Myocyte enhancer factor 2C regulation of hepatocellular carcinoma via vascular endothelial growth factor and Wnt/β-catenin signaling

ORIGINAL ARTICLE

Myocyte enhancer factor 2C regulation of hepatocellularcarcinoma via vascular endothelial growth factorand Wnt/β-catenin signalingXL Bai1,2,7, Q Zhang1,2,7, LY Ye1,2, F Liang3, X Sun1, Y Chen4, QD Hu1, QH Fu1, W Su1, Z Chen5, ZP Zhuang6 and TB Liang1,2

Hepatocellular carcinoma (HCC) is one of the leading malignancies worldwide. Myocyte enhancer factor 2C (MEF2C) wastraditionally regarded as a development-associated factor and was recently reported to be an oncogene candidate. We havepreviously reported overexpression of MEF2C in HCC; however, the roles of MEF2C in HCC remain to be clarified. In this study, HCCcell lines and a xenograft mouse model were used to determine the functions of MEF2C in vitro and in vivo, respectively. Specificplasmids and small interfering RNA were used to upregulate and downregulate MEF2C expression, respectively. Functional assayswere performed to assess the influence of MEF2C on cell proliferation, and VEGF-induced vasculogenic mimicry, migration/invasionas well as angiogenesis. Co-immunoprecipitation was conducted to identify the interaction of MEF2C and β-catenin. Human HCCtissue microarrays were used to investigate correlations among MEF2C, β-catenin and involved biomarkers. MEF2C was found tomediate VEGF-induced vasculogenic mimicry, angiogenesis and migration/invasion, with involvement of the p38 MAPK and PKCsignaling pathways. However, MEF2C itself inhibited tumor growth in vitro and in vivo. MEF2C was upregulated by and directlyinteracted with β-catenin. The nuclear translocation of β-catenin blocked by MEF2C was responsible for MEF2C-mediated growthinhibition. The nuclear translocation of MEF2C was associated with intracellular calcium signaling induced by β-catenin. HCCmicroarrays showed correlations of nuclear MEF2C with the angiogenesis-associated biomarker, CD31, and cytosolic MEF2C withthe proliferation-associated biomarker, Ki-67. MEF2C showed double-edged activities in HCC, namely mediating VEGF-inducedmalignancy enhancement while inhibiting cancer proliferation via blockade of Wnt/β-catenin signaling. The overall effect of MEF2Cin HCC progression regulation was dictated by its subcellular distribution. This should be determined prior to any MEF2C-associatedintervention in HCC.

Oncogene advance online publication, 20 October 2014; doi:10.1038/onc.2014.337

INTRODUCTIONHepatocellular carcinoma (HCC) is the third leading cause ofhuman cancers for which there is currently no satisfactorytreatment.1 Although the mechanisms of initiation and progres-sion of HCC have been widely studied for decades, the details ofthese processes remain to be elucidated.Myocyte enhancer factor 2C (MEF2C) is a member of the MEF2

family, which is known to be a central regulator of celldifferentiation and organogenesis.2 MEF2C is a co-transcriptionfactor, and was previously thought to be predominantly involvedin the differentiation of myocytes, endothelial cells, neurons,lymphocytes, chondrocytes and neural crest cells.2 Recently,however, the oncogenic role of MEF2C has been revealed. Forinstance, previous studies implicated MEF2C as a potentialoncogene in leukemia, colon cancer and pancreatic cancer.3–5

Our previous work showed that MEF2C was upregulated in humanHCC tissues compared with the levels in peritumoral tissues,suggesting a possible role in HCC;6 however, the mechanism isawaiting investigation.

Vascular endothelial growth factor (VEGF) is a classic secretedinducer of angiogenesis under both physiological andpathological conditions.7 The close relationship of VEGF andMEF2C in vascular endothelial cells has been comprehensivelydemonstrated.8 Both VEGF and MEF2C are responsible for vasculardevelopment.9 Furthermore, in human umbilical vein endothelialcells, MEF2C was shown to mediate the transcriptional activationof VEGF on phosphatase of regenerating liver 3 (PRL-3), which isfrequently overexpressed in vascular endothelial cells in manycancers.10 Intriguingly, MEF2C-null mice showed reduced expres-sion of VEGF in cardiac cells during development,11 and MEF2Cwas reported to control the transcriptional level of VEGFexpression in angiogenesis.12 These observations suggested thatthe transcriptional factor MEF2C probably binds to the DNAdomain that promotes vegf transcription. However, the associationbetween VEGF and MEF2C in HCC is less well characterized butmay contribute to the role of MEF2C in HCC.Dysregulation of Wnt/β-catenin signaling is involved in tumor

proliferation, invasion and epithelial–mesenchymal transition,

1Department of Hepatobiliary and Pancreatic Surgery, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China; 2Key Laboratory of CancerPrevention and Intervention, The Second Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China; 3Department of Neurosurgery, The Second AffiliatedHospital, Zhejiang University School of Medicine, Hangzhou, China; 4Department of General Surgery, The Children’s Hospital, Zhejiang University School of Medicine, Hangzhou,China; 5Zhejiang Key Laboratory of Gastro-Intestinal Pathophysiology, Zhejiang Hospital of Traditional Chinese Medicine, Zhejiang Chinese Medical University, Hangzhou, Chinaand 6National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD, USA. Correspondence: Professor TB Liang, Department of Hepatobiliaryand Pancreatic Surgery, The Second Affiliated Hospital, Zhejiang University School of Medicine, 88 Jiefang Road, Hangzhou 310009, China.E-mail: [email protected] authors contributed equally to this work.Received 16 January 2014; revised 15 August 2014; accepted 29 August 2014

Oncogene (2014), 1–9© 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14

www.nature.com/onc

http://dx.doi.org/10.1038/onc.2014.337

mailto:[email protected]

http://www.nature.com/onc

leading to enhanced HCC malignancy.13,14 To date, direct mutualregulation and interaction between MEF2C and β-catenin havenot been reported, with the exception of some secondaryrelationships mediated by Mastermind-like 1 and glycogensynthase kinase 3β, or NKX2-5 in non-cancer cells.15,16

In this study, we explored the roles of MEF2C in HCC cells,showing complex regulation of HCC progression mediated byMEF2C, with the critical involvement of VEGF and β-catenin.Previous reports have shown distinct roles of MEF2C in thecytoplasm and nucleus as well as its regulation mechanism.15,17

Here we show preliminary evidence of double-edged character-istics of MEF2C in HCC progression that are associated with thesubcellular location of MEF2C.

