DNA Methyltransferase Is a Downstream Effector of Cellular Transformation Triggered by Simian Virus 40 Large T Antigen*

This paper tests the hypothesis that DNA methyltransferase plays a causal role in cellular transformation induced by SV40 T antigen. We show that T antigen expression results in elevation of DNA methyltransferase (MeTase) mRNA, DNA MeTase protein levels, and global genomic DNA methylation. A T antigen mutant that has lost the ability to bind pRb does not induce DNA MeTase. This up-regulation of DNA MeTase by T antigen occurs mainly at the posttranscriptional level by altering mRNA stability. Inhibition of DNA MeTase by antisense oligonucleotide inhibitors results in inhibition of induction of cellular transformation by T antigen as determined by a transient transfection and soft agar assay. These results suggest that elevation of DNA MeTase is an essential component of the oncogenic program induced by T antigen.

In mammalian cells 60 -80% of CpG dinucleotides are methylated, forming a pattern that correlates with gene expression (1,2). The DNA 5-cytosine methyltransferase (DNA MeTase) 1 is the enzyme responsible for the establishment of this pattern (3). Several reports have demonstrated that cancer cells bear increased levels of DNA MeTase (4), that increased DNA MeTase is an early event in tumor progression in an animal model of lung carcinogenesis (5), and that regional hypermethylation is characteristic of tumor cells (6,7). However, the extent to which DNA MeTase activity is elevated in cancer cells in vivo is still debatable (8).
A basic question is whether an increase in DNA MeTase activity is a critical downstream component of oncogenic pathways (9) or whether it is an aberrant consequence of the transformed state? One molecular link between DNA MeTase and an oncogenic pathway is the observation that its expression in mouse cells is up-regulated at the transcriptional level by the Ras signaling pathway (9 -11). Several lines of evidence have suggested recently that elevated DNA MeTase might play a causal role in cancer (12). Inhibition of DNA MeTase in Y1 adrenal carcinoma cells by either expressing an antisense to DNA MeTase (13) or by antisense oligonucleotides (14) inhibits tumorigenesis ex vivo. Injection of DNA MeTase antisense oligonucleotides into mice harboring Y1 tumors inhibits tumor growth in vivo (14), and reduction of DNA MeTase level by 5-azaCdR suppresses neoplasia in Min mice (15).
If increased DNA MeTase is a necessary constituent of cellular transformation, it should be induced by diverse oncogenic pathways. SV40 T antigen is one of the most studied viral oncoproteins that can induce frequent tumors when expressed as a transgene in mice (16), can immortalize primary cell lines (17) or transform immortalized cells (18), but transforms primary cells only when expressed in conjunction with Ras or other components of its signaling pathway (19). Thus, T antigen induces a very effective transformation pathway that is complementary to, but different from, the one induced by Ras. Extensive studies have established that T antigen transformation is a consequence of its ability to physically interact with the tumor suppressors pRb (20) and p53 (21) as well as yet non characterized functions. A recent observation has shown that two human SV40-transformed lines express higher levels of DNA MeTase protein than their nontransformed counterparts (22). Is it possible that similar to the Ras signaling pathway, the T antigen-tumor suppressor pathway utilizes DNA MeTase as a downstream effector? This paper tests this hypothesis by determining whether expression of T antigen increases the levels of DNA MeTase mRNA and protein in the cell, defining the level of gene expression regulation at which T antigen acts, and testing whether DNA MeTase plays a causal role in T antigen triggered cellular transformation?

EXPERIMENTAL PROCEDURES
Cell Culture and DNA-mediated Gene Transfer-Balb/c 3T3 cells (ATCC) were maintained as monolayers in Dulbecco's modified Eagle's medium medium which was supplemented with 10% heat-inactivated fetal calf serum (Immunocorp, Montreal). All other media and reagents for cell culture were obtained from Life Technologies, Inc. Cells (1 ϫ 10 6 ) were plated on a 150-mm dish (Nunc) 15 h before transfection. The pZip U19 (ts A58)Tneo T antigen expressing vector (23) (this plasmid expresses high levels of T antigen and is highly transforming in our experience at 37°C, it was not temperature-sensitive in our experience) was introduced into Balb/c 3T3 cells with 1 g of pUCSVneo as a selectable marker by DNA mediated gene transfer using the calcium phosphate protocol. Selection was initiated 48 h after transfection by adding 1 mg/ml G418 (Life Technologies, Inc.) to the medium. G418resistant cells were cloned in selective medium.
