Epigenetic reprogramming during the preimplantation stage erases differentiation-associated epigenetic information from the gamete genome that has been accumulated during gametogenesis, and ultimately brings a pluripotent epigenetic feature to the resulting embryo (Reik et al.,2001; Rideout et al.,2001; Li,2002; Kang et al.,2003). Despite its importance during early development (Reik and Walter,2001), the patterns and mechanisms of epigenetic reprogramming do not seem to be conserved among mammalian species. A typical instance of species-specific reprogramming is seen in DNA methylation reprogramming. Pre-existing 5-methylcytosines (5-MeCs) of the incoming sperm genome are removed in the oocyte cytoplasm in mice (Mayer et al.,2000; Oswald et al.,2000; Santos et al.,2002). Active demethylation of paternally derived genome has been asserted to occur by showing asymmetric methylation states between the male and the female pronuclei in rat, cow, and pig zygotes (Dean et al.,2001). However, subsequent studies on zygotes of other species such as sheep (Beaujean et al.,2004a) and rabbit (Beaujean et al.,2004a; Shi et al.,2004) have reported sustained DNA methylation in sperm-derived pronuclei without active demethylation events in zygotes. This finding highlights the problems of extrapolation of methylation reprogramming patterns among mammalian species and raises questions about the underlying mechanism of DNA methylation reprogramming and its connection with other epigenetic systems, as well as its obligate requirement for early mammalian development. The nonuniform nature of DNA methylation reprogramming is likely to be associated with nonuniformity of other epigenetic modulations that interact with DNA methylation systems. To date, however, no supporting evidence has been reported.
Here, we report additional evidence supporting the hypothesis that epigenetic reprogramming is not conserved during early mammalian development. Our findings show that trimethylated H3-K9 (H3-m3K9), which is absent from the mouse male pronucleus (Liu et al.,2004; Santos et al.,2005; Yeo et al.,2005), is detected in the pig male pronucleus. We also report how the genome-wide H3-m3K9 pattern is created in the sperm-derived chromatins after protamine replacement by histones. We believe that the distinct epigenetic characteristics of pig zygotes could be an additional mammalian model of epigenetic reprogramming that is markedly different to the mouse model. Our findings also contribute to the understanding of the transitional process of chromatin configuration from a gamete type to a zygote type.
Appearance of H3-K9 Methylation in Pig, but Not in Mouse, Sperm-Derived Chromatins
In mammals, shortly after fertilization, the sperm genome undergoes chromatin remodeling and sperm protamine molecules are replaced by histones stored in the oocyte cytoplasm (McLay and Clarke,2003). At this time, the lysine 9 residues of histone H3 molecules recently incorporated into nucleosomes are unmethylated. In agreement with this finding, the maternal and the paternal pronuclei in the mouse zygote have asymmetric H3-m3K9 methylation states (Liu et al.,2004; Santos et al.,2005; Yeo et al.,2005; Fig. 1Aa). By contrast, however, we observed that, in pig zygotes, the male pronucleus contains a substantial level of trimethylated H3-K9 (H3-m3K9) signals, thereby balancing the methylation state of the female pronucleus (Fig. 1Ab).
Paternally Derived Cytosine Methylation Was Not Actively Demethylated but Preserved During Pronucleus Development
The close relationship between H3-K9 methylation and DNA methylation has been observed in diverse biological systems (Tamaru and Selker,2001; Fahrner et al.,2002; Jackson et al.,2002; Fuks et al.,2003; Lehnertz et al.,2003; Tariq et al.,2003; Freitag et al.,2004). Therefore, the different H3-m3K9 states in the two species prompted us to investigate DNA methylation states in sperm-derived chromatins. In contrast to the mouse male pronucleus, in which 5-methylcytosine (5-MeC) levels were markedly reduced to undetectable levels by active demethylation (Fig. 1Bc), the pig male pronucleus was shown to have considerable levels of 5-MeCs that seemed to be almost equivalent to levels in the female pronucleus (Fig. 1Ba). Similar results were observed with polyspermic eggs with multiple sperm-derived pronuclei (Fig. 1Bb) or with in vivo-derived eggs (data not shown). In this way, we were unable to find evidence supporting a genome-wide loss of DNA methylation, or “active-mode demethylation,” in the pig male pronucleus. The observations in Figure 1B indicate a close correlation between H3-K9 trimethylation and DNA methylation in the male pronucleus. In addition, dimethylated H3-K9 (H3-m2K9), which is also asymmetrically distributed between parental pronuclei in mice (Arney et al.,2002; Liu et al.,2004; Santos et al.,2005; Yeo et al.,2005), was present at equivalent levels in both parental sets of chromatins (Fig. 1C). Figure 1D summarizes the species difference in the paternal epigenetic status between the pig and the mouse fertilized oocytes.
