mdm2 overexpression deregulates the transcriptional ... · ment with nutlin-3, an mdm2 antagonist,...

Human Cancer Biology

MDM2 Overexpression Deregulates the TranscriptionalControl of RB/E2F Leading to DNA Methyltransferase3A Overexpression in Lung Cancer

Yen-An Tang1, Ruo-Kai Lin3, Yo-Ting Tsai2, Han-Shui Hsu4, Yi-Chieh Yang2,Chih-Yi Chen5, and Yi-Ching Wang1,2

AbstractPurpose: Overexpression of DNA 50-cytosine-methyltransferase 3A (DNMT3A), which silences genes

including tumor suppressor genes (TSG), is involved inmany cancers. Therefore, we examined whether the

transcriptional deregulation of RB/MDM2 pathway was responsible for DNMT3A overexpression and

analyzed the therapeutic potential of MDM2 antagonist for reversing aberrant DNA methylation status in

lung cancer.

ExperimentalDesign: The regulation ofDNMT3A expression and TSGmethylation status by RB/MDM2

was assessed in cancer cell lines andpatients. The effects ofNutlin-3, anMDM2antagonist, on tumor growth

in relation to DNMT3A expression and TSG methylation status were examined by xenograft model.

Results: We found that RB suppressed DNMT3A promoter activity and mRNA/protein expression

through binding with E2F1 protein to the DNMT3A promoter, leading to the decrease of methylation

level globally and TSG specifically. In addition, MDM2 dramatically induced DNMT3A expression by

negative control over RB. In clinical study, MDM2 overexpression inversely correlated with RB expression,

while positively associating with overexpression of DNMT3A in samples from patients with lung cancer.

Patients with high MDM2 and low RB expression showed DNMT3A overexpression with promoter

hypermethylation in TSGs. Treatment with Nutlin-3, an MDM2 antagonist, significantly suppressed tumor

growth and reduced DNA methylation level of TSGs through downregulation of DNMT3A expression in

xenograft studies.

Conclusions: This study provides the first cell, animal, and clinical evidence that DNMT3A is transcrip-

tionally repressed, in part, by RB/E2F pathway and that the repression could be attenuated by MDM2

overexpression. MDM2 is a potent target for anticancer therapy to reverse aberrant epigenetic status in

cancers. Clin Cancer Res; 18(16); 4325–33. �2012 AACR.

IntroductionIt has long been known that cancer cells undergo

changes in distribution of 50-methylcytosine modifica-tions including the region-specific hypermethylation ofpromoter CpG islands associated with tumor suppressor

genes (TSG; ref. 1). Aberrant promoter hypermethylationof CpG islands in TSGs leads to transcriptional repressionand results in tumorigenesis. DNA methylation is accom-plished by DNAmethyltransferases (DNMT). DNMT3A isresponsible for the de novo methylation that has beenshown to play crucial roles in embryonic development,genomic imprinting, and transcriptional silencing (2, 3).Recent study showed that DNMT3A played an essentialrole in melanoma tumorigenesis through methylatingpromoters of several TSGs (4). In addition, DNMT3Amutations were identified to be associated with poorprognosis in patients with acute myeloid leukemia (AML;refs. 5–9) andmyelodysplastic syndromes (MDS; ref. 10).There are many reports demonstrating the upregulationof DNMT3A mRNA or protein expression levels inhuman tumors such as non–small cell lung carcinoma(NSCLC; ref. 11), acute and chronic myelogenous leu-kemia (12), hepatocellular carcinomas (13), prostate(14), and breast cancer (15). Overexpression of DNMT3Ais also suggested to be associated with poor prognosis inpatients with NSCLC (16). However, the mechanism of

Authors' Affiliations: 1Institute of BasicMedical Sciences, 2Department ofPharmacology, College of Medicine, National Cheng Kung University,Tainan; 3Graduate Institute of Pharmacognosy, Taipei Medical University;4Division of Thoracic Surgery, Institute of Emergency and Critical CareMedicine, Taipei VeteransGeneral Hospital, National Yang-MingUniversitySchool of Medicine, Taipei; and 5Cancer Center, China Medical Universityand Hospital, Taichung, Taiwan, R.O.C.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Corresponding Author: Yi-Ching Wang, Department of Pharma-cology and Institute of Basic Medical Sciences, National ChengKung University, No.1, University Road, Tainan 70101, Taiwan, R.O.C.Phone: 886-6-2353535, ext 5502; Fax: 886-6-2749296; E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-11-2617

�2012 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 4325

overexpression of DNMT3A remains unclear in manycancers.

The tumor suppressor RB protein represses gene tran-scription, required for transition from G1 to S phase, bydirectly binding to the transactivation domain of E2F as acomplex on the promoters of the E2F targeted genes (17,18). Using bioinformatics analyses, we found E2F-bindingsites (conserved sequence is TTTSSCGC, S¼C or G; ref. 19)on DNMT3A promoter. In addition, the RB protein hasbeen shown to be inactivated by gene deletions, mutation,and protein phosphorylation in lung cancer (17). There-fore, it is worthy to analyze whether RB/E2F is involved intranscriptional regulation of DNMT3A gene in cancer cellsand patients with NSCLC.