Tumorigenicity Assay-For analysis of growth in soft agar, 3 ϫ 10 3 cells were seeded in triplicate onto a six well dish (Falcon) in 4 ml of complete medium containing 0.33% agar solution at 37°C (24). Cells were fed with 2 ml of medium every 2 days. Growth was scored as colonies containing Ͼ10 cells, 21 days after plating.
Oligodeoxynucleotide Treatment and Transient Transfection-Cells were plated on tissue culture grade 100-mm dishes at a density of 1 ϫ 10 5 /plate 24 h before treatment. Each treatment was performed in triplicate. The phosphorothioate oligodeoxynucleotides used in this study are HYB101584, which is antisense to a sequence encoding the second putative translation initiation site of DNA MeTase (5Ј-TCT ATT TGA GTC TGC CAT TT-3Ј) and the reverse sequence HYB101585 (5Ј-TTT ACC GTC TGA GTT TAT CT-3Ј) as described in Ref. 14. To determine whether DNA MeTase antisense oligonucleotides inhibit transformation initiated by transient expression of T antigen, Balb/c 3T3 cells were plated at a density of 2.5 ϫ 10 4 /well in a six-well plate 24 h before initiation of oligonucleotide treatment. Oligonucleotides were mixed with 31 l of Lipofectin (Life Technologies, Inc.) and 4 ml of Opti-MEM (Life Technologies, Inc.) were added to the mix. The cells were incubated in the Opti-MEM/oligonucleotide mix for 4 h, following which the mix was replaced with regular growth medium. The cells were treated with 100 nM amounts of either HYB101584, HYB101585, or with Lipofectin carrier alone. 48 h after plating, 2 g of either pZip U19 (ts A58)Tneo, T ant RbϪ or pUC SV neo plasmids were included in the transfection mix with oligonucleotides. The oligonucleotide treatment was then repeated a third time 72 h after plating with oligonucleotides but without plasmid. The cells were then harvested following the third treatment and counted. The viability was determined by trypan blue dye exclusion, and the cells were plated onto soft agar as described above. For RNase protection, methyltransferase activity assay and nearest neighbor analysis, RNA, nuclear extracts, and DNA were prepared from treated cells as described below. To verify equal efficiency of transfection and expression of T antigen under the different conditions, a sample of the transfected cells was plated on glass coverslips, fixed for immunostaining with methanol, and incubated with an SV40 T antigen mouse monoclonal antibody (Santa Cruz number sc-147) for 1 h in phosphate-buffered saline, 0.1% bovine serum albumin. The signal was detected using Texas Red-conjugated secondary anti-mouse monoclonal antibody (Vector number H0724) using standard techniques as recommended by the manufacturer. No difference in transfection efficiency and expression of T antigen was observed in oligonucleotide-treated transfectants. Nuclei were stained with 4Ј,6diamidino-2-phenylindole dihydrochloride (ICN number 157574) (data not shown). For transient transfection of a human DNA MeTase expression plasmid bearing the entire coding sequence of DNA MeTase (codons 1-1616) under the direction of a cytomegalovirus promoter (pcDNA 3.1 His DNA MeTase), 10 g of either pcDNA 3.1 His DNA MeTase or pcDNA His control were mixed with 4 l of Lipofectin, and transfection was performed as described above. Expression of transfected constructs was verified by immunocytochemistry essentially as described above using the Xpress monoclonal antibody (Invitrogen number 46-0528) for detection of His-tagged proteins and Texas Red secondary monoclonal antibody. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole dihydrochloride (data not shown). Cells were harvested for soft agar assay 48 h posttransfection.
DNA and RNA Analyses-Genomic DNA was prepared from pelleted nuclei, and total cellular RNA was prepared from cytosolic fractions. To quantify the relative abundance of DNA MeTase and T antigen mRNA, total RNA (5 g) was blotted onto Hybond Nϩ using the Bio-Rad slot blot apparatus. The filter-bound RNA was hybridized to a 32 P-labeled 0.6-kilobase pair cDNA probe encoding the 5Ј sequences of the mouse DNA MeTase (1-600) (13) and exposed to XAR film (Eastman Kodak Co.). The filter was rehybridized with a T antigen probe, and after removing this probe, the filter was hybridized to an 18 S RNA-specific 32 P-labeled oligonucleotide (13). The autoradiograms were scanned with a Scanalytics scanner (one D analysis), and the signal at each band was determined. The signal obtained at each point was normalized to the amount of total RNA at the same point. For Northern blot analysis, the RNA was transferred to an Hybond Nϩ membrane (Amersham Pharmacia Biotech). The Northern blots were hybridized to a 0.6kilobase pair antisense fragment resulting from a single stranded PCR reaction with a DNA MeTase cDNA fragment (1-600) using the oligonucleotide (TCAATGACAGCTCTCTCTGGTGTGACG) as a 3Ј primer and a single-stranded PCR amplification of c-fos (1440 -1458) using the oligonucleotide (CCCAGCCCACAAAGGTCCAGAATC) as a 3Ј primer as well as an oligonucleotide hybridizing to 18 S rRNA. The autoradiograms were quantified by densitometry (Scanalytics), and hybridization was plotted as DNA MeTase signal standardized to 18 S and compared with 0 h actinomycin D treatment.