Specific Epigenetic Interplay Between DNA Methylation and H3-m3K9 in Pig Paternal Pronucleus
We then investigated whether the differences in histone methylation levels between the two species were confined to H3-m3K9 or were also found for other histone methylation types. Pronucleus-stage eggs were immunostained with various antibodies that can specifically detect di- or trimethylation at certain lysine residues of histone H3 and H4. The resulting histone methylation profile led to the distinctive pig H3-m3K9 pattern (Fig. 2). In both species, methylated H3-K4 was present at equivalent levels in both parental pronuclei, whereas H3-K27, H3-K36, and H4-K20 methylations were restricted, or heavily biased, to the female pronucleus. Thus, in both species, the histone methylation profile was identical for all histones except for H3-m3K9. This result highlights that H3-m3K9, together with DNA methylation, is a distinguishing epigenetic mark among different histone modifications. It provides additional evidence to support the possibility that H3-K9 methylation is associated with DNA methylation during the pronucleus period of mammalian development.
Gradual Establishment of Paternal H3-m3K9 Pattern During the Pronucleus Stage
We examined how the H3-K9 methylation pattern is created on the naked sperm chromatins. Pig zygotes at an earlier pronucleus stage were collected and stained for acetylated H3-K9 (H3-acK9) as a control. The genomic material of mature oocytes, including the first polar body and the metaphase II (MII) -arrested chromosomes, were not stained for H3-acK9. However, 6 hr after fertilization, H3-acK9 signals could be visualized in both parental pronuclei (Fig. 3A).
In contrast to H3-acK9, H3-m3K9 marks were found at high levels in the maternal chromosomes of unfertilized oocytes (Fig. 3B, MII). Interestingly, paternal H3-m3K9 signals developed in a gradual manner: a few spots of H3-m3K9 signal appeared on decondensed male chromatins at approximately 6 hr postfertilization (6 hpf, male pronucleus [♂PN]). At approximately 10 hpf, a diffuse signal was present in the paternal pronucleus. However, the paternal H3-m3K9 signal at this time was markedly weaker than the maternal one.
We examined staining patterns of H3-m3K9 and 5-MeC in parallel at given time points to investigate how the epigenetic marks are functionally related in the pig male pronucleus. Paternal cytosine methylation remained constant at a level equivalent to that of the maternal counterpart while the H3-m3K9 pattern was being established in the male pronucleus (Fig. 3C). The H3-m3K9 spots all developed from chromosomal regions with 5-MeC-dense signals (6 hpf, ♂PN). The spots, although weak, began to spread outward along the contour of the 5-MeC-rich signals (10 hpf). Since then, the paternal H3-m3K9 level rapidly increased, and, at approximately 20 hpf, the paternal H3-m3K9 pattern became almost indistinguishable from that of the maternal pronucleus (see also Fig. 1B). Taken together, our results demonstrate that accumulation of paternal H3-m3K is a progressive process, and ends in coating of most of the male chromosomes before the first mitosis, as shown in Figure 1Ab in which the 4′,6-diamidino-2-phenylindole (DAPI) signal of the paternal pronucleus is completely matched by the H3-m3K9 signal.
