Loss of p53 and Ink4a/Arf Cooperate in a Cell Autonomous Fashion to Induce Metastasis of Hepatocellular Carcinoma Cells Free

Hepatocellular carcinoma (HCC) is a leading cause of cancer mortality worldwide. More than 500,000 people are diagnosed annually with this malignancy, with a 5-year survival rate below 5% (1). HCCs are resistant to current chemotherapeutic intervention strategies, and therefore, surgery remains the only curative option (2). However, surgical resection is not an option in patients with metastasis—most commonly observed in the extrahepatic region of the portal vein, lymph nodes, and lungs—a common finding at the time of diagnosis (3). Therefore, understanding the mechanisms underlying the spread of HCC is of great importance.

HCC is commonly associated with chronic infection by the hepatitis B (HBV) and hepatitis C (HCV) viruses, and the mechanisms by which these viruses contribute to tumor initiation and progression remain critical areas of investigation (2, 4, 5). Previous studies have suggested that protein products of HBV and HCV—the HBx and NS5 proteins, respectively—interact with the p53 tumor suppressor protein and inhibit its function (6, 7). In the cellular genome, mutations are commonly found in TP53, confirming the significance of loss of p53 function in HCC pathogenesis (8). Silencing of the INK4A/ARF locus also occurs in a majority of HCC cases (8–12). Additionally, whereas RB1 is seldom inactivated by mutation, the retinoblastoma tumor suppressor protein is frequently inactivated by overexpression of the oncoprotein gankyrin, which functionally targets both p53 and pRb (13). Thus, the Arf-p53 and p16-pRb tumor suppressor axes are targeted for inactivation in the majority of HCC cases.

Many previous studies have described mouse models for HCC (14–23). We have recently described a mouse model for HCC in which somatic delivery of avian retroviral vectors encoding the mouse polyoma virus middle T antigen (PyMT) to the livers of albumin-tv-a transgenic mice leads to the formation of hepatic tumors (24). Although these tumors can exceed 1 cm in diameter, they rarely metastasize to the lungs. Significantly, we found that whereas induction of tumors in tv-a transgenic mice with germ line Trp53 deletion did not lead to a higher tumor incidence, it led to a higher rate of lung metastases (24). These findings are consistent with the hypothesis that loss of TP53 is a late event associated with tumor progression in HCC, as well as with previously published findings on Trp53 loss and tumor progression in mouse models (25). However, mice with germ line deletion of Trp53 are not suitable for studies of cancer progression because of the rapid occurrence of thymic lymphomas and sarcomas that lead to a shortened life span (26, 27). Furthermore, because all cells within the mouse lack the p53 tumor suppressor, it is not possible to identify cell autonomous requirements for tumor metastasis using these mice. Therefore, mouse models without these deficiencies will be of great benefit to the field.

We report here the development of mouse models for HCC after the somatic delivery of oncogene-bearing avian retroviruses into mice with liver-specific deletion of tumor suppressor genes. We find that PyMT is able to induce metastatic liver tumors in mice with liver-specific deletion of Trp53, suggesting that the effect of p53 on metastasis in our HCC model occurs in a cell autonomous fashion. Furthermore, we find that concomitant loss of Trp53 and Ink4a/Arf accelerates tumorigenesis and the development of metastatic lung lesions. Consistent with this finding, cell lines generated from tumors lacking both Trp53 and Ink4a/Arf exhibited strong migration and invasion capabilities, whereas a cell line generated from a Trp53 null tumor did not. Significantly, despite their poor migration and invasion capabilities, Trp53 null tumor cells, as well as cells lacking Trp53 and Ink4a/Arf, efficiently colonized the lungs of nude mice after i.v. injection, suggesting that the ability to colonize distant sites can occur before the acquisition of invasive capability. Thus, our data collectively show the cell autonomous role for p53 in the metastatic phenotype of HCC cells and illustrate the utility of our model system for the elucidation of the molecular mechanisms underlying HCC metastasis.

