A cell-penetrating ARF peptide inhibitor of FoxM1 in mouse hepatocellular carcinoma treatment

The mouse Foxm1 transcription factor is required for hepatic tumor progression.

We previously showed that conditional deletion of Foxm1 in hepatocytes prior to DEN/PB liver tumor induction is sufficient to inhibit hepatic tumor initiation (8, 31). Here, we determined that Foxm1 is required for hepatic tumor progression. In order to do so, we used the IFN-α/β–regulated Mx-Cre transgene (36) to conditionally knock out (CKO) or delete the Foxm1^fl/fl targeted allele in preexisting liver tumors induced by the DEN/PB exposure (8). We induced HCC in mice with 30 weeks of DEN/PB exposure, and then induced Mx-Cre expression with synthetic double-stranded RNA (dsRNA) to CKO the Foxm1^fl/fl targeted allele. Mice were then subjected to an additional 10 weeks of PB tumor promotion protocol (Figure 1A). To achieve long-term BrdU labeling of the liver tumors, the mice were then given drinking water containing 1 mg/ml of BrdU for 4 days (8, 37). The Mx-Cre transgene efficiently deleted the Foxm1^fl/fl targeted allele, as evidenced by the absence of detectable nuclear staining of FoxM1 protein in liver tumors of dsRNA CKO Mx-Cre Foxm1^–/– mice compared with control liver tumors (Figure 1, B–D).

We used liver sections stained with H&E to determine the number of tumors per square centimeter of liver tissue (Figure 1, E–G). To calculate the area or size of liver tumors, we used micrographs of H&E-stained liver tumor sections taken with an Axioplan 2 microscope (Zeiss) and the AxioVision program (version 4.3; Zeiss). After 40 weeks of DEN/PB exposure, control mice displayed hepatic adenomas and HCCs that were larger than 2 mm² in size (Table 1). Deletion of Foxm1 in preexisting hepatic tumors in dsRNA CKO Mx-Cre Foxm1^–/– mice caused a significant reduction in the number of liver tumors larger than 2 mm² in size compared with control liver tumors after 40 weeks of DEN/PB exposure (Table 1). We next measured tumor cell proliferation by determining the number of hepatic tumor cells that immunostained positive for BrdU incorporation (Figure 1, H–J). Compared with controls, dsRNA CKO Mx-Cre Foxm1^–/– mice displayed an 80% reduction in the number of liver tumor cells that stained positive for BrdU after 40 weeks of DEN/PB treatment (Figure 1K). Taken together, these results indicate that deletion of Foxm1 in preexisting liver tumors significantly diminished proliferation and growth of hepatic cancer cells.

The cell-penetrating WT ARF_26–44 peptide targets the endogenous mouse FoxM1 protein to the nucleolus of hepatic tumor cells.

We previously synthesized a cell-penetrating ARF_26–44 peptide fused to 9 N-terminal D-Arg residues (32, 33), which efficiently transduced into osteosarcoma U2OS cells and inhibited FoxM1b transcriptional activity (8). Treatment of U2OS cells with 12 μM of (D-Arg)₉-ARF_26–44 (WT ARF_26–44) peptide fluorescently tagged with tetramethylrhodamine (TMR) targeted nuclear GFP-FoxM1b green fluorescence colocalized with red WT ARF_26–44 peptide fluorescence in the nucleolus (Figure 2, C and D). In contrast, GFP-FoxM1b protein remained nuclear in U2OS cells when treated with a TMR fluorescently tagged mutant (D-Arg)₉-ARF_37–44 (mutant ARF_37–44) peptide (Figure 2E), which lacked the amino acids 26–37 required to interact with the FoxM1b protein (8). Because Arg-rich sequences are sufficient for nucleolar targeting (38), the mutant ARF_37–44 peptide fluorescence also localized to the nucleolus of U2OS cells (Figure 2F). We did not observe any signal in the absence of the ARF peptide (data not shown).

