cancer research dysregulation of p53/sp1 control leads to...

Published OnlineFirst June 22, 2010; DOI: 10.1158/0008-5472.CAN-09-4161

Molecular and Cellular Pathobiology
Cancer
Research

Dysregulation of p53/Sp1 Control Leads to DNAMethyltransferase-1 Overexpression in Lung Cancer

Ruo-Kai Lin1, Chiu-Yi Wu1, Jer-Wei Chang1, Li-Jung Juan3, Han-Shui Hsu4, Chih-Yi Chen6, Yun-Yueh Lu3,Yen-An Tang2, Yi-Chieh Yang1, Pan-Chyr Yang5, and Yi-Ching Wang1,2

Abstract

Authors'Medicine,Kung UnivAcademiaGeneral HoNational YInternal MeR.O.C.; anTaichung, T

Note: SupResearch

CorresponNational ChTaiwan, R.OE-mail: ycw

doi: 10.115

©2010 Am

www.aacr

Downlo

Overexpression of DNA 5′-cytosine-methyltransferases (DNMT), which are enzymes that methylate thecytosine residue of CpGs, is involved in many cancers. However, the mechanism of DNMT overexpressionremains unclear. Here, we showed that wild-type p53 negatively regulated DNMT1 expression by forming acomplex with specificity protein 1 (Sp1) protein and chromatin modifiers on the DNMT1 promoter. How-ever, the stoichiometry between p53 and Sp1 determined whether Sp1 acts as a transcription activator orcorepressor. Low level of exogenous Sp1 enhanced the repressive activity of endogenous p53 on theDNMT1 promoter whereas high level of Sp1 upregulated DNMT1 gene expression level in A549 (p53wild-type) cells. In H1299 (p53 null) cells, exogenous Sp1 induced DNMT1 expression in a dose-dependentmanner. We also discovered a new mechanism whereby high level of Sp1, via its COOH-terminal domain,induced interaction between p53 and MDM2, resulting in degradation of p53 by MDM2-mediated ubiqui-tination. Clinical data from 102 lung cancer patients indicated that overexpression of DNMT1 was asso-ciated with p53 mutation (P = 0.014) and high expression of Sp1 protein (P = 0.006). In addition, patientswith overexpression of both DNMT1 and Sp1 proteins showed poor prognosis (P = 0.037). Our cell andclinical data provided compelling evidence that deregulation of DNMT1 is associated with gain of tran-scriptional activation of Sp1 and/or loss of repression of p53. DNMT1 overexpression results in epigeneticalteration of multiple tumor suppressor genes and ultimately leads to lung tumorigenesis and poor prog-nosis. Cancer Res; 70(14); 5807–17. ©2010 AACR.

Introduction

CpG island methylation is a common feature of many hu-man cancers and is thought to play an important role in can-cer initiation and progression (1–3). Aberrant promoterhypermethylation of CpG islands associated with tumor sup-pressor gene (TSG) can lead to transcriptional silencing re-sulting in tumorigenesis (4). Epigenetic disorders give rise toseveral significant human diseases including various cancers,neuron disorder, psychosis, and cardiovascular diseases,many of which are mediated by altered DNA methyltransfer-

Affiliations: 1Department of Pharmacology, College ofand 2Institute of Basic Medical Science, National Chengersity, Tainan, Taiwan, R.O.C.; 3Genomics Research Center,Sinica; 4Division of Thoracic Surgery, Taipei Veteransspital, Institute of Emergency and Critical Care Medicine,ang-Ming University School of Medicine; 5Department ofdicine, National Taiwan University Hospital, Taipei, Taiwan,d 6Cancer Center, China Medical University Hospital,aiwan, R.O.C.

plementary data for this article are available at CancerOnline (http://cancerres.aacrjournals.org/).

ding Author: Yi-Ching Wang, Department of Pharmacology,eng Kung University, No. 1, University Road, Tainan 70101,.C. Phone: 886-6-2353535, ext. 5502; Fax: 886-6-2749296;[email protected].

8/0008-5472.CAN-09-4161

erican Association for Cancer Research.

journals.org

on July 27, 201cancerres.aacrjournals.org aded from

ase 1 (DNMT1) expression and activity (4–10). DNMT1 is oneof the major enzymes responsible for DNA methylation (4). Itis a maintenance DNMT and is indirectly involved in de novomethylation (11).Previous reports have shown that DNMT1 is over-

expressed in lung, hepatocellular, acute and chronic mye-logenous leukemia, colorectal, gastric, and breast tumors(12–17). However, the mechanism of overexpression ofDNMT1 remains unclear in many cancers. A study showsthat deletion of p53 in the HCT116 human colon carcinomacell line results in increase of DNMT1 mRNA and protein(18). Another report shows that transiently knocked downwild-type p53 in human B-lymphoblast TK6 cells leads toincrease of DNMT1 expression (19). Notably, previous stud-ies showed that p53 could cooperate with specificity protein1 (Sp1) to repress the promoters of survivin, cyclin B1,cdc25C, RECQ4, and MnSOD genes (20–24). In addition,Sp1 and Sp3 have been reported to increase the activityof DNMT1 promoter by binding physically to their promo-ters in mouse NIH3T3 cells (25).Importantly, the p53 gene is a frequent target of mutation

that mainly resides in the region coding for the DNA bindingdomain resulting in inability to recognize the p53 consensusbinding sites (26). For example, high frequency of p53 muta-tion is reported in non–small-cell lung cancer (NSCLC) pa-tients (27). In addition, DNMT1 is found to be overexpressedin various cancers including lung cancer (12–17). Moreover,

5807

8. © 2010 American Association for Cancer Research.

http://cancerres.aacrjournals.org/

Lin et al.