RNase Protection Assays-RNA was prepared from exponentially growing cells using standard protocols. RNase protection assays were performed as described in Ref. 11 using a 0.7-kilobase pair HindIII-BamHI fragment as a riboprobe. This probe is a genomic fragment bearing exons 3 and 4 of the murine DNA MeTase probe. It therefore protects, as described in Ref. 11, two major bands of 112 and 100 nucleotides corresponding to these exons as well as a number of minor alternatively spliced and alternate initiations as described previously (25). To normalize the signal obtained for DNA MeTase with the amount of total RNA present in each sample and to verify equal loading, the RNA was simultaneously hybridized with a 32 P-labeled riboprobe complementary to 18 S RNA and subjected to RNase digestion and protection (Ambion).
Nuclear Run-on Assays-Nuclei were prepared from 3 ϫ 10 6 exponentially growing 3T3 and T ant 10 transfectants and were incubated with [␣-32 P]UTP (800 Ci/mmol) as described previously (26). An equal concentration of 32 P-labeled RNA samples (1 ϫ 10 6 dpm/sample) was hybridized in triplicate with Hybond-Nϩ filters bearing 20 g of immobilized pSKMet5Ј (containing a genomic fragment bearing exons 2-4 of the murine DNA MeTase gene) (31) (indicated as Met in Fig. 3), pZip U19 (ts A58)Tneo (T ant), and p␥-actin (␥-actin) plasmids and subjected to autoradiography as described previously (26). The intensity of the signals corresponding to 32 P-labeled RNA hybridizing to DNA MeTase transcribed in T antigen expressing transfectants and control cells was determined by scanning densitometry and normalized to the signal obtained for ␥-actin. The ratio of DNA MeTase/actin signal in T antigen transfectants was compared with the value obtained for neo transfectants.
Nearest Neighbor Analysis-Two g of DNA were incubated at 37°C for 15 min with 0.1 unit of DNase, 2.5 l of [␣-32 P]dGTP (3000 Ci/mmol from Amersham Pharmacia Biotech), and 2 units of Kornberg DNA polymerase (Boehringer Mannheim) were then added and the reaction was incubated for an additional 15 min at 30°C as described previously (10). The labeled DNA (8 l) was digested to 3Ј mononucleotides with 70 g of micrococcal nuclease (Amersham Pharmacia Biotech) in the manufacturer's recommended buffer for 10 h at 37°C, spleen phosphodiesterase (Boehringer Mannheim) was then added, and the reaction mixture was incubated for 3 additional h. Equal amounts of radioactivity were loaded on TLC phosphocellulose plates (Kodak), and the 3Ј mononucleotides were separated by chromatography in one dimension (isobutyric acid:H 2 O:NH 4 OH in the ratio 66:33:1). The chromatograms were exposed to XAR film (Kodak), and the autoradiograms were scanned by scanning laser densitometry (Scanalytics one D analysis). Spots corresponding to cytosine and 5-methylcytosine were quantified.
Assay of DNA Methyltransferase Activity-To determine nuclear DNA MeTase levels, cells were harvested immediately posttransfection, and DNA MeTase activity was assayed as described previously (26).