Mouse lines. Albumin-tv-a and albumin-cre transgenic mice, and Trp53 and Ink4a/Arf conditional mutant mice have been previously described (24, 28–30). All animals were kept in specific pathogen-free housing at the University of Massachusetts Medical School with abundant food and water. All experiments were reviewed and approved by the University of Massachusetts Institutional Animal Care and Use Committee.

Virus delivery. The RCAS-GFP and RCAS-PyMT vectors have been previously described (31, 32). DF1 chicken fibroblasts (33, 34) transfected with RCAS vectors were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) in humidified 37°C incubators under 5% CO₂. Cells to be injected were harvested, washed once with PBS, and resuspended in PBS at a final concentration of 2 × 10⁵ cells/μL. About 5 μL of the cell suspension was delivered by injection into the liver parenchyma of 2- or 3-day-old animals using Hamilton syringes attached with 26-gauge needles.

Tumor harvest and histology. Animals were sacrificed with a lethal dose of CO₂ as per institutional guidelines. Tumors, liver tissue, and lungs were removed and either fixed in 10% buffered formalin overnight at room temperature or snap frozen in liquid nitrogen. Fixed tissues were paraffin embedded, and 5-μm sections were placed on sialynated slides.

Immunohistochemistry. Paraffin sections were deparaffinized and rehydrated by passage through Clear-Rite 3 and a graded alcohol series. Endogenous peroxidase activity was inactivated by treatment with 3% hydrogen peroxide after which antigen retrieval was done using heated citric buffer. Slides were blocked for 1 h and then incubated with appropriate primary and secondary antibodies according to manufacturers' instructions. Substrate incubation and color development were done according to the manufacturer's instructions (Vector Labs). Primary antibodies: mouse anti–E-cadherin, 1:50 (BD); mouse anti–β-catenin, 1:100 (BD); TROMA-1 (anti–keratin 8), 1:50 (University of Iowa Developmental Studies Hybridoma Bank); and rat anti-polyoma virus T antigens, 1:500 (gift of Michelle Fluck, Michigan State University, East Lansing, MI).

Immunoblotting. Cell lysates were isolated in radioimmunoprecipitation assay buffer (RIPA) after incubation on ice for 30 min. Protein concentration was determined using the BCA reagent (Pierce) and 20 or 30 μg of protein loaded per well. After transfer to polyvinylidene difluoride (PVDF) or nitrocellulose membranes, primary antibodies were incubated for 1 h at room temperature in TBS, 0.1% Tween 20 (TBS-T), and appropriate blocking reagent, according to the manufacturer's instructions. After washing with TBS-T, blots were incubated with appropriate secondary antibodies (Jackson Immunochemicals) at room temperature for 45 min. Chemiluminescence was done using either enhanced chemiluminescence reagent (Amersham) or Super Signal reagent (Pierce). Primary antibodies: rabbit anti–p-Akt, 1:1,000 (Cell Signaling); rabbit anti–p-Erk, 1:1,000 (Cell Signaling); rabbit anti-Akt, 1:1,000 (Cell Signaling); and rabbit anti-Erk 1:1,000 (Cell Signaling).

Reverse-transcription PCR. RNA was isolated from frozen tumor tissue and tumor cell lines as previously described (24). About 0.5 μg of purified total RNA was used for reverse-transcription PCR (RT-PCR). All RT-PCR reactions were done in 50 μL total reaction volume using the Superscript III One-Step RT-PCR kit (Invitrogen). Primers: Cxcr4: 5′-ACCTCCTCTTTGTCATCACACTCC-3′ and 5′-AACACCACCATCCACAGGCTATCG-3′; p16/p19 (exon 2): 5′-ATGATGATGGGCAACGTTC-3′ and 5′-CAAATATCGCACGATGGTC-3′.

Primers for tv-a, IGF2, snail, and β-actin have been previously described (24, 35). All reactions were done for 35 cycles except for β-actin, which was done for 25 cycles. Bands were quantified using a densitometer.

