Main

Maternal-genome chromatin bears considerable DNA methylation and contains H3K9me2 (ref. 10). Conversely, little DNA methylation remains and no H3K9me2 is present in the chromatin of the paternal genome. We found that DNA methylation of naked DNA and chromatin DNA did not lead to substantial differences in the binding of PGC7 to DNA (Supplementary Figs 1, 2a and 3 and Supplementary Discussion). By contrast, compared with the nucleosomes purified from wild-type embryonic stem (ES) cells, PGC7 showed significantly weaker binding to the nucleosomes purified from ES cells lacking G9a (also known as Ehmt2)—an H3K9me2-specific lysine methyltransferase—in which H3K9me2 was absent, without affecting DNA methylation status (Fig. 1a, b and Supplementary Figs 2b and 4a, b)11. However, PGC7 bound to nucleosomes purified from rescued G9a knockout (G9a−/−) ES cells, which expressed the short or long form of G9a (G9a−/−-G9a(S) and G9a−/−-G9a(L), respectively), with comparable affinity to the nucleosomes of wild-type ES cells (Fig. 1a, b). Next, we conducted a chromatin-immunoprecipitation (ChIP) assay in PGC7-null and PGC7-expressing ES cells using an anti-PGC7 antibody. As shown in Fig. 1c, nucleosomes that were bound by PGC7 specifically contained H3K9me2. H3 with other methylation modifications, however, did not show binding to PGC7. We confirmed the H3K9me2-dependent PGC7 association with chromatin in wild-type, DNA methyltransferase triple-knockout (Dnmt1−/− Dnmt3a−/− Dnmt3b−/−) mutants12 and G9a-null ES cells using a stepwise salt-extraction method13. PGC7 and H3 were similarly extracted from the nuclear pellets of wild-type, Dnmt1−/− Dnmt3a−/− Dnmt3b−/−and G9a (G9a−/−-G9a(S) ES cells (Fig. 1d and Supplementary Fig. 5a, b). By contrast, PGC7 was extracted from nuclear pellets of G9a−/− ES cells under a lower NaCl concentration compared with that in wild-type and Dnmt1−/− Dnmt3a−/− Dnmt3b−/−cells (Fig. 1d and Supplementary Fig. 5a, b), showing that PGC7 binding to chromatin without H3K9me2 was weaker. These two experiments indicate that the association between PGC7 and chromatin is dependent on the presence of H3K9me2.

Figure 1: Preferential binding of PGC7 to H3K9me2-marked chromatin.
figure 1

a, b, Electrophoretic gel-mobility shift assay of His–PGC7. Nucleosomes (approximately 10 μg) purified from wild-type (WT) and G9a−/−ES cells, and G9a−/− ES cells in which G9a−/−-G9a(S) and G9a−/−-G9a(L) forms of G9a were expressed were incubated with increasing concentrations of His–PGC7 (2.5, 5, 10 and 20 μg). The binding mixtures were analysed on agarose gels (a). Ratios of high-molecular-weight (MW) PGC7–nucleosome complex (>5 kb) were determined using NIH Image J (b). Error bars indicate s.d. (n = 3). Binding affinities were significantly different between wild-type and G9a−/− ES cells (*P < 0.005, t-test). His–glutathione S-transferase (GST; 20 μg) was used as a negative control in both experiments. Similar results were obtained in at least three independent experiments and representative results are shown. c, Histone-methylation status of PGC7-containing chromatin. Anti-PGC7 antibody was used to immunoprecipitate chromatin from PGC7−/− ES cells stably expressing PGC7. Immunoprecipitates (IP) and aliquots of the input protein were analysed by immunoblotting with antibodies against various histone modifications. Essentially the same results were obtained in two independent experiments. d, Chromatin-binding status of PGC7 in ES cells. Nuclei were isolated from stable-PGC7-expressing wild-type, Dnmt1−/− Dnmt3a−/− Dnmt3b−/−triple knockout (Dnmt TKO), G9a−/− or G9a−/−-G9a(S) ES cells and were treated with DNase I under various concentrations of NaCl (100, 200, 300, 400 and 500 mM) to separate nuclear extract from nuclear debris. Equivalent amount of aliquots were analysed by immunoblotting with anti-PGC7 or anti-H3 antibodies. e, In vitro assay of the binding of recombinant PGC7 to various H3 tail peptides. Recombinant His–PGC7 and histone H3 tail peptides with various methylation modifications were used in an in vitro peptide-binding assay. Binding characteristics were analysed by immunoblotting with anti-PGC7 antibody. Similar results were obtained in three independent experiments. f, PGC7 competitive peptide-binding assay. N-terminal biotinylated H3K9me2 peptide was immobilized on streptavidin–sepharose beads and incubated with recombinant His–PGC7 in the presence of increasing amounts of unmodified H3 or the indicated methylated H3 peptides. Binding characteristics were analysed by immunoblotting with anti-PGC7 antibody. Similar results were obtained in two independent experiments. g, Concomitant binding of PGC7 to the chromatin loci marked with H3K9me2 in ES cells. A ChIP analysis using the indicated antibodies was conducted in the PGC7−/− and G9a−/− ES cells with or without enforced PGC7 expression. The percentages of each PCR product in the immunoprecipitated sample per those of the input samples are shown (mean and s.d., n = 3). Anti-mouse or -rabbit IgG was used for the negative controls and the signals per total input of the negative controls were <0.12% in all genes examined (data not shown).

