Growth Arrest and DNA Damage 45G Down-Regulation Contributes to Janus /Signal Transducer and Activator of Transcription 3 Activation and Cellular Evasion in

Li Zhang,1 Zhaojuan Yang,1 Aihui Ma,1 Yulan Qu,1 Suhua Xia,1 Dongxu Xu,1 Chao Ge,1 Bijun Qiu,2 Qiang Xia,2 Jinjun Li,1 and Yongzhong Liu1

Growth arrest and DNA damage 45G (GADD45G), a stress sensor with multiple impli- cations in various biological processes, is down-regulated in a broad spectrum of can- cers. However, little is known about the biological effects of GADD45G on hepatocellular carcinoma (HCC) cells and the related mechanisms. In the present study, we found that GADD45G was commonly down-regulated in oncogene-transformed mouse cells and in human and mouse HCC. Ectopic expression of GADD45G robustly elicited senescence in HCC cells and suppressed tumor growth in vivo. Further- more, GADD45G-induced senescence occurred in HCC cells independently of p53, p16INK4a (p16), and retinoblastoma (Rb). Instead, the prompt inhibition of Janus kinase 2 (Jak2), tyrosine kinase 2 (Tyk2), and signal transducer and activator of tran- scription 3 (Stat3) activation was observed in cells undergoing senescence. Impairment of Jak-Stat3 activation caused by GADD45G expression was associated with activation of SH2 domain-containing tyrosine phosphatase-2 (Shp2). Expression of consti- tutively activated Stat3 or human telomerase reverse transcriptase (hTERT), as well as knockdown of Shp2f, efficiently counteracted GADD45G-induced senescence. More important, in clinical HCC specimens, we found that GADD45G expression was inver- sely correlated with phosphorylated Stat3 expression in tumor cells and disease progres- sion. Conclusion: GADD45G functions as a negative regulator of the Jak-Stat3 pathway and inhibits HCC by inducing cellular senescence. The decrease or absence of GADD45G expression may be a key event for tumor cells or premalignant liver cells to bypass cellular senescence. (HEPATOLOGY 2014;59:178-189)

epatocellular carcinoma (HCC) is one of the cence provides a tumor-suppressive mechanism for pre- most malignant and is listed as the venting liver tumorigenesis through the cell-autonomous second-most frequent cause of deaths in regulation of proliferation or by triggering immune sur- H 1 2-5 men and sixth in women worldwide. HCC generally veillance. Consistently, genetic or functional inactiva- develops in patients with liver cirrhosis and chronic infec- tion of senescence-related , such as p53 and p16/ tion with hepatitis B virus (HBV) and hepatitis C virus retinoblastoma (Rb), has been observed in human HCC (HCV). Emerging evidence has shown that cellular senes- specimens.6,7 Because initiation of the senescence program

Abbreviations: Akt, protein kinase; BrdUB, bromodeoxyuridine; DEN, diethylnitrosamine; DOX, doxycycline; GADD45G, growth arrest and DNA damage 45G; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; hTERT, human telomerase reverse transcriptase; IHC, immunohistochemi- cal; IL, interleukin; JAK, Janus kinase; LPCs, liver progenitor cells; mRNA, messenger RNA; qRT-PCR, quantitative real-time polymerase chain reaction; Rb, reti- noblastoma; SA-b-gal, senescence-associated b-galactosidase; SC, subcutaneous(ly); Shp2, SH2 domain-containing protein tyrosine phosphatase 2; siRNA, small interfering RNA; Stat3, signal transducer and activator of transcription 3; SV40 LT, SV40 large T antigen; Tet, tetracycline; TMA, tissue microarray; Tyk2, tyro- sine kinase 2. From the 1State Key Laboratory of Oncogenes and Related , Shanghai Cancer Institute, Renji Hospital, Shanghai Jiaotong University School of Medicine, Shanghai, China; and 2Department of Liver Surgery, Renji Hospital, Shanghai Jiaotong University School of Medicine, Shanghai, China. Received February 25, 2013; accepted July 7, 2013. This work received support by grants from the National Natural Science Foundation of China (no. 81201542), the Natural Science Foundation of Shanghai (no. 12ZR1430100), the Chinese National Key Project (no. 2013ZX10002-011), and the State Key Laboratory of Oncogenes and Related Genes (no. 91-12-03).

