Abstract

Mono-allelic expression of imprinted genes from either the paternal or the maternal allele is mediated by imprinting control regions (ICRs), which are epigenetically marked in an allele-specific fashion. Although, in somatic cells, these epigenetic marks comprise both DNA methylation and histone methylation, the relationship between these two modifications in imprint acquisition and maintenance remains unclear. To address this important question, we analyzed histone modifications at ICRs in mid-gestation embryos that were obtained from Dnmt3L−/− females, in which DNA methylation imprints at ICRs are not established during oogenesis. The absence of maternal DNA methylation imprints in these conceptuses led to a marked decrease and loss of allele-specificity of the repressive H3K9me3, H4K20me3 and H2A/H4R3me2 histone modifications, providing the first evidence of a mechanistic link between DNA and histone methylation at ICRs. The existence of this relationship was strengthened by the observation that when DNA methylation was still present at the Snrpn and Peg3 ICRs in some of the progeny of Dnmt3L−/− females, these ICRs were associated with the usual patterns of histone methylation. Combined, our data establish that DNA methylation is involved in the acquisition and/or maintenance of histone methylation at ICRs.

INTRODUCTION

Epigenetic mechanisms are critical for the correct development of complex organisms (1). These mechanisms define in a heritable manner the distinct functional states of chromatin and rely on the coordinate regulation of different levels of chromatin modifications. For instance, heritably repressed chromatin is often marked by both DNA methylation and a set of ‘repressive’ covalent histone tail modifications, such as tri-methylation of lysine 9 on histone H3 (H3K9me3) and tri-methylation of lysine 20 on histone H4 (H4K20me3). Interplay between these marks is suggested from biochemical and genetic studies conducted in various organisms from fungi to mammals (2,3).

A mechanistic link between histone modifications and DNA methylation was discovered in Neurospora crassa where H3K9me3 is a prerequisite for DNA methylation (4). This connection is unidirectional, since, in DNA methylation mutants, H3K9me3 is unaffected (4). A more complex relationship is observed in Arabidopsis thaliana where DNA methylation of CpG dinucleotides directs H3K9me2 deposition, which, in turn, controls DNA methylation at nearby non-CpG sites (5–7). These findings underscore a self-reinforcing feedback loop for the maintenance of both DNA and histone methylation (8). A complex picture is also emerging from studies in mammalian cells. In mouse embryonic stem (ES) cells, mutations in the Suv39h1/h2 H3K9-specific histone methyltransferases (HMTs) decrease but do not abolish DNA methylation at pericentric heterochromatin (9). However, it is not clear whether this effect is due to the absence of H3K9me or Suv39h1/h2. Recent studies on ES cells lacking the G9a HMT reported decreased DNA methylation at several euchromatic regions and at pericentric heterochromatin due to not the loss of H3K9me per se, but rather, to the absence of G9a (10–12). Perhaps more relevant for a possible mechanistic link between histone and DNA methylation in mammalian cells, is the observation that H4R3me2s, through its interaction with DNMT3A, influences DNA methylation at the human β-globin locus (13). On the other hand, the reciprocal mechanistic link, from DNA to histone methylation, remains poorly documented, and there are discrepant data according to the cell type analyzed. Absence of functional DNA methyltransferases was reported to affect patterns of H3K9me in human HeLa and HCT116 cells (14), but not in mouse ES cells (9,10,12,15). Further characterization of this relationship is important to understand how epigenetic patterns are set up and maintained.

Genomic imprinting, an epigenetic phenomenon, which leads to the mono-allelic expression of genes depending on the parental origin of the allele, provides an attractive model to gain further insights into the interplay between DNA and histone methylation. This allele-specific expression is mediated by essential sequence elements of several kilobases in size called imprinting control regions (ICRs). ICRs are marked by allelic DNA methylation, which is established in the female or the male germ line and, after fertilization, is maintained throughout development. In somatic cells, ICRs are also marked by allelic histone modifications, suggesting that there could be interplay between DNA methylation and histone modifications in imprinting. For instance, the parental allele carrying the DNA methylation imprint is consistently associated with ‘repressive’ histone modifications, including H3K9me3 and H4K20me3, and absence of H3 lysine-4 methylation (16–18).

Germ line acquisition of DNA methylation at ICRs relies on Dnmt3L that belongs to the Dnmt3 family of de novo methyltransferases, but lacks a functional methyltransferase domain (19). Progeny of Dnmt3l−/− females die by embryonic day 10.5. They completely lack maternal DNA methylation at ICRs, resulting in deregulated expression of associated imprinted genes, whereas methylation of the rest of the genome is apparently unaffected (20,21). Further genetic analyses support a model in which Dnmt3L is involved in the establishment of the maternal imprint through its interaction with Dnmt3A (21,22). Recent studies suggest a possible role of chromatin in the recruitment of the Dnmt3L/Dnmt3a complex to ICRs. Indeed, using a biochemical approach, it has been established that Dnmt3L interacts with histone H3 only when it is unmethylated on lysine 4, suggesting that the H3K4me0 signature could be a prerequisite for germ line acquisition of DNA methylation at ICRs (23,24). Whether the repressive histone modifications at ICRs are already present in oocytes and guide the establishment of DNA methylation imprints is currently unknown.

A link between DNA methylation and chromatin structure at ICRs is suggested by one study in which bi-allelic DNA methylation at the differentially methylated U2af1-rs1 region was found to be associated with H3 hypo-acetylation on both parental alleles (25). Of concern, the interaction between DNA methylation and histone repressive marks is suspected to be involved in the somatic maintenance of imprinting (26). So far, however, a mechanistic link between these two marks at ICRs failed to be demonstrated. Recently, we found that the absence of H4K20me3 in mouse embryonic fibroblasts deficient for Suv20h1/h2 does not affect DNA methylation at ICRs (27). In ES cells deficient for the H3K9-specific HMT G9a, DNA methylation levels were reported to be reduced at several ICRs. However, this was not observed in G9a−/− embryos and was likely to be due to the absence of G9a, rather than to a loss of H3K9me (10,12,28,29).

