p19^Arf Inhibits the Invasion of Hepatocellular Carcinoma Cells by Binding to C-terminal Binding Protein Free

Worldwide, 626,000 new cases of hepatocellular carcinoma (HCC) are diagnosed each year, with a survival rate of <5% and an average survival of <1 year after diagnosis (1). To date, surgery remains the only effective approach for treatment of small HCCs; unfortunately, there is currently no reliable therapy for most patients with advanced or metastatic HCC that has spread to the lymph nodes, portal vein, or lungs (2).

Epidemiologic and molecular studies have determined that HCC development is associated with chronic hepatitis B virus and hepatitis C virus infection, alcoholic cirrhosis, metabolic disorders, and environmental factors, such as aflatoxin B1 exposure (3). Furthermore, genetic alterations resulting in the activation of the MYC oncogene, inactivation of tumor suppressor genes, such as TP53, RB1, and INK4A/ARF, and alterations that stimulate the Wnt signaling pathway are particularly important in hepatocarcinogenesis (4). However, little is known regarding the genes and molecular pathways that influence metastasis. Recent in vivo models and gene expression analyses have begun to shed some light on the tumor and host factors regulating metastasis in HCC (5–10).

To elucidate the mechanisms of hepatocarcinogenesis, several mouse models have been generated through the expression of oncogenes or viral genes, the inactivation of tumor suppressor genes, and chemical carcinogenesis. Given the effect of metastasis on the clinical management of HCC patients, models that allow the dissection of metastasis are of great importance. We have recently described a novel HCC mouse model induced by the somatic and sporadic activation of oncogenes, specifically in the liver (8). We have further refined this model to include liver-specific deletion of the Trp53 and Ink4a/Arf tumor suppressor loci (11). Our recent data showed that liver-specific Trp53 deletion induces lung metastases, the formation of which can be accelerated by concomitant deletion of Ink4a/Arf. Furthermore, we showed that mouse HCC cell lines lacking both Trp53 and Ink4a/Arf displayed enhanced migration and invasion capabilities, suggesting that the Ink4a/Arf locus may play a role in regulating these processes in liver cancer cells.

The Ink4a/Arf locus encodes two distinct tumor suppressors, the cyclin-dependent kinase (Cdk) inhibitor p16^Ink4a and a protein translated from an alternative reading frame, Arf (p14^ARF in human and p19^Arf in mouse; ref. 12). Through the inhibition of cyclin D/cdk4, p16 maintains pRb in the hypophosphorylated state and regulates the G₁ transition of the cell cycle (13). On the other hand, p19^Arf inhibits the p53-ubiquitinating function of Mdm2, thereby stabilizing the p53 transcription factor, which regulates genes mediating G₁ cell cycle arrest and apoptosis (14). Arf also has p53-independent functions. It mediates a p53-independent cell cycle arrest by targeting both E2F1 and c-Myc to the nucleolus and preventing their transcription activation functions and also induces p53-independent apoptosis by targeting the transcription factor C-terminal binding protein 2 (CtBP2) for degradation (15–17). CtBP transcription factors have been previously shown to induce an epithelial-to-mesenchymal transition (EMT) and to stimulate cell migration (18–20). Thus, regulation of CtBP activity by Arf may represent an important tumor suppressor function. Significantly, the INK4A/ARF locus is frequently inactivated in HCC, yet its role in the pathogenesis of this disease remains unclear (21–23).

Guided by these findings in HCC and our mouse model, we sought to identify mechanisms by which Arf may regulate HCC metastasis. Using mouse HCC cell lines, we found that RNA interference (RNAi)–mediated loss of Arf enhances cell invasion, and conversely, reintroduction of Arf into cell lines deficient for both Trp53 and Ink4a/Arf inhibits tumor cell invasion. The ability of Arf to inhibit invasion is independent of its interaction with Mdm2 and nucleolar localization. However, it is dependent on Arf's ability to bind to CtBP. Consistent with this, RNAi-mediated depletion of CtBP1 or CtBP2 reduces cell invasion, and ectopic CtBP2 expression enhances tumor cell migration and invasion. Thus, our findings show a new role for the Arf tumor suppressor in HCC progression, one that may be applicable to other tumor types.

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 DMEM supplemented with 10% fetal bovine serum (FBS) and antibiotics. After 48 h, nonadherent cells were removed and adherent cells were washed with PBS and refed with fresh medium containing FBS.

To generate cells with knockdown of p19^Arf, BL185 HCC cells were infected with MLP retroviral vectors encoding a short hairpin RNA (shRNA) targeted to the Arf-specific exon 1β of the Ink4a/Arf locus (24). Infected cells were then placed in DMEM containing 8 μg/mL puromycin (EMD Biosciences). Resistant cells were then analyzed by immunoblot.