5808


Sp1 is also the transcription factor of the DNMT1 gene (25).Previous studies have identified three Sp1 putative bindingsites on the DNMT1 promoter (28). One of the Sp1 bindingsites (+7 to +20) is proximal to the p53 (+30 to +56) putativebinding site predicted by PROMO (29) and TFSEARCH (30)searching tools. Therefore, using the lung cancer cell and clin-ical models, we tested the hypothesis whether alteration of thep53 or Sp1 pathway is involved in DNMT1 overexpression andthe interplay between p53 and Sp1 in the regulation of DNMTactivity in lung cancer.

Materials and Methods

Cell cultureThe A549 (p53 wild-type) and H1299 (p53 null) cell lines

were derived from NSCLC tumors cultured in DMEM(Invitrogen). SL2 cells were an Sp1-deficient cell line andwere maintained in Schneider medium (Life Technologies,Inc.) at 25°C. All media were supplemented with 10% fetalbovine serum (Invitrogen) and 1% penicillin/streptomycin(Invitrogen). All cell lines were purchased from the AmericanType Culture Collection.

Western blotThe cells were lysed on ice. Lysates were centrifuged

and SDS gel loading buffer was added. Samples containingequal amounts of protein (50 μg) were separated on a 6%SDS-PAGE and then electroblotted onto Immobilon-Pmembranes (Millipore Co.). Immunoblotting was per-formed for DNMT1, p53, Sp1, HDAC1, HDAC6, SIRT1,MDM2, ubiquitin, and β-actin, using the conditions de-scribed in Supplementary Table S1.

Dual luciferase assay and promoter deletion assayThe plasmids included pGL4-Long-DNMT1 promoter (nu-

cleotides −254 to +317), pGL4-short-DNMT1 promoter (nu-cleotides −394 to +2), pGL4-exon-DNMT1 promoter(nucleotides −19 to +316), wild-type p53, R273H mutant p53and R248L mutant p53 full-length cDNA expression con-structs, and internal control pGL4-Renilla construct, whichwere transfected into H1299 and A549 cells. The dual lucifer-ase reporter assay (Promega Corp.) was performed 48 hoursafter transfection according to the manufacturer's protocol.The data were represented as the mean ratio of firefly lucifer-ase to Renilla luciferase activity from triplicate experiments.

Real-time reverse transcription-PCRThe mRNA levels of DNMT1 was measured by a real-time

quantitative PCR system using the β-actin gene as an inter-nal control in the H1299 cell line, which was transfected withpCEP4-control vector, pCEP4-WTp53, pCEP4-R248L, orpCEP4-R273H, and in the A549 cell line, which was trans-fected with si-p53 or control siRNA (Invitrogen). Relativequantitation using the comparative Ct method with the datafrom ABI PRISM 7000 (version 1.1 software) was performedaccording to the manufacturer's protocol. The primer nucle-otide sequences for DNMT11 and their PCR conditions weredescribed in Supplementary Table S2.

Cancer Res; 70(14) July 15, 2010

on July 27, 201cancerres.aacrjournals.org Downloaded from

Immunoprecipitation assayCatch and Release Reversible Immuonprecipitation Sys-

tem kit (Millipore) was used for protein-protein interactionanalysis. Two milligrams of cell protein lysates were incubat-ed at 4°C overnight with the appropriate antibodies, includ-ing anti-p53, anti-Sp1, or normal mouse IgG, and 10 μL ofaffinity ligand. Immunoprecipitation was then performed ac-cording to the manufacturer's protocol. Proteins were elutedfrom washed immune complexes and then blotted with ap-propriate antibodies.

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

bated with streptavidin-agarose (Sigma-Aldrich) for 1 hourand then incubated with poly(deoxyinosinic-deoxycytidylicacid) (Amersham, GE), salmon sperm DNA, DTT, and prote-ase inhibitor cocktail for 1 hour. Next, 5′-biotin–labeledoligonucleotides of the DNMT1 promoter and streptavidin-agarose were added and incubated for 1 hour. Finally, oligo-nucleotide-bound proteins were washed and dissolved inSDS sample buffer and then Western blot was performedwith 5′-biotin–labeled oligonucleotides as described in Sup-plementary Table S2.

Chromatin immunoprecipitation (ChIP)-PCR andre-ChIP assayCells were cross-linked with 1% formaldehyde for 15 minutes

at 37°C. Lysate was sonicated on ice to shear DNA and subse-quent steps were performed with the ChIP assay kit (Millipore)according to the manufacturer's instructions. Chromatin wasimmunoprecipitated for 16 hours at 4°C using anti-Sp1, anti-p53, anti-H3K9m3, anti-H3K9K14Ac, anti-RBP2, anti-H3K4me3, and normal IgG (negative control). For the re-ChIPassay, we first used anti-p53 antibody to perform immunopre-cipitation for 3 hours and then removed first antibody. We nextadded anti-Sp1 antibody for the next immunoprecipitation for16 hours at 4°C. The primer nucleotide sequences were listed inSupplementary Table S2.

Patients and sample preparationTissues were collected after obtaining permission from the

appropriate institutional review board and informed consentsfrom recruited patients. Surgically resected tumor samplesfrom 102 patients diagnosed with primary NSCLC admittedto Veterans General Hospital, Taichung and Taipei, were col-lected between 1993 and 2007. Surgically resected tumor sam-ples were immediately snap-frozen and subsequently storedin liquid nitrogen. For the methylation assay, genomic DNAwas prepared using proteinase-K digestion and phenol-chloroform extraction, followed by ethanol precipitation.

Immunohistochemistry and immunofluorescenceassaysParaffin blocks of tumors were sectioned into 5-μm slices

and then processed using standard techniques. Evaluation ofimmunohistochemical staining was conducted blindly with-out prior knowledge of the clinical and pathologic character-istics of the cases. Staining was scored as 3 if >60% tumor

Cancer Research



Dysregulation of p53/Sp1 Leads to DNMT1 Overexpression


cells were immunostaining positive; 2 for 41% to 60%; 1 for11% to 40%; and 0 if <10% cells were positive. For DNMT1,Sp1, and proliferating cell nuclear antigen (PCNA) proteinexpression levels, staining was graded as overexpression ifthe score was 3. For immunofluorescence, we used anti–5-methylcytosine antibody to detect total 5-methylcytosinelevel. 4,6-Diamidino-2-phenylindole dihydrochloride (Sigma)counterstaining was used to justify the cell density. All theantibodies used and their experimental conditions were pro-vided in Supplementary Table S1.