PowerPoint slide

Full size image

We studied the in vitro binding of PGC7 to various histone peptides to see whether PGC7 directly binds to H3K9me2. Although all histone peptides containing residues 1–21 bound to PGC7 to some extent, H3K9me2 bound to PGC7 most strongly (Fig. 1e). The same binding assay using truncated versions of PGC7, namely PGC7ΔC (containing amino acids 1–75) and PGC7ΔN (containing amino acids 76–150), showed that PGC7 bound to H3K9me2 by its amino terminus (Supplementary Fig. 6). Next, we carried out a competitive-binding assay to determine the binding strength of PGC7 and H3K9me2. As shown in Fig. 1f, only the excess amount of H3K9me2, but not that of the other peptides, outcompeted the binding of PGC7 to H3K9me2, indicating that the binding of PGC7 to H3K9me2 was stronger than that of other histone modifications.

PGC7 binding was higher at two representative H3K9me2-enriched loci, Magea2 and Wfdc15a, but not at the control Pou5f1 locus (Fig. 1g). By contrast, such enrichment was not observed under the G9a-null condition (Fig. 1g). These data clearly demonstrate that PGC7 specifically binds to the loci marked with H3K9me2 in vivo. Meanwhile, micrococcal nuclease (MNase) activity was inhibited significantly by the enforced expression of PGC7 and PGC7ΔNES, a nuclear export signal deleted mutant of PGC7 (ref. 8), in wild-type and Dnmt1−/− Dnmt3a−/− Dnmt3b−/−ES cells (Supplementary Fig. 5c–f and Supplementary Discussion). However, neither PGC7 nor PGC7ΔNES reduced MNase sensitivity in G9a−/−ES cells, indicating that the protective function of PGC7 required H3K9me2 (Supplementary Fig. 5a, b, g and h). These results indicate that the protective function of PGC7 is dependent on H3K9me2, but not DNA methylation.

We next asked whether the H3K9me2-dependent binding of PGC7 occurs under physiological conditions during early embryogenesis. Both the paternal and maternal pronuclei stained positively with anti-PGC7 antibody after conventional paraformaldehyde (PFA) fixation (PFA–Triton (PT) condition in Fig. 2a, b and Supplementary Fig. 7). However, the staining pattern was completely different when zygotes were treated with Triton X-100 before PFA fixation14 (Triton–PFA (TP) condition in Fig. 2a, b and Supplementary Fig. 7). Under TP conditions, only the maternal pronucleus was labelled with the anti-PGC7 antibody, indicating that PGC7 in the paternal pronucleus was eluted by Triton X-100. In other words, PGC7 was tightly attached to a type of ‘architecture’ that was present in the maternal, but not the paternal, pronuclei.