To generate cell lines expressing either wild-type or mutant p19^Arf proteins, HCC cell lines were infected with pBabe-puro retroviruses and then placed in DMEM containing 8 μg/mL puromycin (EMD Biosciences). Resistant cells were then analyzed by immunoblot and immunofluoresence.

Cells with ectopic CtBP2 expression were generated by transfecting MM189 cells with pcDNA3-V5-CtBP2 or pcDNA3-V5-CtBP2 and selection of transfected cells in 4 mg/mL G418 (EMD Biosciences).

RNAi-mediated depletion of CtBP1 and CtBP2 was achieved by infecting HCC cells with pLKO-based lentiviruses encoding shRNAs targeting the appropriate mRNA. Infected cells were selected in 8 μg/mL puromycin. To generate knockdown in HCC cells expressing the p19^ArfL46D mutant, which were already puromycin resistant, infection with shRNA containing lentiviruses was performed twice within 24 h. Knockdown was confirmed by immunoblot.

Immunoblotting. Cells were collected and lysed in radioimmunoprecipitation assay buffer containing protease inhibitors. Protein concentration was determined with bicinchoninic acid assay kit (Pierce), and equal amounts of protein were loaded per lane of a polyacrylamide gel. After protein transfer, nitrocellulose membranes were blocked with 5% dry milk in Tris-buffered saline, 0.1% Tween 20 (TBS-T) for 1 h. Primary antibodies were incubated overnight at 4°C in 5% bovine serum albumin (Sigma) in TBS-T, and secondary antibodies were incubated for 1 h at room temperature. Chemiluminescence was performed with Supersignal reagent (Pierce). Primary antibodies are as follows: anti-p19Arf (Abcam), anti-CtBP2 (Upstate Biotechnology), anti–β-actin (1:5,000; Santa Cruz), α-catenin (1:1,000; Becton Dickinson), anti–β-catenin (1:1,000; Becton Dickinson), anti–γ-catenin (1:1,000; Becton Dickinson), anti–E-cadherin (1:1,000; Becton Dickinson), anti–α smooth actin clone A14 (1:1,000; Sigma), anti–N-cadherin (1:1,000; Becton Dickinson), anti-fibronectin (1:1,000; Santa Cruz), anti-vimentin (1:1,000; Lab Vision), and anti-V5 (1:2,000; Covance).

Cell proliferation assay. Cells (10³) were seeded in quadruplicate onto collagen-coated 96-well plates 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 assay. Soft agar assays were performed as previously described (25). The number of colonies larger than 100 μm in diameter present within 20 microscopic fields was counted under a light microscope.

Migration and invasion assays. Cells (2.5 × 10⁴) in 0.5 mL of serum-free DMEM were plated into either control or Matrigel-coated invasion inserts (Becton Dickinson). 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 10 microscope fields at 50× magnification per insert. The percentage invasion was calculated as the number of invading cells divided by the number of migrating cells. The percentage invasion was then normalized to get the invasion index, with the value for the control cell population set to 1. All experiments were performed in duplicate and repeated thrice at a minimum. Data are shown for representative experiments.

Immunofluorescence. Cells on collagen-coated culture slides were fixed with 4% paraformaldehyde for 10 min and permeabilized with 0.1% Triton X-100 in PBS for 15 min at room temperature. After blocking with 10% goat serum in PBS for 1 h, cells were incubated with primary anti-p19^Arf antibody (1:200; Abcam) at 4°C overnight, followed by incubation with appropriate secondary antibodies and rhodamine-phalloidin at room temperature. Slides were mounted with antifade and viewed under a fluorescence microscope.

p19^Arf inhibits cell invasion. We have previously shown that mouse HCC cell lines lacking both Trp53 and Ink4a/Arf have enhanced migration and invasion activity in Boyden chamber assays compared with a cell line that lacks Trp53 alone (11). To determine whether loss of p19^Arf influences tumor cell migration and invasion, we infected the BL185 HCC cell line, which lacks Trp53 but retains functional Ink4a/Arf, with retroviral vectors encoding shRNA against p19^Arf or luciferase as a control (24). Knockdown of p19^Arf was confirmed by immunoblot (Fig. 1A). We then measured the migration and invasion activities of these cells using Boyden chamber assays. We found that depletion of Arf resulted in a 1.5-fold increase in the number of invading cells (Fig. 1C), a result confirmed in independent infections (shRNA 1 and 2; Fig. 1C), and multiple replications of the invasion assays. These data indicate that acute loss of Arf stimulates the in vitro invasion activity of Trp53 null HCC cells, albeit to a lesser extent than that observed in HCC cells with genetic ablation of Trp53 and Ink4a/Arf (data not shown).