Mutation spectrum analysis of the p53 geneTumors were analyzed for sequence alterations of the p53

gene as described previously (31). PCR fragments were se-quenced using the ABI 377 automatic sequencer (PE AppliedBiosystems). Sequencing analysis of the opposite strand forall the alterations was repeated at least once using indepen-dent PCR products.

Methylation-specific PCR for the FHIT, p16INK4a, RARβ,RASSF1A, and hRAB37 genesPromoter methylation status was determined by chemical

treatment with sodium bisulfite and methylation-specificPCR analysis, as previously described (32, 33). Primer se-quences and PCR conditions were listed in SupplementaryTable S2. All PCRs were performed with positive controlsfor both unmethylated and methylated alleles and no DNAcontrol. TSG alteration was defined as having more than threeof five TSGs analyzed showing promoter hypermethylation.

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

analysis. The Pearson χ2 test was used to compare the fre-quency of DNMT1 overexpression between NSCLC patientswith different p53 mutation status, Sp1 expression status,and TSG methylation status, and to study relationships be-tween variables, such as DNMT1 protein expression levelsin tumor samples. Overall survival curves were calculated ac-cording to the Kaplan-Meier method, and comparison wasperformed using the log-rank test. P < 0.05 was consideredto be statistically significant.

Results

Wild-type p53 decreases global 5′-methylcytosine leveland represses DNMT1 promoter activity andexpression levelGlobal 5′-methylcytosine level was changed when p53 was

knocked down in A549 (p53 wild-type) cells or overexpressedin H1299 (p53-null) cells when detected by immunofluores-cence staining using anti–5-methylcytosine antibody at 48hours posttransfection (Supplementary Fig. S1), suggestingchanges in DNMT1 activity on p53 manipulation. To testwhether wild-type p53 can modulate DNMT1 promoter activ-ity, a dual luciferase assay to quantify the regulation ofDNMT1 gene by wild-type p53 in A549 and H1299 lung cancercells was performed by cotransfecting with pGL4-luciferaseand pGL4-DNMT1 promoter vectors. Knockdown of wild-

www.aacrjournals.org


type p53 (si-p53) in A549 cells increased DNMT1 promoteractivity by 3-fold compared with si-control (Fig. 1A, left;P < 0.001). The H1299 cells were cotransfected with pCEP4-control-vector, wild-type p53 (pCEP4-WTp53), or mutant p53(Mut R248L or Mut R273H). The data indicated that wild-type p53 decreased DNMT1 promoter activity to 30% of thepCEP4-control vector (Fig. 1A, right; P < 0.001). However,both p53 mutants R248L and R273H could not repress theDNMT1 promoter as strongly as wild-type p53 (Fig. 1A, right).Real-time reverse transcription-PCR (RT-PCR) and Westernblot confirmed that knockdown of endogenous wild-typep53 in A549 cells could increase DNMT1 mRNA and proteinlevels (Fig. 1B, left). In addition, exogenous expression ofwild-type p53 suppressed DNMT1 mRNA and protein expres-sion in H1299 cells (Fig. 1B, right).To identify the sites in the DNMT1 promoter region that

are required for an optimal response to p53, we generateddeletion constructs of the DNMT1 promoter by shorteningits 5′ or 3′ regions (Fig. 1C, left). The luciferase assay resultsshowed that wild-type p53 mainly repressed DNMT1 genethrough the exon 1 region (−19 to +317), which containsSp1, p53, and E2F putative binding sites predicted byTFSEARCH searching tool (Fig. 1C, right).

Wild-type p53 binds cooperatively with Sp1 at p53 andSp1 binding sites to repress DNMT1 promoter activityTo investigate whether Sp1 and p53 putative binding sites

are necessary for p53 repression of the DNMT1 promoter, wegenerated various DNMT1 promoter constructs containingwild-type p53 and Sp1 binding sites (wt-p53/Sp1) and muta-tion at p53 binding site (mut-p53), mutation at Sp1 bindingsite (mut-Sp1), and mutation at both p53 and Sp1 bindingsites (mut-p53/Sp1). Suppression of DNMT1 promoter activ-ity by p53 was abolished when either p53 or Sp1 binding sitesin the DNMT1 promoter were mutated (Fig. 2A). It suggestedthat both p53 and Sp1 binding sites were necessary for p53transcriptional repression of the DNMT1 promoter.We performed immunoprecipitation assay to determine

whether p53 and Sp1 interact. The results showed that p53and Sp1 formed a complex (Supplementary Fig. S2A). Toidentify whether Sp1 protein and Sp1 putative binding sitewere involved in the interaction between p53 protein andthe DNMT1 promoter, DNA affinity precipitation assay(DAPA) was conducted. The data indicated that p53 couldinteract weakly with the probe containing p53 putative bind-ing site, but bind strongly to the probe with both p53 and Sp1binding sites. In addition, the interaction of p53 with theDNMT1 promoter was further enhanced in the presence ofboth p53 and Sp1 proteins (Fig. 2B, left). The data supportedthat the presence of Sp1 protein and Sp1 binding site couldaugment the binding of p53 to the DNMT1 promoter. In ad-dition, histone deacetylases HDAC1 and HDAC6 could bindto the probe with both p53 and Sp1 binding sites (Fig. 2B,left). The DAPA using probes containing mutation at p53or Sp1 binding sites showed that p53 and Sp1 specificallybound to wild-type p53 and Sp1 binding sites (Fig. 2B, right).Mutation of p53 and Sp1 binding sites significantly attenua-ted the interaction of p53 and Sp1 proteins (Fig. 2B, right).

Cancer Res; 70(14) July 15, 2010 5809



Lin et al.

