Stage-by-Stage Change in DNA Methylation Status of Dnmt1 Locus during Mouse Early Development*

Methylation of DNA is involved in tissue-specific gene control, and establishment of DNA methylation pattern in the genome is thought to be essential for embryonic development. Three isoforms of Dnmt1 (DNA methyl- transferase 1) transcripts, Dnmt1s , Dnmt1o , and Dnmt1p , are produced by alternative usage of multiple first exons. Dnmt1s is expressed in somatic cells. Dnmt1p is found only in pachytene spermatocytes, whereas Dnmt1o is specific to oocytes and preimplantation embryos. Here we determined that there is a tissue-dependent differentially methylated region (T-DMR) in the 5 (cid:1) region of Dnmt1o but not in that of the Dnmt1s/1p . The methylation status of the Dnmt1o T-DMR was dis-tinctively different in the oocyte from that in the sperm and adult somatic tissues and changed at each stage from fertilization to blastocyst stage, suggesting that active methylation and demethylation occur during preimplantation development. The T-DMR was highly methylated in somatic cells and embryonic stem cells. Analysis using Dnmt-deficient embryonic stem cell lines revealed that Dnmt1, Dnmt3a, and Dnmt3b are each partially responsible for maintenance of methylation of Dnmt1o T-DMR. In particular, there are compensatory and cooperative roles between Dnmt3a and Dnmt3b .

Methylation of DNA is involved in tissue-specific gene control, and establishment of DNA methylation pattern in the genome is thought to be essential for embryonic development. Three isoforms of Dnmt1 (DNA methyltransferase 1) transcripts, Dnmt1s, Dnmt1o, and Dnmt1p, are produced by alternative usage of multiple first exons. Dnmt1s is expressed in somatic cells. Dnmt1p is found only in pachytene spermatocytes, whereas Dnmt1o is specific to oocytes and preimplantation embryos. Here we determined that there is a tissuedependent differentially methylated region (T-DMR) in the 5 region of Dnmt1o but not in that of the Dnmt1s/1p. The methylation status of the Dnmt1o T-DMR was distinctively different in the oocyte from that in the sperm and adult somatic tissues and changed at each stage from fertilization to blastocyst stage, suggesting that active methylation and demethylation occur during preimplantation development. The T-DMR was highly methylated in somatic cells and embryonic stem cells. Analysis using Dnmt-deficient embryonic stem cell lines revealed that Dnmt1, Dnmt3a, and Dnmt3b are each partially responsible for maintenance of methylation of Dnmt1o T-DMR. In particular, there are compensatory and cooperative roles between Dnmt3a and Dnmt3b. Thus, the regulatory region of Dnmt1o, but not of Dnmt1s/1p, appeared to be a target of DNA methylation. The present study also suggested that the DNA methylation status of the gene region dynamically changes during embryogenesis independently of the change in the bulk DNA methylation status.
DNA methylation of the vertebrate genome occurs predominantly at cytosine residues in cytosine-guanine dinucleotides (CpGs) 1 (1). About 70% of CpGs are methylated, mainly in the repressive heterochromatin region and in repetitive sequences such as retrotransposable elements (2). Formation of the cell type-specific DNA methylation pattern is one of the epigenetic events accompanying the production of diverse cell types in the body (3). Generally, DNA methylation in gene-containing regions of the genome is inversely correlated with the transcriptional activity of associated genes (4,5) through direct (6) and/or indirect (7-10) mechanisms.
We have previously shown that the rSphk1 (rat sphingosine kinase 1) gene, which has multiple alternative first exons embedded in a CpG island, has a tissue-dependent differentially methylated region (T-DMR), with its methylation status inversely correlated with the expression of one of the rSphk1 mRNA subtypes (11). We have also shown that the rPL-1 (rat placental lactogen 1) gene, exclusively expressed in the placenta and having no CpG island at its 5Ј region, has a T-DMR that is hypomethylated in the placenta compared with other tissues (12). Furthermore, T-DMRs were also found in the 5Ј region of two developmentally essential genes in the mouse: Oct4 (octamer binding transcription factor 4) and Sry (sexdetermining region on the Y chromosome). The T-DMRs of the Oct4 and Sry genes are hypomethylated in cells/tissues in which these genes are expressed (13,14). Thus, DNA methylation-mediated gene silencing is involved in regulation of various genes regardless of the richness of CpGs.