Figure 2: Protection of the maternal genome from DNA demethylation by PGC7 through H3K9me2-containing chromatin in early embryos.
figure 2

a, Schematic diagram of the two pre-treatment procedures: PT and TP conditions. b, After PT or TP treatment, immunostaining was performed with the indicated antibodies (m, maternal pronuclei; p, paternal pronuclei; pb, polar body). PGC7 and H3K9me2 are shown in red and green, respectively; nuclei were stained with DAPI (blue). A total of 38 and 42 zygotes were stained, under PT and TP conditions, respectively. c, d, H3K9 dimethylation after microinjecting Jhdm2a or Jhdm2a mRNA coding for the H1122A mutation (Jhdm2a (H1122A)). Zygotes were injected with Jhdm2a or Jhdm2a(H1122A) mRNA and cultured for 4.5 h in potassium-enriched simplex optimized medium (KSOM). H3K9 dimethylation was analysed using anti-H3K9me2 antibody (H3K9me2, red; DAPI, blue) (c). A total of 32 and 34 zygotes were injected with Jhdm2a and Jhdm2a(H1122A), respectively. A total of 55 non-injected zygotes were analysed as controls. The strength of the H3K9me2 staining in maternal (M) and paternal (P) pronuclei was analysed (d). e, Analysis of PGC7 localization after microinjecting Jhdm2a or Jhdm2a(H1122A) mRNA. Zygotes were stained after TP fixation. PGC7 and histone H3 are shown in red and green, respectively; nuclei are stained with DAPI (blue). f, g, The methylation status of the parental genome after microinjecting Jhdm2a or Jhdm2a(H1122A) mRNA. Zygotes were injected with Jhdm2a mRNA and cultured for 4.5 h in KSOM. Demethylation of the parental genome was analysed using anti-5mC antibody (5mC, green; DAPI, blue) (f). A total of 33 and 34 zygotes were injected with Jhdm2a and Jhdm2a(H1122A) mRNA, respectively, and 41 non-injected zygotes were analysed as controls. The strength of 5mC staining in the maternal and paternal pronuclei was analysed (g). Scale bar, 20 μm.

PowerPoint slide

Full size image

Considering the maternal pronucleus-specific localization of H3K9me2 (Fig. 2b) and the results of our experiments using ES cells (Fig. 1), we hypothesized that chromatin containing H3K9me2 is the crucial structure to which PGC7 strongly binds in the maternal pronucleus. To test this hypothesis we expressed Jhdm2a (also known as Kdm3a), an H3K9 methylation/dimethylation-specific demethylase15, in zygotes to erase H3K9 dimethylation. This involved injection of Jhdm2a polyadenylated messenger RNA and its inactive mutant, which contains a histidine to alanine point mutation (Jhdm2a(H1122A)) in the JmjC domain and does not possess histone-demethylase function (Supplementary Figs 8, 9 and Supplementary Discussion).

As expected, Jhdm2a expression abolished H3K9 methylation and dimethylation (Fig. 2c, d and Supplementary Fig. 10) but did not affect H3K9 trimethylation (Supplementary Fig. 11). Microinjection of Jhdm2a mRNA abolished PGC7 staining of the maternal pronuclei (Fig. 2e and Supplementary Fig. 12), resulting in the well-correlated PGC7 and H3K9me2 staining patterns under the TP condition (Supplementary Fig. 13). These data clearly indicate that the structure to which PGC7 strongly binds in the maternal pronucleus is H3K9me2-containing chromatin. Next, we examined whether DNA methylation was maintained after Jhdm2a expression. As shown in Fig. 3f, g, DNA methylation was not retained after H3K9me2 demethylation by Jhdm2a. In addition, the H3K9me2 and methylated-cytosine staining intensities were well correlated (Supplementary Fig. 14), and DNA methylation in the maternal pronucleus was not maintained at the pronuclear (PN) 5 stage without PGC7, as reported previously8. Embryonic development was impaired by reducing methylated H3K9, and the accessibility of the antibodies to chromatin is shown in Supplementary Figs 15 and 16 and Supplementary Discussion. Taken together, these observations indicate that PGC7 protects methylated DNA by binding to chromatin regions containing H3K9me2.

Figure 3: Remaining H3K9me2 at the DMRs of two paternally imprinted genes, H19 and Rasgrf1 , in mature sperm.
figure 3

a–c, A ChIP–quantitative (q)PCR analysis of mature sperm was conducted using anti-H3 (a), -H3K9me2 (b) and -H3K4me2 (c) antibodies. H19, Dlk1–Gtl2, Rasgrf1, Peg1, Peg3, Peg5, Nanog, Prkacb and Zp4-ps were analysed by qPCR. The mean and s.d. (n = 3) of the percentage of each PCR product in the immunoprecipitated sample compared with that in the input sample is shown. Anti-mouse or -rabbit IgG was used as a negative control. The signals per total input of the negative controls were <0.16% for all genes examined (data not shown).