5810


DAPA showed that p53 binds to the DNMT1 promoterprobe along with HDAC1 and HDAC6 (Fig. 2B), and DNMT1promoter activity is repressed by wild-type p53 (Fig. 1). ChIP-PCR assay was then used to examine whether the repressionof DNMT1 by p53 was mediated by chromatin condensation.The results indicated that H3K9 trimethylation level in-creased along with a decrease in H3K9K14 acetylation level.In addition, an increase in the recruitment of RBP2 lysine de-methylase and a decrease in H3K4 trimethylation level wereobserved for the exon 1 sequence of DNMT1 in response towild-type p53 overexpression in H1299 (Fig. 2C, left). In con-trast, the level of RBP2 lysine demethylase decreased alongwith an increase in H3K4 trimethylation level in p53-knock-



down A549 cells (Fig. 2C, right). The level of histone marksand the recruitment of chromatin modifiers further con-firmed that wild-type p53 transcriptionally suppressed theDNMT1 promoter.

Sp1 is essential for p53-mediated transcriptionalrepression of the DNMT1 promoterWe further conducted re-ChIP assay to test whether p53

and Sp1 simultaneously formed a complex with the DNMT1promoter in H1299 and A549. Anti-p53 and anti-Sp1 antibo-dies were used to perform the first and second immunopre-cipitations, respectively. Re-ChIP assay showed that p53 andSp1 were indeed bound together with the DNMT1 promoter

8. © 2010 American Associat

Figure 1. Wild-type p53 repressesDNMT1 promoter activity. A, dualluciferase assay was done aftercotransfecting with DNMT1promoter firefly luciferase,pGL4-Renilla vectors, si-p53 (left),and various p53 constructs(right) for 48 h. B, real-timeRT-PCR and Western blotconfirmed that knockdown ofendogenous wild-type p53increased the mRNA and proteinexpression of DNMT1 in A549 cells(left). Exogenous expression ofwild-type p53 suppressed DNMT1mRNA and protein expression inH1299 cells (right). C, promoterdeletion constructs of the DNMT1promoter were used to identify thesites in the promoter region thatwere required for an optimalresponse to p53. AP1, NFκB, Sp1,p53, and E2F putative bindingsites were predicted byTFSEARCH searching tool. TheSp1 mutation site for the assay inFig. 2 is indicated by an asteriskat nucleotides +11 and +12 ofDNMT1 gene. All experiments wereperformed at least three times.Columns, mean; bars, SD.*, P < 0.05; **, P < 0.01;***, P < 0.001.

Cancer Research

ion for Cancer Research.




in A549 (Fig. 2D, top) and H1299 (Fig. 2D, bottom) cellsexpressing exogenous p53.To investigate whether Sp1 is essential for p53-mediated

transcriptional repression of the DNMT1 promoter, we ana-lyzed DNMT1 promoter activity by luciferase assay in aD. melanogaster cell line, SL2, which lacks the Sp1 transcrip-tion factor (34). The results showed that p53 alone could notsuppress the DNMT1 promoter in SL2 cells, whereas p53could repress the DNMT1 promoter when exogenous Sp1was expressed in SL2 cells (Fig. 2A, right). The data sup-ported that Sp1 is necessary for p53-mediated transcriptionalrepression of the DNMT1 promoter.

Stoichiometry between p53 and Sp1 determineswhether Sp1 acts as a transcription activator orcorepressorSome reports showed that Sp1 was a transcription activa-

tor of the DNMT1 gene (25, 28). To further dissect the role ofSp1 in regulating DNMT1 expression in relation to p53,



DNMT1 mRNA and protein levels were examined in cells ex-pressing different levels of Sp1 and p53 proteins. We ana-lyzed DNMT1 mRNA and protein levels in cells transfectedwith 0, 0.5, 1, 2, 3, and 4 μg of HA-Sp1 expression plasmidsin A549 cells. The results showed that low level of exogenousSp1 enhanced the ability of endogenous p53 to repressDNMT1 mRNA and protein expression, whereas high levelof exogenous Sp1 induced the expression of DNMT1 (Fig.3A). In addition, Sp1 increased DNMT1 mRNA and proteinexpression in a dose-dependent manner in p53-null H1299cells (Fig. 3B). siRNA knockdown of Sp1 in H1299 cells de-creased the mRNA and protein expression of DNMT1 gene(Fig. 3C). The data indicated that overexpression of Sp1can induce DNMT1 expression.Interestingly, we found that high level of exogenous Sp1

expression depleted endogenous p53 protein in A549 cells.In contrast, knockdown of Sp1 increased p53 protein levelin A549 (Fig. 4A, top). We performed confocal microscopyto determine the p53, Sp1, and DNMT1 protein expression

Figure 2. Wild-type p53cooperates with Sp1 to repress theDNMT1 promoter through bindingto Sp1 and p53 putative bindingsites. A, DNMT1 (wt-p53/Sp1)repression by p53 was abolishedby mutation of either p53 or Sp1binding sites in the DNMT1promoter (mut-p53, mut-Sp1,mut-p53/mut-Sp1; left). In SL2cells (Sp1−/−), p53 could repressDNMT1 promoter activity onlywhen exogenous Sp1 wasexpressed (right). B, DAPA inH1299 cells transfected with eitherSp1 alone or Sp1 plus p53indicated that p53 and Sp1proteins strongly bound to theprobe with both p53 and Sp1binding sites (left). DAPA showedthat the probe with both p53 andSp1 putative sites could be boundby p53, Sp1, HDAC1, andHDAC6 proteins. A probe that hadmutation of p53 or Sp1 bindingsites decreased these interactions(right). C, ChIP-PCR indicatedthat H3K9m3 level increased alongwith a decrease in H3K9K14Aclevel. In addition, an increase inRBP2 lysine demethylaserecruitment and a decrease ofH3K4m3 level were observed inresponse to wild-type p53overexpression in H1299 (left).Inverse levels of RBP2 andH3K4m3 were observed inp53-knockdown A549 cells (right).D, in A549 (top) and H1299 (bottom)transfected with p53, p53 and Sp1formed a complex with the DNMT1promoter in vivo using a re-ChIPassay. Columns, mean; bars, SD.*,P<0.05; **,P<0.01; ***,P<0.001.

