Growth regulation of mouse DNA methyltransferase gene expression.

The steady state level of DNA methyltransferase mRNA is markedly increased as growth-arrested Balb/c 3T3 cells progress into the S phase of the cell cycle. mRNA abundance is reduced to the basal level before termination of DNA synthesis activity. Maintenance DNA methylation activity in nuclear extracts follows a similar pattern with two exceptions. (a) A small peak of DNA methylation activity is detected in early G1 phase. (b) The extinction of DNA methylation activity lags behind the termination of DNA synthesis. Nuclear runon experiments demonstrate that the gene is transcribed in growth-arrested cells, and expression of the gene is post-transcriptionally regulated. We suggest that this mode of regulation of the DNA methyltransferase gene might play an important role in determining and maintaining DNA methylation patterns.

The steady state level of DNA methyltransferase mRNA is markedly increased as growth-arrested Balb/ c 3T3 cells progress into the S phase of the cell cycle. mRNA abundance is reduced to the basal level before termination of DNA synthesis activity. Maintenance DNA methylation activity in nuclear extracts follows a similar pattern with two exceptions. (a) A small peak of DNA methylation activity is detected in early GI phase. ( b ) The extinction of DNA methylation activity lags behind the termination of DNA synthesis. Nuclear runon experiments demonstrate that the gene is transcribed in growth-arrested cells, and expression of the gene is post-transcriptionally regulated. We suggest that this mode of regulation of the DNA methyltransferase gene might play an important role in determining and maintaining DNA methylation patterns.
Although most CpG dinucleotide sequences in vertebrate genomes are methylated at the 5-position of cytosine, a minor fraction remains hypomethylated (1). These nonmethylated sequences are nonrandomly distributed and constitute a pattern of methylation which is gene-and tissue-specific (2). While numerous experiments have demonstrated that specific DNA methylation patterns could be correlated with patterns of gene expression (2) and more recent experiments have shown that DNA methylation might interfere with binding of transacting factors to promoter elements (3), the mechanisms responsible for generating specific methylation patterns are still obscure.
DNA methylation results from interaction between the methyltransferase and its DNA substrate (4); however, about 20% of the CpG sites escape DNA methylation (1). The cDNA encoding for the mouse DNA methyltransferase has been cloned by Bestor et al. (4), but nothing is yet known about the regulation of its expression.
Based on results obtained from the Escherichia coli dam methylase system, we have suggested that maintaining a limiting methylation capacity in the cell coupled with differ-* This work was supported by a grant from the National Cancer Institute of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Research  (1,5). Recent evidence indeed suggests that the level of d a m methylase in E. coli plays an important role in maintaining the origin of replication in a hemimethylated state and regulating DNA replication (6, 7). A similar mechanism may play a role in shaping methylation patterns in vertebrate cells. If the level of DNA methyltransferase activity plays such a role, it should be regulated with respect to DNA synthesis activity. In accordance with this hypothesis, we and others have previously shown that DNA methylation activity correlates with the proliferative activity of cells (8)(9)(10). In this report, we test the hypothesis that regulation of DNA methyltransferase gene expression is associated with the proliferative state of the cell and determine the level at which it is regulated using the mouse fibroblast Balb/c 3T3 cell as a growth-induced cell system.

RESULTS
Balb/c 3T3 cells are a nontransformed murine embryonal mesenchymal line, isolated by Todaro and Green ( l l ) , which can be maintained in a state in which they exhibit stringent density-dependent growth. The ability to obtain large numbers of Balb/c 3T3 cells synchronized at specific portions of the cell cycle have made these cells valuable models for analysis of cellular and molecular events that regulate the progression of the cell cycle (12). The following protocol was used for growth-arresting the cells. Balb/c 3T3 cells were grown to confluency in DMEM' containing 10% fetal calf serum, and the medium was then replaced with DMEM containing 0.5% serum for an additional 2 days. To induce the cells into the cell cycle, we replaced the medium with 10% serum containing DMEM.