In mammals, there are five members of the DNA (cytosine-5) methyltransferase (Dnmt) family: Dnmt1, Dnmt2, Dnmt3a, Dnmt3b, and Dnmt3L (15)(16)(17)(18)(19). Dnmt1 is considered to be a maintenance methyltransferase based on the in vitro enzyme assay in which it preferentially recognized hemimethylated DNA (2,20,21) and on its localization at replication foci in proliferating cells (22,23). Inactivation of Dnmt1 gene in the mouse leads to global loss of methylation and biallelic expression or silencing of imprinted genes (24 -26). Unlike Dnmt1, biological activity of Dnmt2 did not reveal methyltransferase activity specific for CpGs (16), and knock-out of the Dnmt2 gene in the mouse resulted in no detectable abnormality (18). Dnmt3a and Dnmt3b have de novo methyltransferase activity in vitro (17), and a lack of either of them caused extensive perturbations in DNA methylation patterns and premature (Dnmt3a Ϫ/Ϫ ) or prenatal (Dnmt3b Ϫ/Ϫ ) lethality (27).
Three mRNA isoforms, Dnmt1s, Dnmt1p, and Dnmt1o, are expressed from the Dnmt1 locus because of alternative usage of multiple first exons. Dnmt1s is expressed in somatic cells, whereas Dnmt1p and Dnmt1o are exclusively expressed in male and female germ cells, respectively (28). Dnmt1p mRNA, expressed in pachytene spermatocytes, does not produce the active form of Dnmt1 protein, because short open reading frames in the first exon of Dnmt1p (exon 1p) likely interfere with translation of the authentic open reading frame. In contrast, translation from the ATG codon in the first exon of a * This work was supported in part by the Program for Promotion of Basic Research Activities for Innovative Biosciences and by Grants-inaid for Science Research 15080202 and 15208027 (to K. S.) and 16380226 (to S. T.) from the Ministry of Education, Culture, Sports, Science and Technology. 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.
Dnmt1o isoform (exon 1o) is apparently compatible with translation from an in-frame downstream ATG, which produces an N-terminal truncated active Dnmt1 protein (Dnmt1o). Dnmt1o solely exists in growing oocytes and preimplantation embryos until it is subsequently replaced by Dnmt1s at the blastocyst stage (28,29). Dnmt1o protein is of interest because it is excluded from the nucleus at all stages of preimplantation development but the eight-cell stage (28 -31).
The expression of Dnmt1s requires a cis-element in the promoter region and is controlled by transcription factors Sp1 and Sp3 (32). Furthermore, Dnmt1 transcription during cell cycle progression is modulated by two other cis-elements in the promoter region, which are regulated in a complex fashion by E2F and other transcription factors through E2F-Rb-HDACdependent and -independent pathways (33).
Although the regulatory mechanism for spatio-temporal expression of Dnmt1o isoform is poorly understood, the strictly regulated expression of Dnmt1o suggests that there may be tissue-or developmental stage-specific epigenetic marks on the upstream region of the first exon of Dnmt1o isoform. To address this possibility, here we investigate the methylation status of CpGs in the 5Ј upstream region of Dnmt1o and Dnmt1s/1p in germ cells and preimplantation embryos.

MATERIALS AND METHODS
Animal Treatment and Reagents-The experiments were carried out according to the guidelines for the care and use of laboratory animals (Graduate School of Agriculture and Life Sciences, The University of Tokyo). C57BL/6NCrj (B6) mice were purchased from Oriental Yeast Co., Ltd. (Tokyo, Japan) and kept under regulated temperature (22-25°C), humidity (40 -60%), and illumination cycles (14 h light of and 10 h of dark) with ad libitum access to food and water. All of the reagents were purchased from Wako Pure Chemicals (Osaka, Japan) unless otherwise stated.
Collection of Oocytes and Embryos-Adult B6 females (7 weeks old) were injected with 7.5 IU of serum gonadotropin from a pregnant mare (Teikoku Hormone Mfg. Co., Ltd., Tokyo, Japan) and 5 IU of human chorionic gonadotropin (Teikoku Hormone Mfg. Co., Ltd.) with a 48-h interval. Metaphase II oocytes were then collected from the oviducts 20 h after human chorionic gonadotropin injection. The collected metaphase II oocytes were denuded of cumulus cells by incubation in 1 mg/ml hyaluronidase (Sigma-Aldrich, Tokyo, Japan) in phosphate-buffered saline for 5 min at room temperature.