RESULTSActivation of MEF2C by VEGF and corresponding signalingConsidering the potent pro-tumor effects of VEGF and the validityof VEGF-targeted therapy in cancer,18 we first investigatedwhether MEF2C-mediated VEGF signaling in HCC cells as it doesin endothelial cells.8,10 In Huh-7 and Sk-hep-1 cell lines, we foundan obvious concentration- and time-dependent upregulation ofMEF2C and phosphorylated MEF2C (p-MEF2C; Figures 1a and b). InHuh-7 cells treated with 25 ng/ml VEGF for 6 h, the transcriptionalactivity of MEF2C was found to be significantly enhanced inluciferase report assays (Figure 1c).To explore the signaling pathway involved in VEGF-mediated

regulation of MEF2C, we inhibited PKC, JNK, PI3K, ERK1/2 and p38MAPK signaling individually by use of specific pharmacologicalinhibitors. Compared with non-inhibited controls, the upregula-tion of p-MEF2C induced by VEGF was attenuated in both cell linesby the PKC and p38 MAPK inhibitors, whereas inhibition of JNKand ERK1/2 signaling affected only Sk-hep-1 cells and inhibition ofPI3K showed no influence on both cell lines (Figure 1d). Toconfirm these findings, Huh-7 cells was further treated with VEGFin the presence or absence of PKC/p38 MAPK inhibitors. The VEGF-induced overexpression of MEF2C was completely blocked inpresence of either inhibitor (Supplementary Figure S1A and B).Considering decreased VEGF expression in MEF2C-null mice, we

hypothesize a possible positive circuit regulation as VEGF-MEF2C-VEGF. Changes of MEF2C expression positively correlated with VEGFin Huh-7 cells (Supplementary Figure S1C). Moreover, VEGF transientpre-treatment induced subsequent VEGF expression and secretion,which was abrogated by MEF2C knockdown (Figures 1e and f).

VEGF-induced vasculogenic mimicry, invasion and angiogenesiswere mediated by MEF2CVEGF has been reported to influence cancer cells directly byautocrine or paracrine pathways, including promoting vasculogenicmimicry,19,20 migration and invasion.21,22 Based on the VEGF-inducedchange in MEF2C expression, we investigated the role of MEF2C inmediating these pro-tumor effects of VEGF in HCC. Huh-7 and Sk-hep-1 cells exhibited vasculogenic mimicry in the presence of VEGF,and this effect was diminished by MEF2C interference using siRNA(Figure 2a and Supplementary Figure S2A). In vitro transwell andwound-healing assays indicated that VEGF-enhanced cancer cellmigration, which was again blocked by MEF2C siRNA (Figure 2b,Supplementary Figure S2B and C). Additionally, VEGF-MEF2Csignaling was also responsible for Huh-7 cell invasion, withsignificantly increased passage of cells through the Matrigel matrix,an effect that was dramatically inhibited by MEF2C knockdown(Figure 2c). The enhanced invasive ability of cancer cells suggestedincreased expression of matrix degradation-associated factors, suchas matrix metallopeptidase 9 and membrane type 1-matrixmetalloproteinase in HCC cells.23 In Huh-7 cells, which showedepithelial traits, matrix metallopeptidase 9 mRNA levels weresignificantly elevated by VEGF and reduced by MEF2C siRNA

(Figure 2d and Supplementary Figure S2D). The mRNA level oftissue inhibitor of metalloproteinase 2, a negative regulator of matrixdegradation-associated factor balance, showed a similar changingpattern to that of matrix metallopeptidase 9; however, ratio of MMP 9and tissue inhibitor of metalloproteinase 2 mRNAs revealedadvantages with regard to matrix degradation induced by VEGFand inhibited by MEF2C interference (Figure 2e). No such alterationsin any of these factors were detected in mesenchymal-like Sk-hep-1cells or the membrane type 1-matrix metalloproteinase in Huh-7 cells(Figure 2d and Supplementary Figure S2D). Moreover, we assessedthe influence of MEF2C on VEGF-induced angiogenesis in vivo.MEF2C overexpression increased blood perfusion on the surfaces oftumors in mice (Figure 2f), and MEF2C overexpressing xenograftsshowed greater microvessel density as indicated by immunohisto-chemistry (IHC) staining of CD31 (Supplementary Figure S2E).