Cancer Res; 70(14) July 15, 2010 5811



Lin et al.

5812


levels in individual cultured A549 cells transfected with vary-ing amounts of HA-Sp1 plasmids. Data indicated that ongradual increase of Sp1 level, nuclear DNMT1 level increasedalong with an increase in the nuclear to cytoplasmic translo-cation of p53 (Fig. 4A, bottom).

High level of Sp1 induces p53 degradation byMDM2-mediated ubiquitinationp53 protein is known to be degraded in the cytoplasm by

the ubiquitin-mediated proteasomal degradation pathwaymodulated by MDM2 (27). To test the hypothesis that Sp1may be involved in p53 degradation by MDM2, we performedimmunoprecipitation, which showed that an Sp1/p53/MDM2complex could be formed in the presence of p53 (Supplemen-tary Fig. S2B). In addition, when the Sp1 level was decreasedvia siRNA knockdown and then restored by transfection ofSp1 expression plasmid, we found that the ubiquitination lev-el of p53 protein and the level of its binding with MDM2 de-creased (lane 5) and then increased (lane 6) according to Sp1expression level (Fig. 4B). Similar results were obtained incells overexpressing Sp1 (Supplementary Fig. S2C). To con-firm whether Sp1 is essential for the binding of p53 withMDM2, we performed immunoprecipitation in SL2 cells(Sp1−/−). The data indicated that there was no apparent bind-ing between MDM2 and p53 when cells lacked Sp1. However,interaction of MDM2 and p53 was greatly enhanced in SL2cells expressing exogenous Sp1 without apparently changingthe total MDM2 level (Fig. 4C). The data supported that Sp1 isnecessary for MDM2-mediated p53 degradation.To study which region of Sp1 is important for p53 inter-

action and degradation, we constructed HA-Sp1 plasmids



expressing different domains, including long Sp1, NH2-terminal Sp1, and COOH-terminal Sp1, and then performedimmunoprecipitation assay with anti-p53 antibody in cellstransfected with plasmids expressing various Sp1 domains.We found that there was an interaction of long Sp1 andCOOH-terminal of Sp1, with p53 leading to an increase inubiquitination of p53 protein (indicated by red arrows inFig. 4D).

DNMT1 overexpression, multiple TSG methylation, andp53/Sp1 pathway alteration in lung cancer patientsWe further examined the correlation of overexpression of

DNMT1 protein with p53 gene mutation and Sp1 proteinexpression in tumors from 102 lung cancer patients. Immu-nohistochemistry showed that 49% (50 of 102) of the pa-tients had high DNMT1 expression in lung tumor (Fig.5A; Table 1). Sequencing analyses revealed that alterationsof p53 included point mutation and small intragenic dele-tion/insertion at repetitive sequences (Fig. 5B). Statisticalanalysis indicated that p53 gene mutation was significantlyassociated with DNMT1 protein overexpression (Table 1;P = 0.014). We also used immunohistochemistry to examinethe Sp1 protein expression level in these 102 lung tumors.Sp1 protein was highly expressed in lung tumors comparedwith surrounding normal lung tissue (Fig. 5C). Sixty-six pa-tients had Sp1 protein overexpression (Table 1). There wasa strong association between high expression of Sp1 andoverexpression of DNMT1 in lung cancer (Table 1; P = 0.006).The correlation between the protein expression levels of Sp1and DNMT1 remained strong even in patients with wild-typep53 gene (Table 1; P = 0.021).

Figure 3. Sp1-regulated DNMT1 expression is dependent on the p53 gene status and Sp1 protein level. A, low level (0.5 and 1 μg) of exogenous Sp1repressed the expression of DNMT1 mRNA and protein whereas high level (3 and 4 μg) of Sp1 largely induced DNMT1 mRNA and protein expression inA549 cells (p53 wild-type). B, Sp1 increased DNMT1 mRNA and protein expression in a dose-dependent manner in H1299 cells (p53 null). C, si-Sp1in H1299 decreased the mRNA and protein expression of DNMT1 gene. Real-time RT-PCR and Western blots showed the expression levels of genesanalyzed. Columns, mean; bars, SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Cancer Research





We further performed methylation analyses for five TSGs in-cluding p16, RARβ, FHIT, RASSF1A, and hRAB37, which havebeen shown to be frequently hypermethylated in lung cancer(13, 35–37). The data indicated that alteration of DNMT1, p53,and/or Sp1 resulted in patients with multiple TSG hypermethy-lation (Table 1; P = 0.003–0.016). In addition, Kaplan-Meier sur-vival curves stratified by Sp1 overexpression showed that lungcancer patients with both DNMT1 and Sp1 protein overexpres-



sion had a poor postoperative overall survival (Fig. 5D;P = 0.038).The clinical data indicated that overexpression of DNMT1 wasstrongly correlated with aberrant p53/Sp1 pathway.

Discussion

In this study, we found that wild-type p53 could negative-ly regulate the DNMT1 gene by binding with Sp1 protein to

Figure 4. Sp1 enhances p53protein degradation throughMDM2-mediated ubiquitination.A, high level of exogenous Sp1depleted endogenous p53 proteinwhereas knockdown of Sp1increased p53 protein level in A549(top). Confocal microscopyindicated that on gradual increaseof Sp1 level, nuclear DNMT1level increased along withan increase in the nuclear tocytoplasmic translocation of p53protein (bottom). B, Sp1reconstitution assay wasperformed by si-Sp1 followed bytransfection of Sp1 expressionplasmid to restore the Sp1 level.The ubiquitination level of p53 andits binding level with MDM2increased when Sp1 expressionlevel was restored (lane 6).C, MDM2 interacted with p53 morestrongly in SL2 cells withexogenous Sp1 compared withcells without exogenous Sp1(MDM2 panel in lane 4 versus lane3). D, map of HA-Sp1 domainplasmids, including long Sp1[amino acids (A.A.) 8–785], NH2-terminal Sp1 (amino acids 8–618),and COOH-terminal Sp1 (aminoacids 619–785) expressionvectors. We detected aninteraction of long Sp1 and COOH-terminal of Sp1 with p53 leading toan increase in the ubiquitination ofp53 protein (indicated byred arrows).