H3-K9 Trimethylation Is Independent of the DNA Replication Process
Epigenetic states, once established, are transferred to daughter strands by a process associated with DNA replication in somatic cells. We investigated whether establishment of the paternal H3-m3K9 pattern was also associated with progression of DNA replication. Bromodeoxyuridine (BrdU) incorporation was first, and most actively, detected from the male pronucleus at approximately 8 hpf (Fig. 4Aa,Ba), but very little incorporation could be detected by approximately 20 hpf in pig (data not shown). Although the male pronucleus contained many BrdU signals, interestingly, there was no detectable H3-m3K9 signal (Fig. 4Aa). Even when H3-m3K9 spots were visible in the male pronucleus, they were not found at the sites of DNA replication (Fig. 4Ab). This result indicates that BrdU incorporation during DNA replication and trimethylation of H3-K9 residues occurs separately in the pig male pronucleus. However, when the BrdU-treated eggs were stained for H3-acK9 as a control (Fig. 4B), we observed H3-acK9 signals in both parental pronuclei in the absence of BrdU (data not shown). Figure 4Ba shows acetylated female chromatins before the start of DNA replication. Thus, it could be assumed that H3-K9 acetylation begins shortly after fertilization and almost ends before the start of DNA replication. Figure 4C schematically represents the order of individual H3-K9 modification events compared with the DNA replication process during pronucleus development. In summary, there were several important findings regarding epigenetic reprogramming in pig early development. First, H3-m3K9, which is normally absent from mouse sperm-derived pronucleus, was detected in the pig male pronucleus. Second, the de novo establishment of H3-m3K9 pattern is likely to be associated with the preserved DNA methylation states of the pig male pronucleus. Third, the pattern of H3-m3K9 was established in a gradual manner and was not coupled to the DNA replication process.
We found that the level of H3-m3K9 in pig sperm-derived chromatins increases markedly during the pronucleus stage (Fig. 1), which is in contrast to mouse-derived sperm chromatins, which are not modified until the four-cell stage (Liu et al.,2004; Yeo et al.,2005). The distinctive feature of H3-K9 trimethylation is supported by the histone methylation profile in Figure 2; all histone lysine residues examined, except H3-K9, were equally unmodified or minimally methylated in both pig and mouse male pronuclei. The species difference in development of the paternal H3-K9 methylation pattern might simply be due to the activity of certain histone methyltransferases (HMTases) in the oocyte cytoplasm. Various H3-K9-specific HMTases have been characterized to date, including SUVAR39 (Rea et al.,2000; Nakayama et al.,2001), G9A (Tachibana et al.,2002), GLP/Eu-HMTase (Ogawa et al.,2002), SETDB1 (Schultz et al.,2002; Yang et al.,2002; Dodge et al.,2004), and SETDB2 (Mabuchi et al.,2001). Currently, however, it is unknown which activity, or combination of activities, is responsible for development of the paternal H3-m3K9 pattern in the pig and whether the activity responsible is absent from, or inhibited in, the mouse fertilized oocyte.
The initiation process that creates H3-m3K9 speckles on recently remodeled male chromatins might be enzymatically distinct from the propagation/enhancement step that spreads the signals into neighboring regions according to pre-existing H3-m3K9 marks. Therefore, individual steps of H3-m3K9 installment may require different H3-K9 methylation machineries. Of particular interest is the identity of the HMTase responsible for the formation of pronounced H3-m3K9 speckles. This de novo HMTase complex is likely to act on the male chromatins shortly after fertilization, at which time the chromatin structure targeted by the complex may be in a highly decondensed state after chromatin remodeling (Fig. 3B, DNA, ♂PN). Therefore, it might be less likely that the “speckle”-forming HMTase recognizes a certain chromatin structure to seed initial H3-m3K9 signals. Instead, regional DNA methylation signals in the male chromatins may provide a platform from which H3-m3K9 speckles can “bud off,” and this possibility is supported by the observation that H3-m3K9 speckles spread from heavily methylated DNA regions (Fig. 3C, see below). This hypothesis assumes the presence of a factor in the HMTase complex that links the two different epigenetic systems: DNA methylation and H3-K9 trimethylation. Indeed, in support of this hypothesis, of the HMTases specific to H3-K9 residues, SETDB1 and SETDB2 have a methylated DNA-binding domain (MBD). The MBD has been shown to bind 5-MeC–rich DNA regions, but it is not yet known whether the MBDs of SETDB1 and SETDB2 bind methylated DNA to allow 5-MeC–mediated H3-K9 methylation. Nevertheless, our previous genetic study, in which we demonstrated an early lethal phenotype of a setdb1-null mutant and its inability to form embryonic stem cells (Dodge et al.,2004), indicates an important epigenetic role for SETDB1 in early development. At present, the function of SETDB2 is largely unknown.