Isolation and culture of tumor cell lines. Immediately after harvest, tumor tissue was minced and dissociated by pipetting in DMEM. Cells were washed once in sterile PBS and plated in a 10-cm collagen-coated culture dish with DMEM supplemented with 10% FBS and antibiotics. After 48 h, nonadherent cells were removed, and adherent cells were washed with PBS and fed with fresh medium containing FBS.

Cell proliferation assays. Approximately 10³ cells were seeded in quadruplicate onto 96-well plate with collagen coated and incubated at 37°C under 5% CO₂. After 24 h, viable cell numbers were measured in quadruplicate every period for 4 days using CellTiter 96 Aqueous One Solution Cell proliferation assay (Promega) according to the manufacturer's instructions. The proliferation curves were constructed by calculating the mean value of absorbance measurement at 490 nm using a 96-well plate reader.

Soft agar assays. Soft agar colony formation assays were done as previously described (36).

Migration and invasion assays. About 2.5 × 10⁴ cells in 0.5 mL of serum-free DMEM were plated into either control or matrigel-coated invasion inserts (BD). Inserts were then placed in wells with 0.75 mL of DMEM containing 10% FBS as a chemoattractant. After culture for 20 to 24 h at 37°C, cells were fixed with methanol for 8 min at room temperature and stained with Giemsa reagent diluted 5× in H₂O. Cells on the upper sides of the inserts were removed with a cotton swab, and the insert membranes were removed and mounted on glass slides. Cell numbers for migration and invasion were then determined by counting the number of cells present in five microscope fields per insert at 50× magnification. The percent invasion was calculated as the number of invading cells divided by the number of migrating cells. All experiments were done in duplicate and repeated a minimum of thrice. Data are shown for representative experiments.

Tail vein injections. 8- to 10-week-old male nude mice were placed in a tail vein injector and restrainer, the tail was swabbed with ethanol, and 10⁵ cells were resuspended in 300 μL sterile PBS injected into the tail vein. Pressure was applied to the injection site for several seconds to prevent bleeding.

Liver-specific Trp53 deletion induces tumors with metastatic potential. We have previously observed that liver tumors induced after somatic introduction of PyMT to albumin-tv-a mice with germ line deletion of Trp53 metastasized to the lung in ∼40% of cases (24). However, these animals have a reduced life span due to the development of thymic lymphomas and sarcomas, and therefore, the analysis of tumor progression in these mice is compromised. We hypothesized that the delivery of PyMT to mice bearing liver-specific deletion of Trp53 would lead to tumors with a similar phenotype, without the complications of additional nonhepatic tumors, thus providing a system to pursue studies related to tumor progression. We therefore crossed albumin-tv-a transgenic mice to animals bearing conditional mutant alleles (floxed alleles) of the Trp53 tumor suppressor locus, in which loxP sites are located in intron 1 and intron 10 (28), and to the albumin-cre transgenic mouse line (30), to generate triple transgenic mice with liver-specific p53 loss of function. To induce liver tumors in these mice, DF1 chicken fibroblasts producing RCAS-PyMT were injected into the livers of newborn mice as previously described (24). The animals were sacrificed at 6 or 8 months of age and analyzed for tumor development.

Consistent with our published findings (24), we found that 10:18 injected animals sacrificed at 6 months of age had liver tumors, with 7 of these animals bearing tumors >6 mm in diameter (Table 1). The primary tumors displayed a range of histologies, from aggregates of architecturally normal but cytologically atypical hepatocytes reminiscent of large cell change in humans, to frankly anaplastic and poorly differentiated carcinomas with regions of necrosis (Fig. 1A and data not shown). In addition, cholangiocarcinomas consisting of irregularly shaped glandular and micropapillary structures were observed in two tumor-bearing mice, either as a separate lesion (Fig. 1A , right) or as a region within a larger HCC (combined hepatocellular-cholangiocarcinoma; ref. 37). These histologic features resemble those of tumors found in animals with germ line Trp53 deletion, suggesting that the histopathologic features are determined by the tumor cells themselves, and are not appreciably influenced by interactions with p53-deficient stromal and immune cells.