The steady state levels of the DNA methyltransferase mRNA were determined using Northern blot assays of RNA purified from growth-induced cells (13,14) at different time intervals following growth stimulation and hybridized to a methyltransferase cDNA probe (4) (Fig. lA). To determine the amount of total RNA in the different samples, the filter was stripped of radioactivity by boiling in 0.2 x SCC, 1% SDS buffer and hybridized to an 18 S RNA-specific 32P-labeled oligonucleotide (15) using standard hybridization conditions (13) (Fig. 1B). As seen in Fig. lA, the -5-kb DNA methyltransferase mRNA (4) is absent in growth-arrested cells and is induced at 8 h, just before the initiation of the DNA synthetic phase of the cycle. An additional band hybridized to our methyltransferase probe and was increased in its intensity as the cells progressed through the cycle. To determine the relative level of DNA methyltransferase mRNA at different phases of the cycle, serial dilutions of total RNA (10-500 ng) were dot-blotted onto nylon filters, and the filter-bound RNA was hybridized sequentially to a methyltransferase and an 18 S probe. The autoradiograms were scanned, and the signal obtained at each point was normalized to the amount of 18 S RNA in each sample (Fig. 1C). DNA synthesis activity serum-starved Balb/c 3T3 cells and from cells induced with 10% fetal calf serum a t different time points postinduction were subjected to Northern blot hybridization analysis using a n2P-labeled 0.6-kb RamHI fragment encoding the most 5' sequences of the DNA methyltransferase cDNA (4). The second arrow indicates the position of the --5-kb DNA methyltransferase mRNA. B, the filter-bound RNA was rehybridized with an 18 S RNA-specific '"P-labeled oligonucleotide (15). The arrow indicates the position of 18 S RNA. C, relative level of DNA methyltransferase mRNA at different time points following serum induction determined by slot blot hybridization analysis using a '"P-labeled methyltransferase cDNA probe. The signal obtained a t each point was normalized to the amount of total RNA a t each sample determined by hybridization to a 18 S RNAspecific probe. Each value is presented as a mean of three determinations f S.D. D, DNA synthesis in serum-starved and seruminduced Balb/c 3T3 cells a t different time points following induction was assayed by measuring the incorporation of 1 pCi of ['Hlthymidine into aliquots of cells which were withdrawn from the bulk cultures a t the indicated time points and were incubated for an additional 2 h at 37 "C. Each value is presented as means of six determinations 2 S.D.
at each time point was determined by measuring incorporation of 1 pCi of ["]thymidine into DNA, measured as trichloroacetic acid-precipitable counts, in aliquots (0.5 ml) of cells withdrawn from the bulk culture and incubated for an additional 2 h a t 37 "C ( Fig. lo). As observed in Fig. 1 the level of DNA methyltransferase mRNA is positively induced 3 h prior to initiation of DNA synthesis and returns to almost basal levels before the cessation of DNA synthesis. This demonstrates that the steady state level of methylase mRNA is tightly regulated with the proliferative state of the cell.