In the present study, we set out to evaluate the impact of the absence of DNA methylation on histone methylation at ICRs. Specifically, we determined the histone modification pattern in progeny of Dnmt3L−/− females. Our observations provide the first evidence for a mechanistic link between DNA and histone methylation at ICRs. Furthermore, our data support the notion that DNA methylation itself controls the allelic histone modifications at ICRs and suggest a self-reinforcing feedback model in the acquisition and the maintenance of parental imprints.

RESULTS

Dnmt3lm−/− conceptuses lack allele-specific histone methylation at ICRs

In order to determine the role of DNA methylation in the allelic distribution of chromatin modifications at ICRs, we performed immunoprecipitation of native chromatin (ChIP) on pooled 9.5dpc conceptuses (n = 25–30 per ChIP) derived from Dnmt3L−/− females that lack female germ line-derived DNA methylation (we refer to these samples as Dnmt3Lm−/− embryos). To use DNA polymorphisms for allele discrimination, Dnmt3Lm−/− embryos and wild-type controls were (C57BL/6x JF1)F1 intra-specific hybrids. The relative abundance of the maternal and the paternal allele in the antibody-bound fractions was determined using single-strand conformation polymorphisms (SSCP) and allelic real-time PCR approaches. Analyses were conducted at eight ICRs that have female germ line-derived DNA methylation (KvDMR1, Grb10, U2af1-rs1, Zac1, Snrpn, Gnasxl, Nap1l5 and Peg3) (maternal ICRs) and at the H19 ICR, which has paternal germ line-derived methylation and served as a control (paternal ICR) (Supplementary Material, Table S1).

An initial analysis was performed on wild-type (WT) conceptuses using a panel of antisera directed against 16 different covalent histone modifications, of which seven showed consistent allelic enrichment both in the embryo and trophoblast, and were therefore retained for this study (i.e. H3K9ac, H3K4me2, H3K4me3, H3K27me3, H3K9me3, H4K20me3 and H2A/H4R3me2s).

In agreement with earlier work by us and others, the DNA-methylated allele was associated with H3K9me3 and H4K20me3 at all eight maternal ICRs analyzed. In addition, we find for all but one of the ICRs (i.e. Grb10) that the DNA-methylated allele shows preferential precipitation of symmetrical dimethylation at H2A/H4 (H2A/H4R3me2s) (Fig. 1 for embryo samples, Supplementary Material, Fig. S1 for throphoblast samples). Unlike these three marks, however, H3K27me3 is not consistently associated with the DNA-methylated allele, as observed at Nap1L5 ICR (data not shown) and Grb10 ICR (Fig. 1). Conversely, the unmethylated allele was associated with H3K4me2 (Fig. 1) and H3K4me3 (data not shown) at all eight ICRs and with H3 lysine-9 acetylation (H3K9ac) which was, however, not enriched at the Grb10 ICR (Fig. 1).

Figure 1.

Modifications of allelic histone marks in Dnmt3Lm−/− embryos. (A) Representative results following allele-specific analysis of histone marks at ICRs (indicated on the right). Native ChIP followed by SSCP-mediated allelic discrimination of the Input (I), antibody-Bound (B) and -Unbound (U) chromatin fractions was performed on (B57BL/6x JF1) wild-type (WT) (left panels) and Dnmt3Lm−/− (right panels) 9.5dpc embryos. Maternal (M) and paternal (P) alleles are indicated. (B) Allelic real-time PCR analysis. Representative results of the allelic enrichment observed at the KvDMR1 and H19 ICRs in WT (dark blue bars for the paternal allele and light blue for the maternal allele) and Dnmt3Lm−/− (dark purple bars for the paternal allele and light purple for the maternal allele) embryos. In Dnmt3Lm−/− embryos, the maternal ICRs show loss of allelic specificity in comparison to WT embryos, whereas the paternally methylated H19 ICR presents the same pattern in WT and mutant embryos.

Figure 1.

Modifications of allelic histone marks in Dnmt3Lm−/− embryos. (A) Representative results following allele-specific analysis of histone marks at ICRs (indicated on the right). Native ChIP followed by SSCP-mediated allelic discrimination of the Input (I), antibody-Bound (B) and -Unbound (U) chromatin fractions was performed on (B57BL/6x JF1) wild-type (WT) (left panels) and Dnmt3Lm−/− (right panels) 9.5dpc embryos. Maternal (M) and paternal (P) alleles are indicated. (B) Allelic real-time PCR analysis. Representative results of the allelic enrichment observed at the KvDMR1 and H19 ICRs in WT (dark blue bars for the paternal allele and light blue for the maternal allele) and Dnmt3Lm−/− (dark purple bars for the paternal allele and light purple for the maternal allele) embryos. In Dnmt3Lm−/− embryos, the maternal ICRs show loss of allelic specificity in comparison to WT embryos, whereas the paternally methylated H19 ICR presents the same pattern in WT and mutant embryos.

We then conducted a similar analysis on Dnmt3Lm−/− conceptuses. In all instances, the eight maternal ICRs, which lack DNA methylation in these conceptuses (20,21,30), did not show any allelic enrichment for the seven histone modifications analyzed (Fig. 1 for embryo samples, Supplementary Material, Fig. S1 for throphoblast samples). By contrast, the allelic precipitation pattern was unaffected at the H19 ICR, where the paternal germ line-derived methylation was maintained (Fig. 1, Supplementary Material, Fig. S1).

This finding indicates that the failure to acquire DNA methylation imprints leads to the absence of allelic histone modifications at these ICRs.

