p19ARF/p14ARF controls oncogenic functions of signal transducer and activator of transcription 3 in hepatocellular carcinoma

Signal transducer and activator of transcription 3 (Stat3) is activated in a variety of malignancies, including hepatocellular carcinoma (HCC). Activation of Ras occurs frequently at advanced stages of HCC by aberrant signaling through growth factor receptors or inactivation of effectors negatively regulating Ras signaling. Here, we addressed the role of Stat3 in Ras‐dependent HCC progression in the presence and absence of p19ARF/p14ARF. We show that constitutive active (ca) Stat3 is tumor suppressive in Ras‐transformed p19ARF−/− hepatocytes, whereas the expression of Stat3 lacking Tyr705 phosphorylation (U‐Stat3) enhances tumor formation. Accordingly, Ras‐transformed Stat3Δhc/p19ARF−/− hepatocytes (lacking Stat3 and p19ARF) showed increased tumor growth, compared to those expressing Stat3, demonstrating a tumor‐suppressor activity of Stat3 in cells lacking p19ARF. Notably, endogenous expression of p19ARF in Ras‐transformed hepatocytes conveyed oncogenic Stat3 functions, resulting in augmented or reduced HCC progression after the expression of caStat3 or U‐Stat3, respectively. In accord with these data, the knockdown of p14ARF (the human homolog of p19ARF) in Hep3B cells was associated with reduced pY‐Stat3 levels during tumor growth to circumvent the tumor‐suppressive effect of Stat3. Inhibition of Janus kinases (Jaks) revealed that Jak causes pY‐Stat3 activation independently of p14ARF levels, indicating that p14ARF controls the oncogenic function of pY‐Stat3 downstream of Jak. Conclusion: These data show evidence that p19ARF/p14ARF determines the pro‐ or anti‐oncogenic activity of U‐Stat3 and pY‐Stat3 in Ras‐dependent HCC progression. (HEPATOLOGY 2011;)

Signal transducer and activator of transcription 3 (Stat3) is activated in a variety of malignancies, including hepatocellular carcinoma (HCC). Activation of Ras occurs frequently at advanced stages of HCC by aberrant signaling through growth factor receptors or inactivation of effectors negatively regulating Ras signaling. Here, we addressed the role of Stat3 in Ras-dependent HCC progression in the presence and absence of p19 ARF /p14 ARF . We show that constitutive active (ca) Stat3 is tumor suppressive in Ras-transformed p19 ARF2/2 hepatocytes, whereas the expression of Stat3 lacking Tyr 705 phosphorylation (U-Stat3) enhances tumor formation. Accordingly, Ras-transformed Stat3 Dhc /p19 ARF2/2 hepatocytes (lacking Stat3 and p19 ARF ) showed increased tumor growth, compared to those expressing Stat3, demonstrating a tumor-suppressor activity of Stat3 in cells lacking p19 ARF . Notably, endogenous expression of p19 ARF in Ras-transformed hepatocytes conveyed oncogenic Stat3 functions, resulting in augmented or reduced HCC progression after the expression of caStat3 or U-Stat3, respectively. In accord with these data, the knockdown of p14 ARF (the human homolog of p19 ARF ) in Hep3B cells was associated with reduced pY-Stat3 levels during tumor growth to circumvent the tumor-suppressive effect of Stat3. Inhibition of Janus kinases (Jaks) revealed that Jak causes pY-Stat3 activation independently of p14 ARF levels, indicating that p14 ARF controls the oncogenic function of pY-Stat3 downstream of Jak. Conclusion: These data show evidence that p19 ARF /p14 ARF determines the pro-or anti-oncogenic activity of U-Stat3 and pY-Stat3 in Ras-dependent HCC progression. (HEPATOLOGY 2011;54:164-172) See Editorial on Page 9 C onstitutive activation of signal transducer and activator of transcription 3 (Stat3) is frequent in solid cancers and contributes to oncogene-sis. 1,2 Stat3 is considered as an oncogene because (1) Stat3 that constitutively activates transcription (caStat3) provides cellular transformation of NIH3T3 fibroblasts, (2) Stat3 is stimulated by growth-promoting signals, such as activated growth factor receptors, via several Janus kinases (Jaks) or Src, (3) both nonphosphorylated and phosphorylated Stat3 accumulate in many human cancers, and (4) Stat3 contributes to abrogated immune surveillance, leading to enhanced tumor cell growth. [3][4][5][6] However, recent findings in glioblastoma and intestinal tumors support the idea that Stat3 can also act as a tumor suppressor. [7][8][9] In the liver, Stat3 is important for liver regeneration by stimulating hepatic cell proliferation and survival. 