Note that MDM2, an E3 ubiquitin ligase, physicallyinteracts with RB, resulting in ubiquitin–proteasome–mediated degradation of RB (20, 21). Overexpression ofMDM2 has been shown in many human cancers (22).Therefore, our study further examined whether MDM2plays a critical role in regulating theDNMT3A by destabiliz-ing RB protein and analyzed the relationship of MDM2protein with RB and DNMT3A proteins in NSCLC cells,animals, and patients. In addition, therapeutic potential ofMDM2 antagonist to reverse aberrant DNA methylationstatus was evaluated in lung tumor xenograft.

Materials and MethodsCell lines

All cell lines used in this study were purchased from theAmerican TypeCulture Collection. The A549 (RB/p53wild-type) and H1299 cell line (RB wild-type, p53-null) werederived from NSCLC tumors, the 5637 cell line (RB-null,

p53-mutant) was derived from a human bladder carcino-ma, and the Saos2 cell line (RB/p53-null) was derived froma human osteosarcoma. Detailed cell culture conditions aredescribed in Online Supplementary Methods.

Plasmid, RNAi, and transfectionThe plasmids and RNA interference (RNAi) used in the

study are listed in Supplementary Tables S1 and S2. Plas-mids and RNAi transfections were carried out with Lipofe-tamine 2000 (Invitrogen) according to the manufacturer’sprotocol.

Reverse transcriptase PCR assayThe primers used for reverse transcriptase (RT)-PCR and

quantitative RT-PCR analyses are described in Supplemen-tary Table S3. Relative levels of mRNA expression of thetarget genes were determined by the methods provided inOnline Supplementary Methods.

Western blot and immunoprecipitationImmunoblotting and immunoprecipitation (IP) were

conducted using the conditions described in Supplemen-tary Table S4.

Chromatin immunoprecipitation–PCR assayCancer cell lysates were cross-linked with 1% formalde-

hyde and sonicated on ice to shear DNA to lengths between200 and 800 bp. Subsequent steps were carried out usingthe chromatin immunoprecipitation (ChIP) assay kit(Upstate) according to the manufacturer’s instructions.Chromatin was immunoprecipitated by incubating withanti-E2F1, anti-RB, anti-AcH3K9K14, and anti-tri-Me-H3K9 antibodies (Supplementary Table S4) for 16 hoursat 4�C. PCR analysis was conducted using the primers forthe DNMT3A promoter described in Supplementary TableS3. For quantitative ChIP–PCR, total input served as inter-nal control. Relative quantitation using the comparative Ct

method with the data from ABI PRISM 7000 (version 1.1software) was conducted according to the manufacturer’sprotocol.

DNA affinity precipitation assayTwo hundred micrograms of nuclear extracts were incu-

bated with streptavidin–agarose (Sigma-Aldrich) for 1 hourat 4�C in a head-over-head rotor, then incubated with poly(dI-dc) (Amersham), salmon sperm DNA, dithiothreitol,and protease inhibitor cocktail for another 1 hour. Next, 50-biotin–labeled oligonucleotides of DNMT3A promoter andstreptavidin–agarose were added and incubated for 1 hourat room temperature. Finally, oligonucleotide-bound pro-teins were washed five times and dissolved in SDS samplebuffer and then Western blot was conducted with the 50-biotin–labeled oligonucleotides described in Supplemen-tary Table S3.

Quantitative methylation-specific PCRPrimer sequences used for quantitative methylation-

specific PCR (qMSP) of the BLU, FHIT, hRAB37, RARb,

Translational RelevanceThe silencing of tumor suppressor genes (TSG) result-

ing from promoter hypermethylation by DNA 50-cyto-sine-methyltransferases (DNMT) is associated withtumorigenesis. DNMT3A, a key enzyme for de novoDNAmethylation, is overexpressed in many cancers. Here,we show a link between overexpression of DNMT3Aand deregulation of MDM2/RB control in lung cancer.Our results using cell models reveal a novel mecha-nism of RB/E2F1–mediated transcriptional repressionof DNMT3A expression, which can be attenuated byMDM2 expression via degradation of RB. Our clinicalresults show that patients with low RB expression orMDM2overexpression also hadoverexpressedDNMT3Aandpromoter hypermethylation inmultiple TSGs. Treat-ment with Nutlin-3, an MDM2 antagonist, significantlyreduced DNMT3A expression and methylation of TSGs,as well as tumor growth in vivo, suggesting a new mech-anism by which MDM2 regulates epigenetic process incancer cells and that MDM2 is a potent target for anti-cancer therapy to reverse aberrant epigenetic status.

Tang et al.

Clin Cancer Res; 18(16) August 15, 2012 Clinical Cancer Research4326

RASSF1A, and SLIT2 genes are validated in previous stud-ies (23–25) and listed in Supplementary Table S3. Quan-titative methods are described in Online SupplementaryMethods.

Study populationSurgically resected tumor samples from 136 patients

diagnosed with primary NSCLC admitted to Veterans Gen-eral Hospital, Taichung and Taipei, were collected between1993 and 2007 after obtaining appropriate institutionalreview board permission and written informed consentfrom the patients.