To ascertain whether loss of p19^Arf was required for the invasive phenotype of HCC cells lacking Trp53 and Ink4a/Arf, double-null HCC cell lines MM189 and BL322 were infected with a retroviral vector encoding mouse p19^Arf or empty vector as a control and expression was confirmed by immunoblot (Fig. 1B). By immunohistochemistry with an anti-p19^Arf antibody, positive signals were detected in nearly 70% of retrovirally infected MM189 cells, suggesting that p19^Arf was reexpressed in most, but not all, cells (data not shown). Measurement of cell migration and invasion in Boyden chamber assays showed that p19^Arf expression reduced the invasion activity of MM189 cells by 80% (Fig. 1D). p19^Arf expression also led to a more modest decrease in cell migration in most, but not all, experiments (Fig. 1D and data not shown). Consistent with this finding, p19^Arf expression did not significantly impair cell migration in a wound healing assay (data not shown). Thus, p19^Arf specifically inhibits the invasion of HCC cells.

Analysis of vector controls and p19^Arf-expressing MM189 cells showed that p19^Arf expression did not affect the proliferation of these cells (Fig. 1E). Likewise, analysis of cell transformation by soft agar colony formation showed no effect of p19^Arf expression (Fig. 1F and Supplementary Fig. S1). Thus, in MM189 cells, p19^Arf seems to regulate cell invasion without affecting any other features related to cell transformation. To confirm that this phenomenon was not restricted to a single HCC cell line, we performed the migration and invasion assays on BL322 HCC cells infected with either p19^Arf-expressing retrovirus or vector controls. We found a similar effect on migration and invasion in this cell line, suggesting that this phenotype is not cell line specific (Fig. 1G).

Tumor cell migration and invasion are often associated with reorganization of the actin cytoskeleton and a phenomenon called EMT (26, 27). To investigate whether p19^Arf-mediated inhibition of cell invasion occurs via the reversal of an EMT, we analyzed p19^Arf-expressing cells (MM189-WT) and vector controls (MM189-PB) by morphology and expression of epithelial and mesenchymal markers. Immunoblot experiments showed that MM189 cells expressed epithelial proteins, such as α-catenin, β-catenin, γ-catenin, and E-cadherin, and the expression of p19^Arf did not enhance the levels of these proteins (Fig. 2A). Examination of E-cadherin localization by immunofluorescent staining did not reveal any significant differences between MM189-PB and MM189-WT cells (data not shown). Consistent with these findings, MM189 cells did not express mesenchymal markers, such as vimentin, a feature unaffected by p19^Arf expression (Fig. 2A). Analysis of mRNA levels for epithelial and mesenchymal markers showed that these were similarly unaffected by p19^Arf expression, although, interestingly, the HCC cells expressed vimentin mRNA, indicating that regulation occurred at the posttranscriptional level (Supplementary Fig. S2). In agreement with the above data, analysis of the morphology of MM189-PB cells and MM189-WT cells showed that the expression of p19^Arf did not alter cell morphology (Fig. 2B). Finally, analysis of the actin cytoskeleton by phalloidin staining did not show any differences between p19^Arf-expressing cells and vector controls (Fig. 2C).

Inhibition of cell invasion by p19^Arf requires the CtBP interaction domain. To determine the functional domains of p19^Arf that are required for inhibition of cell invasion, we introduced retroviral vectors encoding various p19^Arf mutant proteins into MM189 cells. The introduced mutants were Δ2-10, which impairs Mdm2 binding, Δ8-32, which impairs Mdm2 binding and nucleolar localization, Δ46-53, which blocks CtBP binding, and the L46D point mutant, which also blocks CtBP binding. The locations of these mutations are shown schematically in Fig. 3A. The amino terminal deletions also affect Arf's interaction with c-Myc, E2F1, and Foxm1b. We confirmed the expression of the p19^Arf mutants by immunoblotting with Arf-specific antibodies (Fig. 3B) and showed that the majority of cells expressed the mutant p19^Arf proteins by immunostaining (data not shown).