10 Stat3 is up-regulated and activated in the vast majority of human hepatocellular carcinoma (HCC) specimens 11,12 and is essential for cell growth, survival, tumor dedifferentiation, intratumoral microvessel density, and metastasis of HCC. 13,14 Deactivation of Stat3 by low molecular compounds or inhibition of Stat3 expression by employing RNA-interference approaches enhanced the chemosensitivity of HCC cells and suppressed growth and metastasis of human HCC in xenografted mice. 15,16 Recent findings demonstrate that Stat3 is a critical regulator of liver cancer development and progression through a negative cross-talk with nuclear factor-jB. 17 Notably, the constitutive activation of Stat3 is accompanied by high levels of unphosphorylated Stat3 (U-Stat3), which differs from Tyr 705 phosphorylated Stat3 (pY-Stat3)-mediated gene expression in both its binding partners and mechanism to activate transcription. The formation of U-Stat3 complexes occurs either in the cytoplasm or the nuclear compartment. Its transcriptional targets also differ from those of pY-Stat3 dimers, as other promoters can be modulated by, for example, U-Stat3/nuclear factor-jB heterodimers. 18 The Ras cascade mainly transduces extracellular signals via activated growth factor receptors, resulting in proliferative and anti-apoptotic signals. 19 In HCC, the expression of oncogenic Ras, which is locked in its active form due to the insensitivity against guanosine triphosphatase-activating proteins (GAPs), is rare. 19,20 Yet, aberrant activation of Ras signaling is frequently observed by overexpression of Ras, epigenetic silencing of GAPs by promoter hypermethylation, or by mutations of upstream inducers or downstream effectors. 19,20 The tumor suppressor p14 ARF (the human homolog of p19 ARF ) is an important sensor of hyperproliferative stimuli that restricts cell proliferation through both p53-dependent and p53-independent pathways when activated by sustained mitogenic or oncogenic signals, such as Ras. [21][22][23] Disruption of the p14 ARF -Mdm2-p53 pathway is a very common feature in cancer. 23 Remarkably, p14 ARF is inactivated by promoter hypermethylation in up to 40% of HCC cases. 23,24 In this study, we found that caStat3 acted in a tumorsuppressive manner in Ras-transformed p19 ARFÀ/À hepa-tocytes, whereas the expression of U-Stat3 or loss of Stat3 increased tumor growth. Reciprocal effects of caStat3 and U-Stat3 were observed in Ras-transformed hepatocytes endogenously expressing p19 ARF . In human HCC cells, knockdown of p14 ARF resulted in reduced pY-Stat3 levels upon tumor formation, thus impeding the tumor-suppressive function of Stat3. Activation of pY-Stat3 by Jak was not affected by p14 ARF levels, suggesting that p14 ARF modulates the oncogenic function of Stat3 downstream of Jak in Ras-transformed hepatocytes.

Materials and Methods
Cell Culture. Mouse hepatocyte cell lines were generated by stable retroviral transmission of immortalized p19 ARFÀ/À hepatocytes (MIM-1-4), Stat3 Dhc /p19 ARFÀ/À , or MMH-D3 cells with a construct expressing oncogenic v-Ha-Ras or Stat3 variants (wtStat3, caStat3a, caStat3b, and U-Stat3; Supporting Fig. 1) 8 and cultured on collagen-coated dishes. 25,26 Ras-transformed mouse hepatocytes and human Hep3B, PLC/PRF/5, SW480, and p14 ARF -deficient MCM1 melanoma cells 27 (a kind gift of Dr. Mario Mikula, Medical University of Vienna) were cultivated at 37 C and 5% CO 2 . All cells were routinely screened for the absence of mycoplasma. Details for treatment of HCC cells with cytokines and Jak inhibitors as well as for lentiviral-mediated knockdown of p14 ARF are provided in the Supporting Information.
