DNA methylation has several biological functions in mammals, including parental imprinting (Murphy and Jirtle,2003), X-chromosome inactivation (Okamoto et al.,2004), preservation of chromosomal integrity (Lengauer et al.,1997), gene suppression (Meehan and Stancheva,2001; Bird,2002), and development (Li,2002). Five mammalian genes have been identified as DNA methyltransferase genes, namely Dnmt1, Dnmt2, Dnmt3a, Dnmt3b, and Dnmt3l (Bestor,2000; Chen and Li,2004); the predominant protein form is Dnmt1 (Bestor et al.,1988; Bestor,2000; Goll and Bestor,2005). Because of its widespread expression in somatic tissues, its catalytic preference for hemi-methylated substrates , and its localization to the replication fork during DNA replication , the primary role of Dnmt1 has long been thought to be that of a maintenance methyltransferase. Homozygous mutations of the Dnmt1 gene in mice lead to severe abnormalities in genomic imprinting and X-chromosome inactivation (Li et al.,1993ab), and mutant embryos die prior to the mid-gestation stage (Li et al.,1992).
In addition to the somatic type Dnmt1 protein, Dnmt1s, which is present in most cells, two sex-specific Dnmt1 isoforms are produced from the Dnmt1 gene: Dnm1o and Dnmt1p (Mertineit et al.,1998). These variants are expressed from alternative promoters and differ from Dnmt1s by the substitution of germline-specific versions of exon 1 (Mertineit et al.,1998). Although Dnmt1p mRNA is expressed at a high level in mouse pachytene spermatocytes (Singer-Sam et al.,1990; Trasler et al.,1992), no Dnmt1p protein is detected at this stage (Mertineit et al.,1998). In contrast, Dnmt1o is translated at very high levels during oogenesis (Mertineit et al.,1998) and preimplantation development (Monk et al.,1991; Carlson et al.,1992). The Dnmt1o protein has a relative molecular weight of 175,000 (175 kD) and is translated into a truncated version of Dnmt1s that is missing the N-terminal 118 amino acids (Mertineit et al.,1998). The 190-kD Dnmt1s protein replaces Dnmt1o after implantation of the embryo (Mertineit et al.,1998; Ratnam et al.,2002).
Dnmt1o is a maternal-effect protein that is synthesized in the growing oocyte, stored in the cytoplasm of the mature metaphase II-arrested oocyte, and functions after fertilization to maintain DNA-methylation patterns on alleles of imprinted genes (Mertineit et al.,1998; Howell et al.,2001; Ratnam et al.,2002). Studies have indicated that mouse Dnmt1o is unusual in terms of its localization during preimplantation-stage development: Dnmt1o is sequestered in the peripheral cytoplasm, not in the nucleus, and it moves transiently into the nuclei of eight-cell-stage embryos (Carlson et al.,1992; Mertineit et al.,1998; Cardoso and Leonhardt,1999; Howell et al.,2001; Ratnam et al.,2002). However, efforts have not been successful to obtain additional molecular data that confirm the immunostaining results. Cytoplasmic sequestration of Dnmt1o has been observed only indirectly using immunostaining techniques, so we cannot completely exclude the possibility of false-positiveness of the cytoplasmic Dnmt1o signals. These uncertainties have resulted in controversy over the unusual movement of Dnmt1o (Hirasawa et al.,2008; Kurihara et al.,2008). In this study, we wished to resolve the dispute over the localization of Dnmt1o in oocytes. Towards this end, we used a straightforward, non-immunocytochemical approach to determine Dnmt1o localization in early mouse and pig embryos. Pig embryos were chosen because they displayed an opposing pattern to mouse embryos in terms of DNA methylation reprogramming during early development (Jeong et al.,2007a,b; Park et al.,2007). In addition, we characterized the porcine Dnmt1o transcript, and also analyzed Dnmt1s and Dnmt1o in cloned pig embryos to determine the dominant form and location of Dnmt1 in recipient oocyte cytoplasm.
