transformation and promotes chromosomal and mitochondrial DNA stability in c-myc/TGF-α transgenic mice (Factor et al., 2000). Taken together, these data indicate ROS as the primary carcinogenic agent in the c-myc/TGF-α overexpressing livers and chromosomal instability as the underlying mechanism of accelerated oncogenesis in this model of liver cancer (Factor et al., 1998, 2000).
The morphological appearance of frankly malignant HCCs is accompanied by the inhibition of apoptosis and a higher rate of cellular proliferation in c-myc/TGF-α mice when compared with c-myc corresponding lesions. The c-myc/TGF-α HCCs were characterized by a particularly strong expression of TGF-α and very low apoptotic index in contrast to high levels of apoptosis in peritumorous tissues and c-myc HCCs, indicating TGF-α as a potential survival factor (Santoni-Rugiu et al., 1998). In accordance with this observation, Factor et al. (2001) has recently showed that NFκB-induced survival signaling is activated in preneoplastic and neoplastic lesions of c-myc/TGF-α double transgenic mice, as well as in human liver cancer cell lines through phosphorylation of Iκβ kinase, but not in c-myc single transgenics, suggesting that TGF-α mediates the induction NFκB. Moreover, Akt/protein kinase B-induced NFκB activation occurs during c-myc/TGF-α tumor progression, providing an additional survival advantage for neoplastic hepatocytes (Factor et al., 2001). Since oxidative stress has been reported to activate the NFκB pathway (Baeuerle and Henkel, 1994), it is tempting to speculate that ROS generation might mediate NFκB induction in c-myc/TGF-α HCCs.
The molecular mechanisms responsible for cellular proliferation during c-myc and c-myc/TGF-α liver carcinogenesis has been further investigated by Santoni-Rugiu et al. (1998). Differential levels of cell proliferation in non-tumor and tumor tissues correlated with a stronger induction of cyclin D1 in c-myc/TGF-α and c-myc HCCs and were associated with intense pRb hyperphosphorylation. Severe deregulation of G1-S transition was strengthened by the observed dramatic up-regulation, particularly in the HCCs, of pRb-free E2F1-DP1 and E2F2-DP1 transcription factor heterodimers. Abnormally elevated E2F activity during liver tumor development was further indicated by the transcriptional induction of putative E2F target genes involved in cell cycle progression, including endogenous c-myc, cyclin A, Cdc2, and E2F itself (Santoni-Rugiu et al., 1998).
Since the disruption of the Wingless/Wnt signaling pathway is frequently involved in human and rodent hepatocar-cinogenesis (de La Coste et al., 1998; Yamada et al., 1999), we have examined the role of β-catenin activation in c-myc and c-myc/TGF-α tumor progression. Activation of β-catenin was most frequent in liver tumors from c-myc transgenic mice, whereas it was very rare in faster growing and histologically more aggressive hepatocellular carcinomas developed in c-myc/TGF-α mice (Calvisi et al., 2001). C-myc and c-myc/TGF-α HCCs with nuclear accumulation of β-catenin displayed a higher proliferation rate and tumor size when compared with HCCs without β-catenin activation, suggesting that activation of β-catenin provides proliferative rather than survival advantages in transgenic hepatocarcinogenesis (Calvisi et al., 2001). The reason for the low incidence of β-catenin activation in c-myc/TGF-α tumor liver development is unclear.
One possible explanation is that c-myc/TGF-α neoplastic clones without β-catenin activation possess selective growth advantages over β-catenin positive clones. Accordingly, we have recently found that treatment with phenobarbital significantly increases the rate of β-catenin activation in c-myc/TGF-α HCCs by favoring the growth of clones with an intact β-catenin locus (Calvisi et al., 2004b). Indeed, loss of heterozygosity at the β-catenin locus is a frequent event in untreated c-myc/TGF-α HCCs (Calvisi et al., 2001). Furthermore, only c-myc/TGF-α and TGF-α preneoplastic and neoplastic lesions with loss or reduction in β-catenin levels over-express the Wnt signaling inhibitor SARP2 (Calvisi et al., 2004a), suggesting that TGF-α might suppress β-catenin by up-regulation of SARP2. Notably, since c-myc/TGF-α pre-neoplastic and neoplastic lesions displayed a high rate of genomic instability, but a low rate of β-catenin activation (Sargent et al., 1996, 1999; Calvisi et al., 2004a), our results are consistent with the recent observations that β-catenin activation occurs in a subset of human HCCs with a relatively stable genome (Legoix et al., 1999; Laurent-Puig et al., 2001) and a more favorable prognosis (Hsu et al., 2000).
Therefore, the c-myc and c-myc/TGF-α transgenic mouse models reveal a remarkable similarity with development of human liver cancer. β-catenin activation, in cooperation with a signaling pathway initiated by c-myc, contributes to the development of a subset of HCCs characterized by a high degree of cell differentiation and a low rate of genomic instability in c-myc single transgenic mice. In contrast, gross genomic instability characteristic of c-myc/TGF-α-driven hepatocarcinogenesis, leads to tumors of low-grade differentiation, high malignancy, and reduced survival with an early disruption of growth control including cell cycle progression and apoptosis. In summary, our results indicate that c-myc and c-myc/TGF-α transgenic mice are relevant animal models for dissecting the genetic and molecular pathways leading to human HCC.