MEF2C-inhibited cell proliferation by interfering with Wnt/β-catenin signalingTo investigate the role of MEF2C in tumor growth, we used siRNAand plasmid transfection to decrease and increase MEF2C expres-sion, respectively. In Cell Counting Kit-8 (CCK-8) assays in vitro, wefound that MEF2C inhibited cell proliferation (Figure 3a); this effectwas further confirmed in xenograft models. MEF2C overexpressingxenografts exhibited reduced volume and weight compared withcontrols (Figure 3b and Supplementary Figure S3A). In addition,fewer Ki-67 positive cells were detected by IHC in MEF2Coverexpressing xenografts (Supplementary Figure S3B).Wnt/β-catenin is one of the most important signaling pathways

involved in the regulation of cancer cell proliferation; therefore,we investigated the effects of MEF2C on Wnt/β-catenin signaling.Downregulation of MEF2C slightly enhanced expression of β-catenin, as well as that of its target-gene transcription products,c-myc and cyclin D1 (Figure 3c, left panel). Upregulation of MEF2C,in turn, inhibited expression of these proteins (Figure 3c, rightpanel). In accordance with this, β-catenin transcription activity,assessed in a TOP/FOPFlash plasmid and luciferase reportersystem, correlated inversely with changes in MEF2C protein levels(Figure 3d). As the activation of canonical Wnt signaling is markedby nuclear location of β-catenin, we further assessed thesubcellular distribution of β-catenin. Unexpectedly, MEF2C over-expression reduced the nuclear location of β-catenin, and viceversa (Figure 3e).

Co-expression and interaction between MEF2C and β-cateninGiven the mutual regulation between MEF2C and β-catenin, weinvestigated the potential relationship between these twomolecules in four HCC cell lines (Huh-7, HepG2, Hep3B and Sk-hep-1) and an additional normal liver cell line HL-7702. GenerallyMEF2C and β-catenin were overexpressed in HCC cell linescompared with HL-7702, with a positive correlation observed inthe expression of both proteins (Supplementary Figure S4A). Inhuman HCC tissues derived from surgical resection samples, fiveout of seven patients showed overexpression of MEF2C and β-catenin in tumor tissues compared to the correspondingperitumor controls (Figure 4a). Relative levels of β-catenin proteinwere found to correlate with total MEF2C and its phosphorylatedforms (Figure 4b). Real-time PCR showed similar changes at themRNA level with those of the protein after altering Wnt/β-cateninsignaling (Supplementary Figure S4B), suggesting that MEF2Cexpression is positively regulated by β-catenin. In accordance withthese results, MEF2C transcription activity was enhanced in thepresence of Wnt3a, but was inhibited by siRNA-mediated knock-down of β-catenin (Figure 4c).In Huh-7 cells, MEF2C and β-catenin were found to be co-

located to some extent (Figure 4d). Furthermore, MEF2C formedcomplexes with β-catenin in both HL-7702 and Huh-7 cells(Supplementary Figure S4C). The ratio of β-catenin interacting

Myocyte enhancer factor 2C regulation of hepatocellular carcinomaXL Bai et al

2

Oncogene (2014), 1 – 9 © 2014 Macmillan Publishers Limited

with MEF2C and total β-catenin suggested only partial complexa-tion of β-catenin with MEF2C. The interaction between MEF2C andβ-catenin was then further confirmed in Huh-7 cells by usingdifferent co-immunoprecipitation strategies (Figure 4e). Next, tostudy the influence of VEGF signaling on this complex formationand protein distribution, separate lysates were obtained from thenucleus and cytosol of cells treated with or without VEGF. TheMEF2C/β-catenin complex formed in the nucleus and cytosol;however, VEGF diminished cytosolic MEF2C and MEF2C/β-catenincomplexation (Figure 4f), which was consistent with the observa-tion that VEGF promoted nuclear translocation of MEF2C(Supplementary Figure S5A).

Subcellular distribution of MEF2C and its influences in humanHCC tissuesAlthough MEF2C is a cofactor for transcription factors andpredominantly acts in the nucleus, its location in the cytoplasmhas also been observed under certain conditions.24 Furthermore,

the nuclear translocation of MEF2C is attributed to its nuclearlocation signal.25 We detected cytosolic MEF2C in HCC cell lines aswell as in human HCC tissues. Therefore, we assessed thesubcellular distribution of MEF2C under the conditions of ourstudy. VEGF not only induced MEF2C overexpression but alsopromoted translocation of MEF2C from the cytoplasm to thenucleus in Huh-7 and Sk-hep-1 cells (Supplementary Figure S5A).Moreover, siRNA-mediated knockdown of β-catenin resulted insignificant accumulation of MEF2C in the nucleus, while upregula-tion of β-catenin by Wnt3a led to MEF2C overexpression in thecytoplasm (Figure 5a). We further confirmed that β-catenin-regulated nuclear location of MEF2C was mediated by intracellularcalcium signaling (Supplementary Figure S6-S10).We identified different patterns of MEF2C expression in human

HCC tissues; namely non-expression, nuclear, cytosolic and pan-cellular expression (data not shown). More than one-third ofpatients had nuclear expression of MEF2C, while cytosolicexpression was even more frequent. Intriguingly, only nuclearexpression of MEF2C correlated with CD31 expression (Figure 5b

Figure 1. MEF2C was activated by VEGF through PKC and p38 MAPK signaling. (a) MEF2C and p-MEF2C were upregulated in Huh-7 and Sk-hep-1 cells by VEGF in a concentration-dependent manner. Cells were exposed to different concentrations of VEGF for 6 h. (b) MEF2C andp-MEF2C expression was upregulated by VEGF (25 ng/ml) at different time-points in both cell lines. (c) VEGF (25 ng/ml, 6 h) enhanced MEF2Ctranscription activity in Huh-7 cells. (d) Involvement of PKC and p38 MAPK in VEGF-induced MEF2C overexpression in both cell lines. Cellswere treated with specific concentrations of five signaling pathway inhibitors (25 μM for PI3Ki, 5 μM for p38 MAPKi, 10 μM for JNKi, PKCi andERK1/2i) for 18 h and then treated for 6 h with VEGF (25 ng/ml) or vehicle control. (e) Huh-7 cells were pretreated with VEGF (25 ng/ml) orvehicle after transfected with or without siRNA. VEGF was removed completely 6 h later. Cells were lysed after another 24 h andimmunoblotting was conducted. (f) Huh-7 cells were treated as in (E) and relative concentration of VEGF in medium was determined by ELISAafter 48 h of siRNA transfection.