Dose-response curve determined that only i.p. injection of 5 mg/kg body weight or greater of the TMR fluorescently tagged WT ARF_26–44 peptide was detectable in cytoplasm and nucleolus of hepatocytes and in hepatic mesenchymal cells at 24 hours after administration (Figure 2G; see Methods). After 32 weeks of DEN/PB liver tumor induction, we subjected Foxm1^fl/fl mice to daily i.p. injections of 5 mg/kg body weight of the WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 4 weeks and of WT ARF_26–44 peptide for 8 weeks (Figure 2A). After 4 weeks of treatment with TMR fluorescently tagged ARF peptides, laser confocal microscopy of paraffin-embedded mouse liver tumor sections revealed that ARF peptide fluorescence localized to the hepatocyte cytoplasm and nucleolus (Figure 2, H and I) and was uniformly distributed throughout the liver parenchyma (data not shown). The FoxM1 protein staining in WT ARF_26–44 peptide–treated liver tumor sections was partially localized to the nucleolus in hepatic tumor cells (Figure 2L, black arrows), which was similar to the immunostaining pattern of the nucleolar protein nucleophosmin (NPM; Figure 2J, black arrows). In contrast, mutant ARF_37–44 peptide– or PBS-treated liver tumor cells displayed only nuclear FoxM1 staining (compare Figure 2K and Figure 2M, red arrows). These studies demonstrate that the WT ARF_26–44 peptide reduces in vivo function of FoxM1 by partially targeting the endogenous FoxM1 protein to the nucleolus of hepatic tumor cells.

WT ARF_26–44 peptide diminishes proliferation and size of liver tumors.

We next determined the number of hepatic tumor cells that incorporated BrdU in mice treated with WT ARF_26–44 peptide, mutant ARF_37–44 peptide, or PBS. Significant reduction in BrdU incorporation was found in liver tumors that had been treated with the WT ARF_26–44 peptide for 4 or 8 weeks compared with mouse liver tumors treated with mutant ARF_37–44 peptide or PBS (Figure 3, A–K). Compared with control treatment, treatment with the WT ARF_26–44 peptide for 8 weeks significantly reduced tumor growth and prevented development of HCCs larger than 2 mm² in size (Table 1). These results indicate that treatment with the WT ARF_26–44 peptide is an effective method with which to reduce proliferation and growth of HCCs.

We previously demonstrated that tumor resistance in Foxm1^–/– hepatic tumors was associated with persistent nuclear accumulation of the CDKI protein p27^Kip1 (8). WT ARF_26–44 peptide–treated HCC cells displayed increased nuclear levels of the p27^Kip1 protein, which were similar to those found with dsRNA CKO Mx-Cre Foxm1^–/– liver tumors (Figure 4, B and E). In contrast, p27^Kip1 immunostaining was predominantly cytoplasmic in mutant ARF_37–44 peptide– or PBS-treated mouse HCCs (Figure 4, A, C, D, and F). These studies indicate that the WT ARF_26–44 peptide is effective in reducing FoxM1 function in vivo and that nuclear accumulation of p27^Kip1 protein was associated with reduced hepatic tumor proliferation.

WT ARF_26–44 peptide causes selective apoptosis of hepatic tumor cells.

Analysis of H&E-stained liver tumor sections from mice treated with the WT ARF_26–44 peptide revealed that many of the hepatic adenomas and HCC tumor cells stained red and exhibited disruption of nuclear membrane, which is indicative of apoptosis (Figure 5, A–F). These red-staining cells were found neither in the surrounding normal liver tissue (Figure 5, A–F) nor in hepatic tumors from mice treated with either the mutant ARF_37–44 peptide or PBS (Figure 5, G–L). Furthermore, these apoptotic tumor cells were not apparent in deficient livers in dsRNA CKO Mx-Cre Foxm1^–/– mice (Figure 1, E–G), a finding consistent with our previous tumor studies with Alb-Cre Foxm1^–/– livers (8).