Inasmuch as we have previously demonstrated that DNA methylation activity follows a time course which is concordant with DNA synthesis in rat-regenerating liver and in concanavalin A-induced splenocytes (8), we wondered whether the reductfun of methylase mKNA a t a time when UNA synthesis was still at its peak reflected a different mechanism operating in these cells or whether DNA methylation activity exhibited a different time course than that observed with its mRNA. We determined maintenance DNA methylation activity in nuclear extracts of growth-starved Balb/c 3T3 cells at differ-ent time points following growth induction by serum. Nuclei were isolated from 1-3 X lo6 cells (8), and crude nuclear extracts were prepared by 0.3 M NaCl extraction (8). One pg of nuclear extract was assayed for DNA methyltransferase activity immediately in a 50-pl reaction using 0.1 pg of a synthetic 33-base pair hemimethylated oligonucleotide ( Each value is presented as means of three determinations f S.D. mM KCI, 15 mM NaCI, 15 mM HEPES (pH 7.5), 2 mM EDTA, 2 mM EGTA, 0.15 mM spermine, 0.5 mM spermidine, and 14 mM a-mercaptoethanol buffer and centrifugation a t 1500 X , q (22). T o determine the abundance of methyltransferase, the cytosolic fraction was resuspended in guanidine thiocyanate and subjected to a standard RNA extraction protocol (14). The nuclei were further purified by washing in 1 ml of buffer containing 50% glycerol, 20 mM Tris-HCI (pH 7.9), 75 mM NaCI, 0.5 mM EDTA, 0.85 mM dithiothreitol, 0.125 mM PMSF, and 100 units/ml of RNasin buffer and centrifugation at 10,000 X g for 2 min. The nuclei were resuspended in 70 pl of the same buffer and stored at -70 "C. For a nuclear transcription assay, 35 pl of nuclear suspension for each time point was incubated with 150 pCi of [n-:'"P]UTP (800 Ci/ mmol) in a buffer containing 1 mM each of GTP, CTP, and ATP, 0.3 M (NHJ'SO,, 100 mM Tris-HCI (pH 7.9), 4 mM MgC12, 4 mM MnCI', 50 mM NaCI, 0.4 mM EDTA, 0.5 mM PMSF, 10 mM creatine phosphate, and 20 units of RNasin for 30 min at 28 "C. The labeled RNA was purified with a NAP-5 column (Pharmacia LKB Biotechnology Inc.) following DNase I (100 pg/ml) and proteinase K (100 mg/ml) digestion. For hybridization of RNA to the various probes, 10 pg of pMET (a pGEM-3 plasmid containing a 2.7-kb genomic fragment encoding the 5' region of the DNA methyltransferase gene),' pGEM-3, or pSP6-y-actin (a plasmid containing the cDNA for the human y-actin gene (23)) were immobilized onto a Hybond-N+ nylon filter on a slot blot apparatus using alkaline conditions. The filters were prehybridized by incubation a t 42 "C for 16 h in a solution containing 50 mM HEPES (pH 7), 0.75 M NaCl, 50% formamide, 0.5% SDS, 2 mM EDTA, 10 X Denhardt's solution (22), 200 pg/ml herring sperm DNA, and 10 pg/ml poly(rA) oligoribonucleotide. One X 10" dpm of T-labeled RNA transcribed a t each time point was added to the prehybridization buffer and hybridized with the immobilized plasmids for 72 h a t 42 "C. The filters were washed with 2 x SSC, 1% SDS twice a t room temperature and then three times with 0.2 X SSC, 1% SDS at 55 "C. Autoradiography (Fig. 3 A ) was followed by densitometric analysis of the relative intensity ? S.D. ( n = 3) (Fig. 3C). The abundance of the methyltransferase mRNA in the cytosol at each time point following serum induction was determined as described above (Fig. 3, I3 and C). This experiment demonstrates that while the abundance of the methyltransferase message is markedly induced before the onset of the S phase of the cycle (at 8-13 h postinduction, Fig. 3, R and C), the methyltransferase gene is transcribed in resting cell nuclei and does not follow a similar change in transcription activity. This suggests that post-transcriptional regulation is the major determinant of methyltransferase mRNA abundance in t.he cytosol.