Both parental alleles of maternally methylated ICRs adopt a paternal epigenotype in Dnmt3lm−/− conceptuses

Because the approaches used in Figure 1 provided qualitative results, we performed quantitative analysis by real-time PCR amplification to further characterize the extent to which chromatin patterns at the maternal ICRs are affected. For appropriate comparison between independent experiments, results are presented after normalization with the levels of precipitation obtained at the H19 ICR. We observed a marked difference between permissive and repressive histone modifications (Fig. 2). Permissive histone marks (i.e. H3K9ac, H3K4me2 and H3K4me3) were indeed more abundant, whereas precipitation of the repressive marks H3K27me3, H3K9me3, H4K20me3 and H2A/H4R3me2s was greatly reduced at these ICRs in Dnmt3Lm−/− conceptuses in comparison to WT samples. This pattern was, however, somewhat different at the Grb10 ICR, where more H3K27me3 was precipitated in Dnmt3Lm−/− than in WT embryos, unlike for the other repressive marks (Fig. 2). This finding is consistent with our observation that bivalent chromatin is enriched in both H3K4me2/3 and H3K27me3 on the unmethylated paternal allele of this ICR (31, Fig. 1) and suggests that in Dnmt3Lm−/− embryos both parental alleles are now marked by this bivalent structure. Taken together, our allelic and quantitative analyses indicate that in Dnmt3Lm−/− conceptuses both alleles of the maternal ICRs adopt a paternal epigenotype, characterized by the enrichment of permissive marks and global decrease of repressive marks.

Figure 2.

Repressive histone modifications are decreased at maternal ICRs in Dnmt3Lm−/− embryos. Real-time PCR-based quantification of precipitation levels for the indicated histone modifications at three maternal ICRs (KvDMR1, U2af1-rs1 and Grb10) in WT and Dnmt3Lm−/− embryos. The degree of enrichment was calculated as the ratio between the Bound fraction and the Input fraction and was normalized to the Bound/Input ratio obtained for the H19 ICR, which we showed to be similar between WT and mutant embryos (Supplementary Material, Fig. S2). When data were obtained with too few samples, standard deviations were not determined.

Figure 2.

Repressive histone modifications are decreased at maternal ICRs in Dnmt3Lm−/− embryos. Real-time PCR-based quantification of precipitation levels for the indicated histone modifications at three maternal ICRs (KvDMR1, U2af1-rs1 and Grb10) in WT and Dnmt3Lm−/− embryos. The degree of enrichment was calculated as the ratio between the Bound fraction and the Input fraction and was normalized to the Bound/Input ratio obtained for the H19 ICR, which we showed to be similar between WT and mutant embryos (Supplementary Material, Fig. S2). When data were obtained with too few samples, standard deviations were not determined.

In Dnmt3lm−/− embryos, infrequently acquired DNA methylation at the Snrpn ICR correlates with allelic histone methylation

In the previous study, we reported that methylation imprints can be present in some progeny of Dnmt3L−/− females. Although not ICR specific, this incomplete penetrance of the phenotype due to Dnmt3L deficiency in oocytes is more often detected at the Snrpn and Peg3 ICRs (30). Interestingly, at these two ICRs, allelic histone modifications, comparable to those detected in WT embryos, could be sometimes observed also in Dnmt3Lm−/− embryos (Fig. 3 and Supplementary Material, Fig. S3). Cumulative quantitative analysis of histone modifications at the Snrpn ICR in Dnmt3Lm−/− embryos provided a pattern similar to that observed at others ICRs, although repressive marks were less reduced (Fig. 3A). Further analysis revealed indeed that allelic enrichment for repressive histone modifications could still be detected at this ICR in pooled Dnmt3Lm−/− embryos. This enrichment was more or less marked according to the ChIP assay analyzed, but was nonetheless detected in all of them (Fig. 3B). A similar observation was made in trophoblast samples and at the Peg3 ICR as well (Supplementary Material, Fig. S3). Bisulfite sequencing analysis conducted on the input chromatin used for the ChIP assays detected the presence of substantial maternal DNA methylation (Fig. 3B) further supporting the link between the allelic enrichment of repressive histone modifications and the presence of residual DNA methylation.

Figure 3.

Conservation of imprinted DNA methylation at the Snrpn ICR in Dnmt3Lm−/− embryos correlates with the presence of allelic histone modifications. (A) PCR quantification of precipitation for the indicated histone marks at the Snrpn ICR. As in others maternal ICRs, permissive marks increase, whereas repressive modifications decrease in Dnmt3Lm−/− embryos. When data were obtained from two experiments only, standard deviations were not determined. (B) Allelic histone marks can be retained at Snrpn ICR in Dnmt3Lm−/− embryos as observed after SSCP-mediated allelic discrimination (upper panel) or quantitative allele-specific PCR amplification (lower panel) performed on a representative ChIP assay. Maternal DNA methylation (M) is detected at the Snrpn ICR in Dnmt3Lm−/− embryos used for ChIP as revealed by bisulfite analysis performed on the Input fraction (right panel). (C) Analysis of Dnmt3Lm−/− embryos that lack DNA methylation at the Snrpn ICR (referred to as sDnmt3Lm−/− embryos) (see text for details). sDnmt3Lm−/− embryos did not show any allelic enrichment at the Snrpn ICR (upper panel). Lower panel shows a quantitative PCR analysis for the indicated histone modifications performed on ChIP samples from WT, Dnmt3Lm−/− (the ChIP assay shown in B) and sDnmt3Lm−/− embryos.

Figure 3.

Conservation of imprinted DNA methylation at the Snrpn ICR in Dnmt3Lm−/− embryos correlates with the presence of allelic histone modifications. (A) PCR quantification of precipitation for the indicated histone marks at the Snrpn ICR. As in others maternal ICRs, permissive marks increase, whereas repressive modifications decrease in Dnmt3Lm−/− embryos. When data were obtained from two experiments only, standard deviations were not determined. (B) Allelic histone marks can be retained at Snrpn ICR in Dnmt3Lm−/− embryos as observed after SSCP-mediated allelic discrimination (upper panel) or quantitative allele-specific PCR amplification (lower panel) performed on a representative ChIP assay. Maternal DNA methylation (M) is detected at the Snrpn ICR in Dnmt3Lm−/− embryos used for ChIP as revealed by bisulfite analysis performed on the Input fraction (right panel). (C) Analysis of Dnmt3Lm−/− embryos that lack DNA methylation at the Snrpn ICR (referred to as sDnmt3Lm−/− embryos) (see text for details). sDnmt3Lm−/− embryos did not show any allelic enrichment at the Snrpn ICR (upper panel). Lower panel shows a quantitative PCR analysis for the indicated histone modifications performed on ChIP samples from WT, Dnmt3Lm−/− (the ChIP assay shown in B) and sDnmt3Lm−/− embryos.

