Oncogene (2015) 34, 237–245 & 2015 Macmillan Publishers Limited All rights reserved 0950-9232/15 www.nature.com/onc

ORIGINAL ARTICLE MIF4G domain containing regulates and hepatic carcinogenesis by antagonizing CDK2-dependent p27 stability

CWan1,4, S Hou2,4,RNi2,LLv2, Z Ding3, X Huang2, Q Hang3,SHe3, Y Wang3, C Cheng3,XXGu3,GXu1 and A Shen3

The CDK inhibitor p27kip1 plays crucial roles in cell cycle regulation and cancer progression. Through yeast two-hybrid screening, we identified MIF4G domain containing protein (MIF4GD) as a novel binding partner for p27. The association of MIF4GD and p27 was verified using immunoprecipitation and glutathione S-transferase (GST) pull-down assays. Interaction with MIF4GD led to the stabilization of p27 both in the nucleus and in the in (HCC) cells as a result of suppressed of p27 by CDK2 at threonine187. Serum stimulation decreased the levels of MIF4GD and p27 simultaneously. In addition, MIF4GD overexpression resulted in increased p27 levels and reduced cell proliferation, while knockdown of MIF4GD promoted cell cycle progression with decreased p27 levels in cells. Furthermore, overexpression of MIF4GD reduced colony formation and inhibited xenograft tumor growth in nude mice. Finally, we found that both MIF4GD and p27 were expressed at low levels in HCC tissues compared to non-cancerous tissues, and that low expression levels of MIF4GD and p27 were associated with significantly worse prognosis in HCC patients. Our results suggest that MIF4GD is a potential regulator of p27-dependent cell proliferation in HCC. These findings provide a rational framework for the development of potential HCC therapy by targeting the MIF4GD–p27 interaction.

Oncogene (2015) 34, 237–245; doi:10.1038/onc.2013.536; published online 16 December 2013 Keywords: p27; hepatocellular carcinoma; cell cycle regulation; MIF4GD

INTRODUCTION stage and poorer disease-free survival in patients with HCC.8,11,14 Hepatocellular carcinoma (HCC) represents the fifth most common Previous studies have shown that phosphorylation of p27 protein malignancy and the third leading cause of cancer-related deaths (T187) by the cyclin E/CDK2 complex in mid-to-late 15,16 in the world, especially in Asia and Africa. The high frequency of targets p27 to -mediated degradation, allowing G1–S HCC is mainly attributed to an increasing incidence of alcoholic transition to occur. Notably, hyperactive cyclin E/CDK2 can liver disease, along with hepatitis B or hepatitis C viral infections.1 promote nuclear p27 degradation in cancer cells, partly account- Despite vigorous efforts by both basic scientists and clinical ing for disrupted cell cycling and consequently malignant 9,17 practitioners, the underlying mechanisms of the pathogenesis of proliferation. The molecular switch that modulates p27 HCC remain poorly understood.2 Disruption of cell cycle control is activity in hepatocarcinogenesis is not clearly defined. one of the most frequent alterations in tumor cells, which often Our previous investigations have revealed a correlation results in uncontrolled cell proliferation.3 Therefore, many studies between p27 protein level and hepatocarcinogenesis with the have sought to elucidate the mechanism(s) underlying the underpinning that a ubiquitin–proteasome system is responsible 18,19 uncontrolled cell growth during hepatocarcinogenesis.4,5 These for p27 protein stability during that process. This observation studies have provided key evidence that cell cycle regulators, prompted us to search for that could potentially regulate including pRb, p16INK4A, p21WAF1/CIP1 and p27Kip1, when aberrantly p27 stability and as such influence its activity in hepatocarcino- expressed or malfunctioning, may contribute to the pathogenesis genesis. Using yeast two-hybrid screening, we identified MIF4G of HCC.6–8 domain containing protein (MIF4GD) as a novel binding partner p27Kip1 belongs to the family of cyclin-dependent kinase for p27. MIF4GD, also known as stem-loop interacting protein 1, is inhibitors that encompass several structurally and functionally involved in replication-dependent of related proteins. Cyclin-dependent kinase inhibitors ensure the mRNAs.20,21 Interaction with MIF4GD stabilized p27 by correct timing of CDK activation in different phases of cell cycle antagonizing CDK2-mediated phosphorylation and subsequently and are thus considered tumor suppressors.9,10 p27 directly proteasomal degradation, rendering the cyclin E/CDK2 complex inhibits the cyclin E/CDK2 complex, thereby inducing G1-phase inactive and stalling cell cycle progression in HCC, but not normal arrest and halting cell cycle progression.9,11 Downregulation of cells. Importantly, high MIF4GD expression levels were found to p27 has been detected in various cancers,8,9 including correlate with suppressed cell proliferation and improved HCC, suggesting that p27 may contribute to liver oncogenesis.12,13 prognosis in patients with HCC. Therefore, our finding suggests In addition, it has been reported that reduced p27 expression is that MIF4GD may function as a suppressor of HCC by linking p27 associated with a higher tumor grade, more advanced tumor stabilization to decelerated cell cycle progression.