Of the seven mice bearing tumors >6 mm in diameter, three also had pulmonary metastases, a frequency higher than that observed in wild-type animals (P < 0.05; ref. 24), indicating that the loss of p53 contributes to metastasis in a cell autonomous fashion (Fig. 1B). By histology and staining with anti-cytokeratin 19 (CK19) antibodies, we determined that both HCCs and cholangiocarcinomas display metastatic potential in our mouse model (Fig. 1B and data not shown). Consistent with the human disease, the metastatic cells grew as discrete expansile parenchymal nodules without infiltrating into the surrounding lung parenchyma.

Three of six animals sacrificed at 8 months of age had tumors, with only two of these exceeding 6 mm in diameter and none displaying lung metastases. The absence of higher tumor incidence and metastasis rates in 8-month-old mice suggests that, in our model system, the aggressiveness of the tumor and its metastatic capability are manifested by 6 months of age. Only 2:25 animals infected with RCAS-GFP and sacrificed between 12 and 15 months of age developed liver tumors (Fig. 1C and Table 1).

The loss of E-cadherin is proposed to play a role in metastatic progression in several tumor types, including HCC. We therefore analyzed the expression of E-cadherin and its interacting partner β-catenin in metastatic lung lesions. Of the two lung metastases able to be analyzed, we found retention of membrane-localized E-cadherin and β-catenin in one lesion (consisting of a cholangiocarcinoma; Fig. 2A, bottom) and reduced expression of E-cadherin and β-catenin in the other (consisting of an HCC; Fig. 2A,, top). Nuclear localization of β-catenin, indicative of activation of the canonical Wnt signaling pathway, was observed only in rare cells, predominantly stromal in origin, in the metastatic lesions or the primary tumors (Fig. 2A and data not shown).

Elevated mRNA levels for insulin-like growth factor II (Igf2) are observed in liver tumors induced by PyMT in animals with germ line Trp53 deletion, and IGF-stimulated signaling correlates with HCC pathogenesis and tumor progression (24, 38, 39). To determine whether these changes occur in a cell autonomous fashion, we assayed by RT-PCR the mRNA levels for Igf2 in tumors with liver-specific p53 loss. We observed a 1.6-fold increase in Igf2 mRNA levels in Trp53 null tumors relative to Trp53 wild-type tumors (Fig. 2B). These data suggest that the absence of p53 in PyMT-induced liver tumors enhances Igf2 expression, although the induction is not as robust as we previously observed (24).

Recent reports have shown that the expression of the transcription factor snail in HCC samples correlates with metastasis and poor patient outcome (40, 41). Additional reports have suggested that the expression of the chemokine receptor Cxcr4 is associated with the development of lung metastasis in breast carcinoma cells and may regulate the migration of cholangiocarcinoma cells (42, 43). We therefore examined whether these genes were induced in tumors lacking p53. We found that snail mRNA levels are modestly (2-fold) induced in p53-deficient tumors relative to wild-type tumors (Fig. 2B). We additionally observed that the expression of Cxcr4 is constant in all tumors, even those with wild-type p53 alleles that display low metastatic potential (Fig. 2B). These findings suggest that the loss of p53 cooperates with PyMT to stimulate the expression of some prometastasis genes, such as Snail and Igf2, whereas the expression of other genes is not influenced by p53 status.

Concomitant deletion of Ink4a/Arf accelerates the metastatic phenotype. The Ink4a/Arf locus encodes two tumor suppressor proteins, both of which are commonly inactivated in HCC, and one of which, the p19^Arf tumor suppressor protein, acts upstream of p53 (44). Therefore, to determine whether the loss of the Ink4a/Arf locus stimulates the formation of lung metastases, we delivered RCAS-PyMT to albumin-tv-a mice with germ line deletion of the Ink4a/Arf locus. A total of 5:15 animals sacrificed at 6 months of age had grossly visible liver tumors (Table 1). Furthermore, of the three mice with tumors >5 mm in diameter, only one had metastasis to the lung. Thus, germ line deletion of the Ink4a/Arf locus does not accelerate PyMT-induced liver tumorigenesis (24).