Bisulfite Mapping-Bisulfite mapping was performed as described previously with small modifications (27). Sodium bisulfite, catalog number S-8890, free weight ϭ 104, was used. A 3.6 M solution of sodium bisulfite (ACS grade, Sigma) (pH 5) was prepared fresh each time, and a 20 mM stock of solution of hydroquinone was prepared and stored at Ϫ20°C. 5 g of DNA (digested with EcoRI) were incubated for 15 min at 37°C with 54 l of double distilled H 2 O, 6 l of 3N NaOH. Following this incubation, 431 l of a 3.6 M sodium bisulfite, 1 mM hydroquinone solution was added. 100 l of mineral oil were added to overlay the solution, and the tube was heated at 55°C for 12 h. The bisulfite reaction was recovered from beneath the mineral oil and desalted using the QIA Quick PCR Purification Kit (followed manufacturer's protocol). 6 l of 3 N NaOH were added to the desalted solution, and the tube was incubated for 15 min at 37°C. Following ethanol precipitation (in the presence of 0.3 M NH 4 OAc) the DNA was resuspended in 100 l in double distilled H 2 O. Approximately 50 ng of DNA were used in each of the PCR amplifications. PCR products were used as templates for subsequent PCR reactions utilizing nested primers. The PCR products of the second reaction were then subcloned using the Invitrogen TA Cloning Kit (we followed the manufacturer's protocol), and the clones were sequenced using the T7 Sequencing Kit (Amersham Pharmacia Biotech) (we followed the manufacturer's protocol, procedure C). The primers used for the MyoD first exon (GenBank TM accession M84918) were: MyoD5Ј1, 5Ј-atttaggaattgggatatgga-3Ј (176 -196); MyoD5Ј (nested), 5Ј-ttttttttgtttttttgagat-3Ј (245-265); MyoD3Ј1, 5Ј-ctcatttcacttactccaaaa-3Ј (577-554); and MyoD3Ј (nested), 5Ј-caaacaacacccaaacattc-3Ј (488 -469). The primers used for the DNA MeTase genomic region (GenBank TM accession M84387) were: MET5Ј1, 5Ј-ggattttggtttatagtattgt-3Ј; MET 5Ј (nested), 5Ј-ggaattttaggtttttatatgtt-3Ј; MET3Ј1, 5Ј-ctcttcataaactaaatattataa-3Ј; and MET3Ј (nested), 5Ј-tccaaaactcaacataaaaaaat-3Ј.

RESULTS
To determine whether ectopic expression of T antigen alters the levels of DNA MeTase in the cell, we cotransfected immortalized but nontransformed Balb/c 3T3 cells with a T antigen expression vector pZipneoSV40U19tsA58 (23), and G418-resistant clones were isolated. Quantification of the level of DNA MeTase mRNA as a function of T antigen expression in 24 independent T antigen transfectants by a slot blot analysis (Fig. 1A) shows a strong stimulation of DNA MeTase mRNA by increasing levels of T antigen expression, as assessed by quantifying the hybridization signals obtained with a DNA MeTase probe or T antigen probe and normalizing these signals to the signal observed following hybridization to an 18 S ribosomal RNA probe. Our data show a good correlation between T anti-gen and DNA MeTase expression. The levels of DNA MeTase mRNA in four independent neo clones expressing the selection marker but not T antigen remain significantly lower than the levels observed in the high T antigen expressers (Fig. 1A). These data exclude the possibility that the dramatic increase in DNA MeTase observed in T antigen transfectants is a consequence of clonal variability in expression of this gene.
Two independent clones were chosen to discount the possibility of clonal artifacts; T antigen 10 (high level of T antigen and DNA MeTase expression) and T antigen 22 (medium level of T antigen and DNA MeTase expression) were used for further characterization. T ant 10 was arbitrarily chosen as a representative clone in a category of high T antigen expressers. T ant 22 was also arbitrarily chosen as a representative clone in a category of significantly lower T antigen expressers. The Western blot presented in Fig. 1B demonstrates a significant increase in DNA MeTase protein in these T antigen transfectants that corresponds with the differences observed in T antigen expression. This increase in DNA MeTase protein results in a 5-15-fold increase in DNA maintenance methylation activity as determined using a hemimethylated DNA substrate and [ 3 H]S-adenosylmethionine (18) (data not shown).
The general level of methylation of CpG dinucleotides in the genome of the T antigen transfectants and the control cells was determined by a nearest neighbor analysis (10). As shown in Fig. 2A there is a 1.5 (T ant 22)-to 2-fold (T ant 10) reduction in the population of nonmethylated CpG dinucleotides in T antigen-transfected cells, suggesting that increased DNA MeTase activity results in a detectable increase in genomic DNA methylation.