We used a fluorescence-based TUNEL assay to determine that mouse HCC cells treated with WT ARF_26–44 peptide exhibited a significant, 22% increase in apoptosis (Figure 6, A, B, and E). In contrast, very few apoptotic HCC cells were found after treatment with mutant ARF_37–44 peptide or PBS (Figure 6, C–E). Immunostaining of liver tumor sections with proteolytically cleaved activated caspase-3 protein confirmed this selective apoptosis of mouse HCC cells treated with WT ARF_26–44 peptide with no proapoptotic staining in the adjacent normal liver tissue (Figure 6, F–H). These studies show that the WT ARF peptide selectively induces apoptosis of HCC cells without damaging adjacent normal hepatocytes.

DEN/PB treatment induced highly proliferative HCCs in Arf^–/– Rosa26-FoxM1b Tg mice that were responsive to WT ARF_26–44 peptide treatment.

We previously showed that the ARF tumor suppressor targets the FoxM1b protein to the nucleolus and inhibits its transcriptional function (8). Furthermore, we showed that increased expression of the human FoxM1b cDNA in Rosa26-FoxM1b Tg mice (34) stimulated development and progression of prostate cancers in both TRAMP/Rosa26-FoxM1b and LADY/Rosa26-FoxM1b double-Tg mice (35). In order to develop a new genetic model of HCC that is highly dependent on the FoxM1b transcription factor, we crossed the Rosa26-FoxM1b Tg mice into the Arf^–/– mouse background. After 33 weeks of DEN/PB treatment, Arf^–/– Rosa26-FoxM1b Tg mice developed highly proliferative HCCs, and their HCC cells displayed a proliferation rate of 6,000 BrdU-positive cells per square millimeter tumor (Figure 7J), which is approximately 30-fold greater than that observed in DEN/PB-induced HCCs in WT mice (Figure 3K; 200 BrdU-positive cells per mm² tumor). The DEN/PB-treated Arf^–/– Rosa26-FoxM1b Tg livers also exhibited development of necrosis and fibrosis/cirrhosis (data not shown). These HCC tumor–bearing Arf^–/– Rosa26-FoxM1b Tg mice were subjected to daily treatment with either the WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 4 weeks. In Arf^–/– Rosa26-FoxM1b Tg mice, WT ARF_26–44 peptide treatment resulted in a significant, 84% reduction in BrdU labeling of HCC cells compared with treatment of these mice with either mutant ARF_37–44 peptide or PBS (Figure 7, A–C and J). Red-staining HCC cells with disruption of nuclear membrane indicative of apoptosis were found in H&E-stained liver tumor sections from Arf^–/– Rosa26-FoxM1b Tg mice treated with the WT ARF_26–44 peptide but not from those treated with mutant ARF_37–44 peptide or PBS (Figure 7, D–F). We used a fluorescence-based TUNEL assay to determine that Arf^–/– Rosa26-FoxM1b Tg HCC cells treated with WT ARF_26–44 peptide exhibited a 42% increase in apoptosis (Figure 7K), which was twice as high as in liver tumors from WT mice (Figure 6E). Furthermore, TUNEL-positive cells were restricted to the HCC region (white arrowheads) and were not detected in adjacent normal liver tissue (Figure 7I). In contrast, very few apoptotic HCC cells were found after treatment of Arf^–/– Rosa26-FoxM1b Tg mice with mutant ARF_37–44 peptide or PBS (Figure 7, G, H, and K). These Arf^–/– Rosa26-FoxM1b Tg liver tumor studies show that the WT ARF_26–44 peptide is effective in diminishing BrdU labeling of highly proliferative HCC cells and selectively induces apoptosis of HCC cells in these mice without damaging adjacent normal liver tissue.

WT ARF_26–44 peptide inhibits angiogenesis of the HCC region.