Tumor Formation and Recultivation of Tumor Cells. Briefly, 1 Â 10 6 murine or 5 Â 10 6 human cells in 100 lL of Ringer solution were subcutaneously (SC) injected into severe combined immunodeficient (SCID) mice (Harlan Laboratories, San Pietro, Italy). Tumor volume was determined as previously described. 30 Pulmonary metastatic colonization was analyzed after the injection of 1 Â 10 5 cells/100 lL of Ringer solution into the tail vein of SCID mice.
Orthotopic liver transplantation was performed by the injection of 1 Â 10 6 cells/20 lL of Ringer solution into the spleen of SCID mice. 26 Recovery of tumor cells is provided in the Supporting Information. All experiments were performed according to the Austrian guidelines for animal care and protection.
HCC Induction. To initiate tumor development in the liver, 14-day-old Stat3 fl/fl and Stat3 Dhc mice were intraperitoneally (IP) injected with a single dose of diethylnitrosamine (DEN; 25 mg/kg). At 12 months of age, mice were sacrificed and livers were processed for polymerase chain reaction (PCR) analysis or fixed in 4% formaldehyde for immunohistochemistry.
Immunohistochemistry. Mice were sacrificed and tumors and lungs were fixed as previously described. 26 Next, 4-lm-thick, paraffin-embedded sections were stained with hematoxylin and eosin (H&E). For immunohistochemistry, sections were stained with anti-phospho-Stat3 (Tyr 705 ; Cell Signaling Technology, Beverly, MA) and anti-Stat3 antibodies (Cell Signaling Technology).
Reverse-Transcription PCR. Semiquantitative reverse transcription PCR was performed as previously described. 26 Quantitative PCR was performed with Fast SYBR Green Master Mix (Applied Biosystems, Foster City, CA), according to the recommendations of the manufacturer, and quantified with the 7500 Fast Real-Time PCR System (Applied Biosystems). Arbitrary units were calculated by the dCT method. Primer sequences are provided in the Supporting Information.
Statistical Analysis. Data are expressed as means 6 standard deviation (SD). The statistical significance of differences was evaluated using an unpaired, nonparametric Student t test. Significant differences between experimental groups were *P < 0.05, **P < 0.01, or ***P < 0.005.

Results
Stat3 Represses Tumor Growth of Ras-Transformed p19 ARF2/2 Hepatocytes. We employed an established mouse tumor transplantation model to assess the role of Stat3 during HCC progression. This model is based on the lack of p19 ARF in hepatocytes, which allows immortalization. 26 Nontumorigenic p19 ARFÀ/À hepatocytes (MIM-1-4) were transformed with oncogenic Ras (MIM-R). 30 To study the effect of Stat3 on Ras-dependent tumor growth, caStat3 variants of Stat3a and the natural splice variant, Stat3b, lacking the Ser 727 phosphorylation site 4 and U-Stat3 (lacking both Tyr 705 and Ser 727 phosphorylation sites), were stably expressed in MIM-R hepatocytes (Supporting Fig. 1A,B).
Proliferation kinetics showed no changes between MIM-R hepatocytes and those expressing Stat3 mutants (data not shown). To investigate tumorigenicity, Stat3 mutant hepatocytes were SC injected into SCID mice. MIM-R-caStat3a-and MIM-R-wtStat3derived tumors displayed two-fold reduced volumes, compared to those generated by MIM-R hepatocytes. An even five-fold suppression of tumor growth was observed upon the injection of MIM-R-caStat3b hepatocytes. On the contrary, MIM-R-U-Stat3 cells caused a two-fold increased tumor volume, compared to MIM-R hepatocytes (Fig. 1A). Orthotopic transplantation of MIM-R-wtStat3, MIM-R-caStat3a, and MIM-R-caStat3b hepatocytes led to a strong reduction of HCC formation, whereas MIM-R-U-Stat3 cells exhibited enhanced HCC generation, compared to MIM-R-derived liver tumors (Fig. 1B). Notably, both frequency and size of lung metastases were significantly reduced after tail vein injection of MIM-R-wtStat3, MIM-R-caStat3a, or MIM-R-caStat3b cells. In contrast, MIM-R-U-Stat3 cells showed pulmonary metastasis comparable to MIM-R hepatocytes (Fig. 1C,D).