Immunohistochemistry assayAntibodies used and their experimental conditions are

listed in Supplementary Table S4. Staining was scored asþþþ if >75% tumor cells were immunostaining positive;þþ for 50% to 75%; þ for 25% to 50%; þ/� for 10% to

25% cells, and � if <10% were positive. For DNMT3A orMDM2 protein expression level, the staining was graded as"overexpression" if the scores were þþ and þþþ. For RBprotein expression level, the staining was graded as "lowexpression" if the score was þ/� or �.

Xenograft studiesAthymic nu/nu female mice (BALB/c), 4–5 weeks of age,

were maintained in pathogen free conditions. The animalswere implanted subcutaneously with 5 � 106 A549 cells inone flank per mouse. The tumor size was measured accord-ing to the formula: (length � width2)/2. When tumors hadattained a mass of approximately 50 mm3, animals weretreated intraperitoneally with Nutlin-3 (20 mg/kg) ordimethyl sulfoxide every other day for 3 weeks. Tumorsamples were resected at day 25, weighed, and then sub-jected into immunohistochemical (IHC) and qMSP assays.

Figure 1. RB–mediated repressionofDNMT3Agene expression leads toreduction in promoter methylation ofmultiple TSGs. A, dual luciferaseassay showed the downregulationof DNMT3A promoter activity byectopically expressed RB in 5637cells and the upregulation ofDNMT3A promoter activity by RBknockdown (si-RB) in H1299 cellscompared with control. The Westernblot confirmed ectopically expressedand knockdown of RB in thecorresponding cells. B, RT-PCRexpression analysis of 5637 cellsshowed the reduction in endogenousDNMT3A mRNA expression by RBoverexpression. Knockdown of RBinduced endogenous mRNAexpression of DNMT3A in H1299cells. C, Western blot analysis ofDNMT3A protein expression in cellsoverexpressing RB or with RBknockdown. D, qMSP assays formethylation status in the RARb,FHIT, and RASSF1A promoters inH1299 cells with RB and/or DNMT3Aknockdown are shown (left). Westernblot analysis confirmed thepresence/absence of RB andDNMT3A proteins (right). Allexperiments were carried out atleast 3 times. Results from 1representative experiment areshown. Quantitative data arepresented as mean � SD.P values are as indicated.

DNMT3A Overexpression by RB/MDM2 Deregulation

www.aacrjournals.org Clin Cancer Res; 18(16) August 15, 2012 4327

Statistical analysisThe SPSS program (SPSS Inc.) was used for all statistical

analysis in cell and animal studies. The Pearson X2 test wasused to compare the protein expression levels of DNMT3Awith those of RB andMDM2between patients withNSCLC.P � 0.05 was considered to be statistically significant.

ResultsRB negatively regulates the expression of DNMT3Agene leading to decrease of methylation level globallyand TSG specifically

To test whether RB canmodulate theDNMT3A promoteractivity, a dual luciferase assay was conducted to quantifythe regulation of DNMT3A gene by RB in 5637 bladdercancer cells (RB-null) andH1299 lung cancer cells (RBwild-type). The 5637 cells were transiently cotransfected withpGL4-Renilla and DNMT3A promoter–reporter constructs(�925 to �74 bp upstream from the transcription start siteofDNMT3A gene), and pCMV-SPORT6 empty vector (con-trol) or pCMV-SPORT6-RB (RB). Ectopically expressed RBdecreased the DNMT3A promoter activity (Fig. 1A, left). Incontrast, knockdown of RB (si-RB) increased the DNMT3Apromoter activity compared with transfection of si-controlin H1299 cells (Fig. 1A, right). The repressive effects of RBon promoter activity were confirmed by dual luciferaseassays using an RB-responsive pE2F-luciferase construct

(pE2F-Luc), which contains four repetitive E2F1-bindingsites (Supplementary Fig. S1).

To elucidate whether the endogenous expression ofDNMT3A can be regulated by RB in various cell models,we examined the DNMT3A mRNA expression by RT-PCRand protein expression byWestern blot in A549 andH1299human lung cancer cells (RB wild-type), 5637 cells, andSaos2 human osteosarcoma cancer cells (RB-null). Ectop-ically expressed RB downregulated the DNMT3A mRNA(Fig. 1B) and protein expression (Fig. 1C and Supplemen-tary Fig. S2), while knockdown of RB induced higherDNMT3A mRNA (Fig. 1B) and protein expression (Fig.1C and Supplementary Fig. S3), thus confirming the neg-ative transcriptional effects of RB on DNMT3A gene.

DNMTs are enzymes that methylate the cytosine residueof CpGs of the genome and in promoters of genes includingTSGs. To determine if the level of DNMT3A in cell with RBoverexpression or knockdown correlated with the methyl-ation level of TSGs known to be involved in lung tumor-igenesis, we conducted qMSP assay to analyze the methyl-ation of RARb, FHIT, and RASSF1A genes. The validationand primary data of qMSP are shown in supplementary Fig.S4. The methylation level decreased when RB was over-expressed in 5637 cells (Supplementary Fig. S5), whereasincrease ofmethylationwasobservedwhenRBwasknockeddown in H1299 cells (Fig. 1D). Notably, the elevated

Figure 2. E2F1 positively regulatesDNMT3A gene expression andis required for RB–mediatedrepression of DNMT3A. A, theDNMT3A promoter region wasanalyzed. The predicted E2F siteswere indicated by � and the regionsP1 and P2 indicate fragmentsamplified in ChIP–qPCR assayswith the region analyzed shown inparentheses. B, ChIP–qPCRshowed that RB and E2F1 boundat DNMT3A P2 promoter in cells(left). In cells with RB knockdown,the open chromatin mark (Ac-H3K9K14) was increased, whereasthe repressive chromatin mark(Me-H3K9) was decreased (right)."IgG" is a negative control. C,ChIP–qPCR showed the binding ofRBonly to E2F1-boundP2site (left)and E2F knockdown promotes arepressive chromatin structure(right). Experiments were carriedout at least 3 times. Quantitativedata are presented as mean � SD.�, P < 0.05.