Cancer Res; 70(14) July 15, 2010 5813



Lin et al.

5814


their binding sites on the DNMT1 promoter in lung cancercells. In addition, wild-type p53 modulated Sp1 to act ascorepressor with HDAC1, HDAC6, and RBP2 lysine de-methylase to suppress DNMT1 gene expression when Sp1protein level was low. The stoichiometry between p53 andSp1 may determine whether Sp1 will act as a transcrip-tion activator or corepressor. In addition, we are the firstto report that high level of Sp1 could enhance p53 trans-location to the cytoplasm and p53 degradation by MDM2-mediated ubiquitination through formation of an Sp1/p53/MDM2 triple protein complex. Using in vivo immunopre-cipitation-Western assay, we showed that COOH-terminalSp1 (amino acids 619–785) interacted with endogenousp53 leading to an increase in the ubiquitination of p53protein in vivo, confirming the previous in vitro bindingresults reported by others (38, 39). Clinical data supportedthat overexpression of DNMT1 and methylation at multi-ple TSGs were associated with p53 mutation and high ex-pression of Sp1 protein. Overexpression of both DNMT1and Sp1 proteins could be a poor prognosis factor forlung cancer. Our cell model and clinical study revealed



that deregulation of DNMT1 was associated with a gainof transcriptional activation of Sp1 and/or loss of repres-sion of p53.Our data showed that p53 bound to the DNMT1 pro-

moter through p53 putative binding site. p53 could alsobind to Sp1 putative binding site through interaction withSp1. Both p53 and Sp1 binding sites were required for p53/Sp1–mediated DNMT1 transcriptional repression. OurDAPA and promoter activity assay showed that p53 andSp1 proteins could bind to both p53 and Sp1 putativebinding sites. However, mutation of p53 or Sp1 putativebinding sites abolished the binding of p53 and attenuatedDNMT1 repression by p53. Re-ChIP assay also identifiedthat p53 and Sp1 were bound together on the DNMT1 pro-moter. Our results suggested that p53 formed a complexwith Sp1 and then bound to both p53 and Sp1 putativebinding sites to repress DNMT1 promoter activity whenSp1 was expressed at a low level.Sp1 protein was overexpressed in 65% of lung cancer

patients, suggesting that Sp1 is a highly expressed proteinin lung cancer tissue. In addition, our clinical data showed

8. © 2010 American Associat

Figure 5. Expression levels ofDNMT1 and Sp1 proteins and p53gene mutations in lung cancerpatients. A, representative figuresfor DNMT1-positive (left) andDNMT1-negative (right) nuclearstaining by immunohistochemistryare shown for two patients.B, alterations of p53 gene includedpoint mutation and nucleotidesdeletion. C, representative figuresfor Sp1-positive (left) andSp1-negative (right) nuclearstaining by immunohistochemistry.D, Kaplan-Meier survival curvesfor overall postoperativesurvival showed that lung cancerpatients with overexpression ofDNMT1 protein had a poorprognosis by log-rank test(P = 0.053). Patients with bothDNMT1 and Sp1 overexpressionshow an even poorer prognosis bystratification of Sp1 expressionstatus (P = 0.038).

Cancer Research

ion for Cancer Research.




that overexpression of Sp1 was associated with DNMT1protein overexpression. To ascertain that the regulationof DNMT1 gene by p53/Sp1 was not due to the prolifera-tive capacity of the cells, we performed immunohistochem-istry for PCNA and DNMT1 in 96 NSCLC patients. Thedata indicated that DNMT1 expression was not associatedwith the expression of proliferation index–associated genessuch as PCNA (P = 0.471), suggesting that DNMT1 overex-pression did not occur only in the proliferating cancercells (Supplementary Fig. S3; Supplementary Table S3).Note that COOH-terminal of Sp1 was necessary for the in-teraction and degradation of p53. In addition, high level ofSp1 protein facilitated the degradation of p53 by MDM2-mediated ubiquitination. We further treated the cells withthe MDM2 inhibitor nutlin-3, which is reported to inhibitMDM2-mediated p53 degradation (40). Our data indicatedthat p53 protein was indeed induced by nutlin-3 and re-sulted in a dramatic decrease of DNMT1 protein (Supple-mentary Fig. S4). With these data taken together, we



discovered a new mechanism of Sp1-mediated DNMT1overexpression through the Sp1/p53/MDM2 complex.Our cell and clinical studies provide compelling evidence

that increased DNMT1 protein expression was mediated bySp1 overexpression and/or p53 gene alteration in lung can-cer. Low level of Sp1 protein assisted wild-type p53 in re-pressing DNMT1 promoter activity. However, high level ofSp1 protein acted as a transcription activator of DNMT1gene expression. Our data suggested that the stoichiometrybetween p53 and Sp1 can determine whether Sp1 will act asa transcription activator or corepressor, which is dependenton the expression level of Sp1 protein. In addition, overex-pression of Sp1 could induce wild-type p53 degradationthrough MDM2-mediated ubiquitination leading to DNMT1overexpression. Our clinical data also indicated that pa-tients who had alterations of p53 and overexpression ofSp1 and DNMT1 proteins had increased risk of hypermethy-lation in multiple TSG promoters, confirming that DNMT1overexpression resulted in promoter hypermethylation of

Table 1. The correlation between DNMT1 protein overexpression, p53 alteration, and Sp1 expressionand promoter hypermethylation of multiple TSGs in NSCLC patients

Characteristics
Total DNMT1 protein
Normal expression, n (%)