3

© 2014 Macmillan Publishers Limited Oncogene (2014), 1 – 9

and c and Supplementary Figure S5B), suggesting that MEF2C-associated angiogenesis was related to changes in MEF2C genetranscription. In contrast, cytosolic location of MEF2C correlatednegatively with the expression of Ki-67 and β-catenin (Figure 5dand e and Supplementary Figure S5C), which was consistent withthe role of MEF2C in blocking β-catenin translocation to thenucleus. In another HCC microarray from Shanghai BiochipNational Engineering Research Center (Shanghai, China), expres-sion of cytosolic MEF2C was proven to correlate with maximum

diameter of tumor (Supplementary Figure S5D). Together, thesedata distinguished two independent roles of MEF2C in cancer cellregulation, which were relevant to the subcellular distribution ofMEF2C (Figure 5f).

DISCUSSIONMEF2C has recently been identified as a novel candidateoncogene, although evidence is limited to very few types of

Figure 2. VEGF-induced malignant characteristics in cancer cell was mediated by MEF2C. (a) Treatment of VEGF (25 ng/ml, 6 h) inducedvasculogenic mimicry in Huh-7 cells. Knockdown of MEF2C by siRNA significantly decreased the quantity of vessel-like structures (VLS). (b andc) VEGF (25 ng/ml, 6 h) increased the (c) migration and (d) invasion capacity of Huh-7 cells; this effect was blocked by MEF2C siRNAtransfection. Cell number was calculated as the average of three independent chambers, for which five random fields of vision werecountered. (d and e) After the indicated treatments, the mRNA levels of matrix metallopeptidase 9, tissue inhibitor of metalloproteinase 2 andmembrane type 1-matrix metalloproteinase were assessed by quantitative reverse transcription-PCR. Fold changes were calculated and werecalibrated to β-actin. VEGF increased transcription of (d) matrix metallopeptidase 9, tissue inhibitor of metalloproteinase 2 as well as (e) theratio of matrix metallopeptidase 9 and tissue inhibitor of metalloproteinase 2; these effects were attenuated by MEF2C siRNA in Huh-7 cells.(f) In a Huh-7 xenograft model, MEF2C overexpressing xenografts showed much higher blood perfusion on the surface of tumors comparedwith that of controls. **Po0.01, ***Po0.001.


4


malignancies.3–5 Whether MEF2C is a universal or tissue-specificoncogene remains to be clarified. We previously demonstratedoverexpression of MEF2C in human HCC tissues.6 Based on thisobservation, we studied the roles of MEF2C and the underlyingmechanisms in HCC progression. In contrast to its pure oncogenicrole in acute T-cell leukemia, our study showed a double-edgedrole for MEF2C in HCC. On one hand, MEF2C-mediated VEGFinduction of vasculogenic mimicry, migration and invasion as wellas angiogenesis of cancers; on the other hand, MEF2C inhibitedtumor growth via crosstalk with Wnt/β-catenin signaling. Inaddition, these two roles were mediated by nuclear and cytosolicMEF2C, respectively. Thus, the overall effects of MEF2C over-expression in HCC depend on its predominant subcellular

distribution, as well as the concomitant expression of β-cateninand the amount of VEGF in the microenvironment. Patients wereclinically identified with multi-modal expression of MEF2C. Intypes of HCC with pan-cellular expression, both functions ofMEF2C may be performed simultaneously, which suggests asubtype of HCC that is small in size but with abundantangiogenesis as well as the potential for early metastasis.Although proliferation and invasion are both characteristics of

malignant cancers and together contribute to subsequentmetastasis, these are two independent processes with distinctmechanisms that are not necessarily coupled in cells.26,27 In ourstudy, MEF2C-mediated enhancement of VEGF-induced cellinvasion and inhibition of cell proliferation by reducing canonical

Figure 3. MEF2C-inhibited cancer cell proliferation with concomitant inhibition of Wnt/β-catenin signaling. (a) Knockdown of MEF2C by siRNAincreased cell viability, while ectopic overexpression of MEF2C by plasmid transfection reduced cell viability. (b) Representatives of MEF2Coverexpressing xenografts and vehicle controls. (c) Knockdown of MEF2C increased expression of β-catenin and its target genes, and viceversa. (d) MEF2C interference upregulated β-catenin transcription activity, which was enhanced by overexpression of MEF2C. (e) Huh-7 cellsoverexpressing MEF2C showed predominant cytosolic location of β-catenin, while MEF2C knockdown resulted in predominantly nuclearlocation of β-catenin. **Po0.01, ***Po0.001.


5


Wnt/β-catenin signaling. Similar to the functions of transforminggrowth factor-β,28 these double-edged characteristics highlightthe complicated roles of MEF2C in HCC. Unfortunately, we wereunable to investigate the influence of MEF2C on circulating cancercells and metastatic foci in vivo. Inhibition of cancer proliferationhas been reported to be concomitant with epithelial-mesenchymal transition and to facilitate metastasis.26 In addition,MEF2C is included as a critical factor in a scoring system forpredicting cancer recurrence.29 Taking these reports into con-sideration, we speculated that overexpression of MEF2C maybenefit cancer metastasis in HCC. Thus, subcellular location ofMEF2C, as well as VEGF and β-catenin status, should beinvestigated carefully before adopting MEF2C-associated strate-gies in cancer treatment.Tumor initiation is a multistep process and tumors undergo

complicated alterations. During this period, different changesoccurs sequentially.30–32 Considering the high frequency ofconcomitant overexpression of MEF2C and β-catenin and the factthat β-catenin upregulated MEF2C, while MEF2C inhibited β-catenin, we speculate that the event of β-catenin overexpressionprecedes that of MEF2C. This supposition suggested manifoldfunctions of β-catenin in different steps of cancer development. Inthe early stages of HCC development, β-catenin was upregulatedby some mechanism and induced MEF2C, which in turn inhibitedβ-catenin. However, MEF2C can be induced and activated by thepresence of abundant VEGF in the microenvironment in a β-catenin-independent manner. Furthermore, MEF2C mutationsmay also exist as they do in other diseases,33,34 although theyhave not been verified to date.Consistent with our results, reduced MEF2C has been reported

previously to be associated with nuclear accumulation of β-catenin in breast cancer cells.35 However, the direct regulation ofβ-catenin by MEF2C has not been studied comprehensively. As aco-transcription factor, MEF2C might recruit histone deacetylase 5and deacetylate the β-catenin promoter, leading to inhibition of