PowerPoint slide

Full size image

Next, we confirmed the importance of PGC7 binding to H3K9me2-containing chromatin in a different context. DNA methylation of paternally imprinted genes such as the differentially methylated regions (DMRs) of the Dlk1–Gtl2 domain (where Gtl2 is also known as Meg3), H19 and Rasgrf1 is protected during the active demethylation in normal mouse development16. However, DNA methylation of H19 and Rasgrf1 DMRs was not maintained, whereas that of the Dlk1–Gtl2 DMR was maintained in PN5-stage embryos derived from PGC7-null oocytes. It is reasonable to assume that epigenetic modifications of H19 and Rasgrf1 differ from those of Dlk1–Gtl2, and that epigenetic modifications residing only in the former genes are responsible for the protective function of PGC7. Although the histones in nucleosomes are replaced by protamine during spermiogenesis, a recent epigenomic analysis of human and mouse sperm revealed that histone-containing nucleosomes are preferentially retained at loci of developmental importance, including imprinted gene clusters, homeobox-containing (Hox) gene clusters and Nanog homeobox (Nanog), but not in the intergenic regions of protein kinases, cyclic-AMP-dependent catalytic β (Prkacb), or zona pellucida glycoprotein 4 pseudogene (Zp4-ps)17,18. Thus, we carried out a ChIP analysis of the above-mentioned paternally imprinted genes (Dlk1–Gtl2, H19 and Rasgrf1) and three maternally imprinted genes (paternally expressed genes 1, 3 and 5 (Peg1 (Mest), Peg3, Peg5 (Nnat)), as well as the promoter and intergenic regions of three non-imprinted genes (Nanog, Prkacb and Zp4-ps) in mature mouse sperm.

Histone enrichment was detected at all DMRs of the analysed imprinted genes and at the Nanog promoter region, but not at the Prkacb and Zp4-ps intergenic regions (Fig. 3a) as reported for human sperm. Notably, H3K9me2 was enriched at the H19 and Rasgrf1 DMRs (Fig. 3b), but enrichment was not observed at the DMR of the paternally imprinted domain Dlk1–Gtl2. Considering that the Nanog-promoter region from the paternal genome was demethylated in wild-type zygotes, even in the presence of PGC7 (ref. 19), H3K9me2 should be crucial for the PGC7 protective function. These observations strongly suggest that the remaining H3K9me2 is important for protecting the H19 and Rasgrf1 DMRs against active DNA demethylation after fertilization. It is reasonable to conclude that this protection is mediated by PGC7 binding at the loci containing H3K9me2. H3K4me2 was enriched at all maternally imprinted genes examined but not at the paternally imprinted genes (Fig. 3c and Supplementary Discussion. Taking the dynamics of H3K9me2 into account (see Supplementary Discussion), in the maternal genome and two paternally imprinted loci, the binding of PGC7 to H3K9me2-containing chromatin is critical for the protection of DNA methylation.

We and another group found that the Tet3-mediated conversion of 5mC to 5hmC takes place in the paternal genome but not in the maternal genome9,20,21. We also showed that this conversion occurred in the maternal genome of zygotes derived from PGC7-null oocytes. These observations prompted us to analyse the effects of H3K9me2 on 5hmC status as an analogy to the PGC7-dependent regulation of zygotic 5mC. As shown in Fig. 4a, eradicating H3K9me2 by expressing Jhdm2a induced hydroxymethylation of the maternal genome, and the levels of H3K9me2 exhibited an inverse correlation with those of 5hmC (Supplementary Fig. 17). In addition, RanBP5–mER (RanBP5 fused to a mutated oestrogen receptor) expression, which inhibits PGC7 function by driving the subcellular protein out of the nucleus8 (Supplementary Fig. 18), also induced maternal-genome hydroxymethylation (Fig. 4a). Although it has been reported that acetylation of H3K9 decreases in Jhdm2a−/− testes22, this effect was negligible in our experimental system (Supplementary Fig. 19). Taken together, these results indicate that PGC7 protected 5mC from the Tet3-mediated conversion to 5hmC through binding to H3K9me2-containing chromatin, which is essentially identical to the model in which 5mC is protected from active demethylation.