Introduction of the Δ2-10 and Δ8-32 mutants into MM189 cells led to the efficient inhibition of cell invasion (Fig. 3C). However, introduction of the Δ46-53 and L46D mutants failed to appreciably impair cell invasion, particularly compared with the number of migrating cells (Fig. 3C). This resulted in an increased invasion index for cells expressing these two p19^Arf mutants (Fig. 3C). To rule out the possibility that some effects were due to the random integration of retroviruses, retroviral infections and invasion assays were repeated with independent retroviral stocks, several pools of infected cells with varying levels of p19^Arf expression examined, and consistent results obtained. As was observed with expression of wild-type p19^Arf, expression of the Arf mutants did not affect proliferation of MM189 cells (data not shown). Thus, p19^Arf-mediated inhibition of tumor cell invasion is independent of Mdm2 binding and nucleolar localization but is dependent on the CtBP interaction domain. As was observed after reintroduction of wild-type p19^Arf, introduction of mutant Arf proteins did not alter the expression of epithelial and mesenchymal proteins (Fig. 2A). Thus, p19^Arf-mediated inhibition of cell invasion does not occur via the reversal of an EMT.

CtBP regulates tumor cell invasion. Our finding that the p19^Arf Δ46-53 and L46D mutants fail to impair tumor cell invasion suggested that functional inhibition of CtBP proteins by p19^Arf might be critical in this process. We therefore determined whether inactivation of CtBP reduces tumor cell invasion. Lentiviral shRNA vectors targeting CtBP1 or CtBP2 or vector control were introduced into MM189 cells, the cells selected for puromycin resistance, and their invasion activity was measured. By immunoblotting, we identified two shRNAs targeting each of the CtBP family members that displayed efficient knockdown (Fig. 4A). Measurement of the proliferation of CtBP knockdown cells showed that loss of CtBP proteins did not affect cell proliferation (Supplementary Fig. S3). Analysis of the invasion activity showed that cells with CtBP knockdown had lower invasion activity relative to vector controls (Fig. 4C). We next determined whether depletion of CtBP1 and CtBP2 in MM189 cells expressing the Arf L46D mutant would similarly result in reduced cell invasion. We introduced shRNAs targeting either CtBP1 or CtBP2 into these cells and confirmed knockdown by immunoblotting with anti-CtBP antibodies (Supplementary Fig. S4). We found that depletion of either family member led to a reduced cell invasion index (Fig. 4D). Thus, CtBP family members are required for HCC tumor cell invasion.

To determine whether ectopic expression of CtBP could enhance cell invasion, we introduced a V5-tagged CtBP2 expression construct or vector control into MM189 cells. Detection of the introduced CtBP was confirmed by immunoblotting with anti-V5 antibodies (Fig. 4B). Interestingly, the levels of the ectopic CtBP seemed to be less than the endogenous levels of CtBP2 as determined by immunoblotting with CtBP2-specific antibodies (data not shown). In cell invasion assays, exogenous CtBP2 expression led to a 1.5-fold increase in invasion activity relative to vector controls, supporting a role for CtBP2 in tumor cell invasion (Fig. 4E). Importantly, introduction of a NADH binding–deficient CtBP2 mutant (G189A; ref. 28) failed to stimulate cell invasion (Fig. 4E), although it was expressed at levels similar to wild-type CtBP2 (Fig. 4B). Interestingly, in MM189 cells with enforced p19^Arf expression, ectopic CtBP2 failed to significantly enhance tumor cell migration and invasion, suggesting a dominant effect of p19^Arf on the invasion phenotype (data not shown).

The Ink4a/Arf locus is regarded as one of the most important antitumoral defenses in mammalian systems, and the two protein-coding genes at this locus are frequently independently inactivated in HCC. Studies of human tumors and mouse models have suggested a role for the Arf tumor suppressor in constraining tumor initiation (23, 29, 30). We have recently shown that deletion of the Ink4a/Arf locus enhances the progression of liver tumors induced by polyoma middle T antigen and Trp53 deletion, consistent with findings in a skin carcinogenesis mouse model that indicated that Arf may additionally play a role in tumor progression (31). Furthermore, we showed that mouse HCC cell lines that lacked both Trp53 and Ink4a/Arf displayed enhanced migration and invasion capabilities in cell culture assays compared with a cell line that lacks Trp53 alone (11). We have now shown that p19^Arf specifically impairs tumor cell invasion without affecting either proliferation or soft agar colony formation, pointing to a potential role for Arf in regulating tumor metastasis.

p19^Arf induces cell cycle arrest in a p53-dependent manner and relays signals to p53 after oncogene-induced stress that stimulates the onset of cellular senescence, a potent antitumorigenic checkpoint. However, accumulating data indicate that Arf may additionally have p53-independent tumor suppressor functions (32). Our data show that Arf-mediated inhibition of tumor cell invasion is independent of its interaction with Mdm2, suggesting that regulation of tumor invasion may be another p53-independent tumor suppressor function of Arf. Importantly, this function is also independent of Arf's ability to bind to other interacting proteins, such as E2F1, c-Myc, and Foxm1b, as deletion mutants that interfere with the ability of Arf to bind to these proteins still effectively block invasion.