178 HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 179 in cells may require a reorganization of multiple events by immunochemistry staining. Another set of 45 paired adjusting their strength or functional presence, it is impor- HCC samples from the Qidong Institute tant to obtain more information regarding the mecha- (Qidong, China) were processed for RNA and protein nisms by which liver cells are transformed as well as how analysis. All human materials were approved by the liver tumor cells bypass senescence. institutional ethical review committee. The human Growth arrest and DNA damage 45G (GADD45G) HCC cell lines SK-Hep1 and Hep3B were purchased is a member of the family, which has been from the American Type Culture Collection (Manassas implicated in various biological processes, including VA). The human HCC cell line SMMC-7721, embry- integration of the cellular response to different stresses, onic 293 cells were from the Chinese Academy regulation of development, cell differentiation, and of Sciences (Shanghai, China). Mouse fetal liver pro- survival.8-10 GADD45G expression is broadly down- genitor cells (LPCs) were isolated from p532/2 mice regulated in a number of cancers, such as lymphoma, by the methods described previously13,14 and cultured nasopharyngeal carcinoma, cervical carcinoma, esopha- for the retroviral delivery of H-Ras V12, myristoylated geal carcinoma, carcinoma, and HCC.11,12 Mech- protein kinase B (Akt)1, and c-Myc. anistically, decrease of GADD45G expression in these Animal Model. Male BALB/c nude mice (6-8 tumors is primarily attributed to hypermethylation of weeks of age) received single subcutaneous (SC) flank CpG islands.11 Although a ubiquitous down-regulation injection of 5 or 6 3 106 Sk-Hep1 cells diluted in of GADD45G expression has been observed in various 200 lL of saline. Drinking water was supplemented types of cancer, the molecular mechanisms by with doxycycline (2 mg/mL) to induce GADD45G which GADD45G functions as a tumor suppressor are expression. Tumor growth was monitored by bidimen- still unclear. In the present study, we show that sional measurements using a caliper. Tumor-bearing GADD45G robustly elicits cellular senescence in HCC mice were sacrificed 50 days after inoculation, and cells and remarkably suppresses tumor growth in vivo. then the tumors were removed for further study. All We demonstrate, for the first time, that GADD45G mice studies were conducted in accord with protocols induces HCC cell senescence independently of the approved by the Shanghai Medical Experimental Ani- functional presence of p16, p53, and Rb, and that mal Care Commission. down-regulation of Janus kinase (Jak)/signal transducer Statistical Analysis. Statistical analyses were per- and activator of transcription 3 (Stat3) is the key event formed using SPSS 16.0 statistical software (SPSS, for GADD45G-induced cell senescence and tumor Inc., Chicago, IL). For in vivo studies, the percentage suppression. These findings provide new evidence for of tumor-free mice in each group was analyzed using the involvement of the stress sensor, GADD45G, in Kaplan-Meier’s survival analysis and Tarone-Ware’s sta- the biological connection between cellular senescence tistic was used for comparison of curves between evasion and hepatotumorigenesis. groups. For in vitro studies, a two-tailed t test was used to determine significance. The statistical correla- tion between the clinical parameters of HCC and Materials and Methods GADD45G staining levels in tissue sections was ana- lyzed by the chi-square test, nonparametric test, and Human Tissue Specimens and Cell Lines. A com- one-way analysis of variance. A P value less than 0.05 mercially available tissue microarray (TMA) containing was considered statistically significant. 75 pairs of HCC was used, and the patient informa- Laboratory Methods. See the Supporting Materials tion provided by the manufacturer is listed in the and Methods section for detailed experimental Supporting Data. The TMA sections were used for procedures.

Address reprint requests to: Yongzhong Liu, Ph.D., State Key Laboratory of Oncogenes and Related Genes, Shanghai Cancer Institute, Renji Hospital, Shanghai Jiaotong University School of Medicine, Wenxuan Medical Building, 800 Dongchuan Road, 200240 Shanghai, China. E-mail: [email protected]; fax: 86-21-3420-6283. Copyright VC 2013 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI 10.1002/hep.26628 Potential conflict of interest: Nothing to report. Additional Supporting Information may be found in the online version of this article. 180 ZHANG ET AL. HEPATOLOGY, January 2014

Fig. 1. GADD45G is down-regulated in liver tumorigenic cells and human HCC specimens. (A-C) qRT-PCR analysis of the relative expression of GADD45G in (A) transformed and nontransformed cells, (B) HCC cell lines and the immortalized liver cell line, Thle2, and in (C) nontumor tis- sues (N) and mouse HCC tissues (T) from DEN-treated mice (C57BL/6). Data shown are mean 6 SD from three independent experiments (**P < 0.01; *P < 0.05). (D) Scatter plots of relative expression of GADD45G in nontumor tissues (N) and their paired HCC counterparts (T). Statis- tical data are shown as mean 6 SD (***P < 0.001). (E) Protein levels of GADD45G in 12 representative HCC tissues (T) and paired nontumor tissues (N) were analyzed by western blotting. SD, standard deviation.

Results mouse HCCs induced by diethylnitrosamine (DEN), Identification of GADD45G as an Inducer of Cel- in comparison with noncancerous mouse liver tissues lular Senescence in HCC Cells. To explore the (Fig. 1C). More important, a majority of clinical molecular basis of liver tumor initiation, we used a human HCC samples exhibited losses or substantial genome-wide transcriptional profile analysis to com- decreases in GADD45G expression, as demonstrated pare the transcriptional profile of nontransformed p53- by an examination of transcripts from 45 paired clini- deficient fetal LPCs (p532/2 LPCs) with those of the cal samples (Fig. 1D) and protein analysis of 12 paired three types of oncogene-transformed p532/2 LPCs, HCC samples (Fig. 1E). namely, H-RasV12-LPCs, myristoylated-Akt1-LPCs, We reason that the great consistency in GADD45G and c-Myc-LPCs. We focused on those differentially down-regulation or loss in expression among liver can- expressed genes potentially involved in cell-cycle regu- cer cell lines and primary HCC tissues from mice and lation and found that the stress sensor, GADD45G, a humans may suggest a suppressive function of this member of the GADD45 protein family, was consis- in HCC development. Hence, we extended our tently down-regulated by these oncogenes (Supporting study by observing the influence of ectopic Fig. 1). Down-regulation of GADD45G in these GADD45G expression on three human HCC cell oncogene-transduced cells was further validated by lines (Sk-Hep1, SMMC-7721, and Hep3B). After 4 quantitative real-time polymerase chain reaction (qRT- days of culture of cells with GADD45G expression, PCR) analysis (Fig. 1A). Consistently, we also found the senescence-related morphological changes, includ- that GADD45G messenger RNA (mRNA) expression ing the more flattened, enlarged, and irregular cell was drastically decreased in human liver cancer cell shapes, were readily observed in these cells. Indeed, lines, as compared with the immortalized liver cell approximately 50%-70% of these cells ectopically line, Thle2 (Fig. 1B). Furthermore, an aberrant reduc- expressing GADD45G were stained positively for tion in GADD45G expression was observed in primary senescence-associated marker b-galactosidase (SA-b-gal; HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 181