To obtain embryos, superovulated B6 females were individually caged with a B6 stud male overnight and examined for the presence of a vaginal plug the next morning. Noon of the day on which vaginal plug was observed was designated as embryonic day 0.5 (E0.5). Preimplantation embryos were recovered from oviducts at E0.5 (early one-cell stage), E1.0 (late one-cell stage), E1.5 (two-cell stage), E2.0 (four-cell stage), and E2.5 (eight-cell stage). Some of the eight-cell stage embryos were cultured in 50-l microdrops of potassium-enriched synthetic oviductal medium (KSOM) (34) under mineral oil (Sigma-Aldrich) at 37°C in an atmosphere of 5% CO 2 in air to obtain morula and blastocyst stage embryos. Postimplantation embryos (E7.5) were dissected out of uterine horns. For RNA/DNA preparation, only the morphologically normal oocytes and embryos were pooled and stored in 5 l of diethyl pyrocarbonate-treated phosphate-buffered saline at Ϫ80°C until use. Approximately 150 oocytes and 30 -80 preimplantation embryos were used for DNA preparation. Kidney, liver, and ovary were collected from 60-dayold adult B6 mice Embryonic Stem (ES) Cells-The ES cell line MS12, derived from B6 mouse embryo (35), was a kind gift from Dr. H. Suemori (Kyoto University, Kyoto, Japan). The ES cell line J1, derived from 129S4/SvJae mouse embryo, and its Dnmt-deficient derivatives (Dnmt1 Ϫ/Ϫ , Dnmt3a Ϫ/Ϫ , Dnmt3b Ϫ/Ϫ , Dnmt3a Ϫ/Ϫ , and Dnmt3b Ϫ/Ϫ ) were kindly provided by Dr. E. Li (Novartis Institutes for BioMedical Research). Both alleles (four alleles in the case of the double mutant) were targeted in vitro to produce these Dnmt-deficient ES cells (27). The ES cells were cultured on mitomycin C (Sigma-Aldrich)-treated STO cells in the presence of 1000 units/ml leukemia inhibitory factor (ESGRO®; Chemicon, Temecula, CA) under standard conditions (34).
Before preparation of genomic DNA from ES cells, mitomycin Ctreated STO cells were removed by taking advantage of differences in adherence to the dish surface between STO and ES cells. In brief, ES cells co-cultured with STO cells were trypsinized and replated on fresh tissue culture dishes. After 30 min of incubation at 37°C, nonattached cells were collected and added to fresh tissue culture dishes again. Following a second 30-min incubation at 37°C, nonattached ES cells were recovered and stored at Ϫ80°C until use.
Sodium Bisulfite Genomic Sequencing-Genomic DNA was extracted as described previously (14). Sodium bisulfite genomic sequencing was carried out as described previously (36,37) with slight modification. In brief, EcoRI-digested genomic DNAs (0.5-2.0 g) were denatured in 0.33 M NaOH for 15 min at 37°C. Sodium metabisulfite (pH 5.0) and hydroquinone were added to final concentrations of 2.0 M and 0.5 mM, respectively. After incubation in the dark at 55°C for 12 h, modified DNAs were purified with Wizard DNA Clean-Up system (Promega, Madison, WI). The reaction was stopped by further incubation in NaOH at a final concentration of 0.3 M at 37°C for 15 min followed by ethanol precipitation. Purified DNAs were suspended in 20 l of 10 mM Tris-HCl (pH 8.0) containing 1 mM EDTA. Bisulfite-modified DNA (2 g) was amplified with AmpliTaq Gold (Applied Biosystems, Foster City, CA) and one of the following primer sets (see Fig. 1 and o-2-2, 5Ј-CAACCTTAACAACACAACTAAAATA-3Ј; and s-1, 5Ј-GT-TGGTAAGTAAATTAGAGTTATTT-3Ј and s-2, 5Ј-ACTCCCTCAAACT-CCCAATCAATAA-3Ј. The thermocycling program was 43 cycles of 94°C for 30 s, 56°C for 30 s, and 72°C for 1 min, preceded and followed by 10 min of incubation at 95 and 72°C, respectively. The amplified PCR products were cloned into pGEM-T easy vector (Promega) and sequenced.