CTNNB1 gene transcription.36 Mechanically, histone deacetylase 5has been shown to deacetylate histone 3.37,38 Our study alsoshowed dramatically decreased levels of acetylated histone 3 afterVEGF treatment, which was then inhibited by MEF2C knockdownin Huh-7 cells (data not shown). According to our results,deacetylation of VEGF could be indirectly realized by VEGF-induced nuclear translocation of MEF2C, which then recruitshistone deacetylase 5, resulting in deacetylation of histone 3.The possible interaction of MEF2C and β-catenin has been

indicated by bioinformatic analysis.4 Although attempts to identifydirect binding of MEF2 and β-catenin have been unsuccessful inmyoblasts,17 in our study, the MEF2C/β-catenin complex wasdetected. This discrepancy may due to differences in the cell typesinvestigated, with the possibility of mutations in cancer cells,differences in the affinity of immunoprecipitating and antibodiesand distinct experimental conditions. It is a limitation of this studythat an investigation of the interacting domains was notconducted. Complex formation is not necessary to suppressnuclear translocation, and MEF2C has a nuclear locationsequence.25 Consequently, binding may affect the nuclearlocation sequence of MEF2C; however, this requires furtherstudies. The mechanisms responsible for MEF2C translocationunder the context of aberrant β-catenin expression were still fullof questions, though our data showed β-catenin-limited intracel-lular calcium release could be a reasonable mechanism(Supplementary Figure S6-S10). Hepatitis B virus infection(Supplementary Table S2) might also be the possible correlation;however, these need further investigation.Additionally, our results showed reduced expression of β-

catenin when MEF2C was overexpressed. MEF2C has beenreported to locate in the β-catenin promoter and to inhibit genetranscription by recruiting histone deacetylase 5.36 Thus, MEF2Coverexpression could result in higher frequency of deacetylationin the β-catenin promoter and consequent downregulation ofβ-catenin expression.

Figure 4. Concurrent expression and mutual regulation of MEF2C and β-catenin. (a) Concurrent expression of MEF2C and β-catenin was shownin tissues from five representative patients. (b) Wnt3a treatment (200 ng/ml, 12 h) upregulated MEF2C and p-MEF2C expression; this effect wasinhibited by β-catenin knockdown. (c) Wnt3a treatments enhanced MEF2C transcription activity was enhanced by Wnt3a treatment andreduced by β-catenin siRNA treatment. (d) MEF2C was located in the cytoplasm and nucleus, whereas β-catenin showed pan-cellulardistribution. Of note, co-location of these proteins was observed in the cytoplasm and nucleus. (e) Increased β-catenin expression and MEF2C/β-catenin complexes were detected in Huh-7 cells compared with those in HL-7702 cells. (f) Lysates from the cytoplasm and nucleus wereextracted from Huh-7 cells treated with or without VEGF (25 ng/ml, 6 h). MEF2C/β-catenin complexes were detected in both the cytoplasmand nucleus. *Po0.05, **Po0.01, ***Po0.001.


6


Figure 5. Subcellular distribution of MEF2C and the underlying distinct functions. (a) β-Catenin siRNA increased nuclear location ofMEF2C in Huh-7 cells; this effect was partially reversed in the presence of Wnt3a (200 ng/ml, 12 h). (b) Representatives of manifold patternsof MEF2C and CD31 expression in human HCC tissues. Only nuclear MEF2C expression was associated with high CD31 expression.(c) In human HCC tissue microarrays, nuclear expression of MEF2C strongly and positively correlated with CD31 expression (n= 344).Dash lines show 95% confidence interval. (d) Cytosolic expression of MEF2C significantly and negatively correlated with Ki-67 expression(n= 338). (e) Cytosolic expression of MEF2C negatively correlated with nuclear expression of β-catenin. (f) Scheme of MEF2C function and itsregulation of HCC.


7


In summary, we provide evidence of MEF2C overexpression andits double-edged characteristics in HCC by means of in vitro, in vivoanalyses of HCC cell lines, xenograft mice models as well as clinicalhuman tissues. MEF2C-mediated VEGF-induced vasculogenicmimicry, angiogenesis and invasion, as well as inhibition of β-catenin-induced tumor growth. Our study provides the firstidentification of MEF2C in HCC and a novel understanding of its'oncogenic' function.