Figure 4: Protection of the conversion of 5mC to 5hmC by PGC7 through H3K9me2-containing chromatin in early embryos.
figure 4

a, 5mC and 5hmC status after the microinjection of Jhdm2a or RanBP5–mER mRNA. Zygotes were injected with Jhdm2a or RanBP5–mER mRNA and cultured for 4.5 h in KSOM. The 5mC and 5hmC states were analysed using anti-5mC and anti-5hmC antibodies (5mC, green; 5hmC, red; DAPI, blue). Scale bar, 20 μm. b, c, Flag-tagged Tet3 mRNA was injected into zygotes obtained from wild-type or PGC7−/−female mice with or without Jhdm2a or RanBP5–mER mRNA and cultured for 4.5 h in KSOM. After PT or TP treatment as described in the Fig. 2 legend, immunostaining was performed using an anti-Flag antibody. Tet3 is shown in green; nuclei were stained with DAPI (blue). Ten zygotes obtained from wild-type female mice injected with Flag-tagged Tet3 (Flag–Tet3) mRNA were stained with the anti-Flag antibody under PT conditions; 13 zygotes were injected with Flag–Tet3 mRNA alone and 16 zygotes were injected with Tet3 and Jhdm2a mRNA or with Tet3 and RanBP5–mER mRNA. These embryos were stained with anti-Flag antibody under TP conditions. Zygotes obtained from PGC7−/− female mice were injected with Flag–Tet3 mRNA. A total of 6 and 9 zygotes were stained with the anti-Flag antibody under PT and TP conditions, respectively (b) and the percentage of Flag staining in maternal (M) and paternal (P) pronuclei is shown (c). Scale bar, 20 μm. d, Effect of PGC7 on the chromatin binding of Tet3 in ES cells. Nuclei were isolated from PGC7−/− ES cells transfected with full-length PGC7, Flag–Tet3, PGC7 and Flag–Tet3, PGC7ΔC, and both PGC7ΔC and Flag–Tet3. These nuclei were treated with DNase I under various concentrations of NaCl (100, 200, 300, 400 and 500 mM) to separate nuclear extract from nuclear debris. Equivalent amount of aliquots were analysed by immunoblotting with anti-PGC7, anti-Flag and anti-H3 antibodies as described.

PowerPoint slide

Full size image

Tet3 was detectable in both paternal and maternal pronuclei under PT conditions (Fig. 4b, c). However, Tet3 was detected only in paternal pronuclei under TP conditions (Fig. 4b, c), indicating that Tet3 was tightly bound to a structure that existed in the paternal but not the maternal chromatin. Furthermore, Tet3 was detectable in the maternal pronucleus after microinjecting Jhdm2a or RnaBP5–mER mRNA and was also detectable both in maternal and paternal pronuclei of PGC7-null fertilized eggs under TP conditions (Fig. 4b, c and Supplementary Fig. 20). These data clearly indicate that the tight binding of Tet3 to chromatin was inhibited by intranuclear PGC7 through its binding to H3K9me2-containing chromatin. Although we examined direct binding of PGC7 to Tet3, no obvious binding was observed (data not shown).

To examine the effect of PGC7 on Tet3 in more detail, we carried out a stepwise salt-extraction analysis using the PGC7−/− ES cells. As shown in Fig. 4d, PGC7 considerably reduced the binding of Tet3 to chromatin, which was as tight as that of H3 without PGC7. By contrast, PGC7ΔC did not show this effect despite having a chromatin-binding affinity similar to that of full-length PGC7. Moreover, the expression of PGC7ΔC did not inhibit the endonuclease activity in the MNase assay, like Tet3 binding (Supplementary Fig. 21), indicating that the inhibitory effect of PGC7 on Tet3 was not caused by competitive binding. PGC7 inhibited MNase digestion of the linker part of chromatin by binding to H3K9me2, a modified histone tail, as described in the discussion of the MNase assay of G9a−/− ES cells (Supplementary Fig. 5). PGC7 (∼17 kDa) is a relatively small protein compared to the histone octamer (∼100 kDa) and the deleted carboxy-terminal part of PGC7 (∼8 kDa), which is essential for the inhibitory effect on both Tet3 binding and MNase activity. Therefore, it is probably too small to have a marked steric effect. Meanwhile, it is conceivable that the distribution of H3K9me2, through which PGC7 binds to chromatin, is not sufficiently dense. Considering these points, although we cannot exclude it completely, the possibility of a steric effect seems improbable. Therefore, we prefer the hypothesis that PGC7 inhibits the activity of the enzyme(s) acting on DNA, such as Tet3 and MNase, by means of a change in chromatin configuration.