Angiogenesis is critical to mediating HCC growth, and the endothelial cells of new HCC capillaries exhibit expression of the CD34 protein (39–41). Abundant CD34 staining was found in endothelial cells of HCC regions in PBS or mutant ARF_37–44 peptide–treated mice (Figure 8, A and B) and from dsRNA CKO Mx-Cre Foxm1^–/– mice (Figure 8D). In contrast, expression of the CD34 protein was not detected in the HCC region from WT ARF_26–44 peptide–treated mice (Figure 8C). These results suggest that WT ARF_26–44 peptide treatment was preventing HCC angiogenesis, which was likely caused by apoptosis of new HCC endothelial cells (Figure 6B; small apoptotic cells). In order to determine whether WT ARF_26–44 peptide induces apoptosis of endothelial cells, we treated human microvascular endothelial cells (HMEC-1 cells) for 48 hours with 100 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide or with PBS and then assayed for apoptosis as described in Methods. This analysis revealed that WT ARF_26–44 peptide treatment induced a significant increase in apoptosis of HMEC-1 cells compared with treatment with mutant ARF_37–44 peptide or PBS (Figure 8E). These studies suggest that WT ARF_26–44 peptide is able to induce apoptosis of endothelial cells, which contributes to WT ARF peptide–mediated reduction in HCC angiogenesis.

Reduced levels of the antiapoptotic survivin protein contribute to WT ARF_26–44 peptide–induced HCC apoptosis.

We have shown that FoxM1 regulates transcription of survivin (26), which complexes with aurora B kinase to mediate its proper localization during mitosis (42–44). Survivin is also a member of the inhibitor of apoptosis (IAP) protein family and is selectively overexpressed in tumor cells to prevent their apoptosis (45–48). Mutant ARF_37–44 peptide– or PBS-treated liver tumors displayed abundant nuclear and cytoplasmic staining of survivin protein (Figure 8, E and F), and survivin expression was restricted to the mouse HCC region (data not shown). Nuclear levels of survivin were diminished in HCC regions from WT ARF_26–44 peptide–treated and dsRNA CKO Mx-Cre Foxm1^–/– mice (Figure 8, G and H). Western blot analysis revealed that Foxm1^–/– liver tumors displayed a 60% decrease in expression of survivin protein (Figure 8, I and J), and no apoptosis was detected in these FoxM1-deficient liver tumors (Figure 1, E–G). A more drastic, 90% decrease in survivin protein levels was found in hepatic tumors from WT ARF_26–44 peptide–treated mice (Figure 8, H–J). Our hepatoma data presented below (Figure 9) support the hypothesis that WT ARF_26–44 peptide treatment results in hypomorphic levels of FoxM1 activity, which reduces expression of mitotic regulators to levels that are insufficient to properly execute mitosis leading to apoptosis, whereas depleting FoxM1 levels leads to mitotic arrest.

The nucleolar nucleophosmin/B23 (NPM/B23) protein (49, 50) and the p53 negative regulator, Mdm2 protein (51), associated with the ARF tumor suppressor protein through ARF amino acid sequences 1–15 and 26–37, which partially overlapped with our WT ARF_26–44 peptide sequence. Despite this partial overlap, Western blot analysis enabled us to determine that neither WT ARF_26–44 peptide treatment nor Foxm1 deficiency altered expression of NPM, p53, or Mdm2 proteins in liver tumor extracts (Figure 8, J and K, and data not shown), suggesting that the ARF_26–44 peptide sequence was insufficient to influence their expression levels in vivo. The proapoptotic Bcl-2 family member p53-upregulated modifier of apoptosis (PUMA) is a transcriptional target gene of p53 that is essential for both p53 transcriptional and cytoplasmic-dependent apoptosis (52–55). Western blot analysis showed that neither WT ARF_26–44 peptide nor mutant ARF_37–44 peptide treatment caused significant change in expression of PUMA in liver tumor extracts (Figure 8K), suggesting that this apoptosis of HCC cells did not involve the p53/PUMA proapoptotic pathway.

WT ARF peptide–induced apoptosis of human hepatoma HepG2 cells correlates with diminished expression of survivin, PLK1, and aurora B kinase.