MATERIALS AND METHODSCell culture and reagentsThe HL-7702 and HCC cell lines, Huh-7, Sk-hep-1 and Hep3B were obtainedfrom the Shanghai Institute for Biological Science (Shanghai, China); HepG2was obtained from the American Type Culture Collection (ATCC, Manassas,VA, USA). HCC cells were cultured in appropriate medium with 10% FBSand 1% penicillin/streptomycin, and were maintained in standardconditions. Hl-7702 cells were cultured in RPMI-1640 media with 20% FBS.The plasmid used for ectopic overexpression of MEF2C was a gift from

Professor Xu Qiang from State Key Laboratory of PharmaceuticalBiotechnology, School of Life Sciences, Nanjing University, China. All typesof small interfering RNA (siRNA) were purchased from Santa CruzBiotechnology (Santa Cruz, CA, USA). Recombinant VEGF and signaling-pathway inhibitors (chelerythrine chloride for PKC inhibition, PD98059 forERK1/2 inhibition, SB203580 for p38 MAPK inhibition, SP600125 for JNKinhibition, and LY-294002 hydrochloride for PI3K inhibition) and ionomycinwere purchased from Sigma-Aldrich (St Louis, MO, USA). RecombinantWnt3a was purchased from R&D Systems (Minneapolis, MN, USA). Fluo-4-AM and BAPTA-AM was purchased from Life Technologies (Carlsbad, CA,USA). TOPFlash, FOPFlash and renilla luciferase pRL-TK reporter plasmidswere purchased from Millipore (Billerica, MA, USA) and 3×MEF2C reporterplasmid was a gift from Prof. Ron Prywes from Department of BiologicalScience, Columbia University (Addgene plasmid 32967).

Cell proliferation assaysCancer cell proliferation was evaluated with a Cell Counting Kit-8 (CCK-8,Dojindo, Kumamoto, Japan) according to the manufacturer’s protocol.Following treatments, cells were incubated with CCK-8 solution for 3 h, andabsorbance was measured at 450 nm using a Varioskan Flash MultimodeReader (Thermo Scientific, CA, USA). Relative proliferation capacity wasexpressed as a percentage of the indicated controls.

Transient transfection and luciferase reporter assaysTransfections of siRNA (50 nM) and plasmid were performed usingX-tremeGENE siRNA Transfection Reagent and X-tremeGENE HP DNATransfection Reagent (Roche, Basel, Switzerland), respectively. For lucifer-ase reporter assays, cells were seeded in 6-well plates and TOPFlash,FOPFlash or 3 ×MEF2C reporter plasmids were co-transfected with pRL-TKreporter plasmid. Reporter activity was assayed using the dual luciferasereporter system (Promega, Madison, WI, USA).

Immunoblotting, immunofluorescence staining,immunohistochemistry and co-immunoprecipitationThese experiments were performed as described previously.14 Briefly, cellswere treated as indicated for certain periods, collected, lysed and proteinswere extracted and quantified. Equal amounts of proteins were separatedby sodium dodecyl sulfate-polyacrylamide gel electrophoresis andtransferred to PVDF membranes (Millipore). Specific primary antibodiesand secondary antibodies were sequentially incubated, and membraneswere visualized. Anti-GAPDH antibody was purchased from KangchenBiotechnology (Shanghai, China) and was diluted at 1:5000. Other primaryantibodies were purchased from Cell Signaling Technology (Danvers, MA,USA) and were used at a 1:1000 dilution.For immunofluorescence staining, cells were treated as indicated, and

were treated with 4% polyphosphate formaldehyde and subsequent 0.1%Triton X-100 (Sigma). Cells were incubated with anti-MEF2C or anti-β-catenin (1 : 100) at 4 °C overnight, followed by incubation with fluoresceinisothiocyanate or cyanine 3-conjugated secondary antibodies (1:1000,Invitrogen, Grand Island, NY, USA). Counterstaining with DAPI (1:10,000,Sigma) was carried out before inspection under a fluorescence microscope

(Olympus, Tokyo, Japan). Staining without primary antibody was used asnegative controls.For immunohistochemical analyses, paraffin-embedded tumor tissue

samples from nude mice or patients were cut into 5 μm-thick serialsections. The slides were incubated with CD31 (1:50), Ki-67 (1:100), MEF2C(1:50), or β-catenin (1:100) antibodies. The slides were then incubated withhorseradish peroxidase-conjugated antibodies against rabbit immunoglo-bulin using Histostain-Plus Kit (ZSGB-BIO, Beijing, China) and counter-stained with hematoxylin. Staining without primary antibody was used asnegative controls. The pathological score was evaluated as previouslydescribed.14 The scoring system is illustrated using representative samplesof each score in Supplementary Figure S0.For co-immunoprecipitation, ~ 300 μg of protein extract was incubated

with 1 μg of protein-A/G sepharose (Santa Cruz) at 4 °C for 4 h. The sampleswere then centrifuged, rinsed and were mixed with SDS loading buffer forimmunoblotting.

Quantitative reverse transcription-PCRAs previously described and briefly, total RNA was extracted and reversetranscribed. The primers used were showed in Supplementary Table S1.Real-time PCR was performed on the ABI 7900 Prism HT (AppliedBiosystems, Shanghai, China). Fold change in gene expression changewas calculated using delta delta CT method, followed by normalization toβ-actin. Each treatment was tested in triplicate in three independentexperiments.

Transwell assays and wound healing assaysTranswell chambers (Corning, NY, USA) containing 6.5 mm diameterpolycarbonate filters (8 μm pore) with or without gel matrix (Matrigel; BD,Franklin Lakes, NJ, USA) were used for evaluation of cell invasion ormigration, respectively. Cells were seeded at a density of 5 × 104 perchamber. FBS-free medium was added to the upper compartments, andmedium with 10% FBS was added to the lower compartments of thechambers. Cells on the upper surface of the filters were removed after 48 h,and cells attached to the lower surface were fixed using 4% polyphosphateformaldehyde (Beyotime, Shanghai, China). Stained cells were counted infive randomly selected fields of vision for each chamber. Wound-healingassays were performed in 6-well plates by scratching monolayers of cells at90% confluence. The wound area was quantified after 24 and 48 h usingImage-Pro Plus 6.0 software (IPP; Media Cybernetics, Acton, MA, USA). Allexperiments were repeated three times.