Altogether these observations suggest that, in the pooled Dnmt3Lm−/− embryos (n = 25–30) used to perform ChIP, some had a proper imprint at the Snrpn ICR, constituted of allelic DNA and histone methylation, underscoring the tightness of the link between these two marks.

We next evaluated whether allelic histone methylation could be present at the Snrpn ICR independently of DNA methylation. For this purpose, we took advantage of our previous observation that, when it occurs, DNA methylation imprint is present in both embryonic and extra-embryonic tissues of Dnmt3Lm−/− conceptuses (30). Specifically, by bisulfite sequencing, we selected individual trophoblasts that lacked DNA methylation at the Snrpn ICR (data not shown) and used the corresponding embryos for ChIP analysis. Using this selective approach, we identified a total of 21 Dnmt3Lm−/− embryos (referred to as sDnmt3Lm−/− in Fig. 3) that were divided in three batches to perform independent ChIP assays. Bisulfite sequencing analysis was also conducted on the input fractions to confirm the absence of DNA methylation at the Snrpn ICR of these selected embryos (data not shown). Because of the small number of embryos, we chose to analyze by ChIP only a restricted number of histone marks. In all instances, we did not detect any allelic enrichment of repressive histone modifications at the Snrpn ICR (Fig. 3C). Specifically, quantitative analysis showed that H3K9ac was increased, whereas the repressive H3K9me3, H4K20me3 and H2A/H4R3me2s marks were markedly reduced compared with WT and pooled Dnmt3Lm−/− embryos that had both unmethylated and maternally methylated Snrpn ICRs (Fig. 3C). This result further supports the idea that repressive histone methylation at ICRs is dependent on the presence of DNA methylation.

Altered histone methylation at the Grb10 ICR does not reflect transcriptional activity

The marked decrease of repressive histone methylation we observed at Dnmt3Lm−/− ICRs, which lack DNA methylation, could be either the direct consequence of the absence of DNA methylation or of the biallelic promoter activity detected at most of these ICRs in Dnmt3Lm−/− conceptuses (20,21,30,32). Our observation that H3K9me3 and H4K20me3 are affected also at the Grb10 ICR, however, argues against this possibility as Grb10, unlike the others ICRs we analyzed, is transcriptionally silent in Dnmt3lm−/− conceptuses (30,33). To substantiate this observation, we investigated Grb10 expression through a promoter-specific RT–PCR approach (Fig. 4). It has been previously shown that promoter activity from the Grb10 ICR is limited to neurons (31). As expected, we could not detect transcripts arising from this ICR region in WT or in mutant embryos at the developmental stage analyzed (at which neurons are not yet formed).

Figure 4.

Grb10 promoters are inactive in Dnmt3Lm−/− embryos. (A) Schematic representation of the Grb10 promoters. Po: oocyte-specific promoter; Pm: main promoter. Promoter-specific RT–PCR amplifications were conducted by using primers located in each promoter region in combination with a primer located in the body of the gene (arrows). (B) Promoter-specific RT–PCR products obtained from WT and Dnmt3Lm−/− embryos. (C) Quantitative promoter-specific RT–PCR amplification. Graph shows the level of expression from each promoter in WT and Dnmt3Lm−/− embryos after normalization to Rpl13a. Data are shown as fold enrichment of activity detected from the Grb10 Po in WT embryos.

Figure 4.

Grb10 promoters are inactive in Dnmt3Lm−/− embryos. (A) Schematic representation of the Grb10 promoters. Po: oocyte-specific promoter; Pm: main promoter. Promoter-specific RT–PCR amplifications were conducted by using primers located in each promoter region in combination with a primer located in the body of the gene (arrows). (B) Promoter-specific RT–PCR products obtained from WT and Dnmt3Lm−/− embryos. (C) Quantitative promoter-specific RT–PCR amplification. Graph shows the level of expression from each promoter in WT and Dnmt3Lm−/− embryos after normalization to Rpl13a. Data are shown as fold enrichment of activity detected from the Grb10 Po in WT embryos.

To further discard, a role for transcription across the ICR we performed the same analysis for the main Grb10 promoter (Pm) (31) and for a promoter shown to be active in oocytes (Po) (34), both located upstream of the ICR (Fig. 4). In both cases, we observed that the maternal expression arising from these promoter regions in WT embryos was virtually lost in mutant embryos (Fig. 4). This suggests that, in Dnmt3Lm−/− conceptuses, the decreased levels of repressive histone modifications at the Grb10 ICR are transcription-independent and could be a direct consequence of the absence of DNA methylation.

DISCUSSION

Several lines of evidence suggest that there is a link between DNA and histone methylation in the process of gene silencing in mammals. ICRs could represent a model for such link as one allele is DNA methylated and associated with repressive histone marks, whereas the other is unmethylated and associated with permissive histone marks. However, the precise functional relationship between these marks at ICRs remains unclear. To clarify this important issue, we studied histone modification patterns at ICRs of 9.5dpc Dnmt3Lm−/− conceptuses, which lack female germ line-derived DNA methylation. We observed that the non-acquisition of DNA methylation drastically affected allelic histone methylation patterns, providing the first evidence for a mechanistic link between DNA and histone methylation at ICRs.