Localization of Dnmt1o in early embryos has not been determined in mammals other than rodents. It is thus unknown whether the cytoplasmic localization of Dnmt1o is conserved. Before we addressed this question, we needed to characterize Dnmt1 proteins in pig. We first used Western blotting to investigate the forms of porcine Dnmt1 (pDnmt1) in pig somatic cells. Pig ovarian cumulus cells had a single band of pDnmt1 with a size similar to Dnmt1 proteins found in mouse NIH3T3 cells and bovine cumulus cells (Fig. 1A). When a DsRed2-tagged Dnmt1s (Kim et al.,2002) protein was transiently expressed in primary pig fibroblasts, the signals were detected in the nuclei (Fig. 1 Ba). Immunostaining of Dnmt1 proteins in cultured cells in mammals has shown that they diffusely exist in the nucleus during interphase of the cell cycle and build a punctuate pattern during the early S phase of the cell cycle (Leonhardt et al.,1992). As shown in Figure 1 Ba and 1Bb, the ectopically expressed DsRed2-Dnmt1 fusion proteins formed a punctuated pattern as well as a diffuse one in the nuclei of pig fibroblasts. Coherently, endogenous pDnmt1 proteins also displayed such alternate patterns between the cell cycle (Fig. 1 Bc and Bd). These results indicate that pDnmt1s proteins share features, such as size and cell cycle–dependent localization patterns, with Dnmt1s proteins in other mammals (Bestor,2000).
We next examined pig female germ cells for pDnmt1. Western blotting showed a predominant pDnmt1 protein band in germinal vesicle (GV)–stage pig oocytes (Fig. 2A) that was smaller than the ∼190-kD pDnmt1s observed in the pig cumulus cells (pCc, see also Fig. 1A) and the cumulus-oocyte complex extracts (pCc+GV). We assumed that this was the oocyte form of pDnmt1, i.e., pDnmt1o. As shown in Figure 2B, the pDnmt1o protein was the major form of pDnmt1 during the period of oocyte and embryo development in pig, when no detectable pDnmt1s was visible.
We characterized pDnmt1o at the mRNA level. Rapid amplification of 5′ cDNA ends (5′-RACE) was used to determine that the pDnmt1o transcript had a unique 47-bp sequence at the 5′ end that was absent from the pDnmt1s cDNA sequence (Fig. 3A). Since the pDnmt1o and pDnmt1s transcripts utilize the same exons (except for exon 1), the unique 47-bp sequence in the pDnmt1o transcript was likely encoded by an oocyte-specific exon-1 (exon-1o). Similarly, the 125-bp sequence at the 5′-end of the pDnmt1s transcript is likely encoded by a somatic-specific exon-1 (exon-1s). The 5′-end sequence of pDnmt1o cDNA (47 bp) was shorter than its mouse counterpart (156 bp) (Fig. 3B).
Reverse transcriptase-polymerase chain reaction (RT-PCR) showed that pDnmt1o transcripts were the major form of pDnmt1 transcripts in pig GV oocytes (Fig. 3C). The pDnmt1o transcript level gradually decreased over time, leveling out at later cleavage stages (Fig. 3D). pDnmt1s transcripts were barely detectable in pig GV oocytes and embryos, similar to the level of pDnmt1o transcripts detected in somatic cells. Overall, the RT-PCR results indicate that early pig embryos accumulate Dnmt1o transcripts differently than do mouse embryos: in mice, Dnmt1o transcripts are nearly undetectable in early cleavage-stage embryos whereas Dnmt1s transcripts are abundant (Cardoso and Leonhardt,1999; Ratnam et al.,2002).
The observation that Dnmt1o is localized to the cytoplasm during early mouse development is interesting and has been reported by several independent laboratories (Carlson et al.,1992; Mertineit et al.,1998; Cardoso and Leonhardt,1999; Howell et al.,2001; Ratnam et al.,2002). However, in all of the reports, the observation has been indirect using immunohistochemistry; it was possible that the Dnmt1o signal in the cytoplasm might be a false-positive due to incomplete wash-out of labeled antibodies. We tried repeatedly, without success, to locate Dnmt1o in mouse and pig embryos using several antibodies that recognize different parts of human Dnmt1 (data not shown); this led us to seek an alternative detection technique. To more directly determine the intracellular localization of Dnmt1o in cleavage-stage embryos, we physically pulled out the nuclei from the blastomeres of 2-cell-stage mouse embryos using an aspiration pipette. Nuclei were collected separately from the remaining parts of the cells (Fig. 4A). Subsequent Western blot analysis showed Dnmt1o protein exclusively in the cytoplasmic fraction (Fig. 4B). No Dnmt1o protein was detected in the nuclear fraction of the 2-cell embryos. We used α-tubulin and Hdac2 as positive controls; as expected, these proteins were detected only in the cytoplasmic and nuclear fractions, respectively. This demonstrated conclusively that Dnmt1o is located in the cytoplasm in early mouse embryos.