1School of Public Health, Nantong University, Nantong, People’s Republic of China; 2Department of Gastroenterology, Nantong University, Nantong, People’s Republic of China and 3Department of Immunology, Nantong University, Nantong, People’s Republic of China. Correspondence: Professor A Shen, Department of Immunity, Nantong University, 19 Qixiu Road, Nantong 226001, People’s Republic of China. E-mail: [email protected] 4These authors contributed equally to this work. Received 14 April 2013; revised 9 November 2013; accepted 11 November 2013; published online 16 December 2013 MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 238 RESULTS p27 in two different HCC cell lines without altering the abundance MIF4GD interacts with p27 of . In contrast, depletion of endogenous MIF4GD by short To tackle how p27 activity is regulated during cell cycling, we hairpin RNA (shRNA) accelerated degradation of p27 in Huh7 and performed a yeast two-hybrid screening using p27 as a bait. HepG2 HCC cells (Figures 3c and d). MIF4GD was identified in this process as a potential binding p27 protein is present both in the nucleus and in the cytoplasm partner for p27 (data not shown). The interaction between p27 with distinct functionalities. In the nucleus, p27 serves as a CDK and MIF4GD was verified by reciprocal immunoprecipitations as inhibitor to impede cell proliferation, while cytoplasmic p27 binds p27 was brought down by an anti-MIF4GD from 293T with RhoA to promote cell migration. We then investigated cell lysate and vice versa (Figure 1a). Similarly, we observed that whether MIF4GD could influence p27 subcellular distribution. To MIF4GD interacted with p27 in normal liver tissue and HCC cells, this end, Huh7 cells transfected with MIF4GD-myc were subjected while the interactions of MIF4GD and p27 with another important to cell fractionation followed by western blot analysis. As shown in cell cycle regulator, p53, were not detected (Figure 1b). To Figure 3e, both nuclear p27 and cytoplasmic p27 were increased evaluate whether the association of p27 with MIF4GD was direct, in the presence of ectopic MIF4GD. However, the ratio of nuclear GST pull-down assay was performed. Indeed, GST-p27, but not and cytoplasmic p27 did not alter significantly. Collectively, these GST, precipitated with HA-tagged MIF4GD (Figure 1c). These data suggest that MIF4GD stabilizes p27 protein in HCC cells. observations were corroborated by immunofluorescence assay that showed partial co-localization of p27 and MIF4GD in HCC cells MIF4GD modulates the phosphorylation status of p27 (Figure 1d). The degradation of p27 is mainly regulated through phosphoryla- To further delineate the p27–MIF4GD interaction, we performed tion-dependent ubiquitin–proteasome proteolysis. Three major immunoprecipitation assays employing various p27 and MIF4GD phosphorylation sites have been reported to regulate the stability mutants. As shown in Figure 2, p27 interacted with MIF4GD via a of p27, namely S10, T187 and T198.22 Thus, we examined whether region flanked by the CDK binding domain and the nucleus MIF4GD regulates p27 stability by virtue of modulating the localization signal motif, while MIF4GD interacted with p27 via its phosphorylation status of p27. As shown in Figure 4a, while the MIF4G domain. Taken together, these data suggest that p27 phosphorylation of S10 and T198 sites seemed to be elevated, interacts with MIF4GD in vivo and in vitro, indicating that MIF4GD T187 phosphorylation was reduced. In order to confirm the could potentially influence p27 function. hypothesis that MIF4GD impacts p27 stability in a T187 phosphorylation-dependent manner, we assessed the stability of both wild-type (WT) and phosphorylation-defective (S10A, T187A MIF4GD stabilizes p27 in HCC cells and T198A) p27 proteins either in the absence or in the presence Of great intrigue, the level of endogenous p27 increased of MIF4GD. Indeed, MIF4GD prolonged the half-life of the wild dramatically in MIF4GD-transfected cells (Figure 1d, white type (WT), S10A and T198A p27, but failed to impact the T187A arrowheads). Thus, we investigated the detailed mechanism mutant (Figure 4b). Importantly, MIF4GD suppressed ubiquitina- for how MIF4GD regulates p27 expression in HCC cells and tion of p27, but not that of p53, in Huh7 cells (Figure 4c and found that it did not influence p27 mRNA level or p27 translation Supplementary Figure 3). In agreement with these data, treatment (Supplementary Figures 1 and 2), indicating that MIF4GD with MG132, a proteasome inhibitor, stabilized p27 either in the augmented p27 protein level via a post-translational mechanism. presence or in the absence of MIF4GD (Figure 4d). Combined, Then, we probed p27 protein stability either in the presence or in these findings indicate that MIF4GD may serve as an important the absence of MIF4GD. As shown in Figures 3a and b, link between repressed p27 phosphorylation–ubiquitination and overexpression of MIF4GD significantly extended the half-life of augmented p27 stability.

Figure 1. MIF4GD interacts with p27 in vivo and in vitro.(a) Reciprocal immunoprecipitation of MIF4GD and p27 in 293T cells. Lysates of 293T cells were immunoprecipitated (IP) with anti-MIF4GD, anti-p27 or control IgG. The immunoprecipitates were subjected to western blot analysis with anti-p27 and anti-MIF4GD antibodies. (b) Verification of the interaction between MIF4GD and p27 in HCC cells and normal liver tissue using immunoprecipitation assay. Total cell lysates were isolated from HuH7 cells and human normal liver tissue, and equal amounts of total protein were immunoprecipitated using anti-MIF4GD, anti-p27 antibodies or control IgG (*non-specific band). Immunoprecipitates were resolved by SDS-PAGE and immunoblotted with indicated antibodies. (c) Pull-down of HA-MIF4GD from transfected 293T whole-cell extracts with GST or GST-p27 fusion protein. (d) Immunofluorescence analysis of p27 and MIF4GD in Huh7 cells. Huh7 cells transfected with MIF4GD-myc were subjected to immunofluorescence assay using anti-p27 and anti-myc antibodies.

Oncogene (2015) 237 – 245 & 2015 Macmillan Publishers Limited MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 239

Figure 2. Identification of the domains required for MIF4GD–p27 interaction. (a) Schematic diagram of full-length p27 and its mutants. (b) Immunoprecipitation of HA-tagged MIF4GD with myc-tagged p27 and its mutants. Lysates from 293T cells transfected with HA-MIF4GD or its truncated forms and myc-tagged full-length p27 were immunoprecipitated with anti-myc antibody, and immunoblotted with indicated antibodies. p27 binds MIF4GD with its 94–151aa region. (c) Schematic diagram of full-length MIF4GD and its two structural mutants. (d) Immunoprecipitation of HA-tagged MIF4GD and its mutants with myc-tagged p27. MIF4GD interacts with p27 through its MIF4G domain.