We conducted a comprehensive analysis of the allelic distribution of histone modifications for eight maternally and one paternally methylated ICR in embryos and trophoblasts. We confirmed an allelic enrichment for several histone marks in WT embryos, as previously observed (26). In particular, the methylated allele was associated with a panel of repressive histone marks. In Dnmt3Lm−/− conceptuses, this picture was drastically affected, with the absence of allelic enrichment and a marked decrease, but not a complete loss, of repressive histone modifications. One possible explanation for the residual bi-allelic precipitation of repressive histone marks might be that regions directly bordering the ICRs carry these modifications on both parental chromosomes and are then precipitated together with the ICR. Nevertheless, our study demonstrates that repressive marks are strongly reduced and are no longer allele-specific at maternal ICRs devoid of DNA methylation of Dnmt3Lm−/− conceptuses. At the Grb10 ICR at least, this phenomenon is unlikely to result from a transcription-dependent mechanism. Furthermore, this is not a direct result of the absence of Dnmt3L in oocytes, as suggested by our findings on the Snrpn and Peg3 ICRs. Altogether, our results strongly suggest that DNA methylation is involved in the control of repressive histone marks at ICRs.

In our study, we also confirmed and extended our previous observation that proper methylation imprints can be present at some ICRs, in some progeny of Dnmt3L−/− females (30). This is particularly evident at the Snrpn and Peg3 ICRs. All included, we found Snrpn and Peg3 fully maternally methylated in, respectively, 30% (17 of 58) and 14% (8 of 58) of the Dnmt3Lm−/− conceptuses that we analyzed individually (30 and data not shown). In the absence of Dnmt3L, these two ICRs could be prone to stochastically acquired DNA methylation in the developing oocyte (30). However, bisulfite sequencing of oocytes from Dnmt3L−/− females (20,32) failed to detect DNA methylation at these ICRs, suggesting that DNA methylation is most likely acquired after fertilization, presumably at the pre-implantation stage. This process could be mediated by a specific maternally inherited signal (i.e. a chromatin signature) carried by the oocyte. However, we failed to detect, in WT embryos, histone modifications specific to the Snrpn and Peg3 ICRs and we did not observe allelic enrichment of histone modifications at the Snrpn ICRs, which lack DNA methylation. Nevertheless, we cannot totally exclude the hypothesis that a specific chromatin signature established stochastically at the Snrpn and Peg3 ICRs in the female germ line, or resulting from a sporadic incomplete erasure of the imprint marks in the primordial germ cells, could provide a tag for the acquisition of DNA methylation early after fertilization. Interestingly, Li et al. (35) reported recently that Zfp57 is required to maintain DNA methylation imprints at multiple ICRs and that it also involved in the acquisition of DNA methylation specifically at the Snrpn ICR (35). Intriguingly, the failure to establish Snrpn methylation in oocytes lacking Zfp57 was partially rescued by the expression of Zfp57 during early development (35). Further characterization of the precise role of Zpf57, and its possible relationship with Dnmt3L, should help to determine whether this process relies on a pre-existing chromatin signature or makes use of a different mechanism.

Our results indicate that DNA methylation plays a central role in the control of repressive histone methylation at ICRs, but do not allow us to discriminate whether this occurs during histone methylation acquisition and/or maintenance. If repressive histone marks are acquired at ICRs independently of, and possibly prior to, DNA methylation, our results would indicate that DNA methylation is required only for the maintenance of histone methylation. Alternatively, DNA methylation might be a prerequisite for the acquisition of repressive histone marks. We favor this second hypothesis and therefore we propose a working model where an initial DNA methylation tag triggers a tight mechanistic interplay between the two kinds of epigenetic marks during acquisition and maintenance of imprints (Fig. 5).

Figure 5.

Working model of the interaction between DNA methylation and histone methylation at ICRs. In germ cells, in the absence of H3K4me, the Dnmt3L/3a complex is recruited at ICRs and initiates DNA methylation deposition. Initial DNA methylation and/or the Dnmt3L/3a complex recruit HMTs and lead to acquisition of repressive histone marks at the methylated allele. Subsequent recruitment of Dnmts by these repressive marks and by HMTs creates a positive feedback leading to gradual acquisition of more DNA methylation and repressive modifications. This feedback loop is also involved in the maintenance of the imprint in somatic cells by using NP95 as a platform.

Figure 5.

Working model of the interaction between DNA methylation and histone methylation at ICRs. In germ cells, in the absence of H3K4me, the Dnmt3L/3a complex is recruited at ICRs and initiates DNA methylation deposition. Initial DNA methylation and/or the Dnmt3L/3a complex recruit HMTs and lead to acquisition of repressive histone marks at the methylated allele. Subsequent recruitment of Dnmts by these repressive marks and by HMTs creates a positive feedback leading to gradual acquisition of more DNA methylation and repressive modifications. This feedback loop is also involved in the maintenance of the imprint in somatic cells by using NP95 as a platform.

Absence of methylation on H3K4 (H3K4me0) and transcriptional events, which could create or maintain an open chromatin structure across ICRs, are believed to be important for the acquisition of methylation imprints (23,34). In germ cells, these two events could lead to recruitment of the Dnmt3L/Dnmt3a complex (22,24) and subsequent acquisition of DNA methylation at ICRs. Interaction of Dnmt3a with the HMTs SETDB1/Eset (36) and PRMT5 (13) has been reported, and this could favor subsequent acquisition of the repressive histone marks H3K9me3 and H2A/H4R3me2s, respectively. Studies conducted on Xenopus cell extract suggest that H2A/H4R3me2s could play a role in the acquisition of DNA methylation imprint at the H19 ICR (37). Furthermore, in addition to its interaction with PRMT5, DNMT3a interact also directly with H2A/H4R3me2s itself (13), thus creating a feedback loop. This could explain the gradual nature of acquisition of DNA methylation at ICRs in growing oocytes and dividing male germ cells (38). Self-reinforcing mechanisms are likely to also be involved in the maintenance of methylation imprints in somatic cells following fertilization. The Np95 protein, which is essential to maintain DNA methylation through replication, including at ICRs (39), is likely to be a key factor in this process. It recruits complexes containing DNMT1 and HMTs, including the H3K9 methyltransferases G9a and Glp, to hemi-methylated DNA (39). Furthermore, it has been reported that Np95 also interacts with H3K9me3 (40) providing a self-reinforcing mechanism in which this factor is crucial for faithful maintenance of both DNA and histone methylation during cell division.