RNA Extraction and RT-PCR-Collected oocytes or embryos in 5 l of diethyl pyrocarbonate-treated phosphate-buffered saline were combined with 100 l of TRIzol reagent (Invitrogen), and RNA was isolated according to the manufacturer's instructions. Two g of glycogen (Fermentas, Hanover, MD) were added as a carrier before ethanol precipitation. Isolated RNA was finally dissolved in 10 l of diethyl pyrocarbonate-treated water. The RNA preparations from adult tissues were also performed with the TRIzol reagent.
The embryos were freed of zona pellucida by brief exposure to acidified Tyrode's solution (pH 2.5) and fixed for 15 min at room temperature in a freshly prepared 4% formaldehyde/phosphate-buffered saline. Fixed specimens were incubated in a blocking buffer (3% bovine serum albumin, 0.5% Triton X-100) for at least 1 h. For immunofluorescence detection of Dnmt1o, Dnmt3a, and Dnmt3b, the anti-Dnmt1 antiserum, the anti-Dnmt3a antibody, and the anti-Dnmt3b antibody were diluted 1:50, 1:1000, and 1:1000, respectively, in blocking buffer. Bound antibodies were detected by fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) diluted 1:500 in blocking buffer. Immunostained embryos were counterstained in 10 g/ml 4Ј,6-diamidinophenylindole (Roche Applied Science) in blocking buffer for 30 min at room temperature. For microscopic examination, immunostained embryos were transferred into a drop of mounting medium (Biomeda, Foster City, CA) placed on a glass microscope slide. A coverslip was gently placed on the drop containing embryos to spread the mounting medium and slightly flatten the specimen and was subsequently sealed with clear nail polish. The embryos were then examined using an Axioskop microscope (Carl Zeiss, Tokyo, Japan) equipped for epifluorescence with fluorescein and UV filter sets. The obtained images were processed by deconvolution software IPLab program (M&S, Tokyo, Japan).
Sequence Analysis-The genomic DNA sequence of Dnmt1 locus was obtained from GenBank TM (accession number NT_039472). Sequences of part of exon 1o and its downstream region were substituted with those from another GenBank TM entry (accession number AF175410) because they appeared ambiguous in NT_039472. Content of GCs and CpG frequency were analyzed and plotted with a web-based on-line program CpGPlot (bioweb.pasteur.fr/seqanal/interfaces/cpgplot.html) developed by the European Molecular Biology Open Software Suite.

Genomic Structure and CpG Composition around the First
Exons of Dnmt1-The positions of the three alternative first exons of Dnmt1 gene and CpG sequences are illustrated in Fig.  1. The first exon of Dnmt1o (1o) is located ϳ7 kb upstream of the first exon of Dnmt1s (1s), and the first exon of Dnmt1p (1p) is located 85 bp downstream of 1s. There is a CpG island containing 39 CpGs, which covers 1s and a part of 1p (Ϫ68 to 332 relative to the putative transcription initiation site of Dnmt1s). There are 20 CpGs between 500 and 1500 bp upstream of 1o, whereas there are only two CpGs within 500 bp upstream of 1o (Fig. 1A).
In the present study, we intensively investigated the methylation status of 18 of 20 distal CpGs in the 5Ј region of 1o and of 49 CpGs within and around the CpG island at 1s/1p. The distal 18 CpGs in the 5Ј region of 1o are designated hereafter by alphabetical letters, a to r, to avoid confusion and were divided into three groups (I, II, and III) for convenience to describe dynamic change of DNA methylation status. Groups I, II, and III contain five CpGs (a-e), nine CpGs (f-n), and four CpGs (o-r), respectively (Fig. 1A).