We repeated the experiment with 2-cell pig embryos to determine the location of pDnmt1o protein (Fig. 4C). Western blot analysis revealed that pDnmt1o, like mouse Dnmt1o, was only in the cytoplasmic fraction (Fig. 4D). The controls, nuclear lamin C and α-tubulin, were detected exclusively in the nuclear and cytoplasmic fractions, respectively.
To analyze Dnmt1 expression in early cloned embryos, mature pig oocytes were reconstructed by nuclear transfer (NT), activated, and cultured in vitro. As before, nuclear and cytoplasmic fractions were obtained from cloned 2-cell embryos using a micropipette to remove the nuclei, and pDnmt1 expression was analyzed by Western blotting. pDnmt1o was the only isoform present in cloned embryos (Fig. 5A); no pDnmt1s was detected either in the nuclear or the cytoplasmic fractions. pDnmt1o was detected exclusively in the cytoplasmic fraction of cloned 2-cell embryos as in IVF embryos. Besides, the relative levels of pDnmt1o and pDnmt1s transcripts at the 4- to 8-cell stages did not differ greatly between IVF and cloned embryos if they were corrected for the levels of reference transcripts (β-actin and H2A.Z) (Fig. 5B). As shown in Figure 5C, the enucleation procedure itself slightly reduced the levels of pDnmt1o and H2A.Z transcripts, consistent with the result shown in Figure 5B; mature oocytes (MII) had higher levels of the transcripts than either enucleated oocytes (eN) or oocytes (eC) from which a part of the cytoplasmic content, rather than the nuclear content, had been removed. These data suggest that the recipient oocyte cytoplasm rapidly dominates the expression program of the somatic cell nucleus and re-directs somatic-type Dnmt1 expression to its own pattern fit for early development.
In this study, we identified the oocyte-specific variant of the Dnmt1o protein in early pig embryos, which we termed pDnmt1o. pDnmt1o was similar to mouse Dnmt1o in several ways. First, both Dnmt1o proteins are similar in size, as are their Dnmt1s proteins (Figs. 1A and 2A). Second, both were specifically expressed in oocytes and cleavage-stage embryos (Figs. 2B and 3C). Third, both localized mainly to the cytoplasm, not to the nucleus (Fig. 4). Finally, it appeared that in pigs, as in mice, an oocyte-specific exon 1 is used for transcription initiation (Fig. 3A). The fact that Dnmt1o is so similar in the two species strongly supports the idea that Dnmt1o function is conserved during early mammalian development.
This study provided direct evidence that Dnmt1o proteins are cytoplasmic in oocytes and early embryos; these findings were in agreement with those of other researchers who used immunocytochemical methods. Our approach was to physically separate the nuclei from the remaining part of 2-cell blastomeres (Fig. 4). We detected no Dnmt1o in the nuclear extracts; nevertheless, we cannot completely exclude the possibility that cytoplasmic Dnmt1o proteins are transported into the nucleus and that a very low level of Dnmt1o remains in the nucleus to regulate DNA methylation at imprinted loci (Reik and Walter,2001). A simple estimate from the amount of input proteins and the relative intensities of Dnmt1 bands in the immunoblots between pig eggs and somatic cells revealed that the level of Dnmt1 protein in pig eggs are 4 to 5 orders of magnitude higher than Dnmt1 level in somatic cells (see Monk et al.,1991 for the mouse case). So, a tiny fraction of Dnmt1, which is undetectable by Western blots, may play a role for maintenance of DNA methylation in the nucleus of early embryo. It is worth noting that recent studies reported the presence of Dnmt1s in a very low level in preimplantation-stage mouse embryos (Cirio et al.,2008; Hirasawa et al.,2008; Kurihara et al.,2008) and in the nuclei of the embryos (Cirio et al.,2008; Kurihara et al.,2008). It is difficult to reconcile all of the observations; however, if a mechanism for importing Dnmt1 proteins into the nucleus can be identified, this might explain why exogenously expressed Dnmt1o has sometimes been observed in the nucleus in early mouse embryos (Cardoso and Leonhardt,1999; Margot et al.,2003).