MIF4GD inhibits cell proliferation in HCC cells Fancd2, Heatr1 and Rbl2 (Supplementary Figure 2). These results It has been well documented that nuclear p27 inhibits CDK2 suggest that MIF4GD regulates cell proliferation at least in part by activity and cell proliferation. Because MIF4GD promoted the enhancing p27-dependent inhibition of CDK2 and transcriptional nuclear enrichment of p27, we asked whether MIF4GD could also repression of downstream target . have a role in cell cycle regulation. Consistent with previous reports, p27 levels were decreased gradually with time following MIF4GD decreases colony formation and tumor growth of HCC cells the release of cells from starvation (Figure 5a). Of high interest, As our data have found that MIF4GD could stabilize cellular p27 MIF4GD levels were changed in a similar kinetics. This piece of and reduce the proliferation of HCC cells, we then determined data not only reaffirmed the notion that MIF4GD is a key factor whether MIF4GD could regulate colony formation and in vivo maintaining intracellular p27 levels, but alluded to the possibility tumor growth of HCC cells. A pBabe retrovirus system was that MIF4GD might play a role in regulating cell proliferation. exploited to prepare MIF4GD-stable-expressing HCC cells. After To directly probe the role of MIF4GD in cell cycle regulation, we virus infection and selection with puromycin, the expression of transfected HCC cells with the MIF4GD-overexpressing construct tagged MIF4GD was verified in Bel-7404 cells (Figure 6a). MIF4GD followed by EDU proliferation assay to measure the proliferation of dramatically reduced the size of formed colonies (Figure 6b). HCC cells (Supplementary Figure 4). Overexpression of MIF4GD Furthermore, to confirm the biological effect of MIF4GD in vivo,we remarkably decreased the proportion of EDU-positive cells, injected control or MIF4GD-expressing HCC cells into the flanks of indicating suppressed cell proliferation. In addition, more HCC BALB/c nude mice. We found that MIF4GD overexpression cells were found in the G1 phase in the presence of ectopic resulted in significantly reduced tumor growth accompanied by MIF4GD (Figures 5b and c), while MIF4GD elimination facilitated decreased expression of Ki-67 in xenograft tumors (Figures 6c and G1–S transition of HCC cells (Supplementary Figure 5). Further, d). In addition, the expression of p27 was significantly elevated. MTT assay indicated that the proliferation of Bel-7404 HCC cells Thus, these findings validated MIF4GD as a key modulator of p27 was significantly decreased after MIF4GD overexpression to inhibit tumor growth in HCC cells in vitro and in vivo. (Figure 5d). Together, these data demonstrate that MIF4GD participates in cell cycle regulation by putting a hold on G1–S Expression of MIF4GD is correlated with p27 in HCC samples transition. As our results so far strongly implicated MIF4GD as a potential modulator of p27 and as p27 has a critical role in hepatocarci- MIF4GD increases the activity of p27 to inhibit CDK2 activity and nogenesis, we asked whether MIF4GD levels would be altered in repress transcriptional target genes patients with HCC. Indeed, western blot and immunohistochem- Serving as a CDK inhibitor, p27 allosterically inhibits the activity of istry analyses both demonstrated that MIF4GD protein levels were several CDKs, including CDK2, CDK4 and CDK6. As MIF4GD could significantly downregulated in human HCC samples compared trigger the accumulation of p27 in HCC cells, we investigated with adjacent non-tumor tissues, accompanying a decrease in p27 whether the activity of CDK2 could be influenced by MIF4GD. We levels and an increase in Ki-67 levels (Figures 7a and b). assessed the ability of p27 to inhibit CDK2 using two different Furthermore, linear regression analysis confirmed that expression strategies. First, we probed the p27-binding capacity on CDK2 of MIF4GD in HCC tissues was positively correlated with that of using immunoprecipitation assay, which indicated that the level of p27, but inversely correlated with that of Ki-67 (Figure 7c). CDK2-bound p27 was increased when MIF4GD was overexpressed Notably, low levels of MIF4GD were correlated with reduced p27 (Figure 5e). In addition, in vitro kinase assay using recombinant expression, and histological grade (Po0.001), cirrhosis (P ¼ 0.05) histone H1 as a substrate demonstrated that ectopic expression of and Vein invasion (P ¼ 0.024) in HCC tissues (Supplementary MIF4GD led to suppressed CDK2 activity. Table 1). Recent evidence indicates that, apart from the role as a CDK Finally, we compared the cancer-free survival rate of patients inhibitor, p27 might also serve as a transcriptional repressor to with high MIF4GD expression levels with that of patients with low inhibit the expression of a number of target genes.23 Thus, we MIF4GD levels. As shown in Figure 8a, patients whose MIF4GD determined whether MIF4GD might have any influence on p27 levels were on the lower tertile (p33.63%) had a significantly target genes. MIF4GD overexpression significantly repressed the worse prognosis (median cancer-free survival time: 34 months) mRNA levels of all p27 target genes examined, including Coq5, compared with those whose MIF4GD levels were on the upper

& 2015 Macmillan Publishers Limited Oncogene (2015) 237 – 245 MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 240