Expression of Dnmt1 Isoforms Dnmt1o and Dnmt1s in the Oocytes and Embryos-Transcripts of Dnmt1o and Dnmt1s in oocytes, embryos before and after implantation, and ovary and ES cells were analyzed by RT-PCR (Fig. 2). Dnmt1o transcripts were detected in oocytes, one-and two-cell stage embryos, morula, and ovary, whereas they were not detectable in fourand eight-cell stage embryos, blastocysts, ES cells, and E7.5 embryos. No Dnmt1o transcripts were detected in the kidney and liver of adult mice (data not shown). In contrast, Dnmt1s transcripts were detectable in all cells and tissues examined; then amount of Dnmt1s transcript was high in E7.5 embryos, ovary, and ES cells, whereas it was low in other embryos/cells including the oocytes and preimplantation embryos. Thus, expression of Dnmt1o is not allowed in the restricted embryonic stages including four-cell, eight-cell, and blastocyst stages and embryos after implantation. In this context, the expression of Dnmt1o is more severely controlled compared with Dnmt1s.
It is unclear whether the de novo expression of Dnmt1o occurs during embryogenesis. Various mRNA species transcribed in oocytes are carried over into the preimplantation zygotes, and transcription is limited at early embryonic stages (40,41). Interestingly, the one-cell stage embryos contained higher amounts of Dnmt1o mRNA than did the oocytes. Considering that the amount of ␤-actin mRNA was comparable between the oocytes and one-cell stage embryos, the increase in Dnmt1o transcript seems to be due to de novo synthesis after the fertilization. Similarly, resumption of Dnmt1o mRNA detection at the morula stage likely results from bona fide transcription at this stage rather than carry over from previous stages because transcripts were not detectable at the four-and eight-cell stages. Thus, transcription of Dnmt1o likely occurs in one-cell and morula stage embryos, whereas it is suppressed in four-cell, eight-cell, and blastocyst stage, and postimplantation embryos.
DNA Methylation Status of the 5Ј Regions of 1o and 1s/1p in Germ Cells and Somatic Tissues-DNA methylation status of CpGs in the 5Ј regions of 1o and 1s/1p was investigated in adult somatic tissues and germ cells by the sodium bisulfite sequencing method. At the 5Ј regions of 1o, all of the CpGs examined (a-r) appeared to be hypermethylated in sperm, kidney, and liver, whereas in the oocyte, 13 (f-r) were almost completely unmethylated, and the rest (a-e) were hypermethylated. All CpGs examined in the upstream region of 1s/1p were unmethylated in sperm and were also barely methylated in the oocyte, kidney, and liver (Fig. 3). A complete lack of methylation of CpGs in the 1s/1p upstream region was evident in all of the cells and tissues examined, including early embryos and ES cells (data not shown).
These data indicate that there is a tissue-dependent differentially methylated region (T-DMR) in the 5Ј region of Dnmt1o but not of Dnmt1s/1p. The CpG methylation status of Dnmt1o T-DMR was dependent on developmental stage, and it was different in the oocyte from sperm and adult somatic tissues.

Dynamic Changes in DNA Methylation Status of Dnmt1o T-DMR during Preimplantation Development-We then investigated developmental changes in DNA methylation status of
Dnmt1o T-DMR during preimplantation development (Fig. 4). In the early one-cell stage embryos collected at E0.5, the methylation pattern of CpG a-r was already distinct from that in the sperm and the oocyte; CpGs h-j and o-r were perfectly methylated, whereas other CpGs (f, g, and k-n) were unmethylated. This result clearly showed that site-specific de novo methylation (CpGs h-j and o-r) of the oocyte genome as well as sitespecific active demethylation (CpGs f, g, and k-n) of the sperm genome occurred after fertilization to establish the one-cell stage-specific DNA methylation pattern of Dnmt1o T-DMR. The initial DNA methylation pattern of Dnmt1o T-DMR achieved after fertilization was maintained in the late one-cell stage embryos collected at E1.0.
