dnmt3a plays a role in switches between doxorubicin-induced senescence and apoptosis of colorectal...

DNMT3a plays a role in switches between doxorubicin-inducedsenescence and apoptosis of colorectal cancer cells

Yu Zhang1, Yanyan Gao1, Guoping Zhang2, Shuyan Huang1, Zhixiong Dong1, Chenfei Kong1, Dongmei Su1,

Juan Du1, Shan Zhu2, Qian Liang1, Jianchao Zhang1, Jun Lu1 and Baiqu Huang1

1 The Institute of Genetics and Cytology, The Key Laboratory of Molecular Epigenetics of Ministry of Education (MOE),

Northeast Normal University, Changchun, China2 Biology Section of Guangdong Medical College, Dongguan, Guangdong, China

The DNA-damaging drug doxorubicin (Dox) induces cell senescence at concentrations significantly lower than those required

for induction of apoptosis. At low Dox concentrations, tumor suppressor p53 is activated, which enhances the expression of

p21Waf1/Cip1 (p21). At high concentrations, Dox activates p53 leading to apoptosis without enhancing p21 expression. The

underlying mechanisms and factors that govern the differential effects of Dox in inducing senescence and apoptosis are

unclear. Here, we report that the DNA methyltransferase (DNMT) DNMT3a was upregulated by Dox especially at concentrations

that induced apoptosis in HCT116 colorectal cancer cells, and this process was regulated by p53. Meanwhile, p21 expression

was significantly upregulated at senescence-inducing concentrations and kept low on treatment with apoptosis-inducing

concentrations of Dox. The differential expression of DNMT3a and p21 in response to Dox suggests that DNMT3a may be a

key factor in switches between Dox-induced senescence and apoptosis. Moreover, when DNMT3a was silenced, treatment of

HCT116 cells with apoptosis-inducing concentration of Dox increased the percentage of cells undergoing senescence,

accompanied by upregulation of p21. Contrarily, senescence-inducing concentration of Dox promoted apoptosis rate, and p21

expression was repressed. Surprisingly, no changes in DNA methylation status at p21 promoter were detected at either

ranges of Dox, although DNMT3a and HDAC1 were recruited to p21 promoter at apoptosis-inducing Dox concentration, where

they were present in the same complex. Overall, these data demonstrate that DNMT3a impacts the expression of p21 and

plays a role in determining the Dox-induced senescence and apoptosis in HCT116 cells.

When proliferating cells encounter a genotoxic stress,induced by, for instance the chemotherapeutic drugs, the cellcycle must be arrested immediately to ensure DNA integrity.This event is usually followed by a decision of whether thecells remain arrested in the cell cycle for initiating DNArepair, or execute the apoptotic program. The anthracyclineantibiotic agent doxorubicin (Dox) has been used as a che-motherapeutic drug for >40 years for the treatment of a vari-

ety of malignancies.1 Despite the extensive and long-standingclinical utilization of Dox, the mechanism of its actionremains uncertain and has long been an issue of controversy.Dox can induce at least 3 distinct types of cell death, i.e., se-nescence, apoptosis, and necrosis, in a concentration-depend-ent manner. Specific molecular markers such as p21, acti-vated caspase-3 and activated Akt, were associated with thesedeath modes.2 At low concentrations, the predominant effectof Dox is to initiate senescence, whereas at high Dox concen-trations, tumor cells may undergo cell death.

The p53 tumor suppressor is a transcription factor that isinvolved in the cellular DNA damage response, causing eitherG1 arrest or apoptosis.3 p53 plays an important role in Dox-induced senescence and apoptosis. At low concentrations ofDox, p53 is activated to enhance the expression of cell cycleregulating proteins such as p214; and at high concentrationsof Dox, p53 is activated to induce apoptosis through tran-scriptional upregulation of proapoptotic proteins such as Bax,or BH3 family members such as Puma, in concert with directtargeting of mitochondria and neutralization of antiapoptoticmembers of the Bcl-2 family such as Bcl-2 and Bcl-xL3.5

p21 was initially discovered as a negative regulator of cellcycle progression.6 The cell cycle inhibitory effects of p21may be attributed to its ability to bind CDKs as well as pro-liferating cell nuclear antigen, resulting in inhibition of cellcycle progression.7 Although activation of p21 is important

Key words: DNMT3a, doxorubicin (Dox), p53, p21, senescence,

apoptosis

Grant sponsor: National Natural Science Foundation of China;

Grant numbers: 30671184, 30971613; Grant sponsor: National

Basic Research Program of China; Grant numbers: 2005CB522404,

2006CB910506; Grant sponsor: Program for Changjiang Scholars

Innovative Research Team (PCSIRT) in Universities;

Grant number: IRT0519

DOI: 10.1002/ijc.25365

History: Received 14 Dec 2009; Accepted 16 Mar 2010; Online

5 Apr 2010

Correspondence to: Jun Lu, The Key Laboratory of Molecular

Epigenetics of MOE, Northeast Normal University, Changchun 130024,

China, Tel.: þ86-431-85098729, E-mail: [email protected]; or Baiqu

Huang, The Institute of Genetics and Cytology, Northeast Normal

University, Changchun 130024, China, Tel.: þ86-431-85099798, Fax:

þ86-431-85099768, E-mail: [email protected]