We used the TUNEL assay to determine that human hepatoma HepG2 cells (Figure 9, A–E), PLC/PRF/5 cells that express mutant p53 protein, and p53-deficient Hep3B cells exhibited 50% apoptosis after 24 hours of treatment with 25 μM of WT ARF_26–44 peptide (Figure 9E), whereas only low levels of apoptosis were detected in these cells following treatment with mutant ARF_37–44 peptide or PBS (Figure 9E). Diminished levels of p53 protein through p53 siRNA silencing of HepG2 cells did not influence apoptosis in response to WT ARF_26–44 peptide treatment (Figure 9F). In addition, p53 protein levels were unaltered in HepG2 cells after 24 hours of treatment with WT ARF_26–44 or mutant ARF_37–44 peptide (Figure 9F). Furthermore, protein expression of the p53 downstream proapoptotic target PUMA was unchanged in HepG2 cells in response to increasing concentrations of the WT ARF_26–44 peptide (Figure 9I). These results suggest that WT ARF_26–44 peptide–induced apoptosis was independent of the p53/PUMA proapoptotic pathway (52–55). Moreover, HepG2 cells in which FoxM1 levels were depleted by electroporation of FoxM1 no. 2 siRNA duplexes were resistant to apoptosis in response to WT ARF_26–44 peptide treatment (Figure 9F), suggesting that induction of apoptosis by the WT ARF peptide was dependent on FoxM1 levels.

Tumor cells are known to express high levels of the mitotic regulators PLK1, aurora B kinase, and survivin proteins, which function to prevent their apoptosis (45, 48, 56–62). Previous studies demonstrated that U2OS cells transfected with siFoxM1 #2 duplex were blocked in mitotic progression and exhibited undetectable levels of FoxM1 and its downstream target mitotic regulators PLK1, aurora B kinase, and survivin (26). Consistent with these studies, FoxM1-depleted HepG2 cells exhibited undetectable protein levels of survivin, PLK1, and aurora B kinase (Figure 9G). We electroporated HepG2 cells with siFoxM1 #2 or control p27^Kip1 siRNA (siP27) and grew the cells in culture for 2 days to allow for siRNA silencing, and then 2 × 10⁵ HepG2 cells were plated in triplicate, and viable HepG2 cells were counted at 2, 3, 4, or 5 days following electroporation. These cell-growth studies showed that FoxM1-deficient HepG2 cells were unable to grow in culture and gradually detached from the plate with time in culture (Figure 9H). In contrast, HepG2 cells treated with WT ARF_26–44 peptide exhibited a less severe reduction in levels of survivin (50%), PLK1 (80%), and aurora B kinase (80%) proteins compared with controls (Figure 9I). We also determined a growth curve of HepG2 cells at 1, 2, or 3 days following treatment with WT ARF_26–44 peptide, mutant ARF_37–44 peptide, or PBS. Although the WT ARF_26–44 peptide–treated HepG2 cells displayed 50% apoptosis (Figure 9E), they were able to sustain the number of cells initially plated (2 × 10⁵), suggesting that the WT ARF peptide–treated cells were able to proceed through the cell cycle (Figure 9J). These results are consistent with recent studies in which hypomorphic levels of FoxM1 protein (40% of WT FoxM1 levels) in breast cancer cell lines transfected with a different Foxm1 siRNA duplex reduced expression of mitotic regulators to levels that are insufficient to properly execute mitosis, leading to mitotic catastrophe and apoptosis (63). Based on these findings, we propose the hypothesis that WT ARF_26–44 peptide treatment causes hypomorphic levels of FoxM1 activity, leading to apoptosis, whereas depleting FoxM1 levels results in mitotic arrest. However, we have not ruled out other possibilities.

Mx-Cre–mediated deletion of the Foxm1^fl/fl allele in mouse liver tumors induced by DEN/PB exposure.

Generation of C57BL/6 mice containing Foxm1^fl/fl was described previously (25), and they were bred into the C57BL/6 mouse background for 8 generations. The type I IFN–inducible Mx-Cre Tg C57BL/6 mice (C57BL/6-TgN Mx-Cre mice) were purchased from The Jackson Laboratory. The Mx-Cre Tg C57BL/6 mice were bred with Foxm1^fl/fl C57BL/6 mice, and the offspring were screened for Mx-Cre Foxm1^fl/+ mice. These mice were then backcrossed with Foxm1^fl/fl C57BL/6 mice to generate Mx-Cre Foxm1^fl/fl C57BL/6 mice. Mx-Cre Foxm1^fl/fl C57BL/6 mice were bred with Foxm1^fl/fl C57BL/6 mice to generate a sufficient number of mice for the liver tumor experiments. Charles J. Sherr (St. Jude Children’s Research Hospital, Memphis, Tennessee, USA) provided the Arf^–/– C57BL/6 mice (69). The Arf^–/– C57BL/6 mice were backcrossed with the Rosa26-FoxM1b FVBN Tg mice, which ubiquitously express the human FoxM1b cDNA (34), to generate Arf^–/– Rosa26-FoxM1b Tg mice.

