Regulation of DNA Methyltransferase 1 by the Prb/E2F1 Pathway

Research Article

Regulation of DNA Methyltransferase 1 by the pRb/E2F1 Pathway

Michael T. McCabe,1,2 Joanne N. Davis,1 and Mark L. Day 1,2

1Department of Urology and 2Program in Cell and Molecular Biology, Michigan Urology Center, University of Michigan, Ann Arbor, Michigan

Abstract DNA methylation in mammalian cells is regulated by a family of highly related DNA methyltransferase enzymes (Dnmt-1, Dnmt-3a, Tumor suppressor gene silencing by DNA hypermethylation and Dnmt-3b), which mediate the transfer of methyl groups from contributes to tumorigenesis in many tumor types. This S-adenosylmethionine to the 5V position of cytosine bases in the aberrant methylation may be due to increased expression dinucleotide sequence CpG (3, 4). Dnmt-1 functions as the and activity of DNA methyltransferases, which catalyze the maintenance DNA methyltransferase in mammalian cells and is transfer of methyl groups from S-adenosylmethionine to therefore responsible for accurately replicating genomic DNA cytosines in CpG dinucleotides. Elevated expression of the methylation patterns during the S phase of the cell cycle (5). On the maintenance DNA methyltransferase, DNA methyltransferase other hand, de novo methylation of DNA is believed to be 1 (DNMT-1), has been shown in carcinomas of the colon, lung, performed by Dnmt-3a and Dnmt-3b, which possess both liver, and prostate. Based on the nearly ubiquitous alterations maintenance and de novo DNA methylation activity (4). However, of both DNA methylation and the retinoblastoma protein both groups of enzymes have been shown to exhibit some levels of (pRb) pathway found in human cancer, we investigated a both maintenance and de novo methylation in vitro, suggesting potential regulatory pathway linking the two alterations in that this classification of the DNMTs may be oversimplified (6). murine and human prostate epithelial cells. Analysis of DNA All types of cancers have been shown to possess aberrant DNA methyltransferase levels in RbÀ/À murine prostate epithelial methylation characterized by global genomic hypomethylation cell lines revealed elevated Dnmt-1 levels. Genomic DNA and, yet at the same time, localized areas of hypermethylation sequence analysis identified conserved E2F consensus binding frequently within the promoter regions of tumor suppressor genes sites in proximity to the transcription initiation points of (e.g., E-cadherin, GSTP1, MLH1, and p16INK4a; refs. 7, 8). Confirming murine and human Dnmt-1.Furthermore,theDnmt-1 the importance of DNA methylation in tumorigenesis, studies in promoter was shown to be regulated by the pRb/E2F pathway several tumor types, including those of bladder, kidney, and colon, in murine and human cell lines of epithelial and fibroblast have shown all three DNA methyltransferases to be overexpressed origin. In the absence of pRb, Dnmt-1 transcripts exhibited À/À (9). Overexpression of Dnmt-1 in mouse embryonic stem cells led aberrant cell cycle regulation and Rb cells showed to genomic DNA hypermethylation and loss of imprinting at the aberrant methylation of the paternally expressed gene 3 Igf2 locus similar to that observed in cancer (10). Additionally, (Peg3) tumor suppressor gene. These findings show a link murine (NIH3T3) and human (IMR90/SV40) fibroblasts expressing between inactivation of the pRb pathway and induction of exogenous Dnmt-1 exhibited many hallmarks of transformation, DNA hypermethylation of CpG island–containing genes in including spindle morphology, decreased contact inhibition, tumorigenesis. (Cancer Res 2005; 65(9): 3624-32) increased genomic methylation, increased growth in soft agar, and increased tumorigenicity (11, 12). These data suggest that increased Introduction DNA methyltransferase levels and activity affect the methylation In addition to the extensively studied role of genetic status of genes critical to tumor formation. alterations (i.e., deletions, amplifications, and translocations) in The retinoblastoma protein (pRb) participates in a well- cancer initiation and progression, epigenetic modifications of characterized cell cycle regulatory pathway that is controlled by DNA also play a critical role in these processes through the cyclin-dependent kinases (cdk) and cdk inhibitors (13). In this regulation of gene transcription. DNA methylation represents the pathway, pRb functions to restrict cell cycle progression late in G1 only covalent modification of DNA observed in higher organisms in response to growth inhibitory signals, such as growth factor and has been implicated in the regulation of gene expression depletion or contact inhibition. When the pRb pathway becomes inactivated through any one of several mechanisms (e.g., cyclin D1 during development, genomic imprinting, X chromosome INK4a inactivation, and silencing of parasitic transposable DNA amplification, cdk4 overexpression, or p16 inactivation), elements (see ref. 1 for review). DNA methylation can occur homeostatic balance may be lost, thereby promoting aberrant on cytosines found in CpG dinucleotides and its presence is proliferation of cells leading to tumor formation. This loss of generally associated with transcriptional repression (2). This homeostasis due to inactivation of the pRb pathway also results repression is likely due to the recruitment of chromatin- from aberrant management of other cellular functions regulated by modifying enzymes [i.e., histone deacetylases (HDAC) and SWI/ pRb, including differentiation (14), survival/apoptosis (15), and SNF family proteins] to methylated DNA by proteins containing senescence (16). methyl CpG binding domains, such as MeCP2 and the MBD A role for pRb in tumor development has been convincingly established through several animal models. Although homozygous family proteins. knockout of Rb is embryonic lethal in mice, Rb+/À mice are viable and exhibit a predisposition to develop pituitary and thyroid tumors (17, 18). Additionally, RbÀ/À prostate tissues have been Requests for reprints: Mark L. Day, Department of Urology, University of rescued from embryonic RbÀ/À mice by tissue recombination Michigan, 6219 CCGC, 1500 East Medical Center Drive, Ann Arbor, MI 48109-0944. +/À Phone: 734-763-9968; Fax: 734-647-9271; E-mail: [email protected]. (19). Consistent with the tumor predisposition of Rb À/À I2005 American Association for Cancer Research. mutants, Rb prostate grafts exhibited increased susceptibility