Canc

8. © 2010 American Association

High expression,* n (%)

Overall
102 52 (51.0) 50 (49.0) p53 gene
Wild-type
69 41 (59.4) 28 (40.6) Mutation 33 11 (33.3) 22 (66.7)0.014
Sp1 protein
Normal 36 25 (69.4) 11 (30.6) High 66 27 (40.9) 39 (59.1)0.006
Patients with wild-type p53
69 Sp1 normal 26 20 (76.9) 6 (23.1) Sp1 high 43 21 (48.8) 22 (51.2)0.021
Overall patients for TSG methylation analysis
53† 31 (58.5) 22 (41.5) Patients with p53 gene mutation 18
Hypermethylation of multiple TSGs‡

Hypermethylated
13 3 (23.1) 10 (76.9)0.003
Normal
5 5 (100.0) 1 (00.0) Patients with Sp1 overexpression 30
Hypermethylation of multiple TSGs
Hypermethylated 19 6 (31.6) 13 (68.4)0.008
Normal
11 9 (81.8) 2 (18.2) Patients with alteration of p53/Sp1 10
Hypermethylation of multiple TSGs
Hypermethylated 8 1 (12.5) 9 (87.5)0.016
Normal
2 2 (100.0) 0 (00.0)
NOTE: These results were analyzed by the Pearson χ2 test. P values with significance are shown as superscripts.*For DNMT1 and Sp1 protein expression levels, the staining was graded as overexpression if >60% tumor cells showed positiveimmunostaining in nuclei.†The TSG promoter hypermethylation was examined in 53 patients with available samples.‡TSG alteration was defined as having more than three TSGs showing promoter hypermethylation. TSGs included the p16, RARβ,FHIT, RASSF1A, and hRAB37 genes.

er Res; 70(14) July 15, 2010 5815

for Cancer Research.


Lin et al.

5816


multiple TSGs leading to NSCLC tumorigenesis and poorprognosis.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. Yu-Sun Chang at Chang Gung University (Taoyuan, Taiwan,R.O.C.) for kindly providing the pGL2-DNMT1 promoter constructs; Dr. Jan-JongHung at National Cheng Kung University (Tainan, Taiwan, R.O.C.) for kindly



providing us with the HA-pcDNA3-Sp1 plasmid; and Tsung-Han Yang forassistance with immunohistochemistry.

Grant Support

Grant NSC98-3112-B-006-014-CC1 from the National Science Council andgrantDOH98-TD-G-111-024 fromtheDepartmentofHealth (TheExecutiveYuan,Republic of China; Y-C.Wang), and grantDOH98-TD-G-111-025 from theDepart-ment of Health and grant AS-98-CDA-L12 from Academia Sinica (L-J. Juan).

The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 11/12/2009; revised 03/24/2010; accepted 04/09/2010; publishedOnlineFirst 06/22/2010.

References
1. Esteller M. Dormant hypermethylated tumour suppressor genes:
questions and answers. J Pathol 2005;205:172–80.2. Mompar ler RL. Cancer epigenet ics. Oncogene 2003;22:

6479–83.3. Rauch TA, Zhong X, Wu X, et al. High-resolution mapping of DNA

hypermethylation and hypomethylation in lung cancer. Proc NatlAcad Sci U S A 2008;105:252–7.

4. Egger G, Liang G, Aparicio A, Jones PA. Epigenetics in humandisease and prospects for epigenetic therapy. Nature 2004;429:457–63.

5. Jamaluddin MS, Yang X, Wang H. Hyperhomocysteinemia, DNAmethylation and vascular disease. Clin Chem Lab Med 2007;45:1660–6.

6. Costa E, Dong E, Grayson DR, et al. Reviewing the role of DNA(cytosine-5) methyltransferase overexpression in the corticalGABAergic dysfunction associated with psychosis vulnerability.Epigenetics 2007;2:29–36.

7. Kanai Y. Alterations of DNA methylation and clinicopathological di-versity of human cancers. Pathol Int 2008;58:544–58.

8. Abdolmaleky HM, Smith CL, Faraone SV, et al. Methylomics in psy-chiatry: modulation of gene-environment interactions may bethrough DNA methylation. Am J Med Genet B Neuropsychiatr Genet2004;127B:51–9.

9. Rhee I, Bachman KE, Park BH, et al. DNMT1 and DNMT3b coop-erate to silence genes in human cancer cells. Nature 2002;416:552–6.

10. Milutinovic S, Brown SE, Zhuang Q, Szyf M. DNMT1 knockdown in-duces gene expression by a mechanism independent of the two epi-genetic silencing mechanisms: DNA methylation and histonedeacetylation. J Biol Chem 2004;15:15.

11. Fatemi M, Hermann A, Gowher H, Jeltsch A. Dnmt3a and Dnmt1functionally cooperate during de novomethylation of DNA. Eur J Bio-chem 2002;269:4981–4.

12. De Marzo AM, Marchi VL, Yang ES, et al. Abnormal regulation ofDNA methyltransferase expression during colorectal carcinogenesis.Cancer Res 1999;59:3855–60.

13. Lin RK, Hsu HS, Chang JW, et al. Alteration of DNA methyltrans-ferases contributes to 5′CpG methylation and poor prognosis in lungcancer. Lung Cancer 2007;55:205–13.

14. 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.

15. Saito Y, Kanai Y, Nakagawa T, et al. Increased protein expression ofDNA methyltransferase (DNMT) 1 is significantly correlated with themalignant potential and poor prognosis of human hepatocellular car-cinomas. Int J Cancer 2003;105:527–32.

16. Etoh T, Kanai Y, Ushijima S, et al. Increased DNA methyltransfer-ase 1 (DNMT1) protein expression correlates significantly withpoorer tumor differentiation and frequent DNA hypermethylationof multiple CpG islands in gastric cancers. Am J Pathol 2004;164:689–99.