At 14 days after birth, each mouse in the litter received a single, i.p. injection of the tumor initiator DEN (5 μg/g body weight; catalog N0756; Sigma-Aldrich) to induce liver tumors (8). Two weeks later, male mice were given water containing 0.025% PB tumor promoter for the duration of the experiment (8). At 30 weeks of DEN/PB exposure (which leads to HCC formation), we subjected Mx-Cre Foxm1^fl/fl mice to 3 consecutive i.p. injections (separated by 1 day) of 250 μg of synthetic sRNA polyinosinic-polycytidylic acid [poly(I-C); Sigma-Aldrich] (36) to induce expression of the Mx-Cre transgene and cause deletion of the Foxm1^fl/fl allele in preexisting liver tumors. We continued PB administration in the drinking water for an additional 10 weeks to allow tumor growth. Livers from mice sacrificed by CO₂ asphyxiation were dissected and paraffin embedded for histological staining or immunostaining and for isolation of protein extracts as described previously (8). All protocols used in this study were approved by the animal protocol committee at the University of Illinois at Chicago.

Treatment of mice with DEN/PB-induced liver tumors with WT ARF_26–44 peptide or mutant ARF_37–44 peptide.

Genemed Synthesis Inc. synthesized the WT ARF_26–44 peptide (rrrrrrrrrKFVRSRRPRTASCALAFVN) or mutant ARF_37–44 peptide (rrrrrrrrrSCALAFVN), both of which contain 9 N-terminal D-Arg (r) residues (32, 33) and which are TMR fluorescently tagged (red) at the N-terminus. We chose to i.p. inject mice with this cell-penetrating WT ARF_26–44 peptide to efficiently deliver this peptide to the liver in vivo. In order to determine the effective concentration of dose of the ARF peptide for efficient liver delivery, mice were subjected to i.p. injection of 0.1, 1, 5, or 10 mg/kg body weight of TMR fluorescently tagged WT ARF_26–44 peptide and were sacrificed 24 hours later, after which their livers were dissected, formalin fixed, and paraffin embedded as described previously (25). Liver sections were treated with xylene to remove paraffin wax and then examined by fluorescence microscopy for red peptide fluorescence. This dose-response curve determined that i.p. injection of either equal to or greater than 5 mg/kg body weight of TMR fluorescently tagged WT ARF_26–44 peptide was detectable in cytoplasm and nucleolus of hepatocytes and in hepatic mesenchymal cells at 24 hours after injection. Based on these studies, hepatic tumors were induced in Foxm1^fl/fl mice by 32 weeks of DEN/PB exposure, and then they were subjected to daily i.p. injections of 5 mg/kg body weight of the WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 4 weeks and with WT ARF_26–44 peptide for 8 weeks. After 33 weeks of DEN/PB treatment, Arf^–/– Rosa26-FoxM1b Tg mice were subjected to daily i.p. injections of 5 mg/kg body weight of the WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 4 weeks. Liver tumor–bearing mice were also administered sterile PBS as controls.

BrdU labeling, immunohistochemical staining, and TUNEL apoptosis assay.