Cancer Res 2005; 65: (9). May 1, 2005 3624 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 2005 American Association for Cancer Research. DNMT-1 Transcriptional Regulation by pRb and E2F1 to the development of atypical hyperplasia and cancer in the probed overnight at 4jC with primary antibodies diluted in 2.5% milk/TBST. presence of testosterone and estrogen (19). Therefore, these studies Following washes in 2.5% milk/TBST, membranes were incubated for 1 hour support a critical role for pRb in the prevention of tumor at room temperature with appropriate horseradish peroxidase–conjugated development. secondary antibodies. Following a second set of 2.5% milk/TBST washes, The majority of activities attributed to pRb are believed to membranes were developed with enhanced chemiluminescence detection reagents (Amersham Pharmacia Biotech). Primary antibodies were obtained result from direct interactions between pRb and either transcrip- as follows: actin (Santa Cruz, Santa Cruz, CA, sc-1615), cyclin E (Santa Cruz, tion factors or HDACs (20–23). However, pRb may also function sc-481), Dnmt-1 (Santa Cruz, sc-10221), p16INK4a (Santa Cruz, sc-1207), p19Arf through interactions with components of the DNA methylation (Abcam, Cambridge, MA, Ab80), proliferating cell nuclear antigen (PCNA; machinery, as pRb and Dnmt-1 interacted directly in glutathione Santa Cruz, sc-9857), and pRb (PharMingen, San Diego, CA, 554136). S-transferase pull-down and coimmunoprecipitation experiments RNA extraction and reverse transcription-PCR. Total cellular RNA (24, 25). Furthermore, pRb overexpression greatly diminished the was extracted from f5 Â 106 wtPrE or RbÀ/ÀPrE with RNeasy Mini kit ability of Dnmt-1 to bind DNA and thereby induced genomic (Qiagen, Valencia, CA) according to the manufacturer’s directions. The hypomethylation (25). resulting total RNA (2 Ag) was reverse transcribed using the ThermoScript Based on the well-characterized role of pRb as a transcriptional reverse transcription-PCR (RT-PCR) system (Invitrogen, Carlsbad, CA). repressor of E2F target genes and the observation that DNA Briefly, reverse transcription was done with ThermoScript reverse transcriptase at 55jC for 1 hour using an oligo(dT) primer (50 Amol/L). methyltransferase 1 (DNMT-1) is overexpressed in tumors that 20 PCR was carried out using Taq DNA polymerase (Invitrogen), 0.25 Amol/L frequently possess disrupted pRb pathways, we hypothesized that gene-specific primers, and 2 AL cDNA. Variable numbers of PCR cycles were the DNMT-1 gene might be under the control of pRb and E2F done by denaturation at 94jC (30 seconds), annealing at 59jC (45 seconds), +/+ À/À transcription factors. In this study, we used Rb and Rb and elongation at 72jC (1 minute). PCR products were resolved on 1.2% prostate epithelial cell lines rescued by tissue recombination (26) to agarose gels containing 0.5 ng/mL ethidium bromide (Fisher Biotech, study the effect of pRb pathway inactivation on DNA methyl- Pittsburgh, PA) in 0.5Â Tris-borate EDTA buffer (BioWhittaker). The gene- transferase levels. The mouse and human Dnmt-1 promoters were specific primers used are available on request. found to contain E2F binding sites that were required for Cloning of the Dnmt-1 promoter region. The Dnmt-1 promoter region regulation by pRb and E2F1. In the absence of pRb, the RbÀ/À was amplified from high molecular weight DNA, isolated from wtPrE or cells exhibited increased Dnmt-1 levels, which correlated with the LNCaP cells, using primers designed from publicly available sequences (National Center for Biotechnology Information Entrez accession no. inactivation of a tumor suppressor gene by DNA hypermethylation. AB056445): mDnmt-1 promoter forward 5V-CTCGAGGAAGAGTCCA- AGATGGTGTCCT-3V and reverse 5V-AAGCTTGCAGGTTGCAGACGACAGA- 3V and hDNMT-1 forward 5V-CTCGAGCTTCTCGCTGCTTTATCCCC-3V and Materials and Methods reverse 5V-AAGCTTCTCGGAGGCTTCAGCAGAC-3V. Italicized sequences rep- Cell culture and cell treatments. wtPrE and RbÀ/ÀPrE cell lines were resent nucleotides added to the complementary sequences to generate generated and propagated as described previously (26). Briefly, prostate unique restriction digest sites. The resulting PCR products (1,220 bp tissues obtained through tissue recombination (19) were minced in RPMI for mDnmt-1 and 340 bp for hDNMT-1) were gel purified, cloned into 1640 (BioWhittaker, Walkersville, MD) supplemented with 0.01 mg/mL the pCR2.1 TA cloning vector (Invitrogen), excised from pCR2.1, and bovine pituitary extract (Sigma, St. Louis, MO), 0.5 Ag/mL cholera toxin cloned in the pGL3-Basic luciferase reporter construct (Promega, Madison, (Sigma), 2 Amol/L dexamethasone (Sigma), 5 ng/mL insulin-like growth WI) to generate -1220mDnmt1-pGL3 and -340hDNMT1-pGL3. A unique NheI factor-I (Collaborative Research, Bedford, MA), 10 ng/mL epidermal growth restriction site within the mDnmt1 promoter region was used to generate - factor (Collaborative Research), penicillin/streptomycin (Life Technologies, 734mDnmt1-pGL3 from -1220mDnmt1-pGL3. -236mDnmt1-pGL3 was gen- Grand Island, NY), 2 mmol/L L-glutamine (Life Technologies), and 5 Ag/mL erated by PCR with the reverse primer described above for -1220mDnmt1 and insulin, 5 Ag/mL transferrin, and 5 ng/mL selenious acid in an ITS premix the following forward primer: 5V-CTCGAGGGAGGTGGGTGGCGCCAG-3V. (Collaborative Research). Cultures were later maintained in RPMI 1640 Site-directed mutagenesis was used to mutate the putative E2F binding supplemented with 5% fetal bovine serum (FBS; Hyclone, Logan, UT). sites within the context of the À734mDnmt1-pGL3 construct. QuikChange LNCaP and PC3 (American Type Culture Collection, Manassas, VA) prostate Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA) was used cancer cell lines were maintained under identical conditions. NIH3T3 according to the manufacturer’s protocol. Briefly, primers containing the mouse fibroblast cells were maintained in DMEM containing 10% FBS. desired mutations were generated as follows: E2FA mutant 5V- Inhibition of DNA methyltransferase activity was accomplished by CCAATTGGTTTCCGCGCATTGAAAAAAGCCGGGGTCTCG-3V, E2FB mu- growing cells in 1.0 Amol/L 5-aza-2V-deoxycytidine (5-Aza; Sigma) diluted in tant 5V-CCCCCTCCCAATTGGTTTTCAATGCGCGAAAAAGCCGG-3V,and culture medium for 5 days. Twenty-four hours after passaging cells, culture E2FAB mutant 5V-CCCCTCCCAATTGGTTTTCAATATTGAAAAAAGCC- medium was replaced with medium containing 1.0 Amol/L 5-Aza and was GGGGTCTCG-3V. Italicized bases are mutated from the wild-type replaced with fresh 5-Aza-containing medium 3 days later. Cells for serum- sequence. All plasmids were confirmed to contain the desired mutations free treatments were grown in RPMI 1640 without FBS beginning 24 to 48 by DNA sequencing at the University of Michigan DNA Sequencing Core hours after passaging and continuing for 2 to 5 days. (Ann Arbor, MI). Western blot analysis. wtPrE and RbÀ/ÀPrE cells were harvested by Luciferase reporter assays. Cells were plated into six-well tissue culture mechanical disruption with cell scrapers followed by gentle centrifugation. plates at a density of 1 Â 105 cells per well. Twenty-four hours later, cells Cell pellets were then lysed in appropriate volumes of lysis buffer [50 mmol/L were transiently transfected in triplicate with Tfx-50 (Promega) or FuGene6 Tris (pH 8.0), 120 mmol/L NaCl, 0.5% NP40, 1 mmol/L EGTA, 100 Ag/mL (Invitrogen) according to the manufacturer’s directions, with the designated phenylmethylsulfonyl fluoride, 50 Ag/mL aprotinin, 50 Ag/mL leupeptin, 1.0 combination of expression plasmids at 1.0 Ag (unless otherwise noted), 1.0 mmol/L sodium orthovanadate] for 1 hour on ice. Cellular debris was then Ag luciferase reporter plasmid, and 0.5 Ag pSV-h-galactosidase (pSV-h-Gal; pelleted by centrifugation and supernatants were collected and quantitated Promega) for transfection normalization. Cells were harvested 48 hours using a Bradford protein assay (Bio-Rad, Hercules, CA). Equal amounts of after transfection, lysed, and analyzed for luciferase and h-galactosidase protein were then separated on precast Tris-glycine SDS-polyacrylamide gels activity with luciferase substrate and Galacto-Light Plus h-galactosidase (Novex, Carlsbad, CA) and transferred to Hybond nitrocellulose membrane chemiluminescent reporter assay (Tropix, Foster City, CA). All samples were (Amersham Pharmacia Biotech, Piscataway, NJ). Membranes were blocked measured on a Monolight 2010 luminometer (Analytical Luminescence for 1 hour at room temperature in TBST [10 mmol/L Tris (pH 8.0), 150-500 Laboratory, San Diego, CA). Relative luciferase activity was determined by mmol/L NaCl, 0.1% Tween 20] containing 10% nonfat dry milk before being calculating the ratio between luciferase activity and h-galactosidase activity. www.aacrjournals.org 3625 Cancer Res 2005; 65: (9). May 1, 2005