To test this model of interplay between histone and DNA methylation, we would need first to establish the ontogeny of histone modifications at ICRs during imprint reprogramming in germ cells. Obtaining such information at the time of maternal imprint acquisition, however, is currently difficult, as the available material is too limited for the existing ChIP technologies. Further characterization of the HMTs that regulate ICRs should also be important. Except for Suv4-20h1/h2, which regulates H4K20me3 acquisition at ICRs (27), the identity of the other HMTs involved in the establishment of repressive marks is currently unknown. It seems particularly relevant to determine which HMT controls H3K9me3 acquisition, which is consistently associated with DNA methylation at ICRs in somatic cells. A recent study eliminated Suv39h1/h2 for this role (27). Identification of the HMTs involved will further help to unravel the relationship between histone and DNA methylation at ICRs. The finding that the absence of H4K20me3 does not affect the maintenance of DNA methylation at ICRs (27,41) suggests a redundant role for the multiple repressive marks present on the DNA-methylated allele of ICRs.

Finally, it remains to establish whether the H3K4me0 chromatin signature is sufficient to recruit the DNA methylation machinery to ICRs. Further analysis at the Snrpn and Peg3 ICRs, where our study suggests that this signature is likely to be maintained and recognized at pre-implantation stages, will be crucial to answer this key question.

MATERIALS AND METHODS

Material collection

To make use of DNA polymorphisms for allele discrimination, we crossed homozygous Dnmt3L−/− females (129SvJae x C57BL/6 hybrid genetic background) to wild-type JF1 male mice (Mus musculus molossinus) and called their progeny Dnmt3Lm−/− embryos (the name Dnmt3Lm−/− was chosen to distinguish these embryos from embryos born from wild-type and Dnmt3L+/− mothers). At 9.5dpc, Dnmt3Lm−/− embryos were removed from pregnant mothers. Trophoblast tissue was dissected carefully from the embryo and maternal decidua and then washed in PBS twice to remove contaminating maternal blood. Yolk sac DNA was used for genotyping by PCR amplification as previously described (21). Control 9.5dpc embryos and placentas were obtained by crossing C57BL/6 female mice to JF1 male mice.

DNA extraction and bisulfite sequencing

DNA extraction and bisulfite sequencing on 9.5dpc trophoblast samples were conducted as previously described (30).

RNA extraction and expression analysis

Total RNA extraction and first strand cDNA synthesis were performed as previously described (30). Grb10 RT–PCR products were quantified by real-time PCR amplification with a SYBR Green mix (Qiagen) using an MX3000 apparatus (Stratagene). Ribosomal protein LI3a (Rpl13a) transcript level was used for normalization.

Chromatin immuno-precipitation

Chip on native chromatin was carried out on batches of 25–30 wild-type and Dnmt3Lm−/− embryos and trophoblasts as previously described (42). Results presented in this article were obtained from at least three ChIP assays performed on independent chromatin preparations. Details of the antisera used are described in Supplementary Material, Table S2.

Quantitative analysis of immuno-precipitated DNA by real-time PCR amplification

The Bound and Input ChIP fractions were quantified by real-time PCR amplification with a SYBR Green mix (Qiagen) using an MX3000 apparatus (Stratagene). Background precipitation was determined by performing mock precipitations with a non-specific IgG antiserum and was only a fraction of that observed in the precipitations with the specific antisera (Supplementary Material, Fig. S4). Bound/Input ratios were calculated and were normalized to those for the H19 ICR which we showed to be similar in WT and mutant embryos (Supplementary Material, Fig. S2).

Allelic analysis of immuno-precipitated chromatin

In the antibody-bound and -unbound fractions for each antiserum used, parental alleles were distinguished by direct sequencing of the PCR products or by radioactive PCR amplification performed in the presence of [α33P]-dCTP (1% of total dCTP), followed by electrophoretic detection of SSCPs. Primers used and details of the polymorphisms are given in Supplementary Material, Table S1. Allelic analyses were also conducted by allelic real-time PCR amplification; based on SNPs between the C57BL/6 and JF1 strains, allele-specific primer sets were designed for each analyzed region. Quantitative real-time PCR amplifications were performed with each primer set on Bound and Input fractions. The Bound/Input ratio for each amplification was calculated to obtain the relative allelic enrichment for each analyzed modification.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This project was conducted under the ‘Programme Hubert-Curien /Sakura’, a collaborative programme between France and Japan (project 15928TG). A.H. is the recipient of a PhD studentship from the Association pour la Recherche sur le Cancer (ARC) and L.A.S. of a PhD studentship from the French Ministry of Education and Science. The Montpellier Laboratory (R.F.) acknowledges grant funding from the ‘Agence National de la Recherche’ (ANR), the ‘Institut National de Cancer’ (INCa) and the International Agency for Cancer Research (IACR, UK). This work was supported also by a grant for Child Health and Development, grants from the Japanese Ministry of Education, Science, Sports and Culture (grant 20062010) and the Japanese Ministry of Health and Welfare (H20-002) to K.H.

ACKNOWLEDGEMENTS

We thank W. Dean, M. Weber and Thierry Gostan for valuable advice and critical reading of the manuscript. We are grateful to K. Chebli, C. Jacquet and the Animal Facility for expert animal husbandry.

Conflict of Interest statement. None declared.