Fig. 2A). Furthermore, mRNA levels of the senescence- (Fig. 2E). Immunostaining results showed that induction associated secretory cytokine, interleukin (IL)-8, were of GADD45G in vivo resulted in a substantial reduction also significantly elevated (Supporting Fig. 2). These of Ki-67 expression in tumor cells (Fig. 2F). These data cells were then subjected to both and bromo- clearly demonstrate that GADD45G efficiently induces deoxyuridine (BrdU) incorporation assays. We found HCC cell senescence and suppresses tumor growth that GADD45G expression did not induce cell in vivo. apoptosis, but rather resulted in the substantial inhibi- GADD45G Induces Liver Cancer Cell Senescence tion of cell proliferation in these cells (Fig. 2B). Next, Independently of p53 and p16/Rb Pathways. Mount- we examined whether induction of endogenous ing evidence demonstrates that the major cell-cycle GADD45G expression is contributable to drug-induced regulators, p53, p16, and Rb, are critically implicated cellular senescence. We found that transient MG-132 in the induction of cell senescence.15-17 In the present treatment induced GADD45G expression in Sk-Hep1 study, we found that GADD45G overexpression in cells (Supporting Fig. 3). Down-regulation of p16-deficient, but p53-competent, Sk-Hep1 cells did GADD45G by small interfering RNA (siRNA) remark- not significantly affect the protein levels of p53, , ably alleviated MG132-induced senescence (Supporting p27, or Rb throughout the induction of cell senescence Fig. 3), suggesting a significant role of GADD45G in (Fig. 3A). To explore the possibility that GADD45G- certain drugs- or stresses-triggered cellular senescence. In induced senescence may not necessarily require the addition, expression patterns of two other members of functional presence of p53/p21 and p16/Rb, different the GADD45 family, GADD45A and B, were also strategies were employed in the experiments. We first examined in transformed cells, HCC cells, and tumor transfected human Sk-Hep1 cells with lentiviral vector tissues. Notably, expression of GADD45B was signifi- harboring SV40 large T antigen (SV40 LT), which can cantly down-regulted in human HCC specimens inactivate p53 function by direct binding.18,19 The (Supporting Fig. 4). However, by ectopic expression of efficiency of SV40 LT in inhibiting p53 was validated the GADD45 proteins in three different HCC cell by showing that SV40 LT expression entirely blocked lines, we found that only GADD45G could efficiently up-regulation of p53 and downstream p21 expression decrease cell proliferation and induce cellular senescence in Sk-Hep1 cells treated with the chemotherapy agent, (Supporting Fig. 5). doxorubicin (Fig. 3B). We found clear evidence that Next, we further examined the consequence of SV40 LT expression did not rescue GADD45G- GADD45G-induced liver cancer cell senescence in vivo. induced cell senescence in p16-deficient Sk-Hep1 cells For tumor formation experiments, Sk-Hep1 cells were (Fig. 3C and Supporting Fig. 7A), indicating that infected with lentiviruses carrying a tetracycline (Tet)-in- GADD45G-induced senescence efficiently occurs in ducible GADD45G expression cassette (Tet-GADD45G) cells deficient in both p16 and p53 function. We fur- and the control (Tet-empty), respectively. We verified ther verified roles of p53/p21 and p16/Rb in that doxycycline (DOX)-induced GADD45G expres- GADD45G-induced senescence by knocking down sion efficiently triggered cell senescence in cultured cells, these proteins. SK-Hep1 cells were separately trans- as indicated by SA-b-gal staining (Fig. 2C) and expres- fected with small interfering RNAs (siRNAs) against sion of senescence-related secretory proteins IL-6 and p21, p27, p53, and Rb and then cultured for 3 days in IL-8 (Supporting Fig. 6A,B). Consistently, levels of c- the presence or absence of GADD45G induction. Our H2AX were remarkably increased (Supporting Fig. 6C). results showed that knockdown of these proteins did These cells were SC inoculated into athymic nude mice. not affect induction of cell senescence and cell arrest An inducible expression of GADD45G was achieved by by GADD45G (Fig. 3D and Supporting Fig. 7B). feeding mice with DOX-containing water, and tumor The efficiency of the siRNAs for the down-regulation growth was monitored for 50 days postinjection. of the target proteins was confirmed by western blot- GADD45G expression significantly increased the aver- ting (Supporting Fig. 7C). Moreover, in p532/2 mouse age latency of tumor appearance in vivo. Notably, by LPC-H-RasV12 cells, GADD45G also efficiently day 20 after injection, all mice incubated with cells induced cellular senescence (Supporting Fig. 7D,E). without ectopic GADD45G expression developed visi- Finally, we tested whether the combinatory deficiency ble tumors, whereas only 2 of 5 mice injected with cells of p53, Rb, and p16 could modulate the process of ectopically expressing GADD45G exhibited a palpable senescence initiated by GADD45G. We showed that tumor mass (Fig. 2D). Moreover, the size of the human hepatoma Hep3B cells deficient in both p53 generated tumors with ectopic GADD45G expression and Rb had a similar vulnerability to GADD45G- was remarkably smaller than tumors from control groups induced senescence when treated with siRNA specific 182 ZHANG ET AL. HEPATOLOGY, January 2014