Northern blot analysis. Total RNA was isolated as described above and RNA (10-20 Ag) was electrophoresed on 0.85% agarose gels containing 4% formaldehyde and 1Â MOPS. RNA was then transferred to Duralon-UV nitrocellulose membranes (Stratagene) overnight via capillary action with 20Â SSC at room temperature. Following transfer, RNA was UV cross-linked to the membranes that were then incubated for 30 minutes in prehybridization solution [7% SDS, 0.25 mol/L Na2HPO4 (pH 7.2)]. Radiolabeled probes were generated from RT-PCR products for the specified transcripts of interest. Primer sequences used to generate probe templates are available on request. Probe template was then radiolabeled with [a-32P]dATP with Prime-It II Random Primer Labeling kit (Stratagene). Membranes were then incubated overnight at 65jC in hybridization buffer containing 2.0 Â 106 cpm/mL. The following morning, membranes were washed twice in 20 mmol/L Na2HPO4 (pH 7.2), 5% SDS followed by two washes in 20 mmol/L Na2HPO4 (pH 7.2), 1% SDS at 65jC. Membranes were then exposed to BioMax MS-1 autoradiography film (Kodak, Rochester, NY). Bisulfite DNA sequencing. Total genomic DNA was isolated from wtPrE and RbÀ/ÀPrE using the DNeasy Tissue kit (Qiagen) according to the manufacturer’s directions. Genomic DNA (2 Ag) was then modified with bisulfite with the EZ DNA Methylation kit (Zymo Research, Orange, CA) according to the manufacturer’s directions. Bisulfite-modified DNA (2 AL) was then used as template DNA in a PCR reaction with primers specific for the bisulfite-modified paternally expressed gene 3 (Peg3) CpG island as follows: forward 5V-TTTTGTAGAGGATTTTGATAAGGAGG-3V and reverse 5V-AAATACCTCTTTAAATCCCTATCACC-3V. The resulting PCR product was Figure 1. Dnmt-1 is overexpressed in Rb À/À prostate epithelial cells. A, RNA then electrophoresed on a 1.2% agarose gel and the Peg3 band was excised, was isolated from wtPrE and RbÀ/ÀPrE cells and used for semiquantitative purified, and cloned into pCR2.1. Ligation product was transformed into RT-PCR analysis of endogenous Dnmt-1, Dnmt-3a, and Dnmt-3b mRNA levels. RT-PCR samples were analyzed after 25, 28, 31, and 34 cycles of amplification XL-1 Blue Escherichia coli and at least 10 individual colonies were selected to ensure that samples were within the linear range of amplification when for analysis from Luria-Bertani ampicillin agar plates. Individual clones examined. B, total RNA isolated from wtPrE and RbÀ/ÀPrE cells was analyzed by were then analyzed by DNA sequencing through the University of Michigan Northern blot with a 32P-radiolabeled probe from murine Dnmt-1 cDNA. 28S, DNA Sequencing Core facility. ethidium bromide staining of 28S rRNA before transfer. C, whole cell extracts were prepared from wtPrE and RbÀ/ÀPrE and analyzed by SDS-PAGE and Western blotting with antibodies specific to Dnmt-1 or actin. Results Dnmt-1 is overexpressed in RbÀ/À prostate epithelial cells. Characterization of the Dnmt-1 promoter region. To study Two nearly ubiquitous features of cancer cells are aberrant DNA the transcriptional regulation of Dnmt-1, 734 bp upstream of the methylation patterns and inactivation of the p16/pRb/E2F murine Dnmt-1 gene was cloned into the pGL3-luciferase reporter pathway. To investigate a potential causal link between pRb construct for analysis of promoter activity. Due to some debate inactivation and altered DNA methylation and DNA methyltrans- regarding the transcriptional start site for Dnmt-1, we have ferase regulation, we analyzed the expression of the three DNA denoted the ATG translation start site as +1 to avoid any methyltransferases with known methyltransferase activity in wild- confusion. This region of the promoter exhibited considerable type (Rb+/+) and RbÀ/À cell lines. Wild-type (wtPrE) and RbÀ/À promoter activity comparable with the potent cytomegalovirus prostate epithelial (RbÀ/ÀPrE) cell lines were generated from promoter, which served as a positive control (Fig. 2A). The DNA prostatic tissues rescued from embryonic control (Rb+/+) and sequence responsible for this promoter activity was further RbÀ/À mutant mice by tissue recombination (26). The genotypes narrowed down by analyzing a 236-bp fragment of the Dnmt-1 f of these cell lines have been confirmed previously by genomic gene encompassing 100 bp of DNA sequence both upstream and PCR and Western blot (26). Semiquantitative RT-PCR analysis of downstream of the transcription initiation site. This minimal Dnmt-1, Dnmt-3a, and Dnmt-3b transcripts revealed that whereas promoter retained 60% of the activity of the 734-bp promoter amplification of the Dnmt-3a and Dnmt-3b transcripts was fragment (Fig. 2A), suggesting that significant promoter activity is identical between Rb+/+ and RbÀ/À genotypes under log-phase contributed by this minimal Dnmt-1 promoter. Additionally, these growth conditions the Dnmt-1 transcript exhibited elevated levels Dnmt-1 reporter constructs showed a 2-fold increase in basal À/À in the RbÀ/ÀPrE cell line (Fig. 1A). Furthermore, Northern blot promoter activity in Rb PrE cells compared with wild-type cells analysis of equivalent quantities of total cellular RNA from wtPrE (data not shown), suggesting that these constructs accurately and RbÀ/ÀPrE cells with a 32P-labeled Dnmt-1 cDNA probe represent the activity of the endogenous Dnmt-1 promoter. confirmed a consistent 3- to 4-fold up-regulation of Dnmt-1 The pRb negatively regulates a variety of transcription factors transcript levels in the RbÀ/ÀPrE cells (Fig. 1B). Finally, through the recruitment of transcriptional repressor molecules considering that Dnmt-1 protein levels dictate DNA methyl- (e.g., HDACs). To determine if loss of any of the known pRb transferase activity, Western blot analysis was used to determine transcription factor interactions might be responsible for the if the protein was also up-regulated in RbÀ/ÀPrE. Antibodies elevated levels of Dnmt-1 in RbÀ/ÀPrE, the genomic DNA sequence raised against Dnmt-1 also showed increased expression of the flanking the transcription and translation initiation sites for the 195-kDa Dnmt-1 protein in the RbÀ/À cells compared with the mouse and human Dnmt-1 gene was queried for putative Rb+/+ cells (Fig. 1C). Therefore, in the absence of the pRb, Dnmt-1 transcription factor binding sites with MatInspector version 2.2 mRNA and protein expression was elevated 2-fold or greater due to (27). This DNA sequence analysis identified two conserved altered regulation at the transcriptional level. sequences within the murine and human Dnmt-1 promoters,