REFERENCES

1
Reik
W.
Stability and flexibility of epigenetic gene regulation in mammalian development
Nature
 , 
2007
, vol. 
447
 (pg. 
425
-
432
)
2
Stancheva
I.
Caught in conspiracy: cooperation between DNA methylation and histone H3K9 methylation in the establishment and maintenance of heterochromatin
Biochem. Cell. Biol.
 , 
2005
, vol. 
83
 (pg. 
385
-
395
)
3
Fuks
F.
DNA methylation and histone modifications: teaming up to silence genes
Curr. Opin. Genet. Dev.
 , 
2005
, vol. 
15
 (pg. 
490
-
495
)
4
Tamaru
H.
Zhang
X.
McMillen
D.
Singh
P.B.
Nakayama
J.
Grewal
S.I.
Allis
C.D.
Cheng
X.
Selker
E.U.
Trimethylated lysine 9 of histone H3 is a mark for DNA methylation in Neurospora crassa
Nat. Genet.
 , 
2003
, vol. 
34
 (pg. 
75
-
79
)
5
Jackson
J.P.
Lindroth
A.M.
Cao
X.
Jacobsen
S.E.
Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase
Nature
 , 
2002
, vol. 
416
 (pg. 
556
-
560
)
6
Soppe
W.J.
Jasencakova
Z.
Houben
A.
Kakutani
T.
Meister
A.
Huang
M.S.
Jacobsen
S.E.
Schubert
I.
Fransz
P.F.
DNA methylation controls histone H3 lysine 9 methylation and heterochromatin assembly in Arabidopsis
EMBO J.
 , 
2002
, vol. 
21
 (pg. 
6549
-
6559
)
7
Tariq
M.
Saze
H.
Probst
A.V.
Lichota
J.
Habu
Y.
Paszkowski
J.
Erasure of CpG methylation in Arabidopsis alters patterns of histone H3 methylation in heterochromatin
Proc. Natl Acad. Sci. USA
 , 
2003
, vol. 
100
 (pg. 
8823
-
8827
)
8
Johnson
L.M.
Bostick
M.
Zhang
X.
Kraft
E.
Henderson
I.
Callis
J.
Jacobsen
S.E.
The SRA methyl-cytosine-binding domain links DNA and histone methylation
Curr. Biol.
 , 
2007
, vol. 
7
 (pg. 
379
-
384
)
9
Lehnertz
B.
Ueda
Y.
Derijck
A.A.
Braunschweig
U.
Perez-Burgos
L.
Kubicek
S.
Chen
T.
Li
E.
Jenuwein
T.
Peters
A.H.
Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin
Curr. Biol.
 , 
2003
, vol. 
13
 (pg. 
1192
-
1200
)
10
Dong
K.B.
Maksakova
I.A.
Mohn
F.
Leung
D.
Appanah
R.
Lee
S.
Yang
H.W.
Lam
L.L.
Mager
D.L.
Schübeler
D.
, et al.  . 
DNA methylation in ES cells requires the lysine methyltransferase G9a but not its catalytic activity
EMBO J.
 , 
2008
, vol. 
27
 (pg. 
2691
-
2701
)
11
Epsztejn-Litman
S.
Feldman
N.
Abu-Remaileh
M.
Shufaro
Y.
Gerson
A.
Ueda
J.
Deplus
R.
Fuks
F.
Shinkai
Y.
Cedar
H.
, et al.  . 
De novo DNA methylation promoted by G9a prevents reprogramming of embryonically silenced genes
Nat. Struct. Mol. Biol.
 , 
2008
, vol. 
15
 (pg. 
1176
-
1183
)
12
Tachibana
M.
Matsumura
Y.
Fukuda
M.
Kimura
H.
Shinkai
Y.
G9a/GLP complexes independently mediate H3K9 and DNA methylation to silence transcription
EMBO J.
 , 
2008
, vol. 
27
 (pg. 
2681
-
2690
)
13
Zhao
Q.
Rank
G.
Tan
Y.T.
Li
H.
Moritz
R.L.
Simpson
R.J.
Cerruti
L.
Curtis
D.J.
Patel
D.J.
Allis
C.D.
, et al.  . 
PRMT5-mediated methylation of histone H4R3 recruits DNMT3A, coupling histone and DNA methylation in gene silencing
Nat. Struct. Mol. Biol.
 , 
2009
, vol. 
16
 (pg. 
304
-
311
)
14
Estève
P.O.
Chin
H.G.
Smallwood
A.
Feehery
G.R.
Gangisetty
O.
Karpf
A.R.
Carey
M.F.
Pradhan
S.
Direct interaction between DNMT1 and G9a coordinates DNA and histone methylation during replication
Genes Dev.
 , 
2006
, vol. 
20
 (pg. 
3089
-
3103
)
15
Tsumura
A.
Hayakawa
T.
Kumaki
Y.
Takebayashi
S.
Sakaue
M.
Matsuoka
C.
Shimotohno
K.
Ishikawa
F.
Li
E.
Ueda
H.R.
, et al.  . 
Maintenance of self-renewal ability of mouse embryonic stem cells in the absence of DNA methyltransferases Dnmt1, Dnmt3a and Dnmt3b
Genes Cells
 , 
2006
, vol. 
11
 (pg. 
805
-
814
)
16
Yang
Y.
Li
T.
Vu
T.H.
Ulaner
G.A.
Hu
J.F.
Hoffman
A.R.
The histone code regulating expression of the imprinted mouse Igf2r gene
Endocrinology
 , 
2003
, vol. 
144
 (pg. 
5658
-
5670
)
17
Delaval
K.
Govin
J.
Cerqueira
F.
Rousseaux
S.
Khochbin
S.
Feil
R.
Differential histone modifications mark mouse imprinting control regions during spermatogenesis
EMBO J.
 , 
2007
, vol. 
26
 (pg. 
720
-
729
)
18
Meissner
A.
Mikkelsen
T.S.
Gu
H.
Wernig
M.
Hanna
J.
Sivachenko
A.
Zhang
X.
Bernstein
B.E.
Nusbaum
C.
Jaffe
D.B.
, et al.  . 
Genome-scale DNA methylation maps of pluripotent and differentiated cells
Nature
 , 
2008
, vol. 
454
 (pg. 
766
-
770
)
19
Aapola
U.
Kazunori
S.
Scott
H.S.
Ollila
J.
Vihinen
M.
Heino
M.
Shintani
A.
Kawasaki
K.
Minoshima
S.
Krohn
K.
, et al.  . 
Isolation and initial characterization of a novel zinc finger gene, DNMT3L, on 21q22.3, related to the cytosine-5-methyltransferase 3 gene family
Genomics
 , 
2000
, vol. 
65
 (pg. 
293
-
298
)
20
Bourc'his
D.
Xu
G.L.
Lin
C.S.
Bollman
B.
Bestor
T.H.
Dnmt3L and the establishment of maternal genomic imprints
Science
 , 
2001
, vol. 
294
 (pg. 
2536
-
2539
)
21
Hata
K.
Okano
M.
Lei
H.
Li
E.
Dnmt3L cooperates with the Dnmt3 family of de novo DNA methyltransferases to establish maternal imprints in mice
Development
 , 
2002
, vol. 
129
 (pg. 
1983
-
1993
)
22
Kaneda
M.
Okano
M.
Hata
K.
Sado
T.
Tsujimoto
N.
Li
E.
Sasaki
H.
Essential role for de novo DNA methyltransferase Dnmt3a in paternal and maternal imprinting
Nature
 , 
2004
, vol. 
429
 (pg. 
900
-
903
)
23
Ooi
S.K.
Qiu
C.
Bernstein
E.
Li
K.
Jia
D.
Yang
Z.
Erdjument-Bromage
H.
Tempst
P.
Lin
S.P.
Allis
C.D.
, et al.  . 
DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA
Nature
 , 
2007
, vol. 
448
 (pg. 
714
-
717
)
24
Jia
D.
Jurkowska
R.Z.
Zhang
X.
Jeltsch
A.
Cheng
X.
Structure of Dnmt3a bound to Dnmt3L suggests a model for de novo DNA methylation
Nature
 , 
2007
, vol. 
449
 (pg. 
248
-
251
)
25
Gregory
R.I.
Randall
T.E.
Johnson
C.A.
Khosla
S.
Hatada
I.
O'Neill
L.P.
Turner
B.M.
Feil
R.
DNA methylation is linked to deacetylation of histone H3, but not H4, on the imprinted genes Snrpn and U2af1-rs1
Mol. Cell. Biol.
 , 
2001
, vol. 
21
 (pg. 
5426
-
5446
)
26
Delaval
K.
Feil
R.
Epigenetic regulation of mammalian genomic imprinting
Curr. Opin. Genet. Dev.
 , 
2004
, vol. 
14
 (pg. 
188
-
195
)
27
Pannetier
M.
Julien
E.
Schotta
G.
Tardat
M.
Sardet
C.
Jenuwein
T.
Feil
R.
PR-SET7 and SUV4-20H regulate H4 lysine-20 methylation at imprinting control regions in the mouse
EMBO Rep.
 , 
2008
, vol. 
9
 (pg. 
998
-
1005
)
28
Wagschal
A.
Sutherland
H.G.
Woodfine
K.
Henckel
A.
Chebli
K.
Schulz
R.
Oakey
R.J.
Bickmore
W.A.
Feil
R.
G9a histone methyltransferase contributes to imprinting in the mouse placenta
Mol. Cell. Biol.
 , 
2008
, vol. 
28
 (pg. 
1104
-
1113
)
29
Xin
Z.
Tachibana
M.
Guggiari
M.
Heard
E.
Shinkai
Y.
Wagstaff
J.
Role of histone methyltransferase G9a in CpG methylation of the Prader–Willi syndrome imprinting center
J Biol. Chem.
 , 
2003
, vol. 
278
 (pg. 
14996
-
15000
)
30
Arnaud
P.
Hata
K.
Kaneda
M.
Li
E.
Sasaki
H.
Feil
R.
Kelsey
G.
Stochastic imprinting in the progeny of Dnmt3L−/− females
Hum. Mol. Genet.
 , 
2006
, vol. 
15
 (pg. 
589
-
598
)
31
Sanz
L.A.
Chamberlain
S.
Sabourin
J.C.
Henckel
A.
Magnuson
T.
Hugnot
J.P.
Feil
R.
Arnaud
P.
A mono-allelic bivalent chromatin domain controls tissue-specific imprinting at Grb10
EMBO J.
 , 
2008
, vol. 
27
 (pg. 
2523
-
2532
)
32
Lucifero
D.
La Salle
S.
Bourc'his
D.
Martel
J.
Bestor
T.H.
Trasler
J.M.
Coordinate regulation of DNA methyltransferase expression during oogenesis
BMC Dev. Biol.
 , 
2007
, vol. 
7
 pg. 
36
 