To monitor hepatic cellular proliferation, PB was removed 4 days prior to the completion of the experiment, and mice were placed on drinking water with 1 mg/ml of BrdU for 4 days before they were sacrificed (8, 37). Hepatic tumor cell DNA replication in liver sections was determined by immunohistochemical detection of BrdU incorporation as described previously (8). We used an affinity-purified rabbit polyclonal antibody specific to FoxM1b protein (1:500 dilution), which was generated against amino acids 365–748 of the human FoxM1b protein as described previously (26). We also used the following antibodies specific to the mouse anti-BrdU (Bu20a, 1:100; Dako), rabbit anti–cleaved caspase-3 (5A1, 1:100; Cell Signaling Technology), rabbit anti-survivin (1:250; Novus Biologicals), mouse anti-Kip1/p27 (1:100; BD Biosciences), and rat anti-CD34 (RAM34, 1:100; BD Biosciences) for immunohistochemical detection of 5-μm liver sections using methods described previously (8, 23, 25, 29). To measure apoptosis in mouse livers, we used the TUNEL assay on liver sections using the ApopTag Fluorescein In Situ Apoptosis Detection Kit from Intergen according to the manufacturer’s recommendations. We calculated the mean number (±SD) of TUNEL- or DAPI-positive hepatocyte nuclei per 1,000 cells or ×200 field by counting the number of positive hepatocyte nuclei using 5 different ×200 fields of liver tumor sections from male mice at the indicated times of DEN/PB exposure. We used 5 liver tumor sections from 3 mice to calculate the mean number of BrdU-positive cells (±SD) per square millimeter liver tumor. To calculate the area or size of liver tumors, we used micrographs of H&E-stained liver tumor sections taken by an Axioplan 2 microscope (Zeiss) and the AxioVision program (version 4.3; Zeiss).

Western blot analysis.

For Western blot analysis, 75 μg of total protein extracts prepared from liver were separated on SDS-PAGE and transferred to a PVDF membrane (Bio-Rad) as described previously (25). The following commercially available antibodies and dilutions were used for Western blotting: mouse anti–Plk-1 (F-8, 1:500; Santa Cruz Biotechnology Inc.), mouse anti-p53 (FL-393, 1:500; Santa Cruz Biotechnology Inc.), mouse anti-MDM2 (SMP14, 1:500; Santa Cruz Biotechnology Inc.), mouse anti–β-actin (AC-15, 1:5,000; Sigma-Aldrich), mouse anti–aurora B kinase/AIM-1 (1:1,000; BD Biosciences), rabbit anti-survivin (1:2,000; Novus Biologicals), mouse anti-PUMA (1:1,000; Cell Signaling Technology), and mouse anti-NPM/B23 (1:15,000; Zymed). The primary antibody signals were amplified by HRP-conjugated secondary antibodies (Bio-Rad) and detected with Enhanced Chemiluminescence Plus (ECL Plus; Amersham Biosciences). Western blot analysis was performed with liver extracts from 2–4 mice per indicated time point following DEN/PB treatment, and signal intensities were normalized to β-actin signals. Quantitation of expression levels was determined with tiff files from scanned films using the BioMax 1D software program (Kodak).

Treatment of doxycycline-inducible U2OS C3 cells with WT ARF_26–44 peptide or mutant ARF_37–44 peptide.

We previously reported on the generation and growth of an osteosarcoma U2OS clone C3 cell line (U2OS C3 cells) that allowed doxycycline-inducible (Dox-inducible) expression of the GFP-FoxM1b fusion protein (24). For induced expression of the GFP-FoxM1b fusion protein, we added 1 μg/ml of Dox (Sigma-Aldrich). In order to determine the FoxM1b localization in U2OS C3 cells, they were treated with 12 μM of the TMR fluorescently tagged WT ARF_26–44 or mutant ARF_37–44 peptide for 24 hours. U2OS C3 cells were fixed with 10% buffered formalin (Fisher Scientific) for 20 minutes at room temperature and rinsed with PBS, and cover glasses were mounted with Vectashield Mounting Medium with DAPI (catalog H-1200; Vector Laboratories). Immunofluorescence was detected using an Axioplan 2 microscope.

Treatment of human hepatoma cell lines with WT ARF_26–44 or mutant ARF_37–44 peptide and siRNA transfection.