Tang et al.


methylation levels of TSGs in si-RB H1299 cells were sig-nificantly reduced by knockdown of both DNMT3A and RB(Fig. 1D), suggesting that DNMT3A was responsible for theobservedRB–mediated changes inDNAmethylation.Glob-al 50-methylcytosine level was decreased in RB-overexpres-sing cells and increased in cells with knocked down RBwhen detected by immunofluorescence staining using anti-5-methylcytosine antibody at 72 hours posttransfection(Supplementary Fig. S6). These data suggested changes inDNMT3A activity upon RB manipulation.

RB binds at DNMT3A promoter region through E2F1to form a repressive chromatin structure in vitro andin vivoIt has been known that RB represses gene transcription

through E2F–mediated transcription regulation (17, 18).Using TFSEARCH program (version 1.3; ref. 26) to predictE2F-binding sites, three sites located at�682 to�674,�469to �456, and �166 to �159 were found at DNMT3Apromoter. To verify whether RB/E2F1 complex could bindon these predicted E2F1-binding sites, we conducted ChIP–qPCR assays to assess the binding of RB and E2F1 to theDNMT3A promoter region. The location of these threepredicted E2F sites and the regions examined inChIP–qPCRat designated regions P1 and P2 onDNMT3A promoter areshown in Fig. 2A. In ChIP–qPCR assay of H1299 cells, RBand E2F1 were found to bind together on the DNMT3Apromoter at the P2 region (amplified by primers covering�450 to�127 regions), whereas RB and E2F1 did not bind

at the P1 region (amplified by primers covering �726 to�431 regions; Fig. 2B, left). Importantly, RB binding wasonly seen in E2F1-bound P2 site as evident in the loss of RBbinding at P2 site in the si-E2F1 cells (Fig. 2C, left). DNAaffinity precipitation assay also confirmed that the P2region of DNMT3A (TTTGGAGC at �166 to �159) wasbound by both RB and E2F1. Note that addition of non-biotin–labeled DNMT3A P2 (competitor) or mutation atDNMT3A P2 site (TTTAAAGC, with mutated bases under-lined) decreased the binding efficiency of RB and E2F1 tothe DNMT3A promoter (Supplementary Fig. S7).

In support of a role for the RB protein in epigeneticallyrepressing DNMT3A promoter, ChIP–qPCR assay revealedthat H1299 cells with knocked down RB had increasedabundance of Ac-H3K9K14 along with a decrease of tri-Me-H3K9 at the P2 region ofDNMT3A promoter comparedwith vector control cells (Fig. 2B, right). In the absence ofE2F, the repressive chromatin structure at the P2 region ofDNMT3A promoter was found (Fig. 2C, right). These dataindicated that RB forms a repressive complex at DNMT3Apromoter region in an E2F-dependent manner.

MDM2 attenuates the RB/E2F1–mediatedtranscriptional repression of DNMT3A expressionin cell studies

It has been reported that MDM2 interacts with RB topromote proteasome-dependent protein degradation (20,21). Therefore, the effect ofMDM2on DNMT3A expressionthrough targeting RB was examined in cell model.

Figure 3. MDM2enhancesoverexpressionofDNMT3A.A, IP–Western blot showed that overexpression ofMDM2enhancedubiquitinationofRB inH1299cells(lanes 4 versus 3 in the ubiquitin panel). Cells were transiently transfected with empty vector (�) or MDM2 expression vector (þ) for 24 hours and treated withMG132, a 26S proteasome inhibitor, at final 6 hours. Cell lysates were immunoprecipitated with anti-RB and Western blotted with anti-ubiquitin, anti-RB, oranti-MDM2 antibody. "Input" indicates the input protein lysate and "IgG" is a negative control. B, ectopically expressed MDM2 decreased RB proteinexpression but increasedDNMT3A protein expression in H1299 cells. C, dual luciferase assay showed that ectopically expressedMDM2 increasedDNMT3Apromoter activity in H1299 cells. TheWestern blot confirmed ectopically expressedMDM2 in the corresponding cells. D, knockdownofMDM2 (si-MDM2, left)or treatment with an MDM2 antagonist (Nutlin-3, right) promoted the formation of active hypophospho-RB (ser249/Thr252) and decreased the DNMT3Aprotein level in H1299 cells. Experiments were carried out at least 3 times. Results from one representative experiment are shown. Quantitative data arepresented as mean � SD. P values are as indicated.