Recent studies have demonstrated interactions of SETDB1 with DNA-methylation–associated proteins such as MBD1 (Sarraf and Stancheva,2004) and DNMT3A (Li et al.,2006). In particular, the MBD1-SETDB1 complex has been shown to contribute to maintenance of the pre-existing H3-K9 methylation pattern in a replication-coupled manner in differentiated cells (Sarraf and Stancheva,2004). However, in the pig male pronucleus, DNA replication starts before the H3-m3K9 trimethylation signal is detected and appearance of the early H3-m3K9 speckles do not coincide with appearance of the BrdU spots (Fig. 4). Therefore, it is unlikely that establishment of the H3-m3K9 pattern is mediated by DNA replication processes in the male pronucleus. Instead, our results indicate that de novo H3-K9 methylation in the pig male pronucleus occurs independently of DNA replication. This finding leads to speculation that pig zygotes contain an HMTase complex different from that previously reported to be involved in maintenance of the H3-K9 methylation through a DNA-replication–coupled process in differentiated cells (Sarraf and Stancheva,2004).
Our observation that active demethylation is absent from the pig sperm-derived pronucleus conflicts with previous reports (Dean et al.,2001; Gioia et al.,2005; Fulka et al.,2006). The reason for this discrepancy is unknown. To confirm our observation, we performed several detailed DNA methylation analyses with many pig zygotes at different time points (100–300 zygotes for each time point of 6, 10, 18, 20, and 25 hpf; Y.-K. Kang, personal communication). We consistently found either a substantial amount of 5-MeCs in the male pronucleus or a balanced DNA methylation state between the parental pronuclei from the pig zygotes at different pronucleus stages (Fig. 3C). In addition, our DNA methylation analyses using bisulfite mutagenesis technology indicate that the methylation state of a human single-copy sequence delivered by transgenic pig spermatozoa is maintained during pronucleus development (unpublished observation; Y.-K. Kang, personal communication). Therefore, we are confident that, in pigs, paternally derived DNA maintains its cytosine methylation state during pronucleus development.
It is well known that incoming sperm chromatins are required to erase differentiation-associated epigenetic traces and establish a zygotic epigenetic state to ensure proper development (Reik and Walter,2001). However, mammalian species seem to have adopted different strategies for modulating the DNA methylation and H3-K9 methylation states, as shown in Figure 1. It is conceivable that these different mechanisms to reprogram the incoming sperm chromosomes lead to different epigenetic outcomes, at least temporarily. At present, however, it is difficult to predict the epigenetic consequences and the developmental significance of the species-specific responses of the oocyte cytoplasm to an incoming sperm genome. However, it should be noted that the species-specific characteristic of de novo H3-K9 methylation is linked to the species-specific outcome of DNA-methylation reprogramming (Fig. 1), indicating functional crosstalk between the two different epigenetic systems in reprogramming of sperm chromatins. A possible communication mechanism between these processes could involve the heavy 5-MeC signals on some regions of the male chromatins triggering H3-m3K9 marks, which would also stabilize regional DNA methylation states, ultimately causing spread of the initial H3-m3K9 speckles over all chromatins. Recently, Kishigami et al. reported that, in mice, when a round spermatid was used for micro-insemination, the paternal DNA underwent some demethylation initially but then became re-methylated before the first round of mitosis (Kishigami et al.,2006). It was also shown that the male pronucleus uniquely contains a H3-m3K9 signal of spermatid origin, so the authors suggested that the presence of H3-m3K9 inherited from the spermatid may protect the male chromatins from active DNA demethylation. This observation supports our hypothesis. However, the underlying mechanisms of early epigenetic (re) programming can be determined only through extensive analyses with diverse animal species and different epigenetic elements. In conclusion, we provide evidence to support the hypothesis that epigenetic reprogramming is not conserved among mammalian species, through the epigenetic analyses of pig zygotes for H3-m3K9 states along with DNA methylation states. The distinct epigenetic characteristics of pig zygotes might be worth considering as an additional type of mammalian epigenetic reprogramming that is comparable to the mouse model.