17. Mizuno S, Chijiwa T, Okamura T, et al. Expression of DNA methyl-

transferases DNMT1, 3A, and 3B in normal hematopoiesis and inacute and chronic myelogenous leukemia. Blood 2001;97:1172–9.

18. Peterson EJ, Bogler O, Taylor SM. p53-mediated repression of DNAmethyltransferase 1 expression by specific DNA binding. Cancer Res2003;63:6579–82.

19. Guo Z, Tsai MH, Shiao YH, et al. DNA (cytosine-5)-methyltransferase1 as a mediator of mutant p53-determined p16(ink4A) down-regulation.J Biomed Sci 2008;15:163–8.

20. Esteve PO, Chin HG, Pradhan S. Molecular mechanisms of trans-activation and doxorubicin-mediated repression of survivin gene incancer cells. J Biol Chem 2007;282:2615–25.

21. St Clair S, Manfredi JJ. The dual specificity phosphatase Cdc25C isa direct target for transcriptional repression by the tumor suppressorp53. Cell Cycle 2006;5:709–13.

22. Innocente SA, Lee JM. p53 is a NF-Y- and p21-independent, Sp1-dependent repressor of cyclin B1 transcription. FEBS Lett 2005;579:1001–7.

23. Sengupta S, Shimamoto A, Koshiji M, et al. Tumor suppressor p53represses transcription of RECQ4 helicase. Oncogene 2005;24:1738–48.

24. Dhar SK, Xu Y, Chen Y, St Clair DK. Specificity protein 1-dependent p53-mediated suppression of human manganese su-peroxide dismutase gene expression. J Biol Chem 2006;281:21698–709.

25. Kishikawa S, Murata T, Kimura H, Shiota K, Yokoyama KK. Regula-tion of transcription of the Dnmt1 gene by Sp1 and Sp3 zinc fingerproteins. Eur J Biochem 2002;269:2961–70.

26. Sigal A, Rotter V. Oncogenic mutations of the p53 tumor suppressor:the demons of the guardian of the genome. Cancer Res 2000;60:6788–93.

27. Kruse JP, Gu W. Modes of p53 regulation. Cell 2009;137:609–22.

28. Liu S, Liu Z, Xie Z, et al. Bortezomib induces DNA hypomethylationand silenced gene transcription by interfering with Sp1/NF-κB-dependent DNA methyltransferase activity in acute myeloid leuke-mia. Blood 2008;111:2364–73.

29. The PROMO online tool is available for download for academic andnon-commercial use at: http://alggen.lsi.upc.es/.

30. The TFSEARCH online tool for searching transcription factor bind-ing sites is available at http://www.cbrc.jp/research/db/TFSEARCH.html.

31. Wang YC, Chen CY, Chen SK, et al. High frequency of deletionmutations in p53 gene from squamous cell lung cancer patients inTaiwan. Cancer Res 1998;58:328–33.

32. Leu YW, Rahmatpanah F, Shi H, et al. Double RNA interference ofDNMT3b and DNMT1 enhances DNA demethylation and gene reac-tivation. Cancer Res 2003;63:6110–5.

33. Tzao C, Tsai HY, Chen JT, Chen CY, Wang YC. 5′CpG island hyper-methylation and aberrant transcript splicing both contribute to theinactivation of the FHIT gene in resected non-small cell lung cancer.Eur J Cancer 2004;40:2175–83.

Cancer Research





34. Courey AJ, Holtzman DA, Jackson SP, Tjian R. Synergistic activationby the glutamine-rich domains of human transcription factor Sp1.Cell 1989;59:827–36.

35. Wu CY, Tseng RC, Hsu HS, Wang YC, Hsu MT. Frequentdown-regulation of hRAB37 in metastatic tumor by genetic andepigenetic mechanisms in lung cancer. Lung Cancer 2009;63:360–7.

36. Kim JS, Kim H, Shim YM, et al. Aberrant methylation of the FHITgene in chronic smokers with early stage squamous cell carcinomaof the lung. Carcinogenesis 2004;25:2165–71.

37. Kim DH, Kim JS, Ji YI, et al. Hypermethylation of RASSF1A pro-moter is associated with the age at starting smoking and a poor



prognosis in primary non-small cell lung cancer. Cancer Res 2003;63:3743–6.

38. Johnson-Pais T, Degnin C, Thayer MJ. pRB induces Sp1 activity byrelieving inhibition mediated by MDM2. Proc Natl Acad Sci U S A2001;98:2211–6.

39. Lagger G, Doetzlhofer A, Schuettengruber B, et al. The tumor sup-pressor p53 and histone deacetylase 1 are antagonistic regulators ofthe cyclin-dependent kinase inhibitor p21/WAF1/CIP1 gene. Mol CellBiol 2003;23:2669–79.

40. Vassilev LT, Vu BT, Graves B, et al. In vivo activation of the p53 path-way by small-molecule antagonists of MDM2. Science 2004;303:844–8.

Cancer Res; 70(14) July 15, 2010 5817



2010;70:5807-5817. Published OnlineFirst June 22, 2010.Cancer Res Ruo-Kai Lin, Chiu-Yi Wu, Jer-Wei Chang, et al. Methyltransferase-1 Overexpression in Lung CancerDysregulation of p53/Sp1 Control Leads to DNA

Updated version

10.1158/0008-5472.CAN-09-4161doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2010/06/18/0008-5472.CAN-09-4161.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/70/14/5807.full#ref-list-1

This article cites 38 articles, 15 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/70/14/5807.full#related-urls

This article has been cited by 8 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/70/14/5807To request permission to re-use all or part of this article, use this link

on July 27, 2018. © 2010 American Association for Cancer Research.cancerres.aacrjournals.org Downloaded from


http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-09-4161

http://cancerres.aacrjournals.org/content/suppl/2010/06/18/0008-5472.CAN-09-4161.DC1

http://cancerres.aacrjournals.org/content/70/14/5807.full#ref-list-1

http://cancerres.aacrjournals.org/content/70/14/5807.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/70/14/5807


cancer research dysregulation of p53/sp1 control leads to...

Documents