It is hard to determine whether this hypermethylation indicates complete hypermethylation of a limited set of specific sites or whether it reflects a limited increase in the frequency of methylation of all sites. Comparing the state of methylation of specific sites in different transfectants is an arduous task because of the tendency of cells in culture to exhibit fluctuations in the state of methylation of CpG sites. A previous report has shown, however, that increased ectopic expression of DNA MeTase leads to hypermethylation of CpG islands (28). We therefore studied by bisulfite mapping the state of methylation of the CG-rich region in the first exon of the MyoD gene (Gen-Bank TM accession number M84918) in the higher expressing T antigen clone, T ant 10. CpG islands in general and MyoD in particular have been shown to undergo hypermethylation upon cellular transformation (29). The bisulfite method enables one to look at a large number of copies of the studied gene, thus enabling a representative picture of the state of methylation of each specific site in the population of cells (27). As observed in Fig. 2, B and D, the MyoD region analyzed in our study is hypermethylated in T antigen transfectants relative to neo controls. Of the 18 CpG sites analyzed, a total of 9 sites were found to be hypermethylated in the T ant 10 stable line relative to neo controls (Fig. 2B). When the results for all of the CpG sites in this region in all the sequenced clones (18 for neo, 16 for T ant 10) were pooled and averaged for comparative analysis, the average methylation of CpG sites was found to be 39% for neo transfectants and 71% for T ant 10 transfectants. In order to determine whether this genomic hypermethylation extends to CG sites not contained in CG-rich areas, we chose to map the region upstream to the second exon of the DNA Methyltransferase (dnmt 1) locus. As shown in Fig. 2, C and D, the methylation status of the dnmt 1 locus does not change substantially with T antigen overexpression. Of the seven CpG sites analyzed in this region, only one site (CpG 101) was shown to be partially hypermethylated in the T ant 10 stable line relative to the neo control line. Interestingly, another site (CpG 157) was actually hypomethylated in T ant 10 relative to the neo control. The methylation status of other sites was not substantially different between the clones examined. When the results for all of the CpG sites for all sequenced clones were pooled and averaged, there was no significant difference in the average methylation of CpG sites, contrary to the results for the MyoD locus. One possible interpretation of this data could be that CpG sites contained in CG-rich areas of the genome are preferentially hypermethylated as a result of DNA MeTase induction. However, confirmation of this hypothesis would require substantially more study of many additional regions. Whereas it is unclear whether hypermethylation of specific sites plays a causal role in T antigen-mediated transformation, our results support the hypothesis that T antigen expression stimulates DNA methylation activity in mammalian cells. In accordance with previous observations (30), we show here that an increase in DNA MeTase activity is followed by an increase in overall genomic DNA methylation and that there is localized hypermethylation of at least one CG-rich site.
One of the established mechanisms of transformation by T antigen is its ability to bind and inactivate the tumor suppressor Rb (20). To determine whether induction of DNA MeTase is dependent on Rb inactivation, we took advantage of a previously characterized mutant of T antigen, T ant RbϪ (bearing a  (t test, p ϭ 0.14). D, representative bisulfite mapping sequencing films of T antigen (T Ant 10) and neo (NEO) clones at the MyoD and dnmt 1 loci. One representative sequencing film of bisulfite-treated E107K point mutation in the Rb pocket), which does not bind Rb (31). This mutation has been shown to selectively impair interactions between Rb and T antigen while leaving p53 and p300 interactions with T antigen unimpaired (31). Balb/c transfectants expressing high levels of T ant RbϪ as determined by Northern analysis (Fig. 3, A and B) and Western blot analysis (Fig. 3C) do not express induced levels of DNA MeTase mRNA (Fig. 3, A and B) and protein (Fig. 3C). The variability of the DNA MeTase mRNA levels in T ant RbϪ clones is in the order of the variability in neo clones (compare Fig. 1A and 3A). These results suggest that induction of DNA MeTase is dependent on tumor suppressor inactivation by T antigen. It is yet unclear whether Rb is directly involved in regulating DNA MeTase mRNA levels or whether the changes in DNA MeTase mRNA are a consequence of the multiple downstream effectors of Rb. It is also possible that other proteins that are inactivated by T antigen are involved in regulating DNA MeTase.
Two modes of regulation of DNA MeTase activity have been described previously: first, transcriptional regulation by the Ras signaling pathway (10, 11) and second, posttranscriptional regulation at the level of message stability is involved in cell cycle regulation of DNA MeTase (26). We found no evidence for transcriptional regulation of DNA MeTase by T antigen. First, expression of a previously described DNA MeTase-CAT expression vector bearing a transcriptional regulatory region of the murine DNA MeTase (11) is not induced by coexpression of T antigen but is induced by coexpression of c-Jun, suggesting that T antigen activity is not mediated by induction of AP-1 (data not shown). Second, a nuclear runon assay (Fig. 4A) shows that there is no significant change in the transcriptional rate of the DNA MeTase gene in the presence of T antigen. Quantification of this experiment is presented in the bottom panel of Fig. 4A.