These data show that exogenous expression of caStat3 or U-Stat3 causes anti-or pro-oncogenic effects in p19 ARFÀ/À MIM-R hepatocytes, respectively.
We next analyzed the tumorigenic potential after SC injection into SCID mice. Both, MIM-R-Stat3 Dhc -1 and MIM-R-Stat3 Dhc -2 hepatocytes showed increased tumor development, compared to MIM-R cells, whereas the re-expression of wtStat3 in MIM-R-Stat3 Dhc hepatocytes abolished faster tumor kinetics (Fig. 2). In summary, these results confirm that Stat3 has tumor-suppressive functions in Ras-transformed p19 ARFÀ/À hepatocytes.
Stat3 Acts in a Pro-Oncogenic Manner in p19 ARF -Positive Ras-Transformed Hepatocytes. To investigate a possible impact of p19 ARF deficiency on Stat3 functions, we employed murine MMH-D3 hepatocytes that express endogenous p19 ARF (Fig. 3A). 31 MMH-D3 cells transformed with oncogenic Ras (MMH-R) were further analyzed after the expression of either caStat3b or U-Stat3, as these variants showed the strongest tumor-suppressive or -promoting activities in p19 ARFÀ/À MIM-R hepatocytes, respectively (Fig. 1). SC injection of p19 ARF -positive MMH-R cells expressing caStat3b into mice showed an eight-fold increased tumor formation, compared to MMH-R, whereas the expression of U-Stat3 lowered tumor generation by approximately 1.5-fold (Fig. 3B). After the tail vein injection of cells, MMH-R-caStat3b cells exhibited enhanced lung colonization, whereas MMH-R-U-Stat3 hepatocytes showed lower numbers and a reduced size of lung metastasis (Fig. 3C,D). From these data, we conclude that p19 ARF modulates pro-and anti-oncogenic activities of Stat3 during HCC progression.
Up-Regulation of p19 ARF Is Associated with DEN-Induced Tumor Formation in Stat3 fl/fl Mice. Stat3 was recently shown to be required for DEN-induced HCC development. 17 In accord with these data, we observed a reduction in the number and size of liver tumors in Stat3 Dhc , relative to Stat3 fl/fl , mice after DEN treatment (unpublished data). To confirm that the pro-oncogenic role of Stat3 in this background would correlate with p19 ARF , we analyzed samples of DEN-induced liver tumors. Indeed, p19 ARF was remarkably high during HCC development in Stat3 fl/fl mice, whereas DEN-induced liver tumors of Stat3 Dhc mice showed strongly reduced levels of p19 ARF (Fig.  4A). As expected, activation of pY-Stat3 was observed in tumor sections from DEN-treated Stat3 fl/fl mice (Fig. 4B). These data show that pY-Stat3 activation is linked to the presence of p19 ARF during DEN-induced tumor formation (Fig. 4A), underlining the functional interaction of p19 ARF and pY-Stat3 in tumor growth.
p14 ARF Modulates Stat3 Activation During Human HCC Development. To bridge mouse to human hepatocarcinogenesis, we analyzed an established human HCC cell line for the expression of p14 ARF , the human homolog of p19 ARF . Real-time PCR analysis showed that human Hep3B cells express p14 ARF (Supporting Fig. 3A) and activate Ras/mitogen-activated protein kinase signaling effectors, as described. 32 To investigate the effects of Stat3 in the  presence or absence of p14 ARF , we introduced shorthairpin RNAs (shRNAs) targeted against p14 ARF (sh-p14-1 and sh-p14-2) as well as a mixture of both shRNAs (sh-p14-3) into Hep3B cells. Expression of p14 ARF was almost eliminated after shRNA expression (Supporting Fig. 3B,C). Hep3B and corresponding shRNA cell lines were SC injected into SCID mice to examine tumorigenesis. Knockdown of p14 ARF by the expression of sh-p14-3 was accompanied by a significant down-regulation of pY-Stat3 in vivo, as observed by immunohistochemistry and immunoblotting of tumors ( Fig. 5A-C). Notably, tumor volumes of Hep3B cells expressing sh-p14 were comparable to control cells (Fig. 5D). A persistent down-regulation of p14 ARF was confirmed after the recultivation of cells from SC tumors (Fig. 5E). From these data, we conclude that the tumor-suppressive function of Stat3 in the absence of p14 ARF is circumvented by inhibition of pY-Stat3 in vivo.