In cloned embryos, Dnmt1s has long been suspected to hamper epigenetic reprogramming processes in the recipient oocyte, thereby generating aberrant methylation states in the donor genome (Kang et al.,2001a,2002). Our observations do not support this model: Western blot analysis of extracts of cloned embryos detected no pDnmt1s protein (Fig. 5A). Furthermore, we found no evidence that the donor genome expressed the pDnmt1s gene after transplantation into the recipient oocyte; RT-PCR showed that the pDnmt1s gene was transcribed at a background level and that pDnmt1s transcript levels in cloned embryos were no greater than in IVF embryos (Fig. 5B). It probably indicates that the Dnmt1s promoter of somatic cell origin no longer expresses its transcripts within the recipient oocyte, suggesting that the donor cell nucleus rapidly switches from the somatic form to the oocyte version. Additionally, the relative levels of pDnmt1o transcripts were not different between IVF and cloned embryos when compared with the levels of reference transcripts such as H2A.Z and β-actin. These results suggest that cloned pig embryos carrying a somatic donor genome have a Dnmt1s and pDnmt1o expression profile that is similar to embryos that are fertilized normally.
Using an immunocytochemical approach, we recently showed that mouse and pig embryos have distinct DNA methylation reprogramming patterns (Jeong et al.,2007a,b; Park et al.,2007). Mouse embryos undergo both active and passive DNA demethylation events during preimplantation-stage development, whereas pig embryos do not. Knowing that there was a difference between mice and pigs in terms of methylation reprogramming, we thought it likely that we would find species differences in Dnmt1 levels or localization in early embryos. On the contrary, our findings indicate that early mouse and pig embryos both have an oocyte-specific variant of Dnmt1 that is localized to the cytoplasm. Our results seem to indicate that the cytoplasmic localization of Dnmt1o itself is not the immediate cause of passive DNA demethylation in mouse embryos or responsible for global DNA methylation in pig embryos. A role for Dnmt1o in DNA methylation regulation has been posited based on the observation that Dnmt1o transiently traffics into the nuclei of 8-cell embryos; however, this observation has recently been called into question (Hirasawa et al.,2008). It is not easy to envision a role for cytoplasmic Dnmt1o, but its retention in the cytoplasm may reflect the need to prevent it from maintaining global methylation during early mammalian development.
In summary, we identified an oocyte-specific porcine Dnmt1o protein in preimplantation stages that differed in size from somatic pDnmt1s. Using 5′-RACE, we found that pDnmt1o transcripts in oocytes had a 47-bp 5′-end sequence distinct from pDnmt1s transcripts. Through micromanipulation of 2-cell embryos, we demonstrated that the oocyte-specific Dnmt1o proteins are localized to the cytoplasm in mouse and pig embryos as well as in cloned pig embryos. Cloned pig embryos contain mainly pDnmt1o as the major form of pDnmt1; pDnmt1s was not detected. Our findings strongly suggest that the existence of oocyte-specific Dnmt1o and cytoplasmic retention of Dnmt1o during early development are conserved in mammals.
Pig Oocyte Collection and In Vitro Culture
Experiments were conducted according to The Animal Care and Use Committee guidelines of National Livestock Research Institute, Korea. The in vitro fertilization of mature pig oocyte was performed as reported previously (Jeong et al.,2007a,b). Briefly, pig oocytes were collected from ovaries of slaughtered gilts. The oocytes were cultured in maturation medium with 10 IU/ml hCG and 10 IU/ml PMSG for 22 hours (hr) and then without hCG and PMSG for 22 hr at 38°C in a humidified atmosphere of 5% CO2 in air. Mature oocytes were fertilized in vitro with pig semen supplied by the Darby Pig Artificial Insemination Center (Anseong, Korea). Fertilized oocytes were cultured at 38°C in humidified atmosphere of 5% CO2 in air and, if needed, collected according to standard procedures on optical stages. In vitro maturation medium was BSA-free North Carolina State University (NCSU) 23 medium supplemented with 10% (v/v) porcine follicular fluid, 0.57 mM cystein, 25 μg/ml gentamycin, 10 ng/ml EGF, 10 IU/ml PMSG, and 10 IU/ml hCG. Medium for in vitro fertilization, designated modified Tris-buffered medium (mTBM), consisted of 113.1 mM NaCl, 3 mM KCl, 7.5 mM CaCl2-2H2O, 20 mM Tris (crystallized-free base; Fisher Scientific), 11 mM glucose, 5 mM sodium pyruvate, and no antibiotics. Medium for in vitro culture was NCSU23 supplemented with 4 mg/ml BSA (fatty acid free).