Figure 4. MIF4GD modulates p27 stability through phosphorylation- and ubiquitination-dependent manners. (a) MIF4GD influences the phosphorylation status of p27. Huh7 cells were transfected with pCMV-HA or HA-tagged MIF4GD construct. Twenty-four hours after transfection, cells were lysed and blotted with the indicated antibodies. (b) p27 or its phosphorylation-defective forms were co-transfected with empty vector or HA-tagged MIF4GD into Huh7 cells. Twenty-four hours after transfection, cells were treated with 25 mg/ml CHX for the indicated period of time, and then analyzed for p27 stability by western blot. (c) MIF4GD inhibits the ubiquitination of p27. Huh7 cells were transfected with HA-tagged MIF4GD. Twenty-four hours after transfection, cells were treated with 20 mM MG132 for 6 h. The whole-cell extracts were immunoprecipitated with anti-p27 antibody, and blotted with the indicated antibodies. (d) Proteasome inhibitor MG132 stabilizes p27 regardless of the absence or presence of MIF4GD. Huh7 cells transfected with empty Figure 3. MIF4GD stabilizes p27 in HCC cells. (a) Overexpression of or MIF4GD-myc construct were treated with or without 20 mM MIF4GD enhances the stability of p27 in Huh7 cells. Huh7 cells were MG132 for 8 h. The cells were lysed and blotted with the indicated transfected with pCMV-myc or MIF4GD-myc construct. Twenty-four antibodies. hours after transfection, cells were treated with 25 mg/ml cyclohex- imide (CHX) for the indicated period of time and subjected to western blot analysis. (b) Overexpression of MIF4GD promoted the stability of p27 in Bel-7404 cells. Bel-7404 cells transfected with DISCUSSION pCMV-HA or MIF4GD-HA construct were treated with 25 mg/ml Aberrant regulation of cell cycling inevitably causes lesions that CHX for the indicated period of time followed by western blot when not properly addressed will eventually transform into analysis. (c) Interference of MIF4GD reduces the stability of p27 in malignancies.24 Investigations in the last couple of decades have Huh7 cells. Huh7 cells were transfected with pSilencer 4.1, demonstrated that key factors within the cell cycle machinery are shMIF4GD-1 or shMIF4GD-2 constructs. 48 h after transfection, cells 25,26 were treated with 25 mg/ml CHX for indicated period of time. targeted in the pathogenesis of HCC. In particular, the cyclin/ (d) Knockdown of MIF4GD decreases p27 level in HepG2 cells. CDK inhibitor p27 has frequently been inactivated in patients with HepG2 cells were transfected with pSilencer 4.1, shMIF4GD-1 or HCC either due to repressed or due to accelerated shMIF4GD-2 constructs. Cells were directly lysed and blotted with protein degradation.18,27,28 Here, we report that MIF4GD interacts indicated antibodies 48 h after transfection. (e) MIF4GD doesn’t with and stabilizes p27 protein in HCC cells, shedding new light on influence the cytoplasmic and nuclear distribution of p27. Huh7 cells the regulation of p27 during hepatocarcinogenesis. transfected with pCMV-myc or myc-tagged MIF4GD were subjected Activity of p27 is regulated by a range of post-translational to cell fractionation analysis. Nuclear (N) and cytoplasmic (C) modifications (PTM), including ,29 neddylation30 and fractions were separated with 12% SDS–PAGE and blotted with 22 the indicated antibodies. The ratio of the relative intensities of phosphorylation. Among these modifications, phosphorylation nuclear and cytoplasmic p27 bands in the absence or presence of has been most extensively studied. Several phosphorylation sites MIF4GD-myc was analyzed. have been identified that could regulate the nucleus enrichment and protein stability of p27. For instance, KIS phosphorylates p27 at serine 10 to surpass cell cycle arrest and promote cell two tertiles (median survival time: 120 months). Similarly, high p27 proliferation at the expense of p27 degradation in leukemia expression level could also herald a better outcome of patients cells.31 Akt and PKC have been shown to phosphorylate p27 at with HCC (63 months for p27high patients vs 13.5 months for threonine 198 with disparaging effects. Whereas phosphorylation p27low patients) (Figure 8b). Finally, when both MIF4GD and p27 by Akt is synonymous with nuclear exclusion and reduced activity expression levels were taken into account to evaluate HCC of p27 in breast cancer cells,32 phosphorylation by PKC leads to prognosis, we reached the same conclusion that high MIF4GD/ enhanced p27 stability and cell cycle arrest.33 On the other hand, p27 expression could afford longer cancer-free survival time (60.5 unambiguous evidence has been presented that phosphorylation vs 15 months) (Figure 8c). of threonine 187 by CDK2 causes rapid degradation of p27.34 Here,

Oncogene (2015) 237 – 245 & 2015 Macmillan Publishers Limited MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 241

Figure 5. MIF4GD inhibits CDK2 activity and the proliferation of HCC cells. (a) The expression profiles of MIF4GD and p27 in Bel7404 HCC cells after release from serum starvation. Bel-7404 cells were serum-starved for 72 h, and released with DMEM medium containing 10% FBS for the indicated period of time. The cells were harvested and analyzed for MIF4GD and p27 expression by western blot. (b) Overexpression of MIF4GD increased cellular p27 and reduced the level of PCNA. Bel-7404 cells mock-transfected, or transfected with empty or MIF4GD-HA vector were subjected to western blot analysis using indicated antibodies. (c) Flow cytometric results indicated that ectopic expression of MIF4GD increases the percentage of G1-phase cells and reduces cells in S-phase (*Po0.05, significant difference with the control group). (d) Overexpression of MIF4GD inhibits proliferation of Bel-7404 HCC cells. Bel-7404 cells mock-transfected, or transfected with pCMV-HA or MIF4GD-HA construct were determined for cell proliferation using MTT assay. (e) Overexpression of MIF4GD reduces CDK2 activity. Bel-7404 cells mock-transfected, transfected with empty vector or HA-MIF4GD were subjected to immunoprecipitation assay with anti-CDK2 antibody. The immunoprecipitated CDK2 were probed with bound p27, or alternatively subjected to in vitro kinase assay using recombinant Histone H1 as substrate. we demonstrate that MIF4GD significantly downregulated T187 CDK2 and releases the cells from arrest. However, the factors that phosphorylation of p27 (Figure 4a). In keeping with this determine the final outcome of p27–CDK2 interaction remain observation, we show that whereas other phosphorylation- largely unclear. Our data suggest that MIF4GD enhances the defective p27 mutants (S10A and T198A) were influenced by interaction between p27 and CDK2 (Figure 5e) but reduces CDK2- MIF4GD to the same extent as wild-type p27, T187A mutant was mediated phosphorylation of p27 (Figure 4a). Therefore, MIF4GD refractory to the presence of exogenous MIF4GD (Figure 4b). may function as a bridge between p27 and CDK2, fine-tuning their Therefore, MIF4GD preserves p27 activity by specifically modulat- interaction such that p27 can retain its ability to bind CDK2 and ing CDK2-mediated T187 phosphorylation. inhibit its activity without risking getting phosphorylated by CDK2. The interaction between CDK2 and p27 confers delicate control Structural studies in the future will likely give a definitive answer to cell cycle progression.34 Enhanced interaction stalls the cell to this proposal. cycle but can potentially increase p27 phosphorylation. Our evaluation of HCC patients has led to the conclusion that Subsequent degradation of p27 disrupts its interaction with MIF4GD expression was markedly suppressed in cancerous tissues