p14 ARF Acts Downstream of Jak-Mediated Stat3 Phosphorylation. Next, we analyzed whether de novo RNA and protein synthesis would affect pY-Stat3 activation and whether it would depend on the presence of p14 ARF . Interestingly, the inhibition of either transcription or translation, using actinomycin D or cycloheximide, respectively, reduced pY-Stat3 levels in both Hep3B, as well as in Hep3B-sh-p14_3, cells, while keeping total Stat3 levels unaffected (Fig.  6A,B). Comparable results were obtained by  employing human PLC/PRF/5 hepatoma cells (data not shown). These data suggest that de novo synthesis of an upstream mediator is essential for pY-Stat3 activation in Hep3B hepatoma cells. However, pY-Stat3 activation occurs in a mode independent of p14 ARF expression.
To study the impact of p14 ARF on canonical Jak-Stat signaling, Hep3B cells and those showing a knockdown of p14 ARF were treated with a pan-Jak inhibitor (blocking Jak1, Jak 2, Jak3, and Tyk2 activity). The pan-Jak inhibitor efficiently blunted pY-Stat3 activation upon interleukin-6 (IL-6) treatment independently of p14 ARF expression (Fig. 6C). PLC/PRF/5 hepatoma cells revealed comparable results (data not shown). These findings suggest that Jak activity mainly causes pY-Stat3 activation, irrespective of p14 ARF levels, implicating that the control of oncogenic Stat3 function by p14 ARF occurs downstream of Stat3 phosphorylation.

Discussion
This study shows that Stat3 is able to execute both pro-and anti-oncogenic functions, depending on p19 ARF /p14 ARF expression, during Ras-mediated HCC development. CaStat3 acts in a tumor-suppressive manner in Ras-transformed p19 ARFÀ/À hepatocytes as well as in a tumor-promoting manner in hepatocytes expressing p19 ARF . Strikingly, the Y705F mutant (U-Stat3) showed the opposite effect. In line with these findings, tumors derived from the human HCC cell line Hep3B show reduced pY-Stat3 upon p14 ARF silencing. In this scenario, human HCC cells counteract the tumor-suppressive effects of Stat3, as observed in the murine p19 ARFÀ/À model, and prevent diminished tumor growth. p14 ARF levels in HCC cells affect pY-Stat3 activation in vivo, whereas pY-Stat3 activation mainly induced by Jak seems to be independent of p14 ARF expression in vitro.
p14 ARF is a potential target for inactivation in HCC due to its positive role in p53 stabilization by promoting MDM2 degradation. 33 In accord with our results, the p14 ARF -negative HCC cell lines, HepG2 and PLC/PRF/5, are less sensitive to Stat3 inhibitor NSC 74859 treatment, whereas p14 ARF expressing Huh-7 and SNU-398 cells show reduced cell proliferation after the administration of NSC 74859. 16 Our observations in murine hepatocytes suggest that a tumor-suppressive Stat3 function depends on p19 ARF deficiency, but might be independent of both p16 INK4B and p53 inactivation, because p19 ARFÀ/À hepatocytes express p16 INK4A and show p53 response. 30 In line with this, the human hepatoma Hep3B cells used in this study harbored a p53 mutation that did not affect the response of reduced pY-Stat3 activation in p14 ARF knocked-down cells. 34 Furthermore, pro-oncogenic Stat3 functions were observed in hepatitis B virus-positive HCC cell lines, such as SNU-182 or SNU-387, which express mutated and inactivated p53, 35,36 indicating tumor-promoting Stat3 functions independent of p53.