Ectopically expressed MDM2 induced ubiquitination ofRB (Fig. 3A). Importantly, ectopically expressed MDM2not only decreased RB protein expression but alsoincreased the DNMT3A protein expression (Fig. 3B) andDNMT3A promoter activity (Fig. 3C). In contrast, si-MDM2 increased the total RB and the active form ofhypophosphorylated RB at Ser 249/Thr 252 (27, 28) anddecreased the DNMT3A protein expression (Fig. 3D, left).Inhibition of MDM2 by Nutlin-3, an MDM2 inhibitorwhich has been shown to increase hypophosphorylatedand functional RB (29), also resulted in the downregula-tion of DNMT3A expression (Fig. 3D, right). These datafurther confirmed that RB transcriptionally repressesDNMT3A expression and this regulation can be attenuatedby MDM2 expression.

Overexpression of DNMT3A resulting from alterationof MDM2/RB pathway leads to increased methylationof multiple TSGs in clinical studies

RB–mediated repression of DNMT3A protein has neverbeen studied inprimaryhuman tumor tissue. To analyze thecorrelation between DNMT3A protein overexpression andlow expression of RB protein in patients with lung cancer,

IHC for DNMT3A was conducted on tumor tissue slidesfrom 136 patients with lung cancer. An inverse correlationwas found betweenRB andDNMT3Aprotein expressions inpatientswith lung cancer (P¼0.007; Fig. 4AandTable 1). Inaddition, overexpression of MDM2 in patients with lungcancer was associated with low expression of RB (P ¼0.003), andwaspositively correlatedwithDNMT3Aproteinexpression (P < 0.001; Fig. 4A and Table 1).

Next, we speculated that the patients with alteration in allthree proteins examined, that is, DNMT3A overexpression,RB low expression, and MDM2 overexpression, shouldexhibit high level of DNA methylation of TSGs as opposedto those patients with normal expression in all proteins.Therefore, we conducted MSP analyses (primer sequencesare listed in Supplementary Table S3) for 4 TSGs in tumorspecimens from 26 patients whose DNA was available. Thedata indicated that patients with DNMT3A overexpression,RB lowexpression, andMDM2overexpressionhadmultipleTSGs hypermethylation (Fig. 4B).

Nutlin-3 decreases DNMT3A expression and DNAmethylation level of TSGs, and inhibits tumorgrowth in animals

Because highly expressed MDM2 promoted DNMT3Aoverexpression and aberrant DNA methylation status ofTSGs in lung cancer cell and clinical studies, we speculatedthat the inhibition of MDM2 could be a potent therapeuticstrategy for its ability to decrease DNMT3A expression andDNAmethylation level of TSGs. Therefore,we examined thein vivo antitumor activity of Nutlin-3, anMDM2 antagonist,in relation to DNMT3A level and DNA methylation statusof TSGs. Treatment of A549 tumor-bearing mice with20 mg/kg of Nutlin-3 every other day for 3 weeks signifi-cantly suppressed tumor growth (Fig. 5A). Tumors resectedfrommice at day 25were subjected to IHCandqMSP assays.The data indicated that Nutlin-3 treatment increased theactive form of RB and decreased DNMT3A protein levels(Fig. 5B), leading to a significant decrease of DNA methyl-ation level of TSGs, including RASSF1A, SLIT2, and BLUgenes in resected tumors from treated animals (Fig. 5C).

DiscussionDuring lung tumorigenesis, overexpression of DNMTs,

which catalyze cytosine methylation, is important forsilencing TSGs and leads to cancer formation and poorprognosis. The current study shows a link between over-expression of DNMT3A and deregulation of MDM2/RBcontrol in lung cancer. Our cell line data reveal a mecha-nism that DNMT3A is transcriptionally repressed, in part,by RB/E2F pathway and the repression could be attenuatedbyMDM2 overexpression. Our clinical studies confirm thatpatients with low RB expression or MDM2 overexpressionhad overexpressed DNMT3A and exhibited promoterhypermethylation in multiple TSGs. Moreover, xenograftstudy shows a novel finding that inhibition of MDM2 byNutlin-3 reduces DNMT3A protein level and DNA meth-ylation status of TSGs in vivo and significantly suppressestumor growth in treated animals.

Figure 4. Expression of DNMT3A, RB, and MDM2 proteins and analysisof methylation status in surgically resected NSCLC tumors. A,immunohistochemical analyses of DNMT3A, RB, and MDM2 proteins intwo representative samples frompatientswere shown. DNMT3A positiveimmunoreactivity (þ), RB negative immunoreactivity (�), and MDM2positive immunoreactivity (þ) were found in patient 1, whereas patient 2shows reverse pattern. Original magnification�200. B, MSP analyses for4 TSGs including RARb, hRAB37, RASSF1A, and FHIT in 26 lung cancerpatients whose DNA was available.

Tang et al.


We showed new evidence for a repressive effect of RB onDNMT3A in various human cancer cell lines. The promoterluciferase reporter assay and RT-PCR analysis suggested thatDNMT3A expression is negatively regulated by RB at thetranscriptional level (Fig. 1 and Supplementary Fig. S2).