Can

cerCellBiology

Int. J. Cancer: 128, 551–561 (2011) VC 2010 UICC

International Journal of Cancer

IJC

in mediating p53-dependent cell growth arrest, p21 is notessential for p53-mediated apoptosis.8 p21 can protect humancolorectal cancer cells against p53-dependent and p53-inde-pendent apoptosis induced by natural chemopreventive andtherapeutic agents.9 Dox induces apoptosis by p53-dependentand p53-independent mechanisms, and p53-dependent andp53-independent expression of p21 protects cells from Dox-induced apoptosis.10 In human prostate and colorectal cancercells with wild-type p53, Dox strongly induced p53 expres-sion, while p21 expression was increased at senescence-induc-ing concentrations but decreased at apoptosis-inducing con-centrations of Dox. It was reported that the Dox-inducedapoptosis occurred in parallel with p21 downregulation.11

Why p21 expression is increased at senescence-inducing Doxconcentrations but decreased at apoptosis-inducing concentra-tions has not been fully studied. In a study addressing themechanism of this phenomenon, Massague et al. demon-strated a critical role of Myc and the DNA binding proteinMiz-1, being recruited to the p21 promoter and influencingthe outcome of the p53 response to DNA damage.12 Anotherkey mechanistic study showed that caspase-3 mediated cleav-age of p21 during the apoptotic process resulting in theembark of growth-arrested cells onto apoptosis.13 In addition,Dox was able to induce acetylation of histone H3 at the pro-moter of p21 only at the senescence-inducing concentrationof Dox,14 implicating the involvement of other factors suchas epigenetic modifying enzymes in regulation of p21 tran-scription in response to Dox.

DNA methyltransferases (DNMTs) are responsible for themaintenance of methylation status in the genome, and cata-lyze the transfer of a methyl moiety from S-adenosyl-L-methi-onine (SAM) to the cytosine of a CpG dinucleotide.15 Threedistinct DNMTs, namely, DNMT1, DNMT3a and DNMT3bencoded by different genes, direct DNA methylation in mam-malian cells. DNMT1 generally utilizes hemimethylated DNAas the substrate and is involved in maintenance of methyla-tion, whereas DNMT3a and DNMT3b have been shown tobe required for de novo DNA methylation.16,17 Several studieshave shown that besides affecting DNA methylation, DNMTscan also act as corepressors to silence gene expression, inpart through their association with histone deacetylases(HDACs) that help maintaining chromatin in a compact andsilent state.18,19 DNMTs are highly expressed in many cancercells to silence certain tumor suppressor genes.20,21 Moreover,Dox has been shown to act as a potent inhibitor of DNMT1activity,22 most likely acting through DNA intercalation.DNMT1 is required for conditional apoptosis induced by aparticular concentration of Dox, suggesting that DNMT is oneof the targets of Dox for apoptosis induction in cancer cells.22

Albeit these available data, the molecular events thatdetermine the switch of tumor cells between Dox-inducedapoptosis and senescence remain largely unknown. BesidesDNMT1, roles of other DNMTs in response to Dox areunclear. We show in our study that expression of DNMT3awas upregulated by Dox in colorectal cancer cells with wild-

type p53. DNMT3a was regulated by p53 in response to Dox.Our data indicate that DNMT3a played a role in determiningDox-induced senescence and apoptosis by impacting theexpression of p21.

Material and MethodsCell lines and reagents

Human colorectal cancer cell lines HCT116 and SW480 werecultured in appropriate media with 10% fetal bovine serum,100 U/ml penicillin and 100 lg/ml streptomycin, and kept ina humidified atmosphere of 5% CO2. Genomic DNAs wereextracted using the standard proteinase-K method. Total RNAwas extracted by using the Trizol reagent (TaKaRa, Dalian/China). Doxorubicin (Dox), 7-ethyl-10-hydroxycamptothecin(SN38) and etoposide (VP-16) were obtained from (Sigma, St.Louis/Missouri).

Reverse transcription-polymerase chain reaction

For cDNA synthesis, 1 lg of total RNA was reverse tran-scribed using the RT-systems supplied by Promega. Quantita-tive real-time reverse transcription-polymerase chain reaction(RT-PCR) was carried out on an ABI Prism 7000 SequenceDetection System (Applied Biosystems, Foster City/California),and SYBR Green (TOYOBO, New York/New York) was usedas a double-stranded DNA-specific fluorescent dye. The PCRprimer sequences were as follows. p21: 50-GGATGTCCGTCA-GAACCC-30 (sense) and 50-GCT CCCAGGCGAAGTCA-30

(antisense)23; p53: 50-CCTCCTCA GCATCTTATCCG-30

(sense) and 50-CACAAACACGCACC TCAAA-30 (antisense)24;DNMT3a: 50-GCCCAAGGTCAAG GAGATTATTG-30 (sense)and 50-TCTGCCGCACCTCGTA CAC-30 (antisense)25;DNMT3b: 50-CGACAAGAGGGACAT CTCACG-30 (sense)and 50-CAGAAACTTTGATGGCATCA ATCA-30 (antisense)25;b-actin: 50-TCGTGCGTGACATTAA GGAG-30 (sense) and 50-ATGCCAGGGTACATGGTGGT-30 (antisense).26

Plasmid constructs and transfection

DNMT3a expression plasmid and DNMT3a RNA interfer-ence plasmid were the gifts from Dr. Peter E. Lobie (The Lig-gins Institute, University of Auckland, Auckland, New Zea-land). siRNA targeting p53 sequence (GACTCCAGTGGTAATCTAC)27 were synthesized. Oligonucleotide that repre-sents p53 siRNA was cloned into the pSuper-neo vector(Oligo Engine, Seattle/Washington) between BglII andHindIII sites. Plasmids were transfected using Lipofectami-neTM2000 (Invitrogen, Carlsbad/California).