Cancer Res 2005; 65: (9). May 1, 2005 3626 www.aacrjournals.org

Figure 2. Identification and characterization of putative E2F binding sites within the murine Dnmt-1 promoter. A, murine Dnmt-1 promoter sequences (734 or 236 bp upstream of the ATG) cloned into the pGL3-luciferase reporter construct were transiently cotransfected into wtPrE cells along with pSV-h-Gal. Activities of a promoter-less pGL3 (-) and cytomegalovirus (CMV) promoter-driven pGL3 (+) are also depicted. Columns, mean; bars, 1 SD. B, transcription factor binding site analysis with MatInspector identified three putative E2F binding sites (underlined) in close proximity to the murine Dnmt-1 ATG (box). Alignment of the corresponding human DNMT-1 promoter sequences is also demonstrated. C, site-directed mutagenesis was used to mutate the conserved putative E2F binding sites A, B,orA and B together within the -734mDnmt1-pGL3 construct. Wild-type and mutant plasmids were transiently transfected into murine wtPrE cells with pSV-h-Gal. Columns, mean; bars, 1 SD. which exhibited considerable similarity to the consensus E2F sites were also identified in 5V untranslated regions of the Dnmt-1 binding site [TTT(C/G)GCGC; Fig. 2B]. The two putative murine gene from chicken and Xenopus (data not shown). Dnmt-1 E2F binding sites were compressed into a single To determine if these putative E2F binding sites contributed to overlapping site in human DNMT-1 similar to E2F binding sites the regulation of Dnmt-1 promoter activity, we used site-directed identified in several other cell cycle–regulated genes (e.g., N-myc, mutagenesis to mutate the two conserved E2F sites in the mouse E2F-1, DHFR, PCNA, UDG, and E1a; ref. 28). Putative E2F binding promoter. In contrast to mutation of the E2FA site, which had

Figure 3. Murine and human Dnmt-1 promoters are regulated by the pRb/E2F pathway. A, wtPrE were cotransfected with wild-type, E2FmutA, E2FmutB, or E2FmutAB versions of the -734mDnmt1- pGL3 construct in the presence of 0, 0.1, or 1.0 Ag exogenous E2F1. B, schematic of the -340hDNMT1-pGL3 and -340D61hDNMT1-pGL3 reporter constructs used in (C) and (D). Circles, putative E2F binding sites; arrows, transcription initiation point. Base pair numbering is from the translation start site. C, -340hDNMT1-pGL3 (black columns)or -340D61hDNMT1-pGL3 (white columns) were transiently transfected into human 293T cells in the presence of 1.0 Ag pcDNA3, 1.0 Ag E2F1, or 1.0 Ag PSM-Rb. D, wtPrE, NIH3T3, and human LNCaP cells were transiently cotransfected with mouse or human DNMT1-pGL3 reporter constructs and either 1.0 Ag pcDNA3, 1.0 Ag DN-E2F1, or 1.0 Ag E1a. DN-E2F1 is a fusion protein of the E2F1 DNA binding and dimerization domains to the pRb A/B pocket domain. Columns, mean; bars, 1 SD.

www.aacrjournals.org 3627 Cancer Res 2005; 65: (9). May 1, 2005

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 2005 American Association for Cancer Research. Cancer Research minimal effect on basal Dnmt-1 promoter activity, mutation of the E2FB site led to a 45% reduction in activity (Fig. 2C). When both E2FA and E2FB sites were mutated simultaneously, a 65% loss of activity was observed, suggesting that the E2FB site, perhaps along with the E2FA site, encodes sequences normally bound by a positive transcriptional activator. The Dnmt-1 promoter is regulated by the pRb/E2F pathway. Considering the role of pRb in repression of E2F transactivation and the putative E2F binding sites within the Dnmt-1 promoter of several species, we investigated the possibility that aberrant E2F transcriptional activation was affecting Dnmt-1 levels. Cells lacking Figure 4. Cells lacking pRb fail to maintain proper cell cycle–specific regulation pRb have been shown previously to exhibit aberrant E2F activity, as of Dnmt-1. Wild-type and Rb À/À prostate epithelial cells were signaled to growth À/À arrest by serum withdrawal for 48 hours (time –48 to 0). Cells were then both Rb fibroblasts and epithelial cells exhibit elevated levels of stimulated at time 0 to reenter the cell cycle following replacement of 5% FBS. a subset of E2F target genes, including cyclin E, PCNA, and p107 Total RNA was prepared from cells at the indicated time points and RNA (10 Ag) (26, 29). To explore the role of E2Fs in the regulation of Dnmt-1 from each sample was analyzed by Northern blot with 32P-labeled probes specific for Dnmt-1. promoter activity, exogenous E2F1 was coexpressed in wild-type prostate epithelial cells with the -734mDnmt1-pGL3 reporter possesses the DNA binding domain of E2F1 fused to the pocket construct. E2F1 was used, as it is one of the three activating domain of pRb and is capable of repressing transcription of E2F E2Fs (E2F1-3) and is the most well characterized of the E2F family target genes (32). DN-E2F was capable of repressing the mouse members. Increasing levels of exogenous E2F1 led to increased and human Dnmt-1 promoter reporter constructs (Fig. 3D)as levels of promoter activity, suggesting that E2F1 is capable of well as the well-characterized E2F-responsive DHFR-Luc reporter activating the Dnmt-1 promoter (Fig. 3A). The ability of E2F1 to construct (data not shown) in all cell lines analyzed. Additionally, activate the Dnmt-1 promoter was greatly reduced in the presence the adenovirus E1a oncogene, which binds to and inactivates of a mutated E2FA, E2FB, or E2FAB site (Fig. 3A). These data pocket proteins, was used to generate an increase in free E2F suggest that E2F1 activation of the murine Dnmt-1 gene is partly activity. Adenovirus E1a potently activated transcription of the dependent on the integrity of the two putative E2F binding mouse Dnmt-1, human DNMT-1, and DHFR promoters in all cell sequences. Additionally, similar to the promoter of the well- lines analyzed (Fig. 3D; data not shown). Additionally, Dnmt-1 characterized E2F target Dhfr, the Dnmt-1 promoter failed to promoter activation by E1a was dependent on its pocket protein respond to E2F1 DNA binding (E2F1E132) and transactivation binding activity, as the E1a CXdl (33) mutant, which does not (E2F11-284) mutants (30). interact with pocket proteins, was unable to activate the As mentioned previously, the human DNMT-1 gene exhibited promoter >2- to 3-fold.3 Taken together, these data indicate that putative E2F binding sites that aligned with those found in the the E2F DNA binding domain localizes to both murine and murine Dnmt-1 gene. To determine if the human DNMT-1 human DNMT-1 promoter sequences where it activates transcrip- promoter was also E2F regulated, a region of 340 bp upstream of tion in a pocket protein-regulated manner. the ATG translation start site was cloned into the pGL3-luciferase Dnmt-1 is regulated throughout the cell cycle by the pRb/ reporter construct (Fig. 3B) and transfected into LNCaP human E2F pathway. The DNMT-1 transcript has been shown previously prostate adenocarcinoma cells and 293T human embryonic kidney to be a growth-regulated transcript (9, 34, 35). Based on these cells. In these cell lines, exogenous E2F1 reproducibly activated the observations and our data demonstrating regulation of Dnmt-1 human DNMT-1 promoter 6-fold, whereas a constitutively active promoter activity by pRb and E2F1, we predicted that normal cell unphosphorylatable large pocket of pRb (amino acids 379-928; cycle regulation of Dnmt-1 may be altered in cells lacking pRb. PSM-Rb; ref. 31) repressed DNMT-1 promoter activity (Fig. 3C; data When wtPrE and RbÀ/ÀPrE cell lines were deprived of serum for not shown). To examine if the conserved E2F binding site was 48 hours, both cultures growth arrested similarly with increased required for pRb/E2F-mediated regulation, the 61 bp of 3Vpromoter cell populations found in the G1 phase of the cell cycle and sequence containing this site was deleted from the -340 bp DNMT-1 decreased cell numbers in S phase as shown by flow cytometry promoter to generate -340D61 DNMT-1 (Fig. 3B). When E2F1 was (data not shown). Following 48 hours of serum deprivation, Rb+/+ cotransfected with this -340D61 DNMT-1 promoter, a 67% decrease prostate cells exhibited decreased expression of Dnmt-1 transcripts in activation was observed (Fig. 3C). Similarly, in the absence of the as is typical for other E2F targets (Fig. 4). However, the same E2F binding site, the constitutively active pRb had no effect on period of serum deprivation did not lead to any detectable DNMT-1 promoter activity. Additionally, the -340D61 DNMT-1 decrease in Dnmt-1 transcript in the RbÀ/À cell line (Fig. 4). These construct exhibited a 58% reduction in basal promoter activity data suggest that pRb is required for proper cell cycle regulation of compared with the -340 DNMT-1 construct (data not shown). Dnmt-1 transcription, which may be crucial for maintenance of These data suggest that the human DNMT-1 gene, like the murine normal methylation patterns. Dnmt-1 homologue, is both positively and negatively regulated by Stable overexpression of E2F1 leads to elevated Dnmt-1 the pRb/E2F pathway. levels. Based on aforementioned data, it seemed that Dnmt-1 To further support a role for the pRb/E2F pathway in Dnmt-1 levels were elevated in RbÀ/ÀPrE cells due to increased levels of promoter regulation, additional regulators of this pathway were free E2F1 and loss of pRb/E2F1 repressor complexes. To test this assessed for their ability to control Dnmt-1 promoter activity. hypothesis, we established wtPrE cell lines that stably overex- First, a dominant-negative (DN) E2F construct was cotransfected pressed E2F1 under control of the cytomegalovirus promoter. Two along with the wild-type Dnmt-1 reporter constructs in wtPrE cells, NIH3T3 murine fibroblasts, and LNCaP human prostate adenocarcinoma cells (Fig. 3D). This previously described DN-E2F 3 M.T. McCabe, J.A. Low, M.J. Imperiale, and M.L. Day, unpublished data.