Furthermore, the elevated DNA methylation levels of threeTSGs in knocked down RB H1299 cells were significantlyreduced by knockdown of bothDNMT3A and RB (Fig. 1D).These data confirmed thatDNMT3Awas responsible for theobserved RB–mediated changes in DNAmethylation. It has

Figure 5. Nutlin-3 inhibits A549xenograft growth and effectivelydownregulates DNMT3A and DNAmethylation status of TSGs in tumorxenograft. A, mice bearing theestablished A549 tumor xenograftswere injected intraperitoneally with20 mg/kg of Nutlin-3 or dimethylsulfoxide every other day for 3weeks. The tumor volumes (left) andtumor weight (right) were measured.Points, mean; bars,�SD. �, P < 0.05;��, P < 0.01; ��, P < 0.001. B, tumorsfrom mice were harvested andsubjected to IHC using antibodyagainst active hypophospho-RB(ser249/Thr252) and DNMT3A.Original magnification �200. C,qMSP assay for hypermethylationstatus in the RASSF1A, SLIT2, andBLU promoters was analyzed forresected tumors from mice.Quantitative data are presented asmean � SD. P values are asindicated.

Table 1. The correlation between protein expression of DNMT3A in relation to RB protein expressionin NSCLC patients

DNMT3A protein MDM2 protein

Normalexpression

Overexpression Normalexpression

Overexpression

Characteristics Total N (%) N (%) Total N (%) N (%)

Overall 136 92 (67.6) 44 (32.4) 136 62 (45.6) 74 (54.4)RB protein 136 RB protein 136Low expression 58 32 (55.2) 26 (44.8)0.007 Low expression 58 18 (31.0) 40 (69.0)0.003

Normal expression 78 60 (76.9) 18 (23.1) Normal expression 78 44 (56.4) 34 (43.6)DNMT3A protein 136Overexpression 44 4 (9.1) 40 (90.9)<0.001

Normal expression 92 58 (63.0) 34 (37.0)

Note: These results were analyzed by the Pearson Chi-Square test. The P values with significance are shown as superscripts.



long been known that RB represses transcription by remo-deling chromatin structure through interaction with pro-teins such as HDAC1 and SUV39H1 (30). We showed forthe first time RB–mediated epigenetic suppression viaformation of a repressive chromatin structure withinDNMT3A P2 promoter region in an E2F1-dependentmanner (Fig. 2). To test whether E2F1 can positivelyregulate the DNMT3A protein expression, 5637 andH1299 cell lines were transfected with E2F1 expressionplasmid or siRNA directed against E2F1. DNMT3A andE2F proteins showed a concordant expression pattern,indicating that the E2F1 can positively regulate humanDNMT3A protein expression (Supplementary Fig. S8).Whether the interaction of E2F1 is an essential require-ment for RB–mediated repression of DNMT3A will befurther examined by using mutant RB or E2F1 that lacksbinding activity with its partner.

Importantly, we unraveled a new mechanism by whichMDM2 regulates epigenetic process in cancer cells. As an E3ubiquitin ligase, MDM2 has been reported to control pro-tein stability of RB (20, 21). The significant positive rela-tionship between DNMT3A and MDM2 shown in cellmodels (Fig. 3) and clinical patients (Table 1) may becaused by MDM2–mediated degradation of RB and/orother transcription factors. Therefore, we speculated thatMDM2 can be a potent target for anticancer therapy toreverse aberrant epigenetic status in lung cancer. Nutlin-3,a selective small-molecule antagonist of MDM2, has beenshown to suppress xenograft tumor growth of severalcancers, including osteosarcoma, prostate cancer, Kaposi’ssarcoma-associated herpesvirus lymphomas, retinoblasto-ma, and neuroblastoma, and is currently in phase Iclinical trial for retinoblastoma treatment (31). Our studyshowed that Nutlin-3 significantly suppressed lung tumorgrowth possibly through activation of RB and downregula-tion of DNMT3A, leading to reduction of DNAmethylationof TSGs (Fig. 5). Taken together, we showed that Nutlin-3can be a therapeutic drug for cancer therapy by anew mechanism through reversing aberrant epigeneticalteration.

Overexpression of DNMTs plays a causative role intumorigenesis. Several disruptive mechanisms that controlthe expression of DNMTs have been reported. For example,viral infection activates DNMT1 through the RB pathway(32). TheDNMT1 promoter has been found to contain E2F-binding sites that are required for RB/E2F regulation inmurine epithelial and human osterosarcoma cells (33, 34).In our previous study, the DNMT1 and DNMT3A proteinswere found to be overexpressed in NSCLC in a coordinatemanner (11), suggesting a common regulatory pathway forexpression of DNMT genes. Here, we report that DNMT3Awas transcriptionally suppressed by RB/E2F in patients withNSCLCand cell lines. Several humanmicroRNAs, includingmiR-29 family, miR-148, and miR-143 have been found tobe frequently downregulated in human cancers and lead toincrease expression of DNMT1, DNMT3A, or DNMT3B

because they directly target the 30-UTR of DNMTs(16, 35, 36).