DNA methylation analysis

Bisulfite treatment of genomic DNA was carried out by usingthe CpGenomeTM DNA Modification Kit (Millipore, Consett/UK) according to the manufacture’s instructions. For bisulfitesequencing, primer sequences were synthesized as follows: 50-AGGGAAGTGTTTTTTTG-30 (sense) and 50-CAAAAACACCTATAAA-30 (antisense). PCR products were gel purifiedand cloned into the pMD18-T vector (TaKaRa) according to

Can

cerCellBiology

552 DNMT3a in response to doxorubicin


the manufacture’s instructions. Integrated PCR fragmentswere verified by EcoRI and HindIII digestion, and sequencedwith the M13 forward primer.

Western blotting

Western blotting was performed as described previously.28

The primary antibodies used were rabbit anti-DNMT3a(1:1,000, Cell Signaling Technology (CST), Danvers/Massa-chusetts), rabbit anti-p53 (1:1,000, CST), rabbit anti-p21(1:500, Santa Cruz), rabbit anti-HDAC1 (1:1,000, CST), rab-bit anticleaved caspase-3 (1:500, CST) and mouse anti-b-actin(1:10,000, Sigma).

Senescence-associated galactosidase (SA-b-gal)activity assay

Cytochemical staining for SA-b-galactosidase at pH6.0 wasperformed using a senescence-b-galactosidase Staining Kit(Cell Signaling Technology). All the experiments were repeated3 times, and one of the representative results was shown.

TUNEL assay

HCT116 cells were cultured on the Poly-PrepVR

slides (SigmaCat. # P0425) 24 hr before transfected with or withoutDNMT3a expression plasmid, then treated with the indicatedconcentrations of Dox. TUNEL assays were performed aftercells were treated with Dox for 36 hr following manufacture’sinstructions. DNase I was used as the positive control andthe ‘‘no rTdT Enzyme’’ sample as the negative control.

Caspase-3/7 activity analysis

HCT116 cells were plated at 2 � 103 cells per well on 96-well plates. At 24 hr after treatment with Dox, the Apo-ONEcaspase-3/7 Reagent (Promega, Madison/Wisconsin) wasadded. Cells were incubated for 1 hr at room temperatureprior to record of the fluorescence (485Ex/527Em).

Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assays were per-formed as described previously.28 Immunoprecipitation wasdone with rabbit antiacetyl-H3/H4 (Millipore), rabbit anti-p53(Abcom), rabbit anti-DNMT3a (CST) or rabbit anti-HDAC1(CST) antibody. Primer pairs for p21 proximal promoter were50-ACCAACGCAGGCGAGGGA-30 (sense) and 50-CCGGCTCCACAAGGAACT-30 (antisense); for p21 distal promoter were50-AATTCCTCTGAAAGCTGACTGCC-30 (sense) and 50-AGGTTTACCTGGGGTCTTTAGA-30 (antisense).23

Coimmunoprecipitation assay

Total cell extracts from Dox-treated HCT116 cells were pre-cleared with salmon sperm DNA/protein A-agarose beads(Millipore). Rabbit anti-p53, rabbit anti-HDAC1 or rabbitanti-DNMT3a antibody was added for immunoprecipitation.The precipitates were then subjected to SDS-PAGE followedby transfer onto a PVDF membrane and incubation withanti-p53, anti-HDAC1 or anti-DNMT3a antibody. Samples

were detected using the Super Signal West Pico Chemilumi-nescent Substrate (Thermo, Rockford/Illinois) detectionmethod following the manufacture’s instructions.

ResultsDox upregulated DNMT3a but not DNMT3b in HCT116

colorectal cancer cells

To investigate the roles of Dox in inducing senescence andapoptosis in colorectal cancer cells, we treated HCT116 cellswith various concentrations of Dox. At the senescence-induc-ing concentrations (under 0.2 lM), HCT116 cells exhibitedphenotypic changes that resembled those observed in thecells undergoing senescence, including intensified SA-b-galstaining, while at apoptosis-inducing Dox concentrations(higher than 0.5 lM), the SA-b-gal positive staining cellswere not significantly changed (Fig. 1a). The expression ofmolecular markers associated with Dox-induced apoptosis(activated caspase-3) was detected. The caspase-3/7 activitieswere upregulated starting at 0.5 lM Dox (Fig. 1b) with aconcurrent increase in cleaved caspase-3 protein (Fig. 1c).Meanwhile, the senescence-inducing Dox concentration (0.1lM) brought about a remarkable increase of p21 mRNA andprotein level, a senescence marker, while it was kept low atthe apoptosis-inducing concentration (1 lM) (Figs. 1d and1h). Moreover, the p53 mRNA and protein levels wereincreased along with the increase of Dox concentration, espe-cially at apoptosis-inducing concentration (1 lM) (Figs. 1eand 1h). We then assessed the changes of DNMTs, andfound that both DNMT3a mRNA and protein levels wereslightly increased at 0.1 lM Dox but were significantlyincreased at 1 lM Dox (Figs. 1f and 1h). Meanwhile,DNMT3b mRNA was slightly declined on Dox treatment(Fig. 1g). These findings provide evidence that senescenceand apoptosis were induced at distinct Dox concentrations,and the expression of DNMT3a, but not DNMT3b, wasstimulated by Dox treatment. Furthermore, we used other 2DNA damage drugs 7-ethyl-10-hydroxycamptothecin (SN38)and etoposide (VP-16) for confirmation, and we found thatat apoptosis-inducing concentrations, SN38 (500nM) andVP-16 (10 lM) were also able to upregulate mRNA and pro-tein expression of DNMT3a (Figs. 1i–1k).

p53 was involved in Dox-induced DNMT3a expression

Data described above reveal that expression of p53 andDNMT3a exhibited a similar pattern and tendency ofchanges in HCT116 cells on the treatment with Dox (Figs.1e, 1f and 1h). We were then interested in studying the linkbetween DNMT3a expression and the activation of p53 inthis Dox-inducing cell model. First, we detected the differen-tial expression of p21 and DNMT3a in HCT116 cells withwild-type p53 and SW480 cells with mutant p53.29 We foundthat p21 mRNA was increased significantly at 0.1–0.2 lMDox followed by a sharp decrease starting at 0.5 lM inHCT116 cells, while it was moderately increased from0.2 lM Dox and peaked at 1 lM Dox in SW480 cells (Fig. 2a).