To identify whether the simultaneous loss of the Trp53 and Ink4a/Arf loci would cooperate in HCC tumor formation and metastasis, we generated compound mice transgenic for albumin-tv-a and albumin-cre, and bearing conditional mutant alleles at the Trp53 and Ink4a/Arf loci. Delivery of RCAS-PyMT producer cells led to the formation of tumors in 8:9 animals sacrificed and analyzed at 4 months of age (Table 1). All eight tumor-bearing mice had tumors that exceeded 6 mm in diameter, and metastatic lung lesions were observed in two of these animals. As was the case with Trp53 deficiency, liver-specific deletion of Trp53 and Ink4a/Arf combined with PyMT expression to induce a range of tumor histologies, including HCC (Fig. 3A,, left), cholangiocarcinoma (Fig. 3A,, middle left), and combined hepatocellular cholangiocarcinoma (Fig. 3A,, middle right). In addition, we identified undifferentiated tumors with frankly anaplastic cells (Fig. 3A,, right). About 8:9 animals sacrificed at 6 months of age also had tumors, with seven of these animals bearing lesions larger than 6 mm in diameter. Of these seven mice, five displayed visible metastases to the lungs or diaphragm (Table 1 and Fig. 3B). By histopathology, all of the metastatic lesions in this cohort seemed to be derived from HCCs.

To determine whether the accelerated tumorigenesis reflected enhanced signaling by the PyMT oncoprotein, we examined the activation status of the mitogen-activated protein (MAP) kinase and phosphoinositide-3-kinase (PI3K) signaling pathways by immunoblotting. Intriguingly, we found that the MAP kinase and PI3K signaling pathways are more highly activated in p53-deficient tumors than in those lacking both p53 and Ink4a/Arf (Fig. 3C). These data suggest that the accelerated tumor phenotype in the double null mice is not a consequence of the enhanced activation of these signaling pathways.

To identify whether metastasis of Trp53, Ink4a/Arf double null tumors required the loss of E-cadherin expression, we analyzed metastatic lesions for the presence of E-cadherin and β-catenin. As was observed for Trp53-deficient lesions, E-cadherin levels were reduced in 2:4 metastatic lesions (data not shown), indicating that the loss of membrane-localized E-cadherin and β-catenin is associated with metastasis, but is not required for the metastatic phenotype. As was seen in Trp53 null tumors, double null tumors did not display nuclear accumulation of β-catenin (data not shown).

We next assayed whether these phenotypes were linked to the elevated expression of Igf2, Snail, and Cxcr4. Consistent with our findings in p53-deficient tumors, we found that Cxcr4 mRNA levels were not increased relative to tumors with wild-type tumor suppressor gene loci (Fig. 2B). However, double null tumors displayed further enhancement of Igf2 (2.5-fold) and Snail (3.8-fold) mRNA levels relative to wild-type tumors (Fig. 2B). Given the suspected roles of Igf2 and snail in HCC progression, this finding may represent a potential explanation for the earlier onset of metastatic lesions in the double null animals (38, 40, 41).

Trp53- and Ink4a/Arf-deficient tumor cells invade in vitro. To further characterize the properties of tumor cells lacking Trp53 alone, or both Trp53 and Ink4a/Arf, we established cell lines from primary liver tumors. We were able to establish one cell line, BL185, lacking Trp53, and two cell lines, MM189 and BL322, lacking both Trp53 and Ink4a/Arf. Immunostaining of these cell lines with antibodies against PyMT and keratin 8 confirmed their origin from transformed hepatic epithelial cells (Fig. 4A). Furthermore, analysis by RT-PCR showed the expression of tv-a and the appropriate expression (or lack thereof) of p16 and p19 in these cell lines (Fig. 4B).