Production of Pig and Mouse Embryos
Experiments were conducted according to The Animal Care and Use Committee guidelines of National Livestock Research Institute, Korea. The in vitro fertilization and cultivation of matured pig oocyte was performed as reported previously (Koo et al.,2000). Especially, polyspermy was induced by using spermatozoa up to a concentration of 1 × 106 sperm/ml. Oocytes were incubated with spermatozoa for 6 hr at 39°C in an atmosphere of 5% CO2 in air. Fertilized mouse eggs were obtained as described before (Yeo et al.,2005). Briefly, mouse strain we used was FVB, and after mating, embryos were removed at an appropriate time according to the standard procedure (Hogan,1994). Timing postfertilization was judged by nuclear morphology and relative distance between parental pronuclei (Hogan,1994).
Antibodies and Immunostaining
A monoclonal anti–5-methylcytosine antibody was purchased from Eurogentech (MMS-900S-B; Mayer et al.,2000). Antibodies that specifically recognize di-/tri-methylated H3-K9, trimethylated H3-K4, trimethylated H3-K27, dimethylated H3-K36, trimethylated H4-K20, or acetylated H3-K9 residues were all purchased from Upstate Biotechnologies. Anti-mouse or anti-rabbit goat secondary antibodies (Alexa 488/594-conjugated form, Molecular Probes) were used to detect the individual primary antibodies. Immunostaining procedure appeared elsewhere (Beaujean et al.,2004b). Briefly, embryos were fixed in 4% formaldehyde in phosphate buffered saline (PBS) for 30 min at 4°C, followed by three 20-min washes in 0.02% Twin 20 in PBS and permeabilized by 0.25% Triton X-100 in PBS for 1 hr at room temperature. The embryos were treated with 4 N HCl for 30 min at room temperature, neutralized with 0.1 M Tris-HCl (pH 8.0), and blocked for 1 hr at room temperature in 2% bovine serum albumin, 0.02% Tween 20 in PBS. Primary antibody incubations (diluted by 1:50–500) were carried out in the blocking solution for 1 hr at 37°C, followed by several washes in 0.02% Twin 20 in PBS. Secondary antibodies (diluted to 1:300–500) were incubated for 30 min at room temperature, followed by several washes. Embryos were mounted on Poly-prep slides (Sigma) and, after dry at room temperature, observed with Karl Zeiss Axiovert 200M fluorescence microscope. Most immunostaining experiments were repeated at least three times each with ∼20 for mouse oocytes/zygotes or ∼50 for pig ones. In case that a uniform figure was observed in more than 70% of immunostained samples, we then regarded it as a pattern. Images were captured digitally using different filter sets and merged using Axiovision (v4.5) or Adobe Photoshop software (v7.0).
BrdU Incorporation Assay
Pig oocytes were treated with 10 μM 5-bromo-2′-deoxy-uridine (5-BrdU, Roche) for 30 min at different time points such as 6, 10, or 20 hpf. After 5-BrdU treatment, the embryos were immediately fixed in 4% formaldehyde solution (Sigma). The subsequent staining procedure was the same with that used for staining 5-methylcytosine. Mouse anti-BrdU antibody was purchased from BD Pharmingen and diluted to 1:50 immediately before use.
We thank Dr. S. Jeong for careful reading of the manuscript.