Fig. 2. GADD45G induces cellular senescence and inhibits tumor growth of HCC cells. (A and B) Sk-Hep1, SMMC-7721, and Hep3B cells len- tivirally transduced with GADD45G (pSin-GADD45G) and their control (pSin-Vector) were cultured for 4 days. (A) Representative light microscopy showing SA-b-gal staining (left panel). Percentage of positive cells is depicted (right panel). (B) Cell apoptosis (upper panel) and proliferation (lower panel) were examined by Annexin V/7-aminoactinomycin D and BrdU fluorescence-activated cell sorting staining, respectively. Values rep- resent the mean 6 SD of three independent experiments (*P < 0.05; **P < 0.01). (C) Tet-GADD45G- and Tet-empty-Sk-Hep1 cells were cul- tured for 4 days in the presence (Tet1) or absence (Tet2) of 0.5 lg/mL of DOX for GADD45G induction. Representative images showing SA-b- gal staining (left panel) and the calculation of the percentage of positive cells (right panel) are shown. (D-F) Groups of 5 nude mice were SC injected with 5 3 106 Sk-Hep1 cells as indicated. (D) Kaplan-Meier’s analysis of tumor onset. The group of GADD45G; Tet1 mice versus the control groups of mice (*P < 0.05). (E) Volumes of tumor were measured at the indicated time intervals. Statistical data are shown as mean 6 SD of tumor volumes by day 50 postinjection (**P < 0.01). (F) Tumor sections were subjected to H&E staining and IHC for Ki-67. Original mag- nification, 3400. H&E, hematoxylin and eosin. HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 183 for p16, as compared with cells treated with control at Tyr580 was promptly increased in cells upon siRNA (Fig. 3E and Supporting Fig. 7F). The effi- GADD45G induction (Fig. 5A), whereas the basal ciency of siRNA in p16 knockdown was confirmed by level of Shp1 was undetectable in these cells (data not western blotting (Fig. 3F). Collectively, these results shown). Intriguingly, levels of p-Stat3 and p-Tyk2 and indicate that GADD45G functions as an inducer of p-Jak2 were constantly down-regulated from days 1 to senescence in liver cancer cells through p53- and p16/ 4 in Sk-Hep1 cells with GADD45G expression, coin- Rb-independent mechanisms. cidently with the persistent up-regulation of p-Shp2 Inhibition of the JAK/Stat3 Pathway in and p-Erk expression (Supporting Fig. 8). We further GADD45G-Induced Senescence. The strength or employed the SHP inhibitor, NSC 87877, and siRNA functional presence of multiple signal pathways may against Shp2 to confirm whether Shp2 activation is be fundamentally reorganized in cells undergoing essential for GADD45G-mediated inhibition of Stat3. senescence. Therefore, we analyzed the alterations in Indeed, treatment with NSC 87877 significantly several components of different signaling pathways in increased the levels of phosphorylated Stat3, Jak2, and cells with or without GADD45G overexpression. Intri- Tyk2 in both Sk-Hep1 and LPC-H-RasV12 cells (Fig. guingly, we found that the constitutive phosphoryla- 5B). Furthermore, we found that knockdown of Shp2 tion of Stat3 (Tyr705) was substantially inhibited by by siRNA efficiently abrogated GADD45G-induced GADD45G expression in the human liver cancer cell dephosphorylation of Stat3 (Fig. 5C) and profoundly lines, Sk-Hep1, SMMC-7721, and Hep3B (Fig. 4A). inhibited senescence induction, as indicated by SA-b- More specifically, in Sk-Hep1 cells, we found that lev- gal and cell-cycle analysis (Fig. 5D,E). In accord with els of Tyr705-phosphorylated Stat3 rapidly decreased, the observation, we also found that JAK/Stat3 inhibi- starting within 2 hours after GADD45G induction; tion, either by their inhibitors (LLL12 and WP1066) coincidently, levels of phosphorylated Jak2 and tyro- or the siRNAs targeting Stat3, Jak2, and Tyk2, effec- sine kinase 2 (Tyk2), the upstream activators of Stat3, tively restored GADD45G-induced senescence in cells were also severely down-regulated (Fig. 4B). We pretreated with sodium orthovanadate (Supporting extended our observation to two subclones (numbers 1 Figs. 9 and 10). Collectively, we demonstrate that the and 2) of mouse LPC-H-RasV12 cells, wherein stress sensor, GADD45G, negatively regulates JAK- GADD45G could significantly induce cell senescence Stat3 phosphorylation in the early stage of senescence (Fig. 4C). Consistently, levels of phosphorylated Stat3, induction, whereas inhibition of phosphatase Shp2 Jak2, and Tyk2 were also substantially decreased in entirely rescues GADD45G-induced cell senescence. these cells (Fig. 4D). Moreover, we observed a similar Stat3 Reactivation Counteracts GADD45G- down-regulation in Stat3 activation in colon cancer Induced Cellular Senescence. Next, we tested cells in the presence of GADD45G (data not shown). whether Stat3 reactivation can counterbalance These results suggest that JAK-Stat3 inhibition is an GADD45G-induced senescence in liver tumor cells. A early event in GADD45G-mediated cell arrest and cell constitutively active Stat3 mutant that substitutes cys- senescence. teine residues for A661 and N663 (Ca-Stat3) was len- Considering that Stat3, Jak2, and Tyk2 can be tivirally transduced into Sk-Hep1 cells (Supporting directly dephosphorylated by SH2 domain-containing Fig. 11A).20 Prominently, Ca-Stat3 almost entirely protein tyrosine phosphatases (SHPs), we then tested diminished the cellular senescence and cell-cycle arrest whether inhibition of phosphatase activity is sufficient induced by GADD45G (Fig. 6A,B and Supporting to block GADD45G-mediated senescence. The phos- Fig. S11B). Meanwhile, we found that GADD45G phatase inhibitor, sodium orthovanadate (Na3VO4), expression significantly impaired human telomerase was added to culture 30 minutes before DOX induc- reverse transcriptase (hTERT) expression, and that Ca- tion of GADD45G expression. Strikingly, treatment Stat3 expression partially rescued the inhibitory effects with the phosphatase inhibitor resulted in a dose- (Fig. 6C). Furthermore, overexpression of hTERT dependent attenuation of GADD45G-mediated senes- remarkably attenuated GADD45G-induced cellular cence in both human Sk-Hep1 cells and mouse LPC- senescence (Fig. 6D and Supporting Fig. 11C). H-RasV12 cells (Fig. 4E). Accordingly, the reduced We further substantiated the aforementioned obser- expression in phosphorylated Stat3, Jak2, and Tyk2 vation by in vivo tumor growth experiments. We was largely restored (Fig. 4F). We further examined found that fewer than 30% (2 of 8) of mice developed the status of the phosphatases, Shp1 and 2, in Sk- tumors at day 50 after SC inoculation of Sk-Hep1 Hep1 cells and LPC-H-RasV12 cells after GADD45G cells with the inducible expression of GADD45G induction. Results showed that phosphorylated Shp2 (Tet1 GADD45G/Vector group). However, in the 184 ZHANG ET AL. HEPATOLOGY, January 2014