Mouse Embryo Collection and In Vitro Culture
Mouse oocytes and early embryos were collected from superovulated BCF1 (C57BL/6 × CBA/CA) females as described previously (Yeo et al.,2005; Jeong et al.,2007b). Briefly, female BCF1 mice at 5 weeks of age were injected with 5 IU of pregnant mare's serum gonadotrophin (PMSG), followed by 5 IU of human chorionic gonadotropins (hCG) 48 hr apart and mated with male mice. Successful mating was determined at the following morning by detection of a vaginal plug. Eighteen to 20 hr after hCG injection, mouse zygotes were collected from mouse oviduct and transferred to M2 medium (Sigma Aldrich) containing 0.1% (w/v) hyaluronidase to remove cumulus cells. Embryos were cultured into M16 medium (Sigma Aldrich) at 37°C, 5% CO2 in air and, if needed, collected according to standard procedures on optical stages.
Cell Culture, Transfection, and Immunocytochemistry
NIH3T3 fibroblasts were cultured in Dulbecco's modified Eagle medium (DMEM, Gibco BRL) with 10% heat inactivated fetal bovine serum, 50 μg/ml of streptomycin, and 50 μg/ml of penicillin. Culture was maintained in a humidified atmosphere with 5% CO2 at 37°C. Pig primary fibroblasts were obtained as described before (Kang et al.,2001b). For transfection, fibroblast cells were cultured on a gelatin-coated coverslip to about 70% confluency. Lipofectamine Plus™ Reagent (Invitrogen) was used to transport DsRed2-hDnmt1 expression plasmid (Kim et al.,2002) to cultured cells according to the manufacturer's instructions. Forty-eight hours after transfection, cells were fixed before mounting to examine DsRed2 signals.
For immunostaining cultured cells for endogenous Dnmt1 proteins, pig primary fibroblasts were fixed in PBS containing 4% formaldehyde for 15 min at RT followed by three times of 10-min washes in PBS-0.5% Tween-20 (Fisher). Fixed cells were permeabilized by 0.5% Triton X-100 (MP Biomedicals) for 15 min at RT and blocked for 1 hr at RT in PBS-2% BSA (Sigma Aldrich). Anti-Dnmt1 antibody (1:300, Santacruz) incubation was carried out in the blocking solution for 1.5 hr at 37°C, followed by several washes in PBS-0.5% Tween-20. Alexa Fluor 594-conjugated secondary antibody (1:300, Molecular Probes) was incubated for 1 hr at RT. After several washes, the sample was mounted on slide glass (Marienfeld) with a mounting media containing DAPI (Vectashield). Samples were observed with Karl Zeiss Axiovert 200M fluorescence microscope equipped with Apotome apparatus. Images were captured digitally using different filter sets and merged using Axiovision (v4.5) or Adobe Photoshop software (v7.0).
Somatic Cell Nuclear Transfer and Two-Cell Enucleation
Overall procedures for nuclear transfer including in vitro maturation, enucleation of pig oocytes, electrical fusion, and activation have been described previously (Koo et al.,2000,2001; Kang et al.,2001b). Ear skin fibroblasts were used as donor cells in somatic cell nuclear transfer. For enucleation of two-cell blastomeres, two-cell mouse embryos and two-cell porcine embryos were micromanipulated in M2 medium containing 7.5 μg/ml cytochalasin B (Sigma Aldrich), as described before (Koo et al.,2001).