Cancer Res 2005; 65: (9). May 1, 2005 3628 www.aacrjournals.org

hypermethylation across all 29 CpG dinucleotides in the RbÀ/À PrE cells compared with Rb+/+ cells (Fig. 6C). These data confirm that the Dnmt-1 overexpression in cells lacking pRb can lead to observable alterations in gene expression.

Discussion Inactivation of the pRb has been associated with increased tumor susceptibility in several model systems (17, 19). The precise mechanism underlying this sensitivity to tumorigenesis has not yet been fully elucidated. Traditional views of pRb function would suggest that inactivation of pRb leads to deregulation of E2F target genes, such as cyclin E, thymidine kinase, and DNA polymerase a, which govern control of the cell cycle. However, more recent Figure 5. Stable overexpression of E2F1 leads to accumulation of Dnmt-1 studies of E2F target gene expression in models of Rb inactivation mRNA and protein. wtPrE cells were transfected and selected for G418 antibiotic suggest that although a burst of E2F activity may follow acute loss resistance to stably express mouse E2F1 or pcDNA3 vector control. A, total RNA was isolated from log-phase wtPrE-pcDNA3 and wtPrE-mE2F1 cells and RNA of the Rb gene (39) compensatory mechanisms involving the (10 Ag) from each sample was analyzed by Northern blot with 32P-labeled probes remaining pocket protein family members, p107 and p130, are specific for Dnmt-1, E2F1, or actin. rRNA, ethidium bromide–stained rRNA sufficient to keep the vast majority of E2F target genes in check before transfer. B, protein lysates generated from log-phase wtPrE-pcDNA3 and wtPrE-mE2F1 cells were analyzed by Western blot with antibodies raised against when pRb is inactivated (29, 40–42). Dnmt-1, cyclin E, PCNA, or actin. In this study, we have identified putative E2F binding sites within the promoter region of the mouse and human Dnmt-1 genes, which E2F1-overexpressing cell lines were generated through selection for are sufficient to allow for pRb- and E2F-mediated control of G418 resistance. Northern blot analysis confirmed increased transcription. Rb gene inactivation is frequently observed in tumors expression of the E2F1 transcript in transfected cells (Fig. 5A). of many tissue types and is typically thought to play a causative role Additionally, although we have not been able to identify an in tumorigenesis. To analyze Dnmt-1 expression in a model of Rb antibody capable of detecting the mouse form of E2F1 protein, we gene inactivation, we analyzed the DNA methyltransferase levels in were able to show increased levels of two E2F targets (cyclin E and Rb+/+ and RbÀ/À prostate epithelial cells. These cells were generated PCNA) at the protein level (Fig. 5B). Investigation of Dnmt-1 levels from wild-type and RbÀ/À prostate tissues, which have been shown in these E2F1-overexpressing cells revealed overexpression at both previously to be uniquely sensitive to hormone-induced tumorigen- mRNA transcript and protein levels (Fig. 5A and B) compared with esis specifically in the absence of Rb (19, 26). Whereas no alterations cells transfected with the pcDNA3 control vector. This result were observed for Dnmt-3a or Dnmt-3b, we detected aberrant basal strongly suggested that the Dnmt-1 overexpression observed in the as well as cell cycle–specific regulation of Dnmt-1 in RbÀ/À epithelial RbÀ/ÀPrE cell lines was due transcriptional activation resulting cells compared with wild-type controls. Additionally, this aberrant from elevated levels of active E2F1. regulation of Dnmt-1 in RbÀ/À prostate epithelium was correlated Transcriptional repression of selective genes in RbÀ/ÀPrE with increased DNA hypermethylation-mediated transcriptional cells due to promoter hypermethylation. Elevated levels of DNA silencing of Peg3, a gene involved in cell survival regulation. methyltransferases have been proposed as an important mecha- Overexpression of DNA methyltransferases has been reported in nism involved in tumor initiation and progression (36, 37). The many tumor types, including breast, colon, lung, and prostate mechanism underlying the DNA methyltransferase–associated cancers (36, 43–45). Suggestive of a functional role for DNA tumor susceptibility is likely due to transcriptional repression of methyltransferase overexpression in cancer, increased expression tumor suppressor genes through promoter hypermethylation. To of DNMTs in tumors has been significantly correlated with increased determine if the elevated Dnmt-1 levels in RbÀ/ÀPrE cells had any methylation of CpG islands found within the promoters of tumor effect on gene expression, we analyzed microarray data compar- suppressor genes (36, 37). Even more convincingly, several studies ing wtPrE cells with RbÀ/ÀPrE cells for genes that were strongly have shown that DNA methyltransferase levels and activity were down-regulated in RbÀ/À cells. One candidate gene was Peg3, progressively elevated in samples representing colon cancer which is an imprinted gene whose protein product is required for progression with successively increasing activity levels from normal translocation of the proapoptotic Bax protein from the cytoplasm tissue from a normal patient to normal tissue from a cancer patient to the mitochondria during apoptotic stimuli (38). Peg3 was to polyps and finally to carcinoma (44, 46). down-regulated an average of 22.2-fold across multiple arrays In addition to the descriptive and correlative studies on tumor (data not shown). The strong repression of Peg3 was confirmed samples, several genetic animal models have been used to implicate by semiquantitative RT-PCR in multiple independently derived Dnmt-1 activity in tumor development. Although homozygous RbÀ/ÀPrE cell lines (Fig. 6A). Reexpression of the Peg3 transcript knockout of the Dnmt-1 gene results in embryonic lethality, following treatment of RbÀ/ÀPrE cells with 1.0 Amol/L 5-Aza, a several groups have generated hypomorphic Dnmt-1 alleles that DNA methyltransferase inhibitor, for 5 days suggested that the typically express f50% of normal DNA methyltransferase transcriptional repression of Peg3 was due to DNA methylation activity in heterozygotes (47, 48). Despite this decrease in (Fig. 6B). However, to directly show methylation of the Peg3 gene, DNA methyltransferase levels and activity, these animals remain we used bisulfite modification and DNA sequencing to analyze viable and seem to maintain near normal levels of DNA methylation the methylation status of a 444-bp CpG island encoded within the (48). Interestingly, when these animals with decreased Dnmt-1 promoter, first exon and first intron of this gene (Fig. 6C). levels were cross-bred with animal models of cancer, in- Analysis of 10 to 14 individual clones revealed significant cluding the mismatch repair–deficient MLH1À/À and colon www.aacrjournals.org 3629 Cancer Res 2005; 65: (9). May 1, 2005