We next characterized the phenotypic features of the MM189 and BL185 cell lines in further detail. We found that both cell lines displayed similar growth rates when cultured in the presence of 10% FBS, suggesting that the loss of the Ink4a/Arf locus does not increase the inherent growth rate of p53-deficient tumor cells in culture (Fig. 4C). Interestingly, both cell lines proliferated in medium containing 0.1% serum without undergoing apoptosis (data not shown). When we measured the ability of these cell lines to grow in soft agar, we found that MM189 cells, which lack both Trp53 and Ink4a/Arf, have a greatly enhanced ability to form colonies relative to the BL185 cell line lacking Trp53 alone (Fig. 4D). Thus, the loss of the Ink4a/Arf locus may enhance the transformation of p53-deficient hepatocytes.

An essential property of metastatic tumor cells is the ability to migrate and invade (45). We therefore measured the ability of the MM189 and BL185 cell lines to migrate and invade using cell culture chamber assays. We found that BL185 cells do not migrate or invade in significant numbers in response to serum as a chemoattractant (Fig. 5A). By contrast, MM189 cells, which lack both Trp53 and Ink4a/Arf, display robust migration and invasion in response to serum (Fig. 5A). Similar results were obtained with the BL322 cell line that also lacks both Trp53 and Ink4a/Arf (data not shown). Thus, our results suggest that the loss of both the Trp53 and Ink4a/Arf tumor suppressor loci is required to induce tumor cell migration and invasion in our model system.

We have previously shown that attenuation of PyMT signaling through the PI3K/Akt signaling pathway impairs lung metastasis in our HCC mouse model (24). To identify whether this occurs, in part, due to reduced cell migration and/or invasion, we measured the migration and invasion activities of MM189 cells treated with the compound rapamycin that inhibits mTOR, a key downstream effector of Akt signaling. We found that rapamycin inhibits tumor cell invasion at a concentration of 10 nmol/L (Fig. 5B). Paradoxically, whereas rapamycin does not affect cell migration at this concentration, it progressively stimulates migration at higher concentrations (Fig. 5B). Thus, in our cell culture model system, rapamycin blocks invasion and uncouples invasion from cell migration.

Trp53-deficient tumor cells colonize the lung. The poor in vitro migration and invasion capabilities of the p53-deficient cell line raised the question of how these cells were able to form metastases in vivo. We therefore determined whether these cells are able colonize the lung after introduction into the bloodstream. Injection of 10⁵ BL185 cells into nude mice via the tail vein led to the development of lung lesions, identified upon sacrifice of the animals, within 28 days after injection (Fig. 5C). A similar phenotype was observed with MM189 cells that lack both Trp53 and Ink4a/Arf (Fig. 5C). Quantification of the number of lung lesions showed that BL185 cells formed more lesions than MM189 cells (Fig. 5D, P < 0.001), although the lesions induced by MM189 were on average of larger size. Thus, HCC tumor cells lacking p53 have the capacity to efficiently colonize the lung, a property that is not further enhanced by the loss of Ink4a/Arf.

The high frequency of intrahepatic and distant metastases in HCC is a critical issue in the management of patients with this disease. Currently, surgical resection remains the only curative option, yet the presence of metastases eliminates this treatment option. Thus, identifying factors that correlate with the development of metastatic disease and tumor recurrence is of great importance. Indeed, recently published work has identified tumor cell–dependent and tumor microenvironment–dependent gene expression changes associated with the progression to a metastatic state in HCC (46, 47).