To determine whether the stability of DNA MeTase mRNA is altered as a consequence of T antigen expression, the T antigen transfectants as well as 3T3 controls were treated with actinomycin D (an inhibitor of mRNA transcription) for 0, 1, 4, or 7 h and the level of DNA MeTase mRNA was assessed using a Northern blot analysis (the signal was normalized to the level of 18 S rRNA in each lane). While the levels of mRNA in Balb/c 3T3 cells decline to undetectable levels by 7 h (Fig. 4B) (this result has been repeated in three independent experiments), the profile of decline in DNA MeTase mRNA obtained for the T antigen transfectants is markedly different. The mRNA in T ant 22 remains at 50% of control (0 h) levels after 7 h (Fig. 4B, left panels) and even up to 24 h treatment (data not shown). In T ant 10, the level of DNA MeTase mRNA remains at, or close to, 100% for the duration of the actinomycin D treatment (Fig.  4B, middle panels). The rate of decay of c-fos mRNA (which expression increased 2-3-fold more in T antigen transfectants versus Balb/c 3T3), a gene known to be regulated by stabilization of its message (32), is unchanged (Fig. 4B, bottom). There- fore, T antigen expression affects the stability of DNA MeTase mRNA without affecting the general ability of the cell to degrade mRNA. Further experiments will be required to determine what are the downstream effectors of Rb that modulate the stability of DNA MeTase mRNA and whether the change in stability of the mRNA can account for all of the increased expression of DNA MeTase. Our data suggest, however, that the level of DNA MeTase can be regulated by viral antigens acting on tumor suppressors at a different level than the transcriptional upregulation induced by protooncogenes such as Ras (10,11).
We have utilized DNA MeTase-specific antisense oligonucleotides that we have developed recently (14) to test the hypothesis that DNA MeTase is a critical downstream effector of T antigen-initiated cellular transformation. This oligonucleotide has been shown previously to reduce DNA MeTase mRNA in Y1 carcinoma cells and inhibit tumorigenesis ex vivo and in vivo (14). We first determined that the antisense oligonucleotides can effectively inhibit DNA MeTase in our system. Balb/c cells were treated with 100 nM DNA MeTase phosphorothioate antisense oligonucleotide (HYB101584) (14) as well as a reverse control phosphorothioate oligonucleotide (HYB101585) (14) in the presence of a lipid carrier (Lipofectin) and transiently transfected with pZipneoSV40U19tsA58. We first determined whether DNA MeTase antisense oligonucleotide treatment (for 72 h) reduces the level of DNA MeTase mRNA relative to treatment with the nonspecific oligonucleotide control as described previously (14). Cellular RNA prepared from the treated cells was subjected to an RNase protection assay for DNA MeTase mRNA, using a previously described DNA MeTase-specific riboprobe (11). The results shown in Fig. 5A show that T antigen-transfected Balb/c cells express lower levels of DNA MeTase mRNA following antisense treatment than those treated with equal concentrations of con-trol oligonucleotides. Scanning densitometry values show nearly 90% inhibition of message expression for antisensetreated cells relative to reverse-treated control cells when normalized to the signal obtained for 18 S ribosomal RNA in the same hybridization reaction.
To determine whether this message level inhibition translated into a decrease in DNA methyltransferase activity, an activity assay was performed using nuclear extracts prepared from oligonucleotide-treated cells. As described previously (26), the assay tests the ability of cellular extracts to incorporate a methyl group from tritiated S-adenosylmethionine into a hemi-

FIG. 3. A T antigen mutant that does not bind Rb cannot trigger an increase in DNA MeTase levels.