Can

cerCellBiology

Zhang et al. 553


Figure 1

Can

cerCellBiology



Meanwhile, the DNMT3a mRNA level was increased whentreated with Dox especially at concentrations higher than 0.5lM in HCT116 cells, but it remained unchanged in SW480cells (Fig. 2b). The DNMT3a and p21 protein levels exhibitedbasically the same patterns of change as mRNA in the 2 celllines (Fig. 2c). These results suggest that DNMT3a may beregulated by p53 on Dox treatment. Furthermore, interferingof p53 by a specific siRNA counteracted the effect of 1 lMDox treatment in promoting DNMT3a protein and mRNAlevel in HCT116 cells (Figs. 2e–2g). These results indicatethat p53 played a role in the induction of DNMT3a by Dox.

Roles of DNMT3a in switches between Dox-induced

apoptosis and senescence

The fact that both DNMT3a mRNA and protein levels weresignificantly increased at apoptosis-inducing concentration ofDox (Figs. 1f and 1h) implicates a possible role of DNMT3ain determination of Dox-induced apoptosis and senescence.To validate this, we transfected HCT116 cells with DNMT3asiRNA plasmid, and we found that the number and intensityof SA-b-gal staining cells were increased at 1 lM Dox afterDNMT3a was knocked down (Fig. 3a). Moreover, overexpres-sion of DNMT3a increased the number of TUNEL positivestaining cells at 0.1 lM Dox (Fig. 3b), and upregulated cas-pase-3/7 activities at this concentration (Fig. 3c). Thus, thesedata demonstrate that DNMT3a played a role in switchesbetween apoptosis and senescence in response to Dox.

DNMT3a repressed the expression of p21 at

apoptosis-inducing concentration of Dox

We next examined whether DNMT3a influences the senes-cence marker p21 in this experimental model. We found thatafter transfection with DNMT3a siRNA, p21 protein levelwas upregulated while the cleaved caspase-3 protein wasdownregulated at 1 lM Dox (Fig. 4a). Contrarily, transfec-tion of DNMT3a expression plasmid downregulated p21 pro-tein level but upregulated the cleaved caspase-3 protein at 0.1lM Dox (Fig. 4a). Similar results were obtained at the p21mRNA level (Fig. 4c). Thus, we conclude that DNMT3a par-ticipated in switches of Dox-induced apoptosis and senes-cence through repressing the expression of p21.

DNMT3a and HDAC1 were recruited to p21 promoter at the

apoptosis-inducing Dox concentration

We then intended to determine whether DNMT3a affectedp21 expression as a DNMT. We analyzed the methylationstatus of the CpG sites at p21 promoter after treatment ofDox at different concentrations. We examined 47 CpG sitesfrom �221 upstream to þ214 downstream the transcriptionsite by using bisulfite sequencing method. Unexpectedly, wedid not detect any methylated CpG sites in cells treated ei-ther with senescence-inducing or with apoptosis-inducingDox concentrations (Fig. 5a). Since DNMT3a has been impli-cated to bind HDACs and can be recruited by a sequence-specific repressor to silence gene transcription,19 we wantedto find out whether DNMT3a and HDAC1 could berecruited to p21 promoter. Our ChIP assays showed that theacetylation levels of H3 and H4 at the p21 promoter wereevidently enhanced when treated with 0.1 lM Dox, whilethey were declined when treated with 1 lM Dox (Figs. 5b–5d). Also, the ChIP results revealed that p53 was enriched onp21 promoter at 0.1 lM Dox but decreased at 1 lM Dox(Figs. 5b and 5e). Furthermore, HDAC1 and DNMT3a wererecruited to the promoter at 1 lM but not at 0.1 lM (Figs.5b, 5f and 5g). Apparently, recruitment of HDAC1 was re-sponsible for the changes in histone acetylation at p21 pro-moter. However, although DNMT3a was also recruited top21 promoter, it did not impact the methylation status ofp21 promoter at apoptosis-inducing concentration of Dox.

DNMT3a was associated with HDAC1 but not with p53 at

apoptosis-inducing Dox concentration

DNMT3a was reported to be able to bind HDACs to silencegene transcription,19 and to interact with p53 to repress p53-mediated gene expression.30 We then examined the possibleinteractions among DNMT3a, HDAC1 and p53 on Doxtreatment. Coimmunoprecipitation (CoIP) assays revealedthat with increasing concentrations of Dox, DNMT3a andHDAC1 were coimmunoprecipitated; however, the DNMT3a/p53 or HDAC1/p53 complexes were not detected after treat-ment with different concentrations of Dox (Fig. 6). The CoIPresults suggest that DNMT3a and HDAC1 existed in thesame complex in the model of Dox-induced apoptosis.