repression compared with our RbÀ/À cell lines.4 For example, both cell lines exhibited down-regulation of Akap12 (Ssecks 322), glutathione S-transferase M2, growth arrest specific 5, embigin, protein S, membrane metalloendopeptidase, tissue plasminogen activator, etc. Considering the role of several of these genes in cell survival, signal transduction, and growth control, it is possible that through the combined effect of these transcriptional alterations, RbÀ/À cells and Dnmt-1-overexpressing cells may be more susceptible to transformation and tumor induction. In this study, we focused on the repression of Peg3 by hypermethylation. Peg3 is an imprinted gene with functions that are still not completely understood. However, inactivation of Peg3 by hypermethylation likely confers a survival advantage, as Peg3 regulates the translocation of the proapoptotic Bax from the cytoplasm to the mitochondria (38). Peg3 hypermethylation has been reported in gliomas and reexpression of a Peg3 cDNA in glioma cell lines resulted in a loss of tumorigenicity in nude mice, suggesting that this gene product functions as a tumor suppressor (55). Taken together, this epigenetic alteration would seem to provide RbÀ/À cells with cell survival advantages compared with wild-type cells. Further investigation is necessary to determine if Peg3 transcriptional silencing is strictly required for tumor susceptibility of RbÀ/À epithelium. In support of an E2F-mediated mechanism of Dnmt-1 gene regulation, levels of Dnmt-1 increase during the transition from late À/À Figure 6. Dnmt-1 overexpression in Rb PrE cells is correlated with DNA G1 phase to early S phase in apparent preparation for the S-phase hypermethylation-induced transcriptional silencing of Peg3. A, semiquantitative synthesis of DNA, which is initially unmethylated (56, 57). RT-PCR was done on total RNA lysates from wtPrE and three RbÀ/ÀPrE cell lines with primers specific for Peg3 and HPRT. B, wtPrE and RbÀ/ÀPrE cells Additionally, Friend murine erythroleukemia cells were shown to were treated with 1.0 Amol/L 5-Aza for 5 days. Total RNA was isolated and express three species of DNA methyltransferases, one of which semiquantitative RT-PCR was done as above. C, bisulfite modification and DNA f sequencing was done on genomic DNA isolated from wtPrE and RbÀ/ÀPrE. migrated at 190 kDa similar to Dnmt-1 and increased during log- Following amplification of the Peg3 CpG island by PCR, z10 individual alleles phase growth when many cells were involved in DNA synthesis (34). were cloned and sequenced. Y axis, percentage of total clones sequenced per These observations suggested that E2F transcription factors might cell line showing methylation at each CpG dinucleotide. regulate Dnmt-1 levels considering that E2Fs are known to regulate cancer–prone ApcMin/+ mice, significant decreases in tumor many genes involved in DNA synthesis and S-phase progression (c- formation were observed compared with the single MLH1À/À or myc, E2F-1, p107, cyclin E, cyclin A, cdc2, Dhfr, PCNA, thymidine ApcMin/+ mutants (47, 49, 50). Additionally, Belinsky et al. showed a kinase, etc.). More recently, a study examining cell cycle–specific reduction by almost 50% in carcinogen-induced pulmonary gene expression identified Dnmt-1 as part of a G1-S cycle cluster, hyperplastic and neoplastic lesions by inactivating one Dnmt-1 which included several other E2F target genes, including cdk2, cyclin allele (51). Further repression of lesion formation by 89% was E, MCM3/5/6, PCNA, replication factor C, DNA polymerase I, and obtained when single allele Dnmt-1 inactivation was combined p107 (58). Thus, there is considerable evidence to support the cell with 5-Aza treatment to inhibit remaining DNA methyltransferase cycle–specific regulation of Dnmt-1. activity (51). These animal model findings strongly suggest that Microarray studies have recently become useful for identifying DNA methyltransferase activity plays a causative role in the nearly comprehensive lists of target genes of various transcription formation of these tumors. factors. Recent reports using this technology to identify lists of In vitro studies have also shown a role for Dnmt-1 overexpression pRb (59) or E2F (60) responsive genes have identified Dnmt-1 as being regulated by these proteins. Specifically, an inducible and in the transformation process. As little as a 2-fold overexpression of f exogenous DNMT-1 has been shown to lead to increased DNA constitutively active large pocket of pRb repressed Dnmt-1 2- to methylation within the promoter regions of several tumor 3-fold in a Rat16 cell line, whereas E2F1 induced Dnmt-1 mRNA suppressor genes (i.e., HIC-1, ER, and E-cadherin) and tumorigenic 2.2- to 2.4-fold in Rat-1a cells (59, 60). Dnmt-1 gene regulation transformation of cells as evidenced by altered cellular morphology, seems to be not only under the control of pRb and E2F1 but also increased saturation density, and greatly increased growth in soft may be regulated by all of the activating E2Fs (E2F1-E2F3) and by agar and tumorigenicity (11, 12, 52, 53). Similarly, treatment of the other pocket proteins (p107 and p130). Microarray studies transformed cells with the DNA methyltransferase inhibitor 5-Aza or examining gene expression alterations in response to p130 and with Dnmt-1 antisense oligonucleotides resulted in reexpression of E2F3 found Dnmt-1 mRNA to be down-regulated 3.5- to 4.9-fold previously methylated genes and reversion of the transformed following p130 expression and activated 3.2- to 4.3-fold by E2F3 phenotype (54). Interestingly, microarray analysis of a rat fibroblast (60, 61). Although these studies were not specifically designed to cell line transformed by f3-fold overexpression of the oocyte- identify direct targets of pocket proteins and E2Fs, these data specific Dnmt-1o (53) showed a similar pattern of transcriptional support our findings that Dnmt-1 is a direct transcriptional target of the pocket protein and E2F pathway. We have shown that the Dnmt-1 gene is transcriptionally regulated by 4 M.J. McCabe and M.L. Day, unpublished data. E2F transcription factors and that the normal expression of Dnmt-1