& 2015 Macmillan Publishers Limited Oncogene (2015) 237 – 245 MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 242

Figure 6. MIF4GD inhibits colony formation and in vivo tumorigenicity of HCC cells. (a) Western blot detection of MIF4GD-Flag expression in virus-infected Bel-7404 cells using anti-MIF4GD and anti-Flag antibodies. (b) MIF4GD reduced the colony formation of Bel-7404 cells. For colony formation assay, 200 control or MIF4GD stable-expression Bel-7404 cells were seeded into each well of six-well-plate colonies were stained with crystal violet after 10 days. (c) MIF4GD inhibited tumor growth of Bel-7404 HCC cells in vivo. Control or MIF4GD-stable-expressing Bel-7404 cells were injected s.c. into BALB/c nude mice. The tumor volume were measured and presented (n ¼ 6). (d) Immunohistochemical analysis of control and MIF4GD-overexpressing tumors with p27 and Ki-67 antibodies. Quantitative analysis of p27- and Ki-67-positive rates in each group of tumors is shown on the right side (*Po0.05, significant difference with the control group).

(Figure 7) and that high MIF4GD expression combined with high MIF4GD-SLBP interaction with regard to histone expression and p27 expression is associated with extended survival time cell proliferation in HCC cells remains unclear. (Figure 8). These data indicate that MIF4GD, at least from a Serving as an important cyclin-dependent kinase inhibitors, p27 cumulative point of view, may exert a protective role in has been implicated in the pathology of various human cancers, hepatocarcinogenesis. In addition to the regulation of p27 including HCC. However, unlike many classical tumor suppressors, stability, MIF4GD also played a role in enhancing stem-loop such as p53 and p16, p27 has been rarely found mutated or histone mRNA translation, which has been confirmed in HCC cells deleted in human cancers.10 The aberrant regulation of p27 in (Supplementary Figure 6).35 MIF4GD does not directly bind to cancer cells, to a large extent, takes place at the post- histone mRNAs. Instead, it interacts with a stem-loop binding transcriptional level. Our investigation revealed that MIF4GD protein (SLBP) that binds to the 30-end stem-loop motif of histone could be a novel factor that facilitates the stabilization of p27 by mRNA to regulate 30-end splicing, nuclear export and stability, and modulating CDK2-dependent phosphorylation in HCC cells. the binding of MIF4GD to SLBP might facilitate the stability of However, we found with intrigue that MIF4GD overexpression SLBP or the affinity of SLBP for histone mRNAs.36 Because the failed to impact p27 in non-tumorous L02 hepatocytes expressions of histone proteins are essential for cell cycle progress, (Supplementary Figure 7), implying that the influence of MIF4GD MIF4GD might play a role in cell proliferation and cancer on p27 might be a function of specific intracellular microenviron- progression in this regard. However, the role of SLBP in cell ment. Therefore, we speculate that MIF4GD–p27 interaction could cycle regulation remains controversial.20,37 Expression patterns of protect p27 from degradation in non-tumorous hepatocytes, while SLBP in tumor specimens are largely unknown. Furthermore, down-regulated MIF4GD in HCC cells might disrupt the protective it has been found that histone genes could alternatively produce mechanism mediated by MIF4GD–p27 interaction, which could be 30-end polyadenylated transcripts after SLBP knockdown, rescued following MIF4GD overexpression. As MIF4GD is down- suggesting the existence of a MIF4GD-SLBP-independent regulated in HCC specimens, overexpression of MIF4GD might regulatory mechanism.38,39 Therefore, the exact role of exert a more evident role in promoting p27 accumulation in HCC

Oncogene (2015) 237 – 245 & 2015 Macmillan Publishers Limited MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 243

Figure 7. Correlation of p27 and MIF4GD expression in HCC specimens. (A) The expression of MIF4GD and p27 is correlated in eight representative paired HCC (T) and adjacent normal tissues (N). (B) Immunohistochemical analysis of MIF4GD and p27 in HCC tissues (d, e) and adjacent normal tissues (a, b). Ki-67 (c, f) was used as a positive control for HCC samples. (C) Statistical correlation of MIF4GD, p27 and Ki-67 expressions in HCC samples. C1&C2, both of MIF4GD and p27 are negatively correlated with Ki-67 in HCC samples. C3, MIF4GD is positively correlated with p27 in HCC samples.