The present study provides a possible mechanism ofDNMT3A overexpression in cancer cells (SupplementaryFig. S9). We elucidate that the MDM2 targets RB for deg-radation, thus interfering with the repression of DNMT3A,and resulting in DNMT3A overexpression and abnormal50CpG hypermethylation in various TSGs in lung cancer.Recently, DNMT3A mutations have been identified to beassociated with poor prognosis in patients with AML andMDS (5–10). Most DNMT3A mutations did not changeDNMT3A expression level (5) and only a subset of patientswith AML with R882 mutation in DNMT3A gene showeddecreasedDNAmethylation in genomic regions (5, 7).Notethat we did not find any mutation withinDNMT3A gene inboth A549 and H1299 lung cancer cells (data not shown).In addition, a link between DNMT3A activity and the cellcycle also warrants interrogations. Previously, we reportedthat overexpression of DNMT3A protein was not associatedwith the expression of proliferation index as measured byexpression of proliferating cell nuclear antigen protein inthis cohort of patients (11). We also showed that patientswith lung cancer with alternative splicing ofMDM2mRNAhad significantly worse prognoses than those without (37).Our study provides rationale for developing drugs targetingDNMTs andMDM2 to reverse the aberrant epigenetic statusfor anticancer therapy.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: Y.-A. Tang, R.-K. Lin, Y.-T. Tsai, Y.-C. WangDevelopment of methodology: Y.-T. TsaiAcquisitionofdata (provided animals, acquired andmanagedpatients,provided facilities, etc.): Y.-A. Tang, R.-K. Lin, H.-S. Hsu, C.-Y. ChenAnalysis and interpretation of data (e.g., statistical analysis, biosta-tistics, computational analysis): Y.-A. Tang, R.-K. Lin, Y.-T. TsaiWriting, review, and/or revision of the manuscript: Y.-A. Tang, Y.-C.WangAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): Y.-T. Tsai, Y.-C. YangStudy supervision: Y.-C. Wang

AcknowledgmentsThe authors thank Dr. Jiunn-Liang Ko from Institute of Molecular

Toxicology of Chung Shan Medical University for kindly providing thepcDNA3–MDM2 constructs and National Cheng Kung University Prote-omics Research Core Laboratory for assisting with mass spectrometryanalysis.

Grant SupportThis work was supported in part by grants NSC99-2628-B-006-004-MY3,

DOH99-TD-G-111-021, andDOH99-TD-C-111-003 (Y.-C.Wang) and grantDOH99-TD-C-111-005 (C.-Y. Chen).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

ReceivedOctober 11, 2011; revisedMay 14, 2012; accepted June 16, 2012;published OnlineFirst June 25, 2012.

Tang et al.


References1. Laird PW. Cancer epigenetics. Hum Mol Genet 2005;14:R65–76.2. Fuks F, Burgers WA, Godin N, Kasai M, Kouzarides T. Dnmt3a binds

deacetylases and is recruited by a sequence-specific repressor tosilence transcription. EMBO J 2001;20:2536–44.

3. Okano M, Bell DW, Haber DA, Li E. DNA methyltransferases Dnmt3aand Dnmt3b are essential for de novo methylation and mammaliandevelopment. Cell 1999;99:247–57.

4. Deng T, Kuang Y,Wang L, Li J,Wang Z, Fei J. An essential role for DNAmethyltransferase 3a in melanoma tumorigenesis. Biochem BiophysRes Commun 2009;387:611–6.

5. Ley TJ,Ding L,WalterMJ,McLellanMD, Lamprecht T, LarsonDE, et al.DNMT3A mutations in acute myeloid leukemia. N Engl J Med 2010;363:2424–33.

6. Yamashita Y, Yuan J, Suetake I, Suzuki H, Ishikawa Y, Choi YL, et al.Array-based genomic resequencing of human leukemia. Oncogene2010;29:3723–31.

7. Yan XJ, Xu J, Gu ZH, Pan CM, Lu G, Shen Y, et al. Exome sequencingidentifies somatic mutations of DNAmethyltransferase gene DNMT3Ain acute monocytic leukemia. Nat Genet 2011;43:309–15.

8. Thol F, Damm F, Ludeking A, Winschel C, Wagner K, Morgan M, et al.Incidence and prognostic influence of DNMT3A mutations in acutemyeloid leukemia. J Clin Oncol 2011;29:2889–96.

9. Hou HA, Kuo YY, Liu CY, ChouWC, LeeMC, Chen CY, et al. DNMT3Amutations in acutemyeloid leukemia: stability during disease evolutionand clinical implications. Blood 2012;119:559–68.

10. Walter MJ, Ding L, Shen D, Shao J, Grillot M, McLellan M, et al.Recurrent DNMT3A mutations in patients with myelodysplastic syn-dromes. Leukemia 2011;25:1153–8.

11. Lin RK, HsuHS, Chang JW, ChenCY, Chen JT,Wang YC. Alteration ofDNA methyltransferases contributes to 50CpG methylation and poorprognosis in lung cancer. Lung Cancer 2007;55:205–13.

12. Mizuno S, Chijiwa T, Okamura T, Akashi K, Fukumaki Y, Niho Y, et al.Expression of DNA methyltransferases DNMT1, 3A, and 3B in normalhematopoiesis and in acute andchronicmyelogenous leukemia. Blood2001;97:1172–9.

13. ChoiMS, ShimYH,HwaJY, LeeSK, Ro JY, KimJS, et al. Expression ofDNA methyltransferases in multistep hepatocarcinogenesis. HumPathol 2003;34:11–7.

14. Patra SK, Patra A, Zhao H, Dahiya R. DNA methyltransferase anddemethylase in human prostate cancer.MolCarcinog 2002;33:163–71.

15. Girault I, Tozlu S, Lidereau R, Bieche I. Expression analysis of DNAmethyltransferases 1, 3A, and 3B in sporadic breast carcinomas. ClinCancer Res 2003;9:4415–22.