Figure 1. Induction of DNMT3a expression by Dox in HCT116 cells. (a) Dox-induced senescence in HCT116 cells. Cells incubated with 0.1–2.0

lM Dox for 24 hr, subcultured to fresh medium for another 48 hr, and then stained for SA-b-gal activity to assess the senescence. (b) Dox-

induced caspase3/7 activities in HCT116 cells. Cells incubated with 0.1–2.0 lM Dox for 24 hr before caspase-3/7 activities were detected. (c)

Dox-induced cleaved caspase-3 in HCT116 cells. Cleaved caspase-3 protein was detected by Western blotting in HCT116 cells incubated with

0.1–2.0 lM Dox for 24 hr. (d–h) Dox-induced expression of p21, p53, DNMT3a and DNMT3b in HCT116 cells. Cells were treated with 0.1 or 1.0

lM Dox for the time indicated. p21 (d), p53 (e), DNMT3a (f) and DNMT3b (g) mRNA levels were detected by real-time RT-PCR. (h) The protein

levels of p21, p53, DNMT3a and DNMT3b were detected by Western blotting. (i) SN38 induced expression of DNMT3a in HCT116 cells. Cells

were treated with 50 or 500 nM SN38 for 24 hr. DNMT3a mRNA levels were detected by real-time RT-PCR. (j) VP-16 induced expression of

DNMT3a in HCT116 cells. Cells were treated with 1 or 10 lM VP-16 for 24 hr. DNMT3a mRNA levels were detected by real-time RT-PCR. (k) The

protein levels of DNMT3a induced by SN38 and VP16 were detected by Western blotting. For real-time RT-PCR, b-actin was used as the internal

reference. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Can

cerCellBiology

Zhang et al. 555


Figure 2. Correlation between Dox-induced p53 and DNMT3a expression. (a, b) Differential expression of p21 and DNMT3a induced by Dox in

HCT116 and SW480 cells. Cells were treated with Dox at the indicated concentrations for 24 hr. Expression of p21 (a) and DNMT3a (b) was

assessed by real-time RT-PCR. (c) Cells were treated with Dox at the indicated concentrations for 24 hr. Expression of p21, cleaved caspase-3

and DNMT3a were detected by Western blotting. (d) Real-time RT-PCR and Western blotting verification of the interfering efficiency of p53

siRNA in HEK 293T cells. (e–g) Effect of p53 knockdown on Dox-induced DNMT3a expression. HCT116 cells were transfected with p53 siRNA

plasmid and then treated with 0.1 or 1.0 lM Dox for 24 hr, DNMT3a and p53 protein levels were assessed by Western blotting (e). DNMT3a

and p53 mRNA levels were determined by real-time RT-PCR (f, g). For real-time RT-PCR, b-actin was used as the internal reference.

Can

cerCellBiology



DiscussionDox is an anthracycline agent, which has been used as ananticancer drug for almost 40 years. However, the mecha-nisms of Dox’s action are still an issue of controversy. Treat-ments of cells with low and high concentrations of Doxinduce senescence and apoptosis, respectively. At the senes-cence-inducing concentrations, p53 is activated, which in

turn induces expression of p21 and other genes. It has beenknown that the senescence development is induced by p21 incells with limited DNA damage.31 Expression of p21 corre-lates with activation of p53 at low but not at high Dox con-centrations, however, the mechanisms underlying this differ-ential action remain unclear. Uncontrolled action of DNMTsmay cause silence of many tumor suppressor genes in a

Figure 3. Effect of DNMT3a on switches between apoptosis and senescence in response to Dox. (a) Effect of DNMT3a knockdown on Dox-

induced senescence. HCT116 cells were transfected with control siRNA or DNMT3a siRNA, incubated with 0.1 or 1.0 lM Dox for 24 hr,

replaced with fresh medium and cultured for another 48 hr, and then stained for SA-b-gal activity to detect senescence. (b, c) Effect of

DNMT3a on Dox-induced apoptosis. HCT116 cells were transfected with or without DNMT3a expression plasmid and incubated with or

without 0.1 lM Dox for 36 hr. TUNEL assays were performed to detect apoptotic cells. The ‘‘DNAse I-treated’’ and ‘‘no rTdT’’ samples were

used as positive and negative controls, respectively (b). Caspase-3/7 activities were detected (c). [Color figure can be viewed in the online

issue, which is available at wileyonlinelibrary.com.]

Can

cerCellBiology

Zhang et al. 557


variety of tumors. DNMT inhibitors, such as 5-aza-cytidine(5-aza-dC), have been attracted extensive interests of study,because DNA hypomethylation induced by these inhibitorsmay re-activate the silenced tumor suppressor genes.24 TheDNMT1 protein level was decreased in HCT116 cells treatedwith the apoptosis-inducible concentration of Dox, indicatingthat DNMT1 was required for apoptosis induced by a partic-ular concentration of Dox.22 Our results indicated thatDNMT3b was downregulated after Dox treatment (Fig. 1g),which is similar to that of DNMT1. On the contrary, resultsfrom the present study showed that the expression ofDNMT3a was upregulated especially at the apoptosis-induc-ing concentrations of Dox (Figs. 1f and 1h). Moreover, theexpression pattern of DNMT3a in response to Dox in our ex-perimental model was apparently consistent with that of p53,but was distinct from that of p21 (Fig. 1), and this hasintrigued us to further explore the role of DNMT3a in Dox-induced senescence and apoptosis. In fact, our results indicatethat DNMT3a indeed influenced the Dox-induced apoptosisand senescence (Fig. 3).