Animals, xenograft model and tumor blood perfusion detectionMale Balb/c nude mice (aged 4–6 weeks) were obtained from the ShanghaiLaboratory Animal Center, and maintained under standard pathogen-freeconditions. Animal experiments were authorized by the Animal CareCommittee of Zhejiang University School of Medicine. For xenograft models,Huh-7 cells were transfected with the MEF2C overexpression plasmid or thepcDNA3.1 vehicle plasmid. One million cells of each group were suspendedin 200 μl PBS and injected subcutaneously into either flank of mice. Bloodperfusion detection was performed at 4 weeks post-inoculation. Mice wereanesthetized with 4% chloral hydrate (Sigma) and tumor blood flow wasmeasured at three sites on the surface of each tumor using a Laser DopplerPerfusion and Temperature Monitor (moorVMS-LDF1; Moor Instruments,Devon, UK). All mice were then sacrificed and xenografts were obtained,weighed, and fixed with 10% formaldehyde.

Statistical analysisStatistical calculations were performed using Prism 5 software (GraphPad,San Diego, CA, USA). Statistical analyses were performed using one-wayanalysis of variance or F test following two-tailed unpaired Student’st-tests, as appropriate, unless otherwise specified. Data of in vivo tumorblood perfusion are presented as mean± standard deviation (s.d.) of sixmice. Other data are presented as means± s.d. of three independentexperiments. Correlation analysis of tissue microarrays were performedusing the Spearman method. Best-fit values and 95% confidence intervalswere calculated using linear regression. For all tests, Po0.05 wasconsidered to be statistically significant.

CONFLICT OF INTERESTThe authors declare no conflict of interest.


8


ACKNOWLEDGEMENTSWe thank Professor Xu Qiang (State Key Laboratory of Pharmaceutical Biotechnology,School of Life Sciences, Nanjing University, China) for the generous gift of MEF2Coverexpression plasmid. We appreciate Mr Xie Shangzhi, Mr Hu Liqiang, and MissChen Conglin (The Second Affiliated Hospital, Zhejiang University School of Medicine,China) for their great help in certain experiments. This study was financiallysupported by the National Natural Science Foundation of China (No. 81171884), theNational Key Basic Research Program of China (No. 2014CB542101), the Ministry-Province Co-supportive Project of China, and Innovation and High-Level TalentTraining Program of Department of Health of Zhejiang.

AUTHOR CONTRIBUTIONSXLB, QZ, ZZP and TBL designed the experiments. XLB, QZ, LYY, FL, XS, YC, QDH,QHF, ZC and WS performed the experiments and analyzed the data. TBLsupervised the project. XLB, QZ and TBL wrote the report.

REFERENCES1 Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin 2013; 63:

11–30.2 Potthoff MJ, Olson EN. MEF2: a central regulator of diverse developmental

programs. Development 2007; 134: 4131–4140.3 Homminga I, Pieters R, Langerak AW, de Rooi JJ, Stubbs A, Verstegen M et al.

Integrated transcript and genome analyses reveal NKX2-1 and MEF2C as potentialoncogenes in T cell acute lymphoblastic leukemia. Cancer Cell 2011; 19: 484–497.

4 Li BQ, Huang T, Liu L, Cai YD, Chou KC. Identification of colorectal cancerrelated genes with mRMR and shortest path in protein-protein interactionnetwork. PLoS ONE 2012; 7: e33393.

5 Zhang JJ, Zhu Y, Xie KL, Peng YP, Tao JQ, Tang J et al. Yin Yang-1 suppressesinvasion and metastasis of pancreatic ductal adenocarcinoma by downregulatingMMP10 in MUC4/ErbB2/p38/MEF2C-dependent mechanism. Mol Cancer 2014; 13:130.

6 Bai X, Wu L, Liang T, Liu Z, Li J, Li D et al. Overexpression of myocyte enhancerfactor 2 and histone hyperacetylation in hepatocellular carcinoma. J Cancer ResClin Oncol 2008; 134: 83–91.

7 Hoeben A, Landuyt B, Highley MS, Wildiers H, Van Oosterom AT, De Bruijn EA.Vascular endothelial growth factor and angiogenesis. Pharmacol Rev 2004; 56:549–580.

8 Maiti D, Xu Z, Duh EJ. Vascular endothelial growth factor induces MEF2C andMEF2-dependent activity in endothelial cells. Invest Ophthalmol Vis Sci 2008; 49:3640–3648.

9 Lin Q, Lu J, Yanagisawa H, Webb R, Lyons GE, Richardson JA et al. Requirement ofthe MADS-box transcription factor MEF2C for vascular development. Development1998; 125: 4565–4574.

10 Xu J, Cao S, Wang L, Xu R, Chen G, Xu Q. VEGF promotes the transcription of thehuman PRL-3 gene in HUVEC through transcription factor MEF2C. PLoS ONE 2011;6: e27165.

11 Bi W, Drake CJ, Schwarz JJ. The transcription factor MEF2C-null mouse exhibitscomplex vascular malformations and reduced cardiac expression of angiopoietin1 and VEGF. Dev Biol 1999; 211: 255–267.

12 Shang Y, Doan CN, Arnold TD, Lee S, Tang AA, Reichardt LF et al. Transcriptionalcorepressors HIPK1 and HIPK2 control angiogenesis via TGF-beta-TAK1-dependent mechanism. PLoS Biol 2013; 11: e1001527.

13 White BD, Chien AJ, Dawson DW. Dysregulation of Wnt/beta-catenin signaling ingastrointestinal cancers. Gastroenterology 2012; 142: 219–232.

14 Zhang Q, Bai X, Chen W, Ma T, Hu Q, Liang C et al. Wnt/beta-catenin signalingenhances hypoxia-induced epithelial-mesenchymal transition in hepatocellularcarcinoma via crosstalk with hif-1alpha signaling. Carcinogenesis 2013; 34:962–973.

15 Saint Just Ribeiro M, Hansson ML, Lindberg MJ, Popko-Scibor AE, Wallberg AE.GSK3beta is a negative regulator of the transcriptional coactivator MAML1. NucleicAcids Res 2009; 37: 6691–6700.