HepG2 cells were plated on 100-mm plates in Ham’s F-12 medium supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 2 mM l-glutamine, and 0.5 U Humulin (insulin; Lilly). HepG2 cells were treated with 12 μM, 25 μM, or 50 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 24 hours and used to prepare whole-cell protein extracts using the NP40 lysis buffer as described previously (26), and Western blot analysis was performed as described above. HepG2 cells were treated with 25 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 24 hours, and HepG2 cell apoptosis was determined by TUNEL assay using the ApopTag Fluorescein In Situ Apoptosis Detection Kit from Intergen according to the manufacturer’s recommendations. We calculated the percent apoptosis of HepG2 cells (±SD) by counting the number of TUNEL-positive cells (green fluorescence) per 1,000 nuclei as visualized by DAPI (blue fluorescence) counterstaining.

Human hepatoma PLC/PRF/5 and Hep3B cell lines (ATCC) were grown as a monolayer on 100-mm plates in DMEM supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. Human hepatoma PLC/PRF/5 and Hep3B cells were treated with 25 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 24 hours, and cells undergoing apoptosis were determined by TUNEL assay as described above.

Transfection of human cells with FoxM1 siRNA duplexes (Dharmacon RNA Technologies), named siFoxM1 no. 1 (CAACAGGAGUCUAAUCAAG) and siFoxM1 no. 2 (GGACCACUUUCCCUACUUU), efficiently depleted FoxM1 levels as described previously (26). We also previously used the human p27^Kip1 siRNA duplex (GUACGAGUGGCAAGAGGUGUU) as a control siRNA, which did not influence expression of the FoxM1 gene (26). Human p53 siRNA duplex was purchased from Cell Signaling Technology Inc. These siRNA duplexes were transfected into HepG2 cells using the Nucleofector II apparatus and buffers recommended by the manufacturer (Amaxa Biosystems). Forty-eight hours after electroporation to allow siRNA silencing, HepG2 cells were treated with 25 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide for 24 hours and then examined for apoptosis by TUNEL assay. HepG2 cells were harvested at 72 hours after FoxM1 siRNA or p27 siRNA transfection to prepare protein extracts for Western blot analysis to examine expression of PUMA, survivin, PLK1, and aurora B kinase proteins as described above. HepG2 cells were harvested at 48 hours after p53 siRNA transfection or at 72 hours after p53 siRNA transfection and examined for apoptosis as described above.

Treatment of HMEC-1 cells with WT ARF_26–44 or mutant ARF_37–44 peptide.

HMEC-1 cells (ATCC) were grown as monolayer cultures on 100-mm plates in MCDB 131 medium supplemented with 15% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, 10 ng/ml EGF (Sigma-Aldrich), and 1 μg/ml hydrocortisone (Sigma-Aldrich). HMEC-1 cells were treated for 48 hours with 100 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide and then examined for apoptosis as described above.

Growth curve of HepG2 cells electroporated with FoxM1 or p27^Kip1 siRNA or treated with WT or ARF peptide.

HepG2 cells were electroporated with 100 nM of FoxM1 (FoxM1 no. 2) or p27^Kip1 (siP27) siRNA duplexes (26) using the Nucleofector II apparatus (Amaxa Biosystems) and eletroporation buffers recommended by the manufacturer for HepG2 cells. HepG2 cells were replated for 2 days to allow siRNA silencing of FoxM1 or p27^Kip1 levels, and then 2 × 10⁵ HepG2 cells were plated in triplicate, and viable HepG2 cells were counted at 2, 3, 4, or 5 days following electroporation. Mock-electroporated cells were used as controls. We also plated 2 × 10⁵ HepG2 cells in triplicate, and viable HepG2 cells were counted at 1, 2, or 3 days following treatment with 50 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide. After 2 days in culture, media was replaced with 50 μM of WT ARF_26–44 peptide or mutant ARF_37–44 peptide. PBS-treated cells were used as controls.

Statistics.

We used the Microsoft Excel program to calculate SD and statistically significant differences between samples using the 2-tailed Student’s t test. The asterisks in each graph indicate statistically significant changes with P values calculated by Student’s t test: *P < 0.05, **P ≤ 0.01, and ***P ≤ 0.001. P values less than 0.05 were considered statistically significant.