DISCUSSION
Our paper defines some basic properties of the regulation of DNA methyltransferase gene expression. First, the cell responds to a mitogenic stimulus by adjusting the level of mRNA encoding the DNA methyltransferase. The message levels are induced about 3 h before the onset of DNA synthesis. Second, the induction of mRNA levels is followed by elevation of maintenance methylase activity. This strongly suggests that DNA methyltransferase activity at the onset of the DNA synthesis is determined by the abundance of its mRNA. Third, DNA methyltransferase mRNA is significantly reduced as the methyltransferase activity reaches its peak, suggesting that the methyltransferase is down-regulated M. Szyf, unpublished data. by factors which are present in the later phases of DNA synthesis. One possible mode of regulation might involve feedback regulation by the level of methyltransferase activity. This mode of regulation suggests that it is important to prevent excess synthesis of methyltransferase as is the case in resting cells. Fourth, other modes of regulation of DNA methyltransferase activity must exist as the short peak of methyltransferase activity observed early after serum induction (Fig. 2 A ) is not preceded by an induction of mRNA ( Figs.  1 and 3). Fifth, the level of DNA methyltransferase mRNA is post-transcriptionally determined as the gene is transcribed in resting cells and does not exhibit a significant change in transcription following stimulation of DNA svnthesis. I'osttranscriptional regulation has been shown to he an important mode of regulating other genes involved in the DNA synthesis phase such as thymidine kinase and histone genes (17-19, 24), as well as in controlling the expression of early GI genes such asfos and myc (25,26). However, the DNA methvltransferase gene seems to differ from these other cell cvcle-regulated genes inasmuch as it does not seem to be transcriptionally regulated. Our observations stress the general siEnificance and role of post-transcriptional regulation. As implied hv our Regulation of DNA Methyltransferase Gene Expression results, a factor must exist that regulates either the maturation or stability of the DNA methyltransferase mRNA and is regulated with the proliferative state ~f the cell. This factor might play a more general role in regulation of gene expression during the cell cycle. The DNA methyltransferase gene provides a good model for analyzing the mechanisms involved in regulating message stability with the proliferative state of the cell. One interesting question that remains to be answered is whether cycling cells regulate DNA methyltransferase gene expression with the S phase of the cell cycle as well. The changes in expression of the methyltransferase gene following growth stimulation of quiescent cells may not be a cell cycle effect but rather be related to other signals that determine the proliferative state of the cell. What is the biological role of cell cycle regulation of the methyltransferase gene? DNA methylation patterns have been shown to be of importance in regulating the transcription of genes and maintaining the status of gene expression of somatic cells while differentiation has been associated with changes in the DNA methylation pattern (27-30). Adjusting the abundance of enzyme and the availability of the substrate might play an important role in the maintenance of patterns of methylation. A controlled or aberrant switch in the regulation of DNA methyltransferase activity could also play a role in altering the pattern of methylation. Although many observations stress the limited and site-specific nature of DNA methylation, genome wide changes in methylation occur during differentiation (27-29). Moreover, some changes in DNA methylation patterns involve wide areas of the genome: methylation of CpG islands in tumor cell lines (31) or the hypermethylation of late replicating regions of the genome (32). Cancer cells have been shown to contain higher levels of DNA methyltransferase activity than nontumor cells (33), and senescent cells in culture have been shown to undergo genome wide hypomethylation (34). We have previously suggested that general changes in DNA methyltransferase activity might also result in site-specific changes because of differences in the affinity of different sites to methylation activity reflecting higher organization of the DNA substrate (5, 35, 36). The higher ratio of DNA methyltransferase to DNA synthesis activity late in the S phase might play a role in the hypermethylation of late replicating DNA sequences. The abundance of DNA methyltransferase activity in tumor cell lines might reflect loss of cell cycle regulation in cancer cells as has been observed with other DNA synthesis genes (37).
Regulation of enzymatic activity with the cell cycle suggests in many cases involvement in controlling some aspects of cell cycle progression (16). One exciting possibility is that regulation of DNA methyltransferase activity plays a role in controlling DNA replication. Recent evidence suggests that initiation of DNA replication in E. coli is controlled by the state of methylation of dam sites at the origin of replication (6, 7). We have previously shown that the level of dum methyltransferase in E. coli is limiting and suggested that this might play an important role in maintaining a specific status of methylation (5). Recent evidence has demonstrated that altering the level of methylase in E. coli will deregulate DNA replication (6). The regulation of DNA methyltransferase in mammalian cells is consistent with this attractive model.