A, quantification of a Northern blot analysis of six transfectants with a T antigen bearing a E107K mutation in the Rb binding pocket (31) (T ant RbϪ), a neo clone, and a Balb/c control. The results were plotted as described in Fig. 1A. B, Northern blot analysis of T antigen and DNA MeTase mRNA expression in T antigen and T antigen RbϪ transfectants. Balb/c 3T3 cells were transfected with a T antigen construct bearing a E107K mutation in the Rb pocket (31). RNA (20 g) extracted from 3T3 and T ant 10, a neo control, and T antigen RbϪ stable transfectants was subjected to a Northern blot analysis as described under "Experimental Procedures" and hybridized to either T antigen, 18 S, or DNA MeTase probes as indicated. C, Western blot analysis of T antigen and DNA MeTase expression of nuclear extracts (50 g) prepared from 3T3 cells T ant 10, T ant RbϪ, and neo transfectants. The antibodies were the same as in Fig. 1B.   FIG. 4. T antigen regulates DNA MeTase at the posttranscriptional level. A, run-on assay. Nuclei from 3T3 cells and the T ant 10 transfectants were isolated, and transcription was allowed to resume in the presence of [ 32 P]UTP as described previously under "Experimental Procedures" (26). Equal counts of each reaction were hybridized (representative duplicate experiments are presented in two horizontal panels for 3T3 and T ant 10 transfectants) to Hybond Nϩ membranes (Amersham Pharmacia Biotech) on which ␥-actin, DNA MeTase (Met), and U 19/ts A 58 T antigen (T ant.) plasmids were immobilized. The average hybridization intensities of three separate experiments were determined by scanning densitometry (Scanalytics) and presented in the bottom panel with S.E. values as indicated. No significant difference in MeTase hybridization intensities was detected for 3T3 and T ant 10 transfectants. B, Northern blot analysis of actinomycin D-treated 3T3, T ant 10, and T ant 22 cells. Equal numbers of cells were plated on three 10-cm plates per clone and allowed to grow for 4 days without serum replenishment. After 4 days serum was added with 5 g/ml actinomycin D, and the cells were allowed to grow for 0, 1, 4, or 7 h. RNA was collected, and 10 g was subjected to Northern blot analysis and hybridization with MeTase, 18 S probes as described under "Experimental Procedures." The intensity of the signal hybridizing to DNA MeTase relative to 18 S RNA was determined (by densitometry) for each time point and plotted in the bottom panel. The Northern blots were hybridized to c-fos, and the relative signal hybridizing to c-fos was determined. methylated substrate. Quantification of this assay in Fig. 5B revealed, similarly to the results of the RNase protection assay, a dramatic (9-fold) and highly significant (p Ͻ 0.001) inhibition of DNA methyltransferase activity in antisense relative to reverse oligonucleotide-treated cells.
We then determined whether the inhibition of DNA methyltransferase message levels and activity results in generalized genomic hypomethylation. Nearest neighbor analysis was used to test this hypothesis, as shown in Fig. 5C. The level of methylation of CpG dinucleotides in Balb/c controls is approximately 65% (data not shown), a result similar to that obtained for Balb/c neo control stable transfectants ( Fig. 2A). As expected, reverse oligonucleotide treatment showed no significant change in methylation levels from control Balb/c cells. However, antisensetreated cells showed a remarkable reduction in the level of DNA methylation. Only 20% of the CpGs in the genome were methylated following antisense treatment (Fig. 5C).
Having established that the antisense oligonucleotides effectively inhibited DNA methyltransferase message and activity levels and DNA methylation levels in Balb/c cells, we devised a cell culture assay to test the role of DNA methyltransferase as a downstream effector of T antigen. To show that DNA methyltransferase is directly involved in initiation of the transformed phenotype, we transiently transfected Balb/c 3T3 cells with the pZipneoSV40U19tsA58 T antigen expression vector following pretreatment with either 100 nM antisense or reverse oligonucleotides. As expected, transient transfection of normal Balb/c 3T3 cells by a T antigen expression vector for 48 h results in their transformation as indicated by their ability to grow in an anchorage independent manner on soft agar (Fig.  6A). Transfection with a neo expression vector or a vector directing the expression of a T antigen that is defective in its ability to interact with Rb does not result in cellular transformation similar to the results observed following stable transfection. Thus, since this assay does not require drug selection it allows us to study the initial stages of transformation induced by T antigen. To determine whether DNA MeTase is required for initiation of cell transformation, the cells were pretreated with 100 nM amounts of either DNA MeTase antisense or reverse oligonucleotides 24 h prior to introduction of T antigen into the cells. Treatment was terminated following an additional 48 h exposure to the oligonucleotides, and the cells were plated on soft agar in the absence of any further treatment. As observed in Fig. 6A, exposure of cells at the initial stages of transformation by T antigen to DNA MeTase antisense results in significant inhibition of cellular transformation as determined by a soft agar assay. Since the antisense oligonucleotide is not present during growth on soft agar, it supports the The relative intensities of the DNA MeTase signals standardized by 18 S signals as determined by scanning densitometry were plotted in the right panel. B, nuclear extracts were prepared from reverse control (Rev) and DNA MeTase antisense oligonucleotide (AS)-treated cells and assayed in triplicate for methyltransferase activity as described under "Experimental Procedures." The results presented are an average of triplicate determinations Ϯ S.E. The differences between activities of reverse-and antisense oligonucleotide-treated extracts were determined to be statistically significant (t test, p Ͻ 0.001). C, genomic DNA was prepared from reverse-treated (Rev) and antisense-treated (AS) cells and subjected to nearest neighbor analysis. The relative abundance of nonmethylated and methylated cytosines was quantified (Scanalytics), and the results are plotted as the percentage of nonmethylated cytosines from the total population of cytosines residing in the dinucleotide sequence CpG. The results presented are an average of triplicate determinations Ϯ S.E. The differences between unmethylated cytosines of reverse-and antisense oligonucleotide-treated DNA were determined to be statistically significant (t test, p Ͻ 0.05).