Although 5mC has been the only recognized DNA modification for many years in mammals, several reports have described a novel 5hmC DNA modification2,3. The discovery of 5hmC raises numerous questions, including its function, tissue localization and regulatory mechanisms of modification. The tight differential regulatory patterns of 5mC and 5hmC in paternal and maternal pronuclei imply that 5hmC is important9,20,21. Here, we showed the molecular function of PGC7 during the modification as a first step towards understanding the regulatory mechanisms of 5-hydroxymethylation in DNA. It is noteworthy that regulation of reciprocal DNA methylation and hydroxymethylation was controlled by the same protein, PGC7, through binding to H3K9me2-containing chromatin. Two recent studies have raised the possibility that 5hmC is an intermediate during DNA demethylation and that the Tet family are the critical enzymes for this process23,24. Our conclusion is consistent with this notion, and the inhibition of Tet3 activity through the binding of PGC7 to H3K9me2 would explain the regulatory function of PGC7 in global DNA demethylation during early embryonic development (Supplementary Fig. 22).

Methods Summary

Details of cell culture, gel-shift assay, ChIP–western blotting, sperm collection, chromatin preparation, Plasmids, global DNA methylation analysis, MNase assay, histone peptide-binding assay, stepwise salt extraction, zygote collection and culture, triton treatment of zygotes, immunohistochemistry, ChIP–quantitative PCR and sperm chromatin preparation can be found in Methods.

Online Methods

Cell culture

G9a−/−, Dnmt1−/− Dnmt3a−/− Dnmt3b−/−and PGC7−/− ES cells and their clones were used. PGC7−/− ES cells were derived from blastocysts. These cells were maintained as described previously8.

Gel-shift assay

Purified mononucleosomes were incubated with various amounts of purified histidine-tagged PGC7 (His–PGC7) on ice for 30 min in 50 mM NaCl, 20 mM Tris buffer–HCl (pH 7.5), 2 mM EDTA buffer, 5 mg ml−1 BSA buffer and 5% glycerol. The samples were electrophoresed on a 1% native agarose gel and the chromatin was visualized using ethidium bromide.

ChIP–western blotting

Mononucleosomes prepared from PGC7−/− ES cells and a cell line stably expressing PGC7 were incubated with anti-PGC7 antibody for 3 h at 4 °C. After adding protein G-Sepharose 6 Fast Flow (GE Healthcare), the samples were incubated for another hour at 4 °C, washed and eluted from the beads using SDS sample buffer. The immunoprecipitates were analysed by immunoblotting as described previously8.

Sperm collection

Sperm was obtained from ICR mice aged 10–20 weeks. The cauda epididymides were partially cut open in human tubal fluid (Millipore) and incubated for 1 h at 37 °C in 5% CO2 to allow the sperm to swim out. The sperm pellet was washed with PBS buffer and ChIP–quantitative (q)PCR analysis was performed as described in the Supplementary Information.

Chromatin preparation

Cell pellets were re-suspended in buffer I (300 mM sucrose, 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 0.1 mM EGTA buffer, 15 mM Tris–HCl (pH 7.5), 0.4% tergitol-type NP-40 (NP-40) and 0.5 mM DTT), and an equal volume of buffer II (300 mM sucrose, 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 0.1 mM EGTA, 15 mM Tris–HCl (pH 7.5) and 0.5 mM dithiothreitol (DTT)) was added. After incubation on ice for 10 min, the cell suspensions were layered over buffer III (1.2 M sucrose, 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 0.1 mM EGTA, 15 mM Tris–HCl (pH 7.5) and 0.5 mM DTT) and the nuclear pellets were collected by centrifugation (10,000g, 4 °C, 20 min). To avoid carry over of NP-40, the supernatant was carefully removed at least three times using a new Pasteur pipette at each time. Nuclear pellets were washed in MNase digestion buffer (320 mM sucrose, 50 mM Tris–HCl (pH 7.5), 4 mM MgCl2 and 1 mM CaCl2) and then re-suspended in MNase digestion buffer. Chromatin was released from the nuclear preparations by digestion with 20 U ml−1 MNase (Takara) at 37 °C for 9 min. Digestion was stopped by adding 0.2 mM EDTA to a final concentration of 5 mM on ice. Chromatin preparations were analysed by native agarose gel electrophoresis (Supplementary Fig. 1) and the mononucleosome was used immediately for the gel-shift assay.