Fig. 3. GADD45G induces senescence independently of the p53 and p16/Rb pathways. (A) Tet-GADD45G-Sk-Hep1 cells were cultured with DOX (Tet) for the indicated times. Cell lysates were harvested for analysis of indicated proteins by western blotting. (B) Sk-Hep1 cells were trans- duced with the control vector or the SV40 LT vector, then treated with 100 nM of doxorubicin for 24 hours. Protein levels of p53 and p21 were examined by western blotting. (C) Tet-GADD45G-Sk-Hep1 cells transduced with control (vector) and SV40 LT were cultured for 4 days with (1) or without (2) GADD45G induction, then subjected to SA-b-gal staining. Representative images (left panel) and the percentage of positive cells (right panel) are shown. (D) Tet-GADD45G-Sk-Hep1 cells were transfected with the indicated siRNAs and then grown for assay of SA-b-gal activity (upper panel); a methyl thiazol tetrazolium assay was performed to determine cell viability (lower panel). Data shown are mean 6 SD from three experiments. (E and F) Tet-GADD45G-Hep3B cells transfected with control (Si Con) or p16 targeting siRNA (Si p16) were cultured for 3 days with (2) or without (1) GADD45G induction. (E) Representative images showing SA-b-gal staining (left panel) and the percentage of positive cells (right panel) are shown. (F) Efficiency of p16 siRNA for knockdown was analyzed by western blotting.

presence of Ca-Stat3 expression, 90% (7 of 8) of mice GADD45G-mediated inhibition of tumor growth in (Tet1 GADD45G/Ca-Stat3 group) developed tumors vivo. (Fig. 6D). GADD45-induced delay of tumor onset GADD45G Expression Is Inversely Correlated was significantly rescued by Ca-Stat3 expression. In With Stat3 Phosphorylation in Primary Human addition, the average volume of tumors from the HCCs. To further investigate whether dys-regulated group (Tet1 GADD45G /Ca-Stat3) was comparable GADD45G-mediated regulation of Stat3 is relevant to to that of the control (Tet2) groups, and tumors were human HCC development, we tested the association much larger than in the Tet1 GADD45G/Vector between GADD45G expression and Stat3 activation in group (Fig. 6E). Histological analyses showed that the human HCC samples. We did immunoblotting assay tumor cells expressing both GADD45G and Ca-Stat3 using another set of human HCC samples, and found displayed a robust augmentation in Ki-67 expression, that most of the human liver tumors with decreased or when compared with tumors that expressed absent GADD45G expression displayed activated Stat3 GADD45G alone (Fig. 6F). These results clearly dem- (Fig. 7A). Immunohistochemical (IHC) staining of onstrate that Stat3 activation counteracts the GADD45G and p-Stat3 in an independent cohort of HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 185

Fig. 4. GADD45G inhibits activation of Jak2, Tyk2, and Stat3 in HCC cells. (A) Sk-Hep1, SMMC-7721, and Hep3B cell lines were cultured for 4 hours with or without GADD45G induction. Levels of p-Stat3 (Tyr705) and GADD45G were detected by western blotting. (B) Tet-GADD45G-Sk- Hep1 cells were cultured with DOX for the indicated time. Levels of p-Stat3 (Tyr705), p-Tyk2 (Tyr1054/1055), and p-Jak2 (Tyr1007/1008) were exam- ined by western blotting. (C and D) Tet-GADD45G-LPC-H-RasV12 cells (No.1 and No.2 clones) were grown for 12 days with or without GADD45G induction. (C) Representative images showing SA-b-gal staining and cell morphology (left panel). The percentage of positive cells is depicted (right panel). Data shown are mean 6 standard deviation from three independent experiments. (D) Levels of p-Stat3, p-Tyk2, p-Jak2, and GADD45G were analyzed 4 hours after GADD45G induction by western blotting. (E and F) Tet-GADD45G-Sk-Hep1 and -LPC-H-RasV12 No.2 cells were treated with different concentrations (100, 250, 500, and 1,000 lM) of Na3VO4 for 0.5 hours, then cultured for the rest of the time of the day with or without GADD45G induction. Treatment of Sk-Hep1 and No.2 cells lasted 4 and 12 days, respectively. (E) Percentages of senescent cells are shown. (F) Western blotting analysis of levels of indicated proteins in cells after 4 hours of GADD45G induction.