Further analyses with later stage embryos revealed that all but CpG n of Dnmt1o T-DMR were unmethylated at the twocell stage, suggesting that active demethylation took place again between the one-cell and two-cell stages. At the four-cell stage, a particular set of CpGs (a and f-n) were fully methylated, and other CpGs (b-e and o-r) remained unmethylated, indicating that site-specific de novo methylation occurred again between the two-and four-cell stages. The methylation pattern at the eight-cell stage was also unique and suggested de novo methylation at four CpGs (b-e). Active demethylation occurred again at CpGs a-n, resulting in an almost totally unmethylated pattern of CpGs in the morula in which Dnmt1o transcripts were detected. The methylation status of each CpG showed a nearly all-or-none pattern until the morula stage, suggesting that all blastomeres at each developmental stage before morula possess an identical methylation pattern regarding the Dnmt1o T-DMR. In the blastocyst stage, CpGs h-n became methylated again, whereas other CpGs remained unmethylated. The MS 12 ES cells showed hypermethylated status that is similar to adult somatic tissues. Thus, it is intriguing that the DNA methylation pattern dramatically changes from stage to stage before the implantation.  1s/1p (B). A CpG tention of the Dnmt1o protein has been reported during preimplantation development except for the eight-cell stage (28,31), making Dnmt1o protein an unlikely candidate for regulating formation of the stage-specific methylation pattern of its own gene. To elucidate the possible participation of Dnmt3a and/or Dnmt3b in the methylation of Dnmt1o T-DMR, expression of Dnmt3a and Dnmt3b mRNA and subcellular localization of Dnmt3a and Dnmt3b protein were analyzed by RT-PCR and immunocytochemistry, respectively. Strong expression of Dnmt3a was observed in oocytes, morula, E7.5 embryos, and ES cells, whereas the signal was weak in one-to eight-cell and blastocyst stage embryos (Fig. 5A). Immunocytochemistry revealed that the Dnmt3a protein mainly localized in the nucleus and/or perinuclear region in one-, two-, and four-cell stage embryos (Fig. 5B). In eight-cell, morula, and blastocyst stage embryos, Dnmt3a was mainly detected in the cytoplasm, but a weak signal still remained in the nucleus. Thus, Dnmt3a protein reside at the all-preimplantation cleavage stages.
Dnmt3b mRNA was not detected throughout the preimplantation stages but was detectable in the ovary, E7.5 embryos, and ES cells (Fig. 5A). Dnmt3b protein was, however, detectable by immunocytochemistry in preimplantation embryos (Fig.  5B). Dnmt3b was mainly localized in the cytoplasm and plasma membrane cortex in one-, two-, and four-cell stage embryos, as well as located extensively in cytoplasmic foci in one-and eight-cell stages and morula stage embryos (Fig. 5B). However, unlike Dnmt1, Dnmt3b did not appear to be excluded from the nucleus during preimplantation development. Thus, both Dnmt3a and Dnmt3b proteins existed in the nucleus of preimplantation embryos till the morula stage.  to MS12 ES cells, wild type J1 ES cells showed hypermethylation of Dnmt1o T-DMR (Fig. 6). The Dnmt1 Ϫ/Ϫ ES cells revealed the average level of DNA methylation of the T-DMR was reduced to ϳ30% of that of the wild type. The dependence on Dnmt1 seems to be different among the CpGs; CpGs b, c, g, and q were very dependent on Dnmt1, whereas CpGs d, e, f, h, i, and k were resistant to Dnmt1 deficiency. In the Dnmt3a Ϫ/Ϫ ES cells, CpGs f, g, and k-n showed lower levels of methylation, whereas other CpGs were resistant to the lack of Dnmt3a. Similar dependence was also observed in Dnmt3b Ϫ/Ϫ ES cells, although the extent of demethylation of CpGs f, g, and k-n was more severe in Dnmt3b Ϫ/Ϫ ES cells compared with Dnmt3a Ϫ/Ϫ ES cells. Thus, involvement of Dnmt3a and Dnmt3b is evident in some CpGs of group II, but their participation was limited. In double mutant ES cells (Dnmt3a Ϫ/Ϫ and Dnmt3b Ϫ/Ϫ ), all CpGs, with only one exception, were fully unmethylated, suggesting that either Dnmt3a or Dnmt3b is required for de novo as well as maintenance methylation of the Dnmt1o T-DMR. DISCUSSION Changes in methylation status of Dnmt1o T-DMR and expression of each member of the Dnmt family are illustrated in Fig. 7. The T-DMR, particularly the CpGs of group II, was hypomethylated in the oocyte and embryos when Dnmt1o was expressed, whereas it was hypermethylated in the sperm and somatic tissues. Thus, the developmental stage-specific transcriptional regulation of Dnmt1o seems to involve the epigenetic system by DNA methylation, which determines the turn-on and -off of gene expression. Transcription of Dnmt1s, whose upstream region appeared in this study to be almost completely methylation-free regardless of tissue and developmental stage, is regulated by both chromatin remodeling, involving E2F-Rb-HDAC complex and a combination of transcription factors (32,33). It is also likely that transcriptional regulation of Dnmt1o involves a combination of transcriptional factors in addition to an epigenetic system. Transcriptional regulation of Dnmt1o remains to be elucidated.