16. Fabbri M, Garzon R, Cimmino A, Liu Z, Zanesi N, Callegari E, et al.MicroRNA-29 family reverts aberrant methylation in lung cancer bytargetingDNAmethyltransferases 3A and 3B. ProcNatl AcadSci USA2007;104:15805–10.

17. Burkhart DL, Sage J. Cellular mechanisms of tumour suppression bythe retinoblastoma gene. Nat Rev Cancer 2008;8:671–82.

18. Dyson N. The regulation of E2F by pRB-family proteins. Genes Dev1998;12:2245–62.

19. Kanai M, Tashiro E, Maruki H, Minato Y, Imoto M. Transcriptionalregulation of human fibroblast growth factor receptor 1 byE2F-1.Gene2009;438:49–56.

20. Sdek P, Ying H, Chang DL, Qiu W, Zheng H, Touitou R, et al. MDM2promotes proteasome-dependent ubiquitin-independent degradationof retinoblastoma protein. Mol Cell 2005;20:699–708.

21. Uchida C, Miwa S, Kitagawa K, Hattori T, Isobe T, Otani S, et al.Enhanced Mdm2 activity inhibits pRB function via ubiquitin-dependent degradation. EMBO J 2005;24:160–9.

22. Momand J, Jung D, Wilczynski S, Niland J. The MDM2 gene ampli-fication database. Nucleic Acids Res 1998;26:3453–9.

23. Chang JW, Hsu HS, Ni HJ, Chuang CT, Hsiung CH, Huang TH, et al.Distinct epigenetic domains separated by a CTCF bound insulatorbetween the tandem genes, BLU and RASSF1A. PLoS One 2010;5:e12847.

24. LinRK,HsiehYS, LinP,HsuHS,ChenCY, TangYA, et al. The tobacco-specific carcinogen NNK induces DNA methyltransferase 1 accumu-lation and tumor suppressor gene hypermethylation in mice and lungcancer patients. J Clin Invest 2010;120:521–32.

25. Tseng RC, Lee SH, Hsu HS, Chen BH, Tsai WC, Tzao C, et al. SLIT2attenuation during lung cancer progression deregulates beta-cateninand E-cadherin and associates with poor prognosis. Cancer Res2010;70:543–51.

26. Heinemeyer T, Wingender E, Reuter I, Hermjakob H, Kel AE, Kel OV,et al. Databases on transcriptional regulation: TRANSFAC, TRRD andCOMPEL. Nucleic Acids Res 1998;26:362–7.

27. Guo J, Sheng G, Warner BW. Epidermal growth factor-induced rapidretinoblastoma phosphorylation at Ser780 and Ser795 is mediated byERK1/2 in small intestine epithelial cells. J Biol Chem 2005;280:35992–8.

28. Zhang Z, Li M, Wang H, Agrawal S, Zhang R. Antisense therapytargeting MDM2 oncogene in prostate cancer: effects on proliferation,apoptosis, multiple gene expression, and chemotherapy. Proc NatlAcad Sci U S A 2003;100:11636–41.

29. Du W, Wu J, Walsh EM, Zhang Y, Chen CY, Xiao ZX. Nutlin-3 affectsexpression and function of retinoblastoma protein: role of retinoblas-toma protein in cellular response to nutlin-3. J Biol Chem 2009;284:26315–21.

30. Giacinti C, Giordano A. RB and cell cycle progression. Oncogene2006;25:5220–7.

31. Secchiero P, Bosco R, Celeghini C, Zauli G. Recent advances in thetherapeutic perspectives ofNutlin-3. Curr PharmDes 2011;17:569–77.

32. McCabe MT, Low JA, Imperiale MJ, Day ML. Human polyomavirusBKV transcriptionally activates DNA methyltransferase 1 through thepRb/E2F pathway. Oncogene 2006;25:2727–35.

33. Kimura H, Nakamura T, Ogawa T, Tanaka S, Shiota K. Transcription ofmouse DNAmethyltransferase 1 (Dnmt1) is regulated by both E2F-Rb-HDAC-dependent and -independent pathways. Nucleic Acids Res2003;31:3101–13.

34. McCabe MT, Davis JN, Day ML. Regulation of DNA methyltransferase1 by the pRb/E2F1 pathway. Cancer Res 2005;65:3624–32.

35. Garzon R, Liu S, Fabbri M, Liu Z, Heaphy CE, Callegari E, et al.MicroRNA-29b induces global DNA hypomethylation and tumor sup-pressor gene reexpression in acute myeloid leukemia by targetingdirectly DNMT3A and 3B and indirectly DNMT1. Blood 2009;113:6411–8.

36. Ng EK, Tsang WP, Ng SS, Jin HC, Yu J, Li JJ, et al. MicroRNA-143targets DNA methyltransferases 3A in colorectal cancer. Br J Cancer2009;101:699–706.

37. WangYC,LinRK, TanYH,ChenJT,ChenCY,WangYC.Wild-typep53overexpression and its correlationwithMDM2andp14ARFalterations:an alternative pathway to non-small-cell lung cancer. J Clin Oncol2005;23:154–64.



mdm2 overexpression deregulates the transcriptional ... · ment with nutlin-3, an mdm2 antagonist,...

Documents