75 pairs of HCC specimens demonstrated that a regions, whereas only 25.3% (19 of 75) of HCC decreased expression of GADD45G protein accompa- samples were positive for GADD45G staining (P < nied a higher p-Stat3 staining in continuous tissue sec- 0.001). In contrast, p-Stat3 staining was positive (score, 4- tions from the same patient in HCC sections (Fig. 7B; 12) in approximately 4% (3 of 75) of nontumor sections Supporting Table 2). Positive staining of GADD45G and in 49.3% (37 of 75) of tumor areas (Fig. 7C). Statisti- (defined as a score greater than 4) was detected in cal analysis showed a significant difference in p-Stat3 approximately 89.3% (67 of 75) of adjacent nontumor expression in HCC regions between GADD45G-negative 186 ZHANG ET AL. HEPATOLOGY, January 2014

Fig. 5. Shp2 is required for GADD45G-mediated inhibition of the Jak/Stat3 pathway. (A) Tet-GADD45G-Sk-Hep1 (upper panel) and Tet- GADD45G-LPC-H-RasV12 No.2 cells (lower panel) were cultured for the indicated times with (Tet1) or without (Tet2) GADD45G induction. Protein levels of p-Shp2 (Tyr580) were measured by western blotting. (B) Tet-GADD45G-Sk-Hep1 (left panel) and Tet-GADD45G No.2 (right panel) cells were treated with NSC-87877 (10 lM) for 4 hours with or without GADD45G induction. Cell extracts were analyzed by western blotting using the indicated primary antibodies. (C) Sk-Hep1 cells were transfected with control siRNA (Si Con) or with siRNA targeting Shp2 (Si Shp2), then cul- tured with or without GADD45G induction for 4 hours. Western blotting analysis of the indicated proteins is shown. (D) After the treatment of indicated siRNA, Tet-GADD45G-Sk-Hep1 cells were grown for 4 days with or without GADD45G induction. Representative images showing SA-b- gal staining (left panel) and the percentage of positive cells (right panel) are shown. (E) After the indicated treatment for 24 hours, Tet- GADD45G-Sk-Hep1 cells were harvested for analysis of the percentages of S-phase cells by fluorescence-activated cell sorting. Data shown are mean 6 standard deviation from three independent experiments (**P < 0.01). and -positive groups (Fig. 7D, left). When categorized cells. In line with our observation, compelling evidence with p-Stat3 expression in HCC sections, the positive demonstrates that Stat3 hyperactivation functions group had significantly lower GADD45G expression than downstream of many pro-oncogenic signals, including the negative group (Fig. 7D, right). Moreover, oncogenes and growth factors, to promote tumor GADD45G expression was significantly lower in poorly development.21-23 In human HCC, Stat3 is constitu- or moderately differentiated tumor tissues than in well- tively activated and correlated with tumor progres- differentiated tissues (Fig. 7E). Overall, the IHC scores of sion.23-25 Therefore, decreased GADD45G expression, GADD45G and p-Stat3 in HCC cells were inversely cor- concurrently with Stat3 activation in human HCC, related in the cohort. Altogether, these results indicate that strongly suggests the functional relevance of GADD45G expression is inversely correlated with Stat3 GADD45G-mediated senescence in attenuating HCC phosphorylation and tumor progression in human HCC. development. Loss of function of the key effectors in senescence Discussion induction, such as p53, Rb, and INK4a-ARF, frequently occurs in primary HCC. Recently, cellular senescence The present study demonstrates that GADD45G has been programmed in the absence of activation of functions as a senescence inducer and has a robust the key cycle checkpoints.5,26-28 For example, p16/RB- capability for inhibiting Jak-Stat3 activation in HCC and p53-independent senescence is elicited by oncogenic HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 187

Fig. 6. Stat3 reactivation inhibits GADD45G-induced cellular senescence and restores tumor growth in vivo. (A-C) Tet-GADD45G-Sk-Hep1 cells stably expressing Ca-Stat3 and the control cells were cultured for 4 days with or without GADD45G induction. (A) Representative images showing SA-b-gal activity (left panel) and the percentage of positive cells (right panel) are shown. (B) Percentage of S-phase cells was measured by fluorescence-activated cell sorting analysis. (C) hTERT mRNA expression was measured by qRT-PCR. The data shown are mean 6 standard devi- ation from three independent experiments (*P < 0.05; **P < 0.01). (D) Tet-GADD45G-Sk-Hep1 cells with or without hTERT overexpression were cultured for 4 days under the indicated treatment. Morphology and SA-b-gal activity (left panel) and the percentage of positive cells (right panel) are shown (**P < 0.01). (E and F) Tet-GADD45G-Sk-Hep1 cells (6 3 106) with or without Ca-Stat3 expression were SC injected into the flanks of nude mice. Mice were fed with drinking water supplemented with (Tet1) or without (Tet2) DOX. (E) Kaplan-Meier’s analysis of tumor onset (left panel, n 5 8 mice for each group; the group of GADD45G; Tet1 mice versus the control groups of mice; *P < 0.05). Volumes of tumor were measured at indicated time intervals (right panel). (F) Tumor sections were subjected to H&E staining and IHC against p-Stat3, Ki-67; origi- nal magnification, 3400. H&E, hematoxylin and eosin.