Figure 8. Kaplan–Meier survival curves in 108 HCC patients according to MIF4GD, p27kip1 expression and MIF4GD/p27kip1 expression. (a) Overall survival curves of low MIF4GD expression patients (p33.63%) vs high MIF4GD expression patients(433.63%). (b) Overall survival curves of low p27 expression patients (p39.42%) vs high p27 expression patients (439.42%). (c) Overall survival curves of low MIF4GD/p27 expression patients (MIF4GDp33.63% and p27p39.42%) vs other patients. cells, compared to non-tumorous hepatocytes. These findings constructed in the pPC86 vector was used for yeast two-hybrid screen. suggest that targeting MIF4GD may preferentially affect p27 level Vectors were introduced into MaV203 cells and the colonies were screened and proliferation of HCC cells rather than normal hepatocytes, on triple dropout plates (-Trp/-Leu/-His) containing 25 mM 3AT (Sigma, St highlighting that MIF4GD might be a valuable therapeutic target Louis, MO, USA). Prey plasmids were isolated from the positive clones and of HCC treatment. sequenced to determine the cDNAs identified in the screen. MIF4GD was identified in a prey clone of p27 C-terminal fragment. In summary, we identified MIF4GD as a novel binding partner for p27. MIF4GD stabilizes p27 and enhances its activity by modulating CDK2-mediated phosphorylation of p27. As high Cell culture, DNA constructs and transient transfection MIF4GD expression predicts a better outcome in HCC patients, a The Bel-7404, HuH7 and HepG2 human HCC cell lines, human hepatic L02 rigorous exploration of agents that can upregulate MIF4GD will cells and human embryonic kidney 293T cell lines were obtained from the prove an efficient strategy to combat HCC. Institute of Cell Biology, Academic of China, Shanghai, China, and cultured in 5% CO2 at 37 1C in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (Hyclone, Logan, UT, USA). The cDNA encoding MIF4GD, p27 and their truncated forms were MATERIALS AND METHODS amplified from a human embryonic liver cDNA library using PCR and Yeast two-hybrid screening subcloned into pCMV-HA, pCMV-myc and pEGFP-N3 constructs using Human full-length p27 or C-terminal fragment of p27 (92-198aa) was appropriate restrictive enzymes. The shMIF4GD targeting sequences (50- subcloned in frame with the GAL4 DNA-binding domain into pDBLeu GCAGGACGCATGTGCTACG-30, and 50-CGCCAGCTGCCCACAAGTA-30) are vector (Invitrogen, Carlsbad, CA, USA). A human liver cDNA library cloned into pSilencer 4.1 vector. Transfection was performed using

& 2015 Macmillan Publishers Limited Oncogene (2015) 237 – 245 MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 244 lipofectamine 2000 transfection reagent (Invitrogen) according to the Preparation of stable expression clones and colony formation manufacturer’s protocol. assays Flag-tagged MIF4GD was cloned into pBABE-puro vector and retroviruses Immunoblotting, cell fractionation and immunoprecipitation were packed in 293T cells by transfecting pBABE-Flag-MIF4GD or pBABE- puro constructs combined with VSV-G (pG1-VSV-G)- and gag-pol (pGPZ9)- Tissue and cell samples were promptly homogenized in a homogenization expressing plasmids. Bel-7404 cells were infected with the virus and buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 5 mM selected with 1 mg/ml puromycin (Sigma) to get stable MIF4GD-expressing EDTA, 5 mM EGTA, 15 mM MgCl2,60mM b-glycerophosphate, 0.1 mM sodium or control cells. Colony formation assay was carried out by plating virus- orthovanadate, 0.1 mM NaF and complete protease inhibitor cocktail infected Bel-7404 cells into a six-well plate (200 cells/well). Ten days later, (Roche Diagnostics, Basel, Switzerland), and the detection of protein level clones were stained with 0.5% crystal violet. was carried out as previously described.40 All antibodies used were purchased from Santa Cruz Biotechnology except the following ones: anti- pT187-p27 (Sigma), anti-GFP (Abmart, Shanghai, China), anti-b-actin Xenograft mouse model (Sigma) and anti-flag M2 (Sigma). Five-week-old female BALB/c nude mice were obtained from Shanghai Cell fractionation assay was performed using the NE-PER nuclear and SLAC Center and raised in a pathogen-free condition. A total of cytoplasmic extraction kit (Pierce, Rockford, IL, USA) according to the 2  106 Bel-7404-MIF4GD or Bel-7404-control cells were resuspended in manufacturer’s protocol. 200 ml PBS and injected subcutaneously into the nude mice. The tumor For immunoprecipitation assays, 0.5 mg of total cell lysates or 2 mg of volume was measured for 5 days with a vernier caliper and calculated human normal liver tissue was incubated with 1 mg primary antibodies or using the following formula: volume (mm3) ¼ length  width  control rabbit IgG (Bioworld Technology, Louis Park, MN, USA) at 4 1C height  0.52. were killed 40 days after injection. The experimental overnight. Twenty microliters of Protein G-Sepharose (Sigma, 1:1 slush in protocol was approved by the Committee on Animals Care and Use of PBS) was then added for 2 h at 4 1C with rocking. The precipitates were Nantong University. washed four times with homogenization buffer and boiled for 5 min with SDS sample buffer. Patients and tissue samples and immunohistochemistry analysis HCC tissues were obtained from 108 patients with HCC (mean age, 49; GST pull-down assay range, 21–72 years) who underwent hepatic surgical resection without The p27 coding sequence was subcloned into pGEX-6p-1 construct. pGEX- preoperative systemic chemotherapy at the Surgery Department, the 6p-1 and pGEX-6p-1-p27 were transformed into BL21(DE3) bacteria and Affiliated Hospital of Nantong University. The study group consisted of 84 induced with 0.1 mM IPTG for 12 h at 25 1C. Equivalent amounts of GST or male and 24 female patients. Histological grades were classified to well- GST–p27 fusion proteins were bound to glutathione-sepharose 4B (GE differentiated tumors (grade I; n ¼ 18), mood-differentiated cancers (grade Healthcare, Little Chalfont, UK) and incubated with 0.5 mg 293T cell lysate II; n ¼ 72), and poor-differentiated tumors (grade III; n ¼ 18). The follow-up overexpressing MIF4GD-HA. The complexes were washed five times with time was 5 years for 108 patients, ranging from 1 to 89 months. Tissue binding buffer (10 mM Tris-HCl pH 8.0, 200 mM NaCl, 0.5% NP-40, 1 mM samples were immediately processed after surgical removal. For histolo- EDTA, 1 mM DTT) and resolved with 12% SDS-PAGE for immunoblotting or gical examination, all tumorous and surrounding non-tumorous tissue Coomassie blue staining. portions were fixed in formalin and embedded in paraffin. Informed consent was obtained from all patients. Immunohistochemical assay was performed as previously described.40 Flow cytometric analysis For flow cytometric analysis, cells were fixed with 70% methanol in PBS at Statistical analysis À 20 1C overnight, then incubated with 1 mg/ml RNase A for 30 min at ± 37 1C in PBS and stained with 50 mg/ml of propidium iodide. The DNA The results were expressed as mean s.e. Statistical analysis was contents were analyzed by a FACScan flow cytometer (Becton Dickinson, performed using the State 7.0 software package. The association between Lincoln Park, NJ, USA). MIF4GD and p27 expression and clinicopathological features was analyzed using the w2 test. MIF4GD and p27 expression was studied using the Spearman rank correlation test because the data were not normally In vitro kinase assay distributed. For analysis of survival data, Kaplan–Meier curves were Immunocomplexes precipitated with anti-CDK2 antibody were incubated constructed, and the log-rank test was performed. with 1 mg histone H1 (Upstate Biotechnology Inc, Lake Placid, NY, USA) and 100 mM ATP containing 5 mCi [g-32P]ATP (Free Biotech, Beijing, China) in 50 ml of kinase buffer (50 mM Tris-HCl, pH 7.2, 10 mM MgCl2,1mM DTT) for CONFLICT OF INTEREST 30 min at 30 1C. The samples were resolved by SDS-PAGE followed by The authors declare no conflict of interest. autoradiography.