p53 is known to be activated by a variety of antiprolifera-tive or proapoptotic molecules triggered by DNA-damagingagents through its stabilization and accumulation in the nu-cleus, and binding to multiple DNA-response elements.32 p53regulates multiple responses to genotoxic stress by eithertranscriptional activation or repression of target genes encod-ing proteins involved in cell cycle control (e.g., p21), DNA

repair (e.g., gadd45) and apoptosis (e.g., Bax, Bcl2 and survi-vin).32 Our experiments revealed that DNMT3a and p53exhibited similar patterns of expression in response to Dox(Figs. 1e, 1f and 1h); and in p53 mutant SW480 cells, Doxfailed to induce DNMT3a (Figs. 2b and 2c). These facts implythat there may be some relevance between p53 and DNMT3ain response to Dox. And indeed, we found that when interfer-ing with p53 expression, the DNMT3a expression induced byDox was counteracted (Figs. 2e and 2f). Although p53 playeda role in Dox-induced DNMT3a, our CoIP experiments didnot detect the association of DNMT3a and p53. We speculatethat at the apoptosis-inducing Dox concentrations, DNMT3awas involved in repressing the p21 expression, while p53 con-ducted the p53-dependent apoptosis.

p21 plays an essential role in growth arrest after DNAdamage.33 A number of recent studies have pointed out thatin addition to being an inhibitor of cell proliferation, p21acts as an inhibitor of both p53-dependent and p53-inde-pendent apoptosis in a number of systems.34 In the model ofDox-induced senescence, p53 was activated to enhance p21expression; while at a relatively high Dox concentration, p53-mediated apoptosis through transcriptional upregulation ofproapoptotic proteins but failed to enhance p21 expression.Under apoptotic circumstances, caspase-mediated cleavage ofp21 usually results in promotion of the p53-dependent apo-ptosis.13 We and others found that both p21 protein andmRNA levels were kept low on the treatment with apoptosis-

Figure 4. Effect of DNMT3a on expression of p21 and cleaved caspase-3 in response to Dox. (a) HCT116 cells were transfected with

DNMT3a siRNA or DNMT3a expression plasmid, treated with 0.1 or 1.0 lM Dox for 24 hr before DNMT3a, p21 and cleaved caspase-3

protein levels were assessed by Western blotting. (b) Real-time RT-PCR verification of the interfering efficiency of DNMT3a siRNA in HCT116

cells. (c) HCT116 cells were transfected with DNMT3a siRNA plasmid, treated with 0.1 or 1.0 lM Dox, p21, and the mRNA level was

detected by real-time RT-PCR. For real-time RT-PCR, b-actin was used as the internal reference.

Can

cerCellBiology



Figure 5. Recruitment of DNMT3a and HDAC1 to the p21 promoter on Dox treatment. (a) The methylation status of the CpG sites at p21

promoter after treatments with various Dox concentrations. Schematic illustration of the region of p21 promoter analyzed by sodium

bisulfite DNA sequencing. The arrow denotes the transcription start site (þ1). Sodium bisulfite DNA sequencing of p21 in HCT116 cells

treated with 0.1 or 1.0 lM Dox. Circles in each horizontal row represent the analysis of a single clone of bisulfite-treated DNA of the 64

CpG sites in the regions. Open and solid circles represent unmethylated and methylated CpG sites, respectively. (b–g) DNMT3a and HDAC1

were recruited to p21 promoter at apoptosis-inducing concentration of Dox. HCT116 cells were treated with indicated concentrations of Dox

for 24 hr and harvested for ChIP assays. Samples were immunoprecipitated with anti-H3, anti-H4, anti-p53, anti-DNMT3a or anti-HDAC1

antibodies, and the precipitated DNA fragments were amplified by using PCR (b) and real-time PCR (c–g).

Can

cerCellBiology

Zhang et al. 559


inducing concentration of Dox (Figs. 1d and 1h), implicatingthat there may be some other factors involved in repressionof p53-dependent induction of p21 transcription. It wasreported that c-Myc and the DNA binding protein Miz-1were recruited to the p21 promoter and influenced theresponse of p53 to DNA damage.12 Our results show thatDNMT3a was increased especially at the apoptosis-inducingDox concentrations and p21 was significantly increased atthe senescence-inducing concentrations. The expression ofDNMT3a exhibited a relatively opposite pattern to that ofp21 in this system. This prompted us to investigate whetherDNMT3a suppressed the transcription of p21 to participatein switches between apoptosis and senescence in response toDox. Our results indicate that DNMT3a influenced the Dox-induced apoptosis and senescence and affected the expressionof p21 (Figs. 3 and 4).

DNMT3a is the DNMT responsible for transferring amethyl group to the 5-position of the cytosine ring in theCpG islands.35 In addition to transcriptional silencing through

direct methylation of CpG islands of the promoters,DNMT3 proteins were also shown to recruit other transcrip-tional repressors to silence the target genes.19,36,37 Moreover,myc was able to recruit DNMT3a to methylate the p21 pro-moter thereby silencing p21 expression.38 To find out theroles of DNMT3a in regulating p21 expression in our exper-imental model, we examined the methylation status of p21promoter on different concentrations of Dox treatment. Sur-prisingly, we did not detect apparent changes in methylationstatus at p21 promoter under either senescence- or apopto-sis-inducing Dox concentrations (Fig. 5a). We then turnedour attention to histone acetylation, because Dox wasreported to induce acetylation of histone H3 at p21 pro-moter at senescence-inducing concentration14; and DNMT3awas able to bind deacetylases to silence transcription.19 Wethen examined whether DNMT3a was involved in regulatingthe histone acetylation level of p21 promoter. Our ChIPassays demonstrate that DNMT3a and HDAC1 wereenriched at p21 promoter under apoptosis-inducing Doxconcentration (Figs. 5b, 5f and 5g), which in turn reducedthe acetylation level of histones H3 and H4 at the p21 pro-moter (Figs. 5b–5d). Surprisingly, in HCT116 cells, wedetected the recruitment of DNMT3a to p21 promoter with-out Dox treatment. We speculate that in HCT116 cells, thelow DNMT3a level may help maintaining the low histoneacetylation level at p21 promoter without the associationwith HDAC1. However, this will need further investigationsto validate.