RAS signaling, wherein transforming growth factor beta nant liver or HCC cells. Notably, using a mouse model, signaling is required.26 Down-regulation of p300 his- Tront et al. demonstrated that GADD45A, a member tone acetyltransferase activity induces senescence inde- of the GADD45 family, suppresses Ras-driven mam- pendently of p53, p21, and p16 function.28 We found mary tumorigenesis by triggering cell apoptosis and that GADD45G-induced senescence did not require the senescence,29 although we found that GADD45A and presence of p53, p16, and Rb in HCC cells. These B did not induce senescence in various HCC cell lines. findings highlight that, in addition to classical initiators These findings suggest that identification of a new of cell senescence, such as p53, Rb, and INK4a-ARF, senescence inducer is significant for understanding the other proteins may trigger cellular senescence to inhibit mechanism of tumor development. tumor development in an independent manner or as a Cells undergoing senescence require a highly coordi- complementary mechanism. Dys-regulation of these nated regulation of multiple signaling pathways. We nonclassical senescence inducers may be essential for found that GADD45G expression promptly resulted efficient tumorigenesis. In this regard, GADD45G in the dephosphorylation of Stat3 in different HCC down-regulation is considered to be one of the impor- cell lines as well as other types of tumor cell lines tant events that lead to senescence evasion in premalig- (data not shown). Phosphorylated Tyk2 and Jak2 were 188 ZHANG ET AL. HEPATOLOGY, January 2014