Immunofluorescence assay ACKNOWLEDGEMENTS Forty-eight hours after transfection, cells were fixed with 4% paraformal- We thank Professor Joel Belasco of New York University School of Medicine for the dehyde in PBS for 1 h at room temperature and later permeabilized with generous gift of pCL-6E1 and pCL-6E1.HSL constructs. This work was supported by 1% Triton X-100 in PBS for 20 min. The cells were then blocked with 1% Grants from National Basic Research Program of China (973 Program, No. BSA in PBS for 1 h and incubated with primary antibodies overnight at 4 1C. 2012CB822104, and No. 2011CB910604), the National Natural Science Foundation Secondary antibody incubation was performed using Alexa Fluor 568 or of China (No. 31070723, No. 31270802, No. 81272708 and No. 81300720), and a Alexa Fluor 488-conjugated goat anti-mouse, or Alexa Fluor 568- project funded by the Priority Academic Program Development of Jiangsu Higher conjugated goat anti-rabbit antibodies (Invitrogen) for 1 h at room Education Institutions (PAPD). temperature. In all, 5 mg/ml DAPI was added to stain the nucleus. The slides were mounted and visualized using a Nikon confocal microscope (Nikon, Melville, NY, USA). REFERENCES 1 Thun MJ, DeLancey JO, Center MM, Jemal A, Ward EM. The global burden of cancer: priorities for prevention. Carcinogenesis 2010; 31: 100–110. Cell proliferation assays 2 Finn RS. Development of molecularly targeted therapies in hepatocellular carci- Cell proliferation was measured by using MTT assay. Briefly, cells were noma: where do we go now? Clin Cancer Res 2010; 16: 390–397. plated into a 96-well plate with a density of 2 Â 103 cells/well. For 3 Malumbres M, Barbacid M. Cell cycle, CDKs and cancer: a changing paradigm. Nat proliferation measurement, 20 ml MTT (5 mg/ml) was added into each well Rev Cancer 2009; 9: 153–166. and incubated at 37 1C for 4 h. Then, the medium was removed and 200 ml 4 Feitelson MA, Sun B, Satiroglu Tufan NL, Liu J, Pan J, Lian Z. Genetic mechanisms DMSO was added to dissolve the formazan crystals. The plate was read of hepatocarcinogenesis. Oncogene 2002; 21: 2593–2604. at 570 nm using a microplate reader (ELx808, Bio-Tek Instruments, 5 Coleman WB. Mechanisms of human hepatocarcinogenesis. Curr Mol Med 2003; 3: Milan, Italy). 573–588.