Results from our CoIP assays verified that DNMT3a andHDAC1 existed in the same complex, but neither DNMT3anor HDAC1 was detected to be associated with p53 (Fig. 6),despite a previous finding that DNMT3a interacted with p53to repress p53-mediated gene expression.30 Apparently, fur-ther investigations will be required to identify the factor(s)that recruits DNMT3a and HDAC1 to p21 promoter.

In summary, we have established in our study thatDNMT3a was upregulated by Dox, especially at the apopto-sis-inducing concentrations, in HCT116 colorectal cancercells. DNMT3a played a role in switches between senescenceand apoptosis induced by Dox, and impacted the p21expres-sion. These data provide useful information and basis for fur-ther insight into the function and mechanisms of DNMTs indrug-induced apoptosis and senescence.

AcknowledgementsWe are grateful to Dr. Peter E Lobie for kindly providing DNMT3a expres-sion plasmid and DNMT3a RNA interference plasmid.

References

1. Gewirtz DA. A critical evaluation of themechanisms of action proposed for theantitumor effects of the anthracyclineantibiotics adriamycin and daunorubicin.Biochem Pharmacol 1999;57:727–41.

2. Rebbaa A, Zheng X, Chou PM, Mirkin BL.Caspase inhibition switches doxorubicin-induced apoptosis to senescence. Oncogene2003;22:2805–11.

3. El-Deiry WS, Harper JW, O’Connor PM,Velculescu VE, Canman CE, Jackman J,

Pietenpol JA, Burrell M, Hill DE, Wang Y,et al. WAF1/CIP1 is induced in p53-mediated G1 arrest and apoptosis. CancerRes 1994;54:1169–74.

4. Chang BD, Swift ME, Shen M, Fang J,Broude EV, Roninson IB. Molecular

Figure 6. DNMT3a and HDAC1 existed in the same complex.

HCT116 cells were treated with 0.1 or 1.0 lM Dox for 24 hr.

Whole-cell extracts were prepared and immunoprecipitated with

anti-IgG, anti-p53 or anti-HDAC1 antibodies, and incubated with

protein A agarose beads. Immunoprecipitates were subjected to

immunoblotting analysis with anti-p53, anti-HDAC1 and anti-

DNMT3a antibodies. Western blots of the whole-cell extracts

detected with anti-DNMT3a, anti-p53, anti-HDAC1 and anti-b-actin

antibodies.

Can

cerCellBiology



determinants of terminal growth arrestinduced in tumor cells by achemotherapeutic agent. Proc Natl Acad SciUSA 2002;99:389–94.

5. Schuler M, Green DR. Transcription,apoptosis and p53: catch-22. Trends Genet2005;21:182–7.

6. Roninson IB. Oncogenic functions oftumour suppressor p21(Waf1/Cip1/Sdi1):association with cell senescence andtumour-promoting activities of stromalfibroblasts. Cancer Lett 2002;179:1–14.

7. Xiong Y, Hannon GJ, Zhang H, Casso D,Kobayashi R, Beach D. p21 is a universalinhibitor of cyclin kinases. Nature 1993;366:701–4.

8. Deng C, Zhang P, Harper JW, Elledge SJ,Leder P. Mice lacking p21CIP1/WAF1undergo normal development, but aredefective in G1 checkpoint control. Cell1995;82:675–84.

9. Mahyar-Roemer M, Roemer K. p21 Waf1/Cip1 can protect human colon carcinomacells against p53-dependent and p53-independent apoptosis induced by naturalchemopreventive and therapeutic agents.Oncogene 2001;20:3387–98.

10. Bunz F, Hwang PM, Torrance C,Waldman T, Zhang Y, Dillehay L,Williams J, Lengauer C, Kinzler KW,Vogelstein B. Disruption of p53 in humancancer cells alters the responses totherapeutic agents. J Clin Invest 1999;104:263–9.

11. Martinez LA, Yang J, Vazquez ES,Rodriguez-Vargas Mdel C, Olive M, HsiehJT, Logothetis CJ, Navone NM. p21modulates threshold of apoptosis inducedby DNA-damage and growth factorwithdrawal in prostate cancer cells.Carcinogenesis 2002;23:1289–96.

12. Wu S, Cetinkaya C, Munoz-Alonso MJ,von der Lehr N, Bahram F, Beuger V,Eilers M, Leon J, Larsson LG. Mycrepresses differentiation-induced p21CIP1expression via Miz-1-dependent interactionwith the p21 core promoter. Oncogene2003;22:351–60.

13. Jin YH, Yoo KJ, Lee YH, Lee SK. Caspase3-mediated cleavage of p21WAF1/CIP1associated with the cyclin A-cyclin-dependent kinase 2 complex is aprerequisite for apoptosis in SK-HEP-1cells. J Biol Chem 2000;275:30256–63.

14. Rebbaa A, Zheng X, Chu F, Mirkin BL.The role of histone acetylation versus DNAdamage in drug-induced senescence andapoptosis. Cell Death Differ 2006;13:1960–7.

15. Bestor TH. The DNA methyltransferases ofmammals. Hum Mol Genet 2000;9:2395–402.

16. Okano M, Li E. Genetic analyses of DNAmethyltransferase genes in mouse modelsystem. J Nutr 2002;132:2462S–5S.

17. Okano M, Bell DW, Haber DA, Li E. DNAmethyltransferases Dnmt3a and Dnmt3bare essential for de novo methylation andmammalian development. Cell 1999;99:247–57.