Modulation of pro-or anti-oncogenic Stat3 functions through tumor suppressors has been described in different cancers. In glioblastoma, deficiency in PTEN induces a malignant transformation of astrocytes upon Stat3 knockout, arguing for anti-oncogenic functions of Stat3. 37 Recently, we also described a dual role of Stat3 in Apc Min/þ mice, where Stat3 promotes early microadenoma formation, whereas Stat3 deficiency in intestinal epithelial cells increased later stage carcinoma progression, associating with nuclear b-catenin and impaired Ceacam1 expression. 9 In addition, we showed that caStat3 blocked c-myc-induced transformation of p53 À/À mouse fibroblasts. 8 These findings indicate that Stat3 functions are modulated by various tumor suppressors.
Despite the inability of oncogenic Ras to drive Stat3 tyrosine phosphorylation or nuclear translocation, Ras transformation was found to be impaired in the absence of Stat3. 38 Similar results were obtained upon mouse mammary tumor progression, showing that Stat3 is indispensable for the metastasis of ErbB2-activated cancer cells to the lung. 39 In contrast, our data suggest that Stat3 is not required for Ras transformation of hepatocytes in the absence of p19 ARF (Fig. 2). Furthermore, Stat3 regulates metabolic functions in mitochondria, requiring Ser 727 phosphorylation that supports Ras-dependent malignant transformation. 38 Our data exclude an involvement of Ser 727 phospho-Stat3 in the dual role of Stat3 in Ras-transformed hepatocytes, as expression of the C-terminally truncated Stat3b either suppressed or promoted tumor growth dependent on p19 ARF expression similarly to gain-of-function studies using full-length Stat3a (Figs. 1 and 3B-D). U-Stat3, harboring the Y705F mutation, abrogated tumor suppression and even enhanced tumor formation (Fig. 1), probably driven by the expression of a gene set specific for U-Stat3 and its putative interaction partners. Because p14 ARF /p19 ARF is known to interact with a multitude of proteins from different functional classes, 21 it is conceivable that a putative factor, designated ARF-X, is involved in the U-Stat3driven transcriptional control, as hypothesized in Fig.  7. The occupation of ARF-X by p14 ARF /p19 ARF could be responsible for the different outcome in the presence of p14 ARF /p19 ARF (Fig. 1 versus Fig. 3B-D). Jak activity, which might be crucially involved in pY-Stat3 activation of HCC cells, 40 is not altered by p14 ARF in human HCC cells in vitro (Fig. 6). However, pY-Stat3 is affected in tumors generated by p14 ARF knockeddown Hep3B cells to overcome tumor-suppressive actions ( Fig. 5A-C). Presumably, the in vivo environment, including tumor-stroma interactions, allows the tumor to act in a manner distinctive from malignant cells in vitro. In this scenario, the identification of ARF-X is to be the focus of future experiments with highest priority.
Activation of Stat3, occurring in the majority of HCC patients, suggests a critical role in liver cancer. 11,12 Stat3 is considered a potential target for therapeutic intervention, as Stat3 inhibition represses experimental tumors, but shows little side effects. 41 We provide the first evidence that p14 ARF determines whether Stat3 acts with pro-or anti-oncogenic func-tion in HCC cells. The link of Stat3 with p14 ARF might be of prognostic value for HCC therapy. Treatment of pY-Stat3-positive HCC patients showing loss of p14 ARF with Stat3 inhibitors could have adverse effects on cancer progression, thus opening new aspects for individualized medicine. In the presence of p14 ARF , active Stat3, either caused by Jak phosphorylation or by constitutive activation (ca) via dimerization, modulates target genes and promotes tumorigenesis (red arrows). p14 ARF sequesters an unknown factor, termed ARF-X. (B) In the absence of p14 ARF , U-Stat3 interacts with ARF-X to drive an oncogenic program (red arrow). U-Stat3 is generated by the suppression of Jak-mediated Stat3 activation. The caStat3 causes tumor suppression (browncolored arrow) by modulating an alternative set of Stat3-specific target genes.