Cancer Res 2005; 65: (9). May 1, 2005 3630 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 2005 American Association for Cancer Research. DNMT-1 Transcriptional Regulation by pRb and E2F1 is kept in check by the pocket proteins. In the absence of pRb, cells Acknowledgments exhibit increased transcription of the Dnmt-1 gene leading to Received 6/18/2004; revised 12/20/2004; accepted 2/4/2005. accumulation of Dnmt-1 protein. These increased levels of Dnmt-1 Grant support: NIH grant R01 DK-61488 (M.L. Day), NIH Cancer Biology Training were correlated with aberrant DNA hypermethylation-mediated Fellowship grant 5T32 CA09676 (M.T. McCabe), and American Cancer Society-Clyde Dixon Trust (PF-03-252-01-TBE). transcriptional silencing of a tumor suppressor gene, which may The costs of publication of this article were defrayed in part by the payment of page À/À predispose Rb cells to tumorigenesis. Perhaps this ability of charges. This article must therefore be hereby marked advertisement in accordance Dnmt-1 to be regulated by the pRb/E2F pathway explains the with 18 U.S.C. Section 1734 solely to indicate this fact. We thank Peggy Farnham, Douglas Cress, Eric Knudsen, and Jean Wang for apparent lack of mutation or amplification of the Dnmt-1 gene in providing plasmids and reagents and Michael Imperiale and Colleen Doyle for critical cancers, as pRb inactivation is nearly ubiquitous in cancer. review of this article.

References hormone-induced carcinogenesis in Rb-deficient pros- 35. Kimura H, Nakamura T, Ogawa T, Tanaka S, Shiota K. tate tissue. Cancer Res 2000;60:6008–17. Transcription of mouse DNA methyltransferase 1 1. Eden S, Cedar H. Role of DNA methylation in the 20. Nevins JR, Chellappan SP, Mudryj M, et al. E2F (Dnmt1) is regulated by both E2F-Rb-HDAC-dependent regulation of transcription. Curr Opin Genet Dev transcription factor is a target for the RB protein and and -independent pathways. Nucleic Acids Res 2003; 1994;4:255–9. the cyclin A protein. Cold Spring Harb Symp Quant Biol 31:3101–13. 2. Bird AP, Wolffe AP. Methylation-induced repression— 1991;56:157–62. 36. Belinsky SA, Nikula KJ, Baylin SB, Issa JP. Increased belts, braces, and chromatin. Cell 1999;99:451–4. 21. Brehm A, Miska EA, McCance DJ, Reid JL, Bannister cytosine DNA-methyltransferase activity is target-cell- 3. Bestor T, Laudano A, Mattaliano R, Ingram V. Cloning AJ, Kouzarides T. Retinoblastoma protein recruits specific and an early event in lung cancer. Proc Natl and sequencing of a cDNA encoding DNA methyltrans- histone deacetylase to repress transcription. Nature Acad Sci U S A 1996;93:4045–50. ferase of mouse cells. The carboxyl-terminal domain of 1998;391:597–601. 37. Etoh T, Kanai Y, Ushijima S, et al. Increased DNA the mammalian enzymes is related to bacterial 22. Luo RX, Postigo AA, Dean DC. Rb interacts with methyltransferase 1 (DNMT1) protein expression restriction methyltransferases. J Mol Biol 1988;203: histone deacetylase to repress transcription. Cell 1998; correlates significantly with poorer tumor differentia- 971–83. 92:463–73. tion and frequent DNA hypermethylation of multiple 4. Okano M, Xie S, Li E. Cloning and characterization of a 23. Magnaghi-Jaulin L, Groisman R, Naguibneva I, et al. CpG islands in gastric cancers. Am J Pathol 2004;164: family of novel mammalian DNA (cytosine-5) methyl- Retinoblastoma protein represses transcription by 689–99. transferases. Nat Genet 1998;19:219–20. recruiting a histone deacetylase. Nature 1998;391: 38. Deng Y, Wu X. Peg3/Pw1 promotes p53-mediated 5. Bestor TH. The DNA methyltransferases of mammals. 601–5. apoptosis by inducing Bax translocation from cytosol Hum Mol Genet 2000;9:2395–402. 24. Robertson KD, Ait-Si-Ali S, Yokochi T, Wade PA, Jones to mitochondria. Proc Natl Acad Sci U S A 2000;97: 6. Pradhan S, Bacolla A, Wells RD, Roberts RJ. Recom- PL, Wolffe AP. DNMT1 forms a complex with Rb, E2F1 12050–5. binant human DNA (cytosine-5) methyltransferase. I. and HDAC1 and represses transcription from E2F- 39. Sage J, Miller AL, Perez-Mancera PA, Wysocki JM, Expression, purification, and comparison of de novo responsive promoters. Nat Genet 2000;25:338–42. Jacks T. Acute mutation of retinoblastoma gene and maintenance methylation. J Biol Chem 1999;274: 25. Pradhan S, Kim GD. The retinoblastoma gene function is sufficient for cell cycle re-entry. Nature 33002–10. product interacts with maintenance human DNA 2003;424:223–8. 7. Herman JG. Hypermethylation of tumor suppressor (cytosine-5) methyltransferase and modulates its acti- 40. Dannenberg JH, van Rossum A, Schuijff L, te Riele H. genes in cancer. Semin Cancer Biol 1999;9:359–67. vity. EMBO J 2002;21:779–88. Ablation of the retinoblastoma gene family deregulates 8. Baylin SB, Herman JG. DNA hypermethylation in 26. Day KC, McCabe MT, Zhao X, et al. Rescue of G(1) control causing immortalization and increased cell tumorigenesis: epigenetics joins genetics. Trends Genet embryonic epithelium reveals that homozygous dele- turnover under growth-restricting conditions. Genes 2000;16:168–74. tion of the retinoblastoma gene confers growth factor Dev 2000;14:3051–64. 9. Robertson KD, Uzvolgyi E, Liang G, et al. The human independence and immortality, but does not influence 41. Sage J, Mulligan GJ, Attardi LD, et al. Targeted DNA methyltransferases (DNMTs) 1, 3a and 3b: epithelial differentiation or tissue morphogenesis. J Biol disruption of the three Rb-related genes leads to loss of coordinate mRNA expression in normal tissues and Chem 2002;20:20. G(1) control and immortalization. Genes Dev 2000; overexpression in tumors. Nucleic Acids Res 1999;27: 27. Quandt K, Frech K, Karas H, Wingender E, Werner T. 14:3037–50. 2291–8. MatInd and MatInspector: new fast and versatile tools 42. McCabe MT, Azih OA, Day ML. pRb-independent 10. Biniszkiewicz D, Gribnau J, Ramsahoye B, et al. for detection of consensus matches in nucleotide growth arrest and transcriptional regulation of E2F Dnmt1 overexpression causes genomic hypermethyla- sequence data. Nucleic Acids Res 1995;23:4878–84. target genes. Neoplasia 2005;7:141–51. tion, loss of imprinting, and embryonic lethality. Mol 28. Kel AE, Kel-Margoulis OV, Farnham PJ, Bartley 43. Girault I, Tozlu S, Lidereau R, Bieche I. Expression Cell Biol 2002;22:2124–35. SM, Wingender E, Zhang MQ. Computer-assisted analysis of DNA methyltransferases 1, 3A, and 3B in 11. Wu J, Issa JP, Herman J, Bassett DE Jr, Nelkin BD, identification of cell cycle-related genes: new targets sporadic breast carcinomas. Clin Cancer Res 2003;9: Baylin SB. Expression of an exogenous eukaryotic DNA for E2F transcription factors. J Mol Biol 2001;309: 4415–22. methyltransferase gene induces transformation of NIH 99–120. 44. el-Deiry WS, Nelkin BD, Celano P, et al. High 3T3 cells. Proc Natl Acad Sci U S A 1993;90:8891–5. 29. Hurford RK Jr, Cobrinik D, Lee MH, Dyson N. pRB expression of the DNA methyltransferase gene char- 12. Vertino PM, Yen RW, Gao J, Baylin SB. De novo and p107/p130 are required for the regulated expression acterizes human neoplastic cells and progression methylation of CpG island sequences in human of different sets of E2F responsive genes. Genes Dev stages of colon cancer. Proc Natl Acad Sci U S A fibroblasts overexpressing DNA (cytosine-5)-methyl- 1997;11:1447–63. 1991;88:3470–4. transferase. Mol Cell Biol 1996;16:4555–65. 30. Ma Y, Yuan J, Huang M, Jove R, Cress WD. Regulation 45. PatraSK,PatraA,ZhaoH,DahiyaR.DNA 13. Livingston DM, DeCaprio JA, Ludlow JW. Does the of the cyclin D3 promoter by E2F1. J Biol Chem methyltransferase and demethylase in human prostate product of the RB-1 locus have a cell cycle regulatory 2003;278:16770–6. cancer. Mol Carcinog 2002;33:163–71. function? Princess Takamatsu Symp 1989;20:187–90. 31. Knudsen ES, Wang JY. Dual mechanisms for the 46. Issa JP, Vertino PM, Wu J, et al. Increased cytosine 14. Lipinski MM, Jacks T. The retinoblastoma gene inhibition of E2F binding to RB by cyclin-dependent DNA-methyltransferase activity during colon cancer family in differentiation and development. Oncogene kinase-mediated RB phosphorylation. Mol Cell Biol progression. J Natl Cancer Inst 1993;85:1235–40. 1999;18:7873–82. 1997;17:5771–83. 47. Laird PW, Jackson-Grusby L, Fazeli A, et al. 15. Day ML, Foster RG, Day KC, et al. Cell anchorage 32. Sellers WR, Rodgers JW, Kaelin WG Jr. A potent Suppression of intestinal neoplasia by DNA hypome- regulates apoptosis through the retinoblastoma tumor transrepression domain in the retinoblastoma protein thylation. Cell 1995;81:197–205. suppressor/E2F pathway. J Biol Chem 1997;272:8125–8. induces a cell cycle arrest when bound to E2F sites. 48. Li E, Bestor TH, Jaenisch R. Targeted mutation of the 16. Bringold F, Serrano M. Tumor suppressors and Proc Natl Acad Sci U S A 1995;92:11544–8. DNA methyltransferase gene results in embryonic oncogenes in cellular senescence. Exp Gerontol 2000; 33. Moran E, Zerler B, Harrison TM, Mathews MB. lethality. Cell 1992;69:915–26. 35:317–29. Identification of separate domains in the adenovirus 49. Trinh BN, Long TI, Nickel AE, Shibata D, Laird PW. 17. Jacks T, Fazeli A, Schmitt EM, Bronson RT, Goodell E1A gene for immortalization activity and the DNA methyltransferase deficiency modifies cancer MA, Weinberg RA. Effects of an Rb mutation in the activation of virus early genes. Mol Cell Biol 1986;6: susceptibility in mice lacking DNA mismatch repair. mouse. Nature 1992;359:295–300. 3470–80. Mol Cell Biol 2002;22:2906–17. 18. Harrison DJ, Hooper ML, Armstrong JF, Clarke AR. 34. Bestor TH, Ingram VM. Growth-dependent expres- 50. Eads CA, Nickel AE, Laird PW. Complete genetic Effects of heterozygosity for the Rb-1t19neo allele in the sion of multiple species of DNA methyltransferase in suppression of polyp formation and reduction of mouse. Oncogene 1995;10:1615–20. murine erythroleukemia cells. Proc Natl Acad Sci U S A CpG-island hypermethylation in Apc(Min/+) Dnmt1- 19. Wang Y, Hayward SW, Donjacour AA, et al. Sex 1985;82:2674–8. hypomorphic Mice. Cancer Res 2002;62:1296–9. www.aacrjournals.org 3631 Cancer Res 2005; 65: (9). May 1, 2005