Fig. 7. GADD45G expression in human HCC is inversely associated with Stat3 activation and clinical parameters. (A) Western blotting analysis of p-Stat3 (Tyr705) and GADD45G in human HCC (T) and adjacent nontumor tissues (N). (B) Representative images of p-Stat3 and GADD45G IHC staining. Original magnification, 3200 (upper panel); 3400 (middle panel). (C) Box-and-whisker plots of the staining. The immunoreactive score is shown as mean 6 SD (***P < 0.001). (D) Statistical analysis of p-Stat3 expression in HCC sections negative (score, 0-3) and posi- tive (score, 4-12) for GADD45G staining (left panel; P 5 0.0084). p-Stat3-negative HCC sections exhibited a significantly higher GADD45G staining (right panel; P 5 0.0094). The immunoreactive score is shown as mean 6 SD. (E) Analysis of GADD45G expression in well, moder- ately, or poorly differentiated HCC samples. The immunoreactive score is shown as mean 6 SD (**P < 0.01). SD, standard deviation. down-regulated upon GADD45G induction. Coinci- cells. One interesting aspect of the study is that dently, GADD45G efficiently activated the phospha- GADD45G suppresses hTERT expression. This obser- tase, Shp2, which dephosphorylates Stat3, Tyk2, and vation is supported by previous studies that show that Jak2. Alterations in activities of Shp2 and Stat3 are hTERT expression confers cellular senescence resist- functionally relevant in that either down-regulation of ance.32,33 In fact, we found that restoration of hTERT Shp2 or restoration of Stat3 activity can efficiently was sufficient to block GADD45G-induced senescence. counteract GADD45G-induced senescence. Impor- The clinical relevance of GADD45G down- tantly, a recent study demonstrates that hepatic defi- regulation or absence in HCC was further addressed in ciency of Shp2 in mice dramatically promotes DEN- this study. We found that GADD45G was down- induced HCC development and that decreased Shp2 regulated in approximately 75% of HCCs and was sig- expression occurs in certain cases of human HCC nificantly correlated with tumor staging. More impor- specimens.30 Interestingly, a recent study also describes tant, GADD45G levels in tumor tissues were inversely a role of Stat3 in driving cell senescence in nontrans- correlated with Stat3 activation in HCC. In summary, formed cells.31 Therefore, the cell context-dependent this study provides novel insight into a function of roles of Stat3 in cellular senescence remain a paradox GADD45G in inhibiting the Jak-Stat3 pathway and in the field, most likely because of the diversities in attenuating growth of HCC cells through induction of Stat3 targets or in cross-talk with other signals in these tumor cell senescence. The present findings identify a HEPATOLOGY, Vol. 59, No. 1, 2014 ZHANG ET AL. 189 link between stress sensor-initiated senescence and HCC 17. Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic inhibition that may enhance our understanding of ras provokes premature cell senescence associated with accumulation of p53 and p16INK4a. Cell 1997;88:593-602. human hepatotumorigenesis and lead to the develop- 18. Jiang D, Srinivasan A, Lozano G, Robbins PD. SV40 T antigen abro- ment of novel treatment strategies. gates p53-mediated transcriptional activity. Oncogene 1993;8:2805- 2812. 19. Bargonetti J, Reynisdottir I, Friedman PN, Prives C. Site-specific bind- References ing of wild-type p53 to cellular DNA is inhibited by SV40 T antigen and mutant p53. Genes Dev 1992;6:1886-1898. 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global 20. Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, Pestell RG, cancer statistics. CA Cancer J Clin 2011;61:69-90. Albanese C, Darnell JE, Jr. Stat3 as an oncogene. Cell 1999;98: 2. Kang TW, Yevsa T, Woller N, Hoenicke L, Wuestefeld T, Dauch D, 295-303. et al. Senescence surveillance of pre-malignant hepatocytes limits liver 21. Al Zaid Siddiquee K, Turkson J. STAT3 as a target for inducing cancer development. Nature 2011;479:547-551. apoptosis in solid and hematological tumors. Cell Res 2008;18:254- 3. Xue W, Zender L, Miething C, Dickins RA, Hernando E, 267. Krizhanovsky V, et al. Senescence and tumour clearance is triggered by 22. Lesina M, Kurkowski MU, Ludes K, Rose-John S, Treiber M, Kloppel p53 restoration in murine liver carcinomas. Nature 2007;445:656-660. G, et al. Stat3/Socs3 activation by IL-6 transsignaling promotes pro- 4. Menard L, Taras D, Grigoletto A, Haurie V, Nicou A, Dugot-Senant gression of pancreatic intraepithelial neoplasia and development of pan- N, et al. In vivo silencing of Reptin blocks the progression of human creatic cancer. Cancer Cell 2011;19:456-469. hepatocellular carcinoma in xenografts and is associated with replicative 23. He G, Karin M. NF-kappaB and STAT3—key players in liver inflam- senescence. J Hepatol 2010;52:681-689. mation and cancer. Cell Res 2011;21:159-168. 5. Senturk S, Mumcuoglu M, Gursoy-Yuzugullu O, Cingoz B, Akcali 24. Pilati C, Amessou M, Bihl MP, Balabaud C, Nhieu JT, Paradis V, et al. KC, Ozturk M. Transforming growth factor-beta induces senescence in Somatic mutations activating STAT3 in human inflammatory hepato- cellular adenomas. J Exp Med 2011;208:1359-1366. hepatocellular carcinoma cells and inhibits tumor growth. HEPATOLOGY 2010;52:966-974. 25. Lee TK, Castilho A, Cheung VC, Tang KH, Ma S, Ng IO. CD24(1) 6. Tannapfel A, Busse C, Weinans L, Benicke M, Katalinic A, Geissler F, liver tumor-initiating cells drive self-renewal and tumor initiation et al. INK4a-ARF alterations and p53 mutations in hepatocellular car- through STAT3-mediated NANOG regulation. Cell Stem Cell 2011;9: cinomas. Oncogene 2001;20:7104-7109. 50-63. 7.NishidaN,GoelA.Geneticandepigenetic signatures in human hepato- 26. Cipriano R, Kan CE, Graham J, Danielpour D, Stampfer M, Jackson cellular carcinoma: a systematic review. Curr Genomics 2011;12:130-137. MW. TGF-beta signaling engages an ATM-CHK2-p53-independent 8. Gierl MS, Gruhn WH, von Seggern A, Maltry N, Niehrs C. RAS-induced senescence and prevents malignant transformation in GADD45G functions in male sex determination by promoting p38 sig- human mammary epithelial cells. Proc Natl Acad Sci U S A 2011;108: naling and Sry expression. Dev Cell 2012;23:1032-1042. 8668-8673. 9. Zerbini LF, Libermann TA. Life and death in cancer. GADD45 alpha 27. Lin HK, Chen Z, Wang G, Nardella C, Lee SW, Chan CH, et al. and gamma are critical regulators of NF-kappaB mediated escape from Skp2 targeting suppresses tumorigenesis by Arf-p53-independent cellu- programmed cell death. 2005;4:18-20. lar senescence. Nature 2010;464:374-379. 10. Campanero MR, Herrero A, Calvo V. The histone deacetylase inhibitor 28. Prieur A, Besnard E, Babled A, Lemaitre JM. p53 and p16(INK4A) trichostatin A induces GADD45 gamma expression via Oct and NF-Y independent induction of senescence by chromatin-dependent alteration binding sites. Oncogene 2008;27:1263-1272. of S-phase progression. Nat Commun 2011;2:473. 11. Ying J, Srivastava G, Hsieh WS, Gao Z, Murray P, Liao SK, et al. The 29. Tront JS, Hoffman B, Liebermann DA. Gadd45a suppresses Ras-driven stress-responsive gene GADD45G is a functional tumor suppressor, mammary tumorigenesis by activation of c-Jun NH2-terminal kinase with its response to environmental stresses frequently disrupted epige- and p38 stress signaling resulting in apoptosis and senescence. Cancer netically in multiple tumors. Clin Cancer Res 2005;11:6442-6449. Res 2006;66:8448-8454. 12. Sun L, Gong R, Wan B, Huang X, Wu C, Zhang X, et al. GADD45g- 30. Bard-Chapeau EA, Li S, Ding J, Zhang SS, Zhu HH, Princen F, et al. amma, down-regulated in 65% hepatocellular carcinoma (HCC) from Ptpn11/Shp2 acts as a tumor suppressor in hepatocellular carcinogene- 23 chinese patients, inhibits cell growth and induces cell cycle G2/M sis. Cancer Cell 2011;19:629-639. arrest for hepatoma Hep-G2 cell lines. Mol Biol Rep 2003;30:249-253. 31. Kong X, Feng D, Wang H, Hong F, Bertola A, Wang FS, Gao B. 13. Zender L, Xue W, Cordon-Cardo C, Hannon GJ, Lucito R, Powers S, Interleukin-22 induces hepatic stellate cell senescence and restricts liver et al. Generation and analysis of genetically defined liver carcinomas fibrosis in mice. HEPATOLOGY 2012;56:1150-1159. derived from bipotential liver progenitors. Cold Spring Harb Symp 32. Dickson MA, Hahn WC, Ino Y, Ronfard V, Wu JY, Weinberg RA, Quant Biol 2005;70:251-261. et al. Human keratinocytes that express hTERT and also bypass a 14. Yang Z, Zhang L, Ma A, Liu L, Li J, Gu J, Liu Y. Transient mTOR p16(INK4a)-enforced mechanism that limits life span become immor- inhibition facilitates continuous growth of liver tumors by modulating the tal yet retain normal growth and differentiation characteristics. Mol maintenance of CD1331 cell populations. PLoS One 2011;6:e28405. Cell Biol 2000;20:1436-1447. 15. Collado M, Serrano M. Senescence in tumours: evidence from mice 33. Harada H, Nakagawa H, Oyama K, Takaoka M, Andl CD, Jacobmeier and humans. Nat Rev Cancer 2010;10:51-57. B, et al. Telomerase induces immortalization of human esophageal 16. Kuilman T, Michaloglou C, Mooi WJ, Peeper DS. The essence of keratinocytes without p16INK4a inactivation. Mol Cancer Res 2003;1: senescence. Genes Dev 2010;24:2463-2479. 729-738.