Instead, our data show that the effect of p19^Arf on tumor cell invasion occurs via a p53-independent mechanism involving CtBP. Mutant Arf proteins that fail to bind to CtBP do not inhibit cell invasion, and shRNA-mediated ablation of CtBP reduces invasion in HCC cells expressing CtBP binding–deficient Arf proteins. Guo et al. have previously shown that genetic disruption of Arf in mouse embryonic fibroblasts (MEF) enhances cell motility in a p53-dependent manner through the stimulation of Rac1 activity (33). Our data indicate that Arf expression in HCC cells does not alter cell morphology or the actin cytoskeleton, as assessed by phalloidin staining. This suggests that the activity of Rac1 or other Rho family GTPases involved in cytoskeletal remodeling may not be affected by Arf expression, although this has not been formally tested.

Additional studies showed that activated Rho GTPases could stimulate invasion by Trp53 null, but not Arf null, MEFs, suggesting that p19^Arf regulates cell migration but not invasion (34). Thus, our experiments are the first to show that Arf inhibits tumor cell invasion via a p53-independent pathway. The differences between the previous studies and ours may reflect the different cell types used — epithelial-derived tumor cell lines versus transformed mouse fibroblasts. Importantly, in a related study, we have also shown that p14^Arf inhibits the hypoxia-induced migration of H1299 human lung carcinoma cells in a p53-independent and CtBP-dependent manner (35), and previous studies have indicated that CtBP can regulate the migration of this cell line, although cell invasion was not assayed in these studies (20). Significantly, the leucine 46 residue, which is critical for the Arf-CtBP interaction, is one of six invariant residues conserved across several species, including human, mouse, chicken, pig, and opossum (32). Thus, the Arf-CtBP interaction likely mediates a highly conserved tumor suppressor function of Arf and may have implications for the dissemination of human tumors.

CtBPs play a critical role in cellular regulation by binding to a variety of transcriptional repressors important for development and tumorigenesis (36). CtBPs have also been suggested to play a role in inhibiting anoikis (19). However, the mechanisms by which Arf inhibits CtBP function remain unclear. Our findings suggest that Arf may impede CtBP function, yet the mechanisms by which this might occur remain unclear, although potential links are beginning to emerge. Although we have observed that Arf expression does not influence the steady-state levels of CtBP, Arf, but not a mutant defective for CtBP binding, stimulates the degradation of CtBP in response to cellular stresses, such as UV irradiation (Supplementary Fig. S5; ref. 17). Therefore, it is possible that Arf stimulates the degradation of CtBP under stress conditions relevant during cell migration and invasion, such as detachment from the basement membrane. In addition, recent findings from the Grossman laboratory indicate that Arf can inhibit CtBP-mediated transcriptional repression of a target promoter.⁶

CtBP has also been previously shown to induce EMT (18, 19). EMT, first identified as a critical process during normal embryonic development, involves the down-regulation of epithelial cell markers, such as E-cadherin, and the induction of mesenchymal markers, such as vimentin, and is postulated to be involved in tumor cell invasion and metastasis (27). However, our data indicate that our HCC cell lines invade without undergoing EMT and E-cadherin protein levels are not significantly altered in cells by reintroduction of p19^Arf. A similar phenomenon has been recently described by Christofori and colleagues in connection to tumor cell invasion stimulated by the mucin-like protein podoplanin (37). EMT has been postulated to a transient process. Thus, an alternate possibility is that Arf may prevent the transient induction of CtBP and the consequent transient reduction of E-cadherin levels, in response to proinvasion stimuli that would allow a tumor cell to initiate invasion. A similar model has been postulated for metastasis of colorectal carcinomas (38).

Thus, our findings indicate a potential new role for Arf in tumor invasion. Further exploration of the pathways involved in this process may yield promising new targets for therapeutic intervention.

Grant support: Burroughs Wellcome Fund Biomedical Sciences Career Development award (B.C. Lewis), American Liver Foundation Liver Scholar award (B.C. Lewis), NIH grants CA121171 (B.C. Lewis) and CA89548 (S.R. Grossman), and Our Danny Fund (S. Paliwal).

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.

We thank Nabeel Bardeesy for mutant p19^Arf retroviral constructs, Scott Lowe for the p19^Arf-specific shRNA construct, and members of the Lewis and Grossman laboratories for thoughtful discussion.