Plasmids

The Jhdm2a complementary DNA was cloned into pcDNA4mycHisA (Invitrogen). The Jhdm2a(H1122A) mutant was generated by PCR-based mutagenesis and confirmed by sequencing. The primers used to generate the mutant are described in Supplementary Table 2.

Global DNA methylation analysis

A previous protocol for global DNA methylation analysis25 was used with slight modifications. Genomic DNA was isolated from various ES cells with proteinase K and RNase A, followed by phenol/chloroform extraction and ethanol precipitation. A 2-μg aliquot of genomic DNA was digested with 50 U of methylation-sensitive HpaII or the methylation-insensitive isoschisomer MspI for 16–18 h at 37 °C. The digested genomes were purified by phenol/chloroform extraction and ethanol precipitation, and 250 ng of purified DNA was labelled with 3H-deoxycytidine triphosphate (dCTP) at 56 °C for 1 h using a single-nucleotide extension reaction. Undigested genomic DNA served as a background control. The samples were applied to DE-81 ion-exchange filters and washed three times with 0.5 M Na3PO4 buffer (pH 7.0) at room temperature. The filters were then dried and processed for scintillation counting.

MNase assay

Living cells were permeabilized on ice with 0.02% l-α-lysolecithin (Sigma) in 150 mM sucrose, 35 mM HEPES–NaOH (pH 7.4), 5 mM KHPO4, 5 mM MgCl2 and 0.5 mM CaCl2 for 90 s, followed by digestion with 3 U ml−1 MNase (Takara) in 150 mM sucrose, 50 mM Tris–HCl (pH 7.5), 50 mM NaCl and 2 mM CaCl2 at room temperature for 0, 2, 4, 6, 8, 10, 30 and 60 min. Digestion was stopped by adding EDTA to a final concentration 5 mM, on ice. DNA was purified by phenol/chloroform extraction and electrophoresed on 0.8% native agarose gels.

Histone peptide-binding assay

Biotinylated histone peptides were purchased from Upstate Biotechnology. In brief, biotinylated histone peptides (0.5 μg) were incubated with streptavidin beads (GE Healthcare) in binding buffer (50 mM Tris–HCl (pH 8.0), 300 mM NaCl and 0.1% NP-40) for 1 h at 4 °C with rotation. After extensive washing, the beads were incubated with 1 μg recombinant-purified His–PGC7 for 2 h at 4 °C with rotation. 10- and 100-fold unbiotinylated histone peptides were added for the competitive-binding assay. After extensive washing, bound PGC7 protein was analysed by SDS–PAGE and immunoblotting with anti-PGC7 antibody.

Stepwise salt extraction

ES cells were suspended in nuclear isolation buffer (10 mM Tris–HCl (pH 7.5), 60 mM KCl, 15 mM NaCl, 1 mM DTT, 1.5 mM MgCl2, 1 mM CaCl2, 250 mM sucrose, 10% glycerol, 1 mM DTT and 0.15% NP-40) on ice for 10 min. Nuclear pellets were treated with 100 U ml−1 DNase I (Takara) in nuclear isolation buffer with increasing amounts of NaCl (100, 200, 300, 400 and 500 mM) at 25 °C for 20 min and on ice for 10 min. After the incubation, EDTA was added to a final concentration of 5 mM, and the sample was incubated on ice for 10 min. Nuclear extracts were separated from pellets by centrifugation.

Zygote collection and culture

Female B6D2F1 mice >8 weeks old were superovulated by injecting 5 U of human chorionic gonadotropin 48 h after injecting 5 U of pregnant mare serum gonadotropin, and then mated with male mice. Fertilized eggs were collected from the oviduct, placed in 100-μl drops of KSOM (Millipore) and cultured at 37 °C in an atmosphere of 5% CO2.