18. Geiman TM, Sankpal UT, Robertson AK,Zhao Y, Zhao Y, Robertson KD. DNMT3Binteracts with hSNF2H chromatinremodeling enzyme. HDACs 1 and 2, andcomponents of the histone methylationsystem. Biochem Biophys Res Commun2004;318:544–55.

19. Fuks F, Burgers WA, Godin N, Kasai M,Kouzarides T. Dnmt3a binds deacetylasesand is recruited by a sequence-specificrepressor to silence transcription. Embo J2001;20:2536–44.

20. Robertson KD, Uzvolgyi E, Liang G,Talmadge C, Sumegi J, Gonzales FA, JonesPA. The human DNA methyltransferases(DNMTs) 1, 3a and 3b: coordinate mRNAexpression in normal tissues andoverexpression in tumors. Nucleic AcidsRes 1999;27:2291–8.

21. Schneider-Stock R, Ocker M. Epigenetictherapy in cancer: molecular backgroundand clinical development of histonedeacetylase and DNA methyltransferaseinhibitors. IDrugs 2007;10:557–61.

22. Yokochi T, Robertson KD. Doxorubicininhibits DNMT1, resulting in conditionalapoptosis. Mol Pharmacol 2004;66:1415–20.

23. Su D, Zhu S, Han X, Feng Y, Huang H,Ren G, Pan L, Zhang Y, Lu J, Huang B.BMP4-Smad signaling pathway mediatesadriamycin-induced premature senescencein lung cancer cells. J Biol Chem 2009;284:12153–64.

24. Jiemjit A, Fandy TE, Carraway H, BaileyKA, Baylin S, Herman JG, Gore SD.p21(WAF1/CIP1) induction by 5-azacytosine nucleosides requires DNAdamage. Oncogene 2008;27:3615–23.

25. Lu R, Wang X, Chen ZF, Sun DF, TianXQ, Fang JY. Inhibition of the extracellularsignal-regulated kinase/mitogen-activatedprotein kinase pathway decreases DNAmethylation in colon cancer cells. J BiolChem 2007;282:12249–59.

26. Zhang Y, Huang S, Dong W, Li L, Feng Y,Pan L, Han Z, Wang X, Ren G, Su D,Huang B, Lu J. SOX7, down-regulated incolorectal cancer, induces apoptosis andinhibits proliferation of colorectal cancercells. Cancer Lett 2009;277:29–37.

27. Kim TK, Lee JS, Jung JE, Oh SY, Kwak S,Jin X, Lee SY, Lee JB, Chung YG, ChoiYK, You S, Kim H. Interferon regulatoryfactor 3 activates p53-dependent cell

growth inhibition. Cancer Lett 2006;242:215–21.

28. Wang X, Pan L, Feng Y, Wang Y, Han Q,Han L, Han S, Guo J, Huang B, Lu J. P300plays a role in p16(INK4a) expression andcell cycle arrest. Oncogene 2008;27:1894–904.

29. Chang BD, Broude EV, Dokmanovic M,Zhu H, Ruth A, Xuan Y, Kandel ES,Lausch E, Christov K, Roninson IB. Asenescence-like phenotype distinguishestumor cells that undergo terminalproliferation arrest after exposure toanticancer agents. Cancer Res 1999;59:3761–7.

30. Wang YA, Kamarova Y, Shen KC, Jiang Z,Hahn MJ, Wang Y, Brooks SC. DNAmethyltransferase-3a interacts with p53and represses p53-mediated geneexpression. Cancer Biol Ther 2005;4:1138–43.

31. Wells SI, Francis DA, Karpova AY,Dowhanick JJ, Benson JD, Howley PM.Papillomavirus E2 induces senescence inHPV-positive cells via pRB- and p21(CIP)-dependent pathways. Embo J 2000;19:5762–71.

32. Vousden KH, Lu X. Live or let die: thecell’s response to p53. Nat Rev Cancer2002;2:594–604.

33. Roninson IB. Tumor cell senescence incancer treatment. Cancer Res 2003;63:2705–15.

34. Li CH, Tzeng SL, Cheng YW, Kang JJ.Chloramphenicol-induced mitochondrialstress increases p21 expression andprevents cell apoptosis through a p21-dependent pathway. J Biol Chem 2005;280:26193–9.

35. Jair KW, Bachman KE, Suzuki H, TingAH, Rhee I, Yen RW, Baylin SB, SchuebelKE. De novo CpG island methylation inhuman cancer cells. Cancer Res 2006;66:682–92.

36. Deplus R, Brenner C, Burgers WA,Putmans P, Kouzarides T, de Launoit Y,Fuks F. Dnmt3L is a transcriptionalrepressor that recruits histone deacetylase.Nucleic Acids Res 2002;30:3831–8.

37. Di Croce L, Raker VA, Corsaro M, Fazi F,Fanelli M, Faretta M, Fuks F, Lo Coco F,Kouzarides T, Nervi C, Minucci S, PelicciPG. Methyltransferase recruitment andDNA hypermethylation of target promotersby an oncogenic transcription factor.Science 2002;295:1079–82.

38. Brenner C, Deplus R, Didelot C, Loriot A,Vire E, de Smet C, Gutierrez A, Danovi D,Bernard D, Boon T, Pelicci PG, Amati B,et al. Myc represses transcription throughrecruitment of DNA methyltransferasecorepressor. Embo J 2005;24:336–46.

Can

cerCellBiology

Zhang et al. 561


dnmt3a plays a role in switches between doxorubicin-induced senescence and apoptosis of colorectal...

Documents