16 Riazi AM, Takeuchi JK, Hornberger LK, Zaidi SH, Amini F, Coles J et al. NKX2-5regulates the expression of beta-catenin and GATA4 in ventricular myocytes. PLoSONE 2009; 4: e5698.

17 De Angelis L, Borghi S, Melchionna R, Berghella L, Baccarani-Contri M, Parise Fet al. Inhibition of myogenesis by transforming growth factor beta isdensity-dependent and related to the translocation of transcription factor MEF2to the cytoplasm. Proc Natl Acad Sci USA 1998; 95: 12358–12363.

18 Ellis LM, Hicklin DJ. VEGF-targeted therapy: mechanisms of anti-tumour activity.Nat Rev Cancer 2008; 8: 579–591.

19 Seftor RE, Hess AR, Seftor EA, Kirschmann DA, Hardy KM, Margaryan NV et al.Tumor cell vasculogenic mimicry: from controversy to therapeutic promise.Am J Pathol 2012; 181: 1115–1125.

20 Kirschmann DA, Seftor EA, Hardy KM, Seftor RE, Hendrix MJ. Molecular pathways:vasculogenic mimicry in tumor cells: diagnostic and therapeutic implications. ClinCancer Res 2012; 18: 2726–2732.

21 Goel HL, Mercurio AM. VEGF targets the tumor cells. Nat Rev Cancer 2013; 13:871–882.

22 Zhou L, Wang DS, Li QJ, Sun W, Zhang Y, Dou KF. Downregulation of the Notchsignaling pathway inhibits hepatocellular carcinoma cell invasion by inactivationof matrix metalloproteinase-2 and -9 and vascular endothelial growth factor.Oncol Rep 2012; 28: 874–882.

23 Maatta M, Soini Y, Liakka A, Autio-Harmainen H. Differential expression of matrixmetalloproteinase (MMP)-2, MMP-9, and membrane type 1-MMP in hepatocellularand pancreatic adenocarcinoma: implications for tumor progression and clinicalprognosis. Clin Cancer Res 2000; 6: 2726–2734.

24 Chen SL, Wang SC, Hosking B, Muscat GE. Subcellular localization of the steroidreceptor coactivators (SRCs) and MEF2 in muscle and rhabdomyosarcoma cells.Mol Endocrinol 2001; 15: 783–796.

25 Borghi S, Molinari S, Razzini G, Parise F, Battini R, Ferrari S. The nuclear localizationdomain of the MEF2 family of transcription factors shows member-specificfeatures and mediates the nuclear import of histone deacetylase 4. J Cell Sci 2001;114: 4477–4483.

26 Evdokimova V, Tognon C, Ng T, Sorensen PH. Reduced proliferation andenhanced migration: two sides of the same coin? Molecular mechanisms ofmetastatic progression by YB-1. Cell Cycle 2009; 8: 2901–2906.

27 Hugo HJ, Pereira L, Suryadinata R, Drabsch Y, Gonda TJ, Gunasinghe NP et al.Direct repression of MYB by ZEB1 suppresses proliferation and epithelial geneexpression during epithelial-to-mesenchymal transition of breast cancer cells.Breast Cancer Res 2013; 15: R113.

28 Akhurst RJ, Derynck R. TGF-beta signaling in cancer--a double-edged sword.Trends Cell Biol 2001; 11: S44–S51.

29 Oka M, Hamamoto Y, Okabe N. Evaluating system for predicting cancer return,Patent Number: CN100398663(C), granted on 02 July 2008.

30 Liu Y, Sanchez-Tillo E, Lu X, Huang L, Clem B, Telang S et al. Sequential inductionsof the ZEB1 transcription factor caused by mutation of Rb and then Ras proteinsare required for tumor initiation and progression. J Biol Chem 2013; 288:11572–11580.

31 Fearon ER, Vogelstein B. A genetic model for colorectal tumorigenesis. Cell 1990;61: 759–767.

32 Jen KY, Song IY, Banta KL, Wu D, Mao JH, Balmain A. Sequential mutationsin Notch1, Fbxw7, and Tp53 in radiation-induced mouse thymic lymphomas.Blood 2012; 119: 805–809.

33 Bienvenu T, Diebold B, Chelly J, Isidor B. Refining the phenotype associated withMEF2C point mutations. Neurogenetics 2013; 14: 71–75.

34 Novara F, Rizzo A, Bedini G, Girgenti V, Esposito S, Pantaleoni C et al. MEF2Cdeletions and mutations versus duplications: a clinical comparison. Eur J MedGenet 2013; 56: 260–265.

35 Yang M, Chen J, Su F, Yu B, Lin L, Liu Y et al. Microvesicles secreted bymacrophages shuttle invasion-potentiating microRNAs into breast cancer cells.Mol Cancer 2011; 10: 117.

36 Zhao JX, Yue WF, Zhu MJ, Du M. AMP-activated protein kinase regulatesbeta-catenin transcription via histone deacetylase 5. J Biol Chem 2011; 286:16426–16434.

37 McGee SL, van Denderen BJ, Howlett KF, Mollica J, Schertzer JD, Kemp BE et al.AMP-activated protein kinase regulates GLUT4 transcription by phosphorylatinghistone deacetylase 5. Diabetes 2008; 57: 860–867.

38 Tate CR, Rhodes LV, Segar HC, Driver JL, Pounder FN, Burow ME et al.Targeting triple-negative breast cancer cells with the histone deacetylaseinhibitor panobinostat. Breast Cancer Res 2012; 14: R79.

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


9


Documents

Myocyte enhancer factor 2C regulation of hepatocellular carcinoma via vascular endothelial growth factor and Wnt/β-catenin signaling