Triton treatment of zygotes

Zygotes were treated with Triton X-100, similar to a previous report with minor modifications14. Zygotes were treated with 0.2% Triton X-100 in PBS for 45–60 s until the perivitelline space was eliminated. Immediately after Triton treatment, the zygotes were washed with PBS at least five times and then fixed in 4% PFA. After washing with PBS, immunostaining was performed as described below.

Immunohistochemistry

Fertilized eggs were washed with PBS, fixed for 15 min in 4% PFA in PBS at room temperature and permeabilized with 0.2% Triton X-100 in PBS for 20 min at room temperature. The eggs were blocked for 1 h in 5% normal goat serum in PBS at room temperature and incubated overnight at 4 °C with primary antibodies as shown in Supplementary Table 1. The following day, the eggs were washed three times with 0.05% Tween20 in PBS and staining was detected by incubating the eggs with secondary antibodies as shown in Supplementary Table 1. Nuclei were stained with 1 μg ml−1 DAPI. 5mC, 5hmC and DNA staining was performed as described previously8. Immunofluorescence was visualized using an LSM510 confocal laser scanning microscope (Carl Zeiss).

ChIP–qPCR

PGC7−/−, PGC7−/−–PGC7, G9a−/−and G9a−/−–PGC7 ES cells (3 × 107) were treated with 1% PFA for 8 min at room temperature. After quenching the PFA crosslinking reaction with 200 mM glycine, the fixed cells were washed with PBS. The cells were suspended in radio immunoprecipitation assay (RIPA) buffer (50 mM Tris–HCl (pH 8) 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.5% deoxycholate and 0.1% SDS) and sonicated to an average fragment size of 200–1000 bp. Solubilized chromatin was clarified by centrifugation for 10 min at 15,000 r.p.m. and 4 °C. The supernatant was pre-cleared with protein G-Sepharose beads, which were pre-blocked with salmon sperm DNA and BSA, at 4 °C for 1 h. The pre-cleared chromatin was incubated with anti-PGC7, anti-H3 and anti-H3K9me2 antibodies for 14–18 h at 4 °C. Immune complexes were bound to pre-blocked protein G-Sepharose beads for 2 h at 4 °C. The beads were washed with RIPA buffer, high-salt wash buffer (20 mM Tris–HCl (pH 8), 500 mM NaCl, 1 mM EDTA, 1% NP-40, 0.5% deoxycholate and 0.1% SDS), LiCl wash buffer (250 mM LiCl, 20 mM Tris–HCl (pH 8), 1 mM EDTA, 1% NP-40 and 0.5% deoxycholic acid (DOC)), and once with Tris–EDTA. Immune complexes bound to protein G beads were suspended in elution buffer (20 mM Tris–HCl (pH 8), 300 mM NaCl, 1 mM EDTA and 0.5% SDS) and incubated for 6 h at 65 °C. After incubation, the samples were treated with 30 μg ml−1 RNase A for 1 h at 37 °C, and 100 μg ml−1 proteinase K for 8 h at 56 °C. DNA was extracted with phenol/chloroform and precipitated with ethanol plus Dr.GenTLE (Takara) as a carrier. Precipitated DNA was re-suspended in 40 μl of water and analysed by qPCR using the specific primers shown in Supplementary Table 2.

Sperm chromatin preparation

Sperm pellets were washed with PBS, suspended in lysis buffer (0.1% SDS, 0.5% Triton X-100 in PBS) and incubated on ice for 20 min. After centrifugation, the sperm pellets were washed with PBS, suspended in 0.05% l-α-lysolecithin (Sigma) in PBS and incubated on ice for 15 min. The sperm pellets were collected by centrifugation, suspended in 10 mM DTT in PBS and incubated on ice for 10 min. After centrifugation, sperm pellets were fixed with 1% PFA in PBS for 8 min at room temperature. Next, 2.5 M glycine was added to a final concentration of 0.2 M, and incubation was continued for an additional 10 min at room temperature. After washing with PBS, the pellet was suspended in RIPA buffer (20 mM HEPES–NaOH (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.5% DOC and 0.1% SDS) and disrupted by sonication using a Bioruptor (Diagenode). ChIP–qPCR was conducted as described above using the specific primers shown in Supplementary Table 2.