51. Belinsky SA, Klinge DM, Stidley CA, et al. Inhibition 55. Kohda T, Asai A, Kuroiwa Y, et al. Tumour suppressor as revealed from DNA microarray analysis. Mol Cell Biol of DNA methylation and histone deacetylation prevents activity of human imprinted gene PEG3 in a glioma cell 2001;21:4684–99. murine lung cancer. Cancer Res 2003;63:7089–93. line. Genes Cells 2001;6:237–47. 59. Markey MP, Angus SP, Strobeck MW, et al. Unbiased 52. Slack A, Cervoni N, Pinard M, Szyf M. DNA 56. Papadopoulos T, Pfeifer GP, Hoppe F, Drahovsky D, analysis of RB-mediated transcriptional repression methyltransferase is a downstream effector of cellular Muller-Hermelink HK. Proliferation-associated expres- identifies novel targets and distinctions from E2F transformation triggered by simian virus 40 large T sion of DNA methyltransferase in human embryonic action. Cancer Res 2002;62:6587–97. antigen. J Biol Chem 1999;274:10105–12. lung cells. Virchows Arch B Cell Pathol Incl Mol Pathol 60. Polager S, Kalma Y, Berkovich E, Ginsberg D. E2Fs 53. Bakin AV, Curran T. Role of DNA 5-methylcytosine 1989;56:371–5. up-regulate expression of genes involved in DNA transferase in cell transformation by fos. Science 57. Vogel MC, Papadopoulos T, Muller-Hermelink HK, replication, DNA repair and mitosis. Oncogene 1999;283:387–90. Drahovsky D, Pfeifer GP. Intracellular distribution of 2002;21:437–46. 54. MacLeod AR, Szyf M. Expression of antisense DNA methyltransferase during the cell cycle. FEBS Lett 61. Russo G, Claudio PP, Fu Y, et al. pRB2/p130 to DNA methyltransferase mRNA induces DNA 1988;236:9–13. target genes in non-small lung cancer cells iden- demethylation and inhibits tumorigenesis. J Biol 58. Ishida S, Huang E, Zuzan H, et al. Role for E2F in tified by microarray analysis. Oncogene 2003;22: Chem 1995;270:8037–43. control of both DNA replication and mitotic functions 6959–69.

Cancer Res 2005; 65: (9). May 1, 2005 3632 www.aacrjournals.org

Michael T. McCabe, Joanne N. Davis and Mark L. Day

Cancer Res 2005;65:3624-3632.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/9/3624

Cited articles This article cites 59 articles, 27 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/9/3624.full#ref-list-1

Citing articles This article has been cited by 13 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/9/3624.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/65/9/3624. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.