In a previous report, focusing on the genome-wide methylation status of a gene area containing CpG islands, we proposed that the changes in DNA methylation level of a gene area are not parallel to that of bulk DNA containing repetitive sequences (3). In agreement with the proposal, the changes in DNA methylation status of Dnmt1o T-DMR were clearly different from the generally accepted concept that mammalian development is accompanied by two major waves of genomewide demethylation and remethylation: one during germ cell development and the other after fertilization (42)(43)(44)(45). Most of the previous studies have suggested that the genome-wide demethylation after fertilization occurs passively, that is, by the lack of maintenance methylation following DNA replication and cell division (43,46), although another study has reported that replication-independent demethylation may also occur during early embryogenesis (47). Obviously, however, DNA methylation status of the Dnmt1o T-DMR changed and showed a unique methylation pattern at any given stage of preimplantation development. All of these changes should occur by de novo methylation and active demethylation. It is clear that dynamic and finely tuned regulation of methylation of Dnmt1o T-DMR exists. The DNA methylation status of Dnmt1o T-DMR in the fertilized egg was quite different from that of sperm and oocyte. This strongly suggests that fertilization is associated with de novo methylation as well as active demethylation at specific sites. The paternal allele of Igf2 gene has been reported to be actively demethylated in the one-cell zygote shortly after fertilization, whereas methylation on the maternal allele remained unchanged or in some cases showed further de novo methylation (48). In the zygote, both active demethylation and de novo methylation activities occur and apparently operate before replication commences.
Recent studies have created controversies with regard to the roles of Dnmt1 and Dnmt3b in maintaining methylation patterns in human cancer cell lines. Rhee et al. (49,50) showed that targeted disruption of either DNMT1 or DNMT3B in HCT116 colorectal carcinoma cells caused only minor reduction in CpG methylation, whereas disruption of both genes resulted in severe loss of methylation, suggesting that the two enzymes have redundant functions in maintaining CpG methylation. In contrast, Robert et al. (51) showed that selective depletion of DNMT1 alone in HCT116 and other human cancer cell lines resulted in global and gene-specific demethylation, a finding consistent with the current view that DNMT1 is the major maintenance enzyme. Furthermore, both Dnmt1 and Dnmt3 were required for stable maintenance of global methylation in the mouse ES cells (52). We also have found that Dnmt3a and 3b are required for maintenance of DNA methylation at some specific loci (53). In the present study, Dnmt1, Dnmt3a, and Dnmt3b all appeared to be required for maintaining the DNA methylation pattern of Dnmt1o T-DMR in ES cells to some extent. Interestingly, however, the T-DMR was fully demethylated in Dnmt3a/3b double mutant ES cells, but demethylation was only partially observed in Dnmt3a Ϫ/Ϫ or Dnmt3b Ϫ/Ϫ ES cells. These data clearly suggested that Dnmt1 alone is not able to exhibit full maintenance methyltransferase activity and that Dnmt3a and Dnmt3b are also components of this activity at least in ES cells, as we have previously suggested by genomewide methylation analysis of Dnmt-deficient ES cells (53). In addition, Dnmt3a and/or Dnmt3b may be able to work as a maintenance methyltransferase without functional Dnmt1 at least on the Dnmt1o T-DMR, because methylation of the T-DMR remained in Dnmt1 Ϫ/Ϫ ES cells. It is, however, also possible that Dnmt3a and/or Dnmt3b remethylate the T-DMR after replication of the genome at each cell cycle. The presence of Dnmt3a and Dnmt3b in preimplantation embryos also supports their contribution to the methylation of Dnmt1o T-DMR.
In conclusion, the 5Ј region of Dnmt1o is a target for DNA methylation, whereas Dnmt1s/p do not have T-DMR. The methylation status of Dnmt1o T-DMR changes during every stage from fertilization to blastocyst stage. The dynamic change in DNA methylation is caused by the active processes of methylation and demethylation of CpGs.