Western Blot Analysis
Porcine and bovine follicle cells were obtained from ovarian follicles through aspiration using syringe needles and cultured for 48 hr before collection for protein analysis. These follicle cells and NIH3T3 cells were lysed in a buffer containing 0.5% NP-40, 50 mM Tris-Cl (pH 8.0), 10% glycerol, 0.1 M EDTA (pH 8.0), and 15 mM NaCl and boiled in SDS loading buffer for 5 min, separated on 6% SDS polyacrylamide gel, and blotted onto a nitrocellulose membrane (BioRad). For protein analysis of pig GV oocytes, cumulus-oocyte complexes were treated with hyaluronidase to completely remove cumulus cells attaching to zona pellucidae before protein collection. Membranes were blocked with 5% (w/v) skim milk in PBS for 1 hr at RT. Anti-Dnmt1 (1:1,000, Santacruz), anti-Tubulin (1:5,000, Sigma), anti- Hdac2 (1:1,000, Santacruz), anti-Lamin A/C (1:1,000, Santacruz) antibodies were incubated overnight at 4°C. The horseradish peroxidase (HRP)-conjugated secondary antibodies was added for 1 hr at RT. Immunoreactivity was detected using the BM Chemiluminescence Blotting Substrate Kit (Roche).
5′-RACE experiment was performed using total RNA isolated from porcine GV-stage oocytes. Porcine oocytes were lysed to extract total RNA using RNeasy mini kit (Qiagen) after removal of cumulus cells. First-strand cDNA synthesis was performed using 5′/3′ RACE kit (Roche) according to the manufacturer's instruction. Primary anti-sense primer for first-strand synthesis of porcine Dnmt1 cDNA was 5′-GACCTTTAGGTTGACTTCTGTGC-3′ corresponding to the coding region from +910 to +888 downstream from the translation start site (Accession no: NM_001032355). After poly (A) tailing to the first-strand cDNA, first-round PCR was performed using oligo dT-anchor primer provided by the manufacturer and porcine Dnmt1-specific primer 2, 5′-ATGGTGGTCTGCCTGGTAGTTTGC-3′, corresponding to the coding region from +548 to +525. The same oligo dT-anchor primer was used for the next PCR. First PCR products were taken as templates for second-round PCR using anchor primer and nested Dnmt1-specific primer 3, 5′-ACTAGATGAGACTTCGGCTGA-3′ corresponding to the coding region from +510 to +490. Another set of primers for third-round PCR was the anchor primer and nested to specific primer 4, 5′-TGACTTTAGCCAGGTAGCG-3′, corresponding to the coding region from +292 to +274. PCR was performed using pfu DNA polymerase (SolGent). The same protocol was used for first-, second-, and third-round PCRs, except for annealing temperatures. First-round PCR condition was 35 cycles of 95°C for 30 s, 58°C for 30 s, 72°C for 1 min 30 s, and 72°C for 5 min. Annealing temperatures for the second- and the third-round PCRs were 60° and 55°C, respectively. Finally, PCR products were introduced into pGEM-T easy vector (Promega) and sequenced.
Total RNAs were isolated from oocytes and embryos using RNeasy mini kit (Qiagen). First-strand cDNAs were synthesized with SuperScript™ III first-strand synthesis system (Invitrogen) according to the manufacturer's instruction. Primers with which we used to discriminate porcine Dnmt1 variants were 5′-TGACTTTAGCCAGGTAGCG-3′ as a common anti-sense primer, and 5′-CGCTCGGGAGCTGACGTG-3′ and 5′-GGGCGTTCTCACTGCCTGA -3′ as sense primers specific to oocyte- and somatic-form Dnmt1 cDNA respectively. Primers for H2A.Z were 5′-TGTTTGAGCTTCAGCAGAATTCGA-3′ and 5′-GGCATCCTTTAGACAGTCTTCTG-3′. Primers for β-actin were 5′-ACATCAAGGAGAAGCCTCACG-3′ and 5′-AGGGGCGATGATCTTGATCTTCA-3′. PCR conditions were 35 cycles of 95°C for 30 s, 60°C for 30 s, 72°C for 40 s, and 72°C for 5 min. The PCR products were analyzed on a 2% agarose gel.
We thank Prof. G. D. Kim for DsRed2-Dnmt1 expression plasmid.