Therefore, the development of the experimental model systems, both in vivo and in vitro, that allow the dissection of the molecular mechanisms involved in HCC metastasis is of primary importance. We have previously shown the feasibility of modeling metastatic HCC after the somatic and sporadic activation of oncogenes (24). In this paper, we described the refinement of this system to use tissue-specific deletion of tumor suppressor genes. This modification allows us to clearly assess contributions that occur in a cell autonomous manner, and we show here that the effects of Trp53 and Ink4a/Arf deficiency on HCC metastasis occur in a cell autonomous fashion. Furthermore, our data suggest that whereas the loss of p53 function enhances the metastatic potential of HCCs, it may not increase the ability of the tumor cells to migrate and invade. A critical caveat for this latter finding is that to date, we have analyzed only a single cell line derived from a Trp53 null tumor. Importantly, however, the p53-deficient cell line efficiently colonizes the lung after introduction into the blood stream. This is consistent with the hypothesis that tumor cells develop at an early stage many of the traits required for metastasis (48). It also suggests that in p53-deficient HCCs, the rate-limiting step may be invasion through the basement membrane and intravasation to enter the circulation, because once in the bloodstream, these cells home effectively to the lungs and establish metastatic lesions.

In HCC patients, p53 function is frequently compromised by point mutation and loss of heterozygosity of the wild-type allele (8). Olive et al. (49) have recently shown that p53 point mutants commonly found in human tumors display gain-of-function properties. Therefore, it would be intriguing to determine whether the presence of p53 point mutants will enhance the formation of metastasis in our model system and increase the migration and invasion capabilities of cell lines derived from these tumors.

One advantage of our experimental system is the ability to analyze the role of specific domains of oncoproteins and downstream signaling molecules through the rapid generation and introduction of RCAS viruses encoding mutant proteins. We previously exploited this feature to show the importance of PI3K signaling in the induction of metastasis by PyMT (24). Significantly, we have now observed that inhibition of this pathway with rapamycin reduces the invasion of HCC cell lines, suggesting that one potential mechanism by which PI3K signaling contributes to HCC metastasis may be the induction of tumor cell invasion. Importantly, activating mutations in the catalytic subunit of PI3K are common in HCC (50), and therefore, our results suggest that one potential consequence of these activating mutations is the enhancement of cell invasion. Thus, our new results confirm and extend our prior in vivo observations on the role of PI3K signaling in HCC metastasis and highlight the utility of our experimental model systems for the elucidation of the mechanisms involved in HCC metastasis.

Our data also point to potential roles for the p16 and p19 tumor suppressors in HCC progression. Inactivation of the locus encoding these proteins is a common finding in HCC; however, the roles of these tumor suppressors in the initiation and progression of this disease remain unknown (9–12). Our results indicate that these tumor suppressors may act to enhance both tumor initiation and tumor progression. Indeed, analysis of the TP53 and INK4A/ARF loci in human HCC samples suggested that the loss of these loci might act cooperatively in this disease, although the specific roles of p16^INK4A and p14^ARF remain to be elucidated (12). Our data suggest that p16 and p19 may act to constrain tumor growth. Concomitant loss of Trp53 and Ink4a/Arf led to the development of palpable PyMT-induced liver tumors at 4 months of age, a phenotype not observed with Trp53 deletion alone. Our results also suggest that p16 and p19 may influence tumor progression by impeding cell invasion, although the mechanisms by which this occurs remain under investigation.

We also observed increased mRNA levels for the transcription factor snail in p53 and Ink4a/Arf double-deficient tumors relative to p53 wild-type tumors that have low metastatic potential. Intriguingly, snail mRNA levels are elevated in invasive subclones of the noninvasive BL185 cell line,⁵

Thus, our results describe a new system for the generation and analysis of HCC invasion and metastasis. The examination of these processes using our model should significantly aid our understanding of malignant progression in this disease.

Grant support: Liver Scholar Award from the American Liver Foundation; Career Development Award in the Biomedical Sciences from the Burroughs Wellcome Fund; NIH grant CA121171 to B.C. Lewis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The authors thank Anton Berns for Trp53 and Ink4a/Arf conditional mutant mouse strains; Kirsten A. Hubbard and Jennifer Morton for critical reading of the manuscript; and members of the Lewis Lab for helpful discussion. The authors apologize to colleagues whose studies were not cited due to space constraints.