DISCUSSION
Our manuscript shows that similar to the Ras oncogenic pathway, SV40 large T antigen elevates cellular DNA MeTase mRNA and protein levels (10,11). This activity of T antigen is dependent on its interaction with the tumor suppressor pRb, suggesting that overexpression of DNA MeTase lies downstream to inactivation of the tumor suppressor Rb. It is unclear yet whether Rb has a direct effect on DNA MeTase or whether these effects are mediated through interactions with other cell cycle regulators. It is also unclear whether other proteins that interact with T antigen are also involved in regulating DNA MeTase. Whereas up-regulation by the Ras oncogene involves transcriptional activation, T antigen up-regulates DNA MeTase at the posttranscriptional level.
This paper also tests the hypothesis that an increase in DNA MeTase levels is critical for cellular transformation triggered by T antigen. The fact that DNA MeTase has been shown to be critical for two different pathways of oncogenesis is consistent with a central role for DNA MeTase in cellular transformation (9,12). DNA MeTase is critical for the initial stages of transformation by T antigen, since inhibition of DNA MeTase at the time of introduction of T antigen into the cells by transient transfection prevents cellular transformation.
What might be the possible role of DNA MeTase in cellular transformation? One hypothesis that has been supported by a growing body of evidence is that an increase in DNA MeTase might lead to an aberrant increase in methylation and inactivation of tumor suppressors (6), suggesting that increased levels of DNA MeTase cause cellular transformation by ectopic methylation of tumor suppressor genes. In contrast, our data surprisingly shows that DNA MeTase expression is controlled by tumor suppressors. Therefore activation of DNA MeTase might be the downstream target of a transformation program triggered by tumor suppressors. An alternative hypothesis is that the DNA MeTase protein is directly involved in controlling cellular growth (9,12). Such a model is consistent with the requirement for increased DNA MeTase protein in both Ras and T antigen pathways. An interesting observation is that methyl deficient diets, which cause hypomethylation of DNA, also result in induction of DNA MeTase levels and are known to be carcinogenic (34). This observation could be explained, if the critical event for oncogenesis is not hypermethylation of DNA but rather high levels of DNA MeTase protein. Recent data showing that DNA MeTase binds proliferating cell nuclear antigen in the replication fork, and competes with p21 WAF1 for the binding site, lends support to the hypothesis that DNA MeTase protein plays a direct regulatory role in the replication fork (22).
The fact that transient transfection of DNA MeTase is sufficient to induce cellular transformation and that transient inhibition of DNA MeTase at the time of induction of cellular transformation by T antigen is sufficient to prevent transformation is consistent with a direct role for the DNA MeTase protein in transformation. The cloned DNA MeTase is very inefficient in de novo methylation and it is improbable that transient expression of the enzyme results in significant hypermethylation of a tumor suppressors (28). Overexpression of DNA MeTase can result in hypermethylation of CG islands, but this is a slow process which takes about 40 -50 passages (6 -7 months in culture) (28). Vertino et al. (28) have shown previously that there was no detectable change in the methylation of CG islands in transfectants expressing greater than nine fold of DNA MeTase levels up to 20 -23 population doublings. Whereas it is generally accepted that many tumor suppressors are inactivated by methylation, the data presented here are consistent with the hypothesis that DNA MeTase levels are controlled by tumor suppressors.
While additional experiments are required to support or nullify the hypothesis that an increase in DNA MeTase is directly involved in cellular transformation, the fact that a viral oncogene can affect the cellular DNA methylation machinery raises new directions and possibilities for understanding cellular transformation processes.