Oncogene (2015) 237 – 245 & 2015 Macmillan Publishers Limited MIF4GD stabilizes p27 to regulate HCC progression C Wan et al 245 6 Naka T, Toyota N, Kaneko T, Kaibara N. Protein expression of p53, p21WAF1, and 23 Pippa R, Espinosa L, Gundem G, Garcia-Escudero R, Dominguez A, Orlando S et al. Rb as prognostic indicators in patients with surgically treated hepatocellular p27Kip1 represses transcription by direct interaction with p130/E2F4 at the carcinoma. Anticancer Res 1998; 18: 555–564. promoters of target genes. Oncogene 2012; 31: 4207–4220. 7 Qin LF, Ng IO. Exogenous expression of p21(WAF1/CIP1) exerts cell growth 24 Pietras EM, Warr MR, Passegue E. Cell cycle regulation in hematopoietic stem cells. inhibition and enhances sensitivity to cisplatin in hepatoma cells. Cancer Lett J Cell Biol 2011; 195: 709–720. 2001; 172: 7–15. 25 Greenbaum LE. Cell cycle regulation and hepatocarcinogenesis. Cancer Biol Ther 8 Matsuda Y. Molecular mechanism underlying the functional loss of cyclinde- 2004; 3: 1200–1207. pendent kinase inhibitors p16 and p27 in hepatocellular carcinoma. World J 26 Andrisani OM, Studach L, Merle P. signatures in hepatocellular carcinoma Gastroenterol 2008; 14: 1734–1740. (HCC). Semin Cancer Biol 2011; 21: 4–9. 9 Lee J, Kim SS. The function of p27 KIP1 during tumor development. Exp Mol Med 27 Qin J, Wang Z, Wang Y, Ma L, Ni Q, Ke J. JAB1 expression is associated with inverse 2009; 41: 765–771. expression of p27(kip1) in hepatocellular carcinoma. Hepatogastroenterology 10 Chu IM, Hengst L, Slingerland JM. The Cdk inhibitor p27 in human cancer: 2010; 57: 547–553. prognostic potential and relevance to anticancer therapy. Nat Rev Cancer 2008; 8: 28 Fornari F, Gramantieri L, Ferracin M, Veronese A, Sabbioni S, Calin GA et al. MiR- 253–267. 221 controls CDKN1C/p57 and CDKN1B/p27 expression in human hepatocellular 11 Zhou Q, He Q, Liang LJ. Expression of p27, cyclin E and cyclin A in hepatocellular carcinoma. Oncogene 2008; 27: 5651–5661. carcinoma and its clinical significance. World J Gastroenterol 2003; 9: 2450–2454. 29 Perez-Luna M, Aguasca M, Perearnau A, Serratosa J, Martinez-Balbas M, Jesus 12 Nguyen H, Gitig DM, Koff A. Cell-free degradation of p27(kip1), a G1 cyclin- Pujol M et al. PCAF regulates the stability of the transcriptional regulator and dependent kinase inhibitor, is dependent on CDK2 activity and the proteasome. cyclin-dependent kinase inhibitor p27Kip1. Nucleic Acids Res 2012; 40: 6520–6533. Mol Cell Biol 1999; 19: 1190–1201. 30 Podust VN, Brownell JE, Gladysheva TB, Luo RS, Wang C, Coggins MB et al. 13 Chen L, Yuan D, Wang GL, Wang Y, Wu YY, Zhu J. Clinicopathological significance A Nedd8 conjugation pathway is essential for proteolytic targeting of p27Kip1 of expression of Tspan-1, Jab1 and p27 in human hepatocellular carcinoma. by ubiquitination. Proc Natl Acad Sci USA 2000; 97: 4579–4584. J Korean Med Sci 2010; 25: 1438–1442. 31 Nakamura S, Okinaka K, Hirano I, Ono T, Sugimoto Y, Shigeno K et al. KIS induces 14 Tannapfel A, Grund D, Katalinic A, Uhlmann D, Kockerling F, Haugwitz U et al. proliferation and the cell cycle progression through the phosphorylation of Decreased expression of p27 protein is associated with advanced tumor stage in p27Kip1 in leukemia cells. Leuk Res 2008; 32: 1358–1365. hepatocellular carcinoma. Int J Cancer 2000; 89: 350–355. 32 Motti ML, De Marco C, Califano D, Fusco A, Viglietto G. Akt-dependent T198 15 Sherr CJ, Roberts JM. CDK inhibitors: positive and negative regulators of G1-phase phosphorylation of cyclin-dependent kinase inhibitor p27kip1 in breast cancer. progression. Genes Dev 1999; 13: 1501–1512. Cell Cycle 2004; 3: 1074–1080. 16 He G, Kuang J, Huang Z, Koomen J, Kobayashi R, Khokhar AR et al. Upregulation of 33 De Vita F, Riccardi M, Malanga D, Scrima M, De Marco C, Viglietto G. PKC- p27 and its inhibition of CDK2/cyclin E activity following DNA damage by a novel dependent phosphorylation of p27 at T198 contributes to p27 stabilization and platinum agent are dependent on the expression of p21. Br J Cancer 2006; 95: cell cycle arrest. Cell Cycle 2012; 11: 1583–1592. 1514–1524. 34 Sheaff RJ, Groudine M, Gordon M, Roberts JM, Clurman BE. Cyclin E-CDK2 is a 17 Nan KJ, Jing Z, Gong L. Expression and altered subcellular localization of the regulator of p27Kip1. Genes Dev 1997; 11: 1464–1478. cyclin-dependent kinase inhibitor p27Kip1 in hepatocellular carcinoma. World J 35 Wu L, Fan J, Belasco JG. direct rapid deadenylation of mRNA. Proc Natl Gastroenterol 2004; 10: 1425–1430. Acad Sci USA 2006; 103: 4034–4039. 18 Huang X, Qian X, Cheng C, He S, Sun L, Ke Q et al. Expression of Pirh2, a p27(Kip1) 36 Bansal N, Zhang M, Bhaskar A, Itotia P, Lee E, Shlyakhtenko LS et al. Assembly of ubiquitin ligase, in hepatocellular carcinoma: correlation with p27(Kip1) and cell the SLIP1-SLBP complex on histone mRNA requires heterodimerization and proliferation. Hum Pathol 2011; 42: 507–515. sequential binding of SLBP followed by SLIP1. Biochemistry 2013; 52: 520–536. 19 Zhao Y, Tang Q, Ni R, Huang X, Wang Y, Lu C et al. Early mitotic inhibitor-1, an 37 Zhao X, McKillop-Smith S, Muller B. The human histone regulator anaphase-promoting complex/cyclosome inhibitor, can control tumor cell pro- HBP/SLBP is required for histone and DNA synthesis, cell cycle progression and liferation in hepatocellular carcinoma: correlation with Skp2 stability and degra- cell proliferation in mitotic cells. J Cell Sci 2004; 117: 6043–6051. dation of p27(Kip1). Hum Pathol 2012; 44: 365–373. 38 Sullivan E, Santiago C, Parker ED, Dominski Z, Yang X, Lanzotti DJ et al. Drosophila 20 Cakmakci NG, Lerner RS, Wagner EJ, Zheng L, Marzluff WF. SLIP1, a factor required stem loop binding protein coordinates accumulation of mature histone mRNA for activation of histone mRNA translation by the stem-loop binding protein. Mol with cell cycle progression. Genes Dev 2001; 15: 173–187. Cell Biol 2008; 28: 1182–1194. 39 Narita T, Yung TM, Yamamoto J, Tsuboi Y, Tanabe H, Tanaka K et al. NELF interacts 21 Gorgoni B, Andrews S, Schaller A, Schumperli D, Gray NK, Muller B. The stem-loop with CBC and participates in 3’ end processing of replication-dependent histone binding protein stimulates histone translation at an early step in the initiation mRNAs. Mol Cell 2007; 26: 349–365. pathway. RNA 2005; 11: 1030–1042. 40 Ke Q, Ji J, Cheng C, Zhang Y, Lu M, Wang Y et al. Expression and prognostic role of 22 Vervoorts J, Luscher B. Post-translational regulation of the tumor suppressor Spy1 as a novel cell cycle protein in hepatocellular carcinoma. Exp Mol Pathol p27(KIP1). Cell Mol Life Sci 2008; 65: 3255–3264. 2009; 87: 167–172.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

& 2015 Macmillan Publishers Limited Oncogene (2015) 237 – 245