33
Schulz
R.
Woodfine
K.
Menheniott
T.R.
Bourc'his
D.
Bestor
T.
Oakey
R.J.
WAMIDEX: a web atlas of murine genomic imprinting and differential expression
Epigenetics
 , 
2008
, vol. 
3
 (pg. 
89
-
96
)
34
Chotalia
M.
Smallwood
S.A.
Ruf
N.
Dawson
C.
James
K.
Dean
W.
Kelsey
G.
Transcription is required for establishment of germline methylation marks at imprinted genes
Genes Dev.
 , 
2009
, vol. 
23
 (pg. 
105
-
117
)
35
Li
X.
Ito
M.
Zhou
F.
Youngson
N.
Zuo
X.
Leder
P.
Ferguson-Smith
A.C.
A maternal-zygotic effect gene, Zfp57, maintains both maternal and paternal imprints
Dev. Cell
 , 
2008
, vol. 
15
 (pg. 
547
-
557
)
36
Li
H.
Rauch
T.
Chen
Z.X.
Szabó
P.E.
Riggs
A.D.
Pfeifer
G.P.
The histone methyltransferase SETDB1 and the DNA methyltransferase DNMT3A interact directly and localize to promoters silenced in cancer cells
J. Biol. Chem.
 , 
2006
, vol. 
281
 (pg. 
19489
-
19500
)
37
Jelinic
P.
Stehle
J.C.
Shaw
P.
The testis-specific factor CTCFL cooperates with the protein methyltransferase PRMT7 in H19 imprinting control region methylation
PLoS Biol.
 , 
2006
, vol. 
4
 pg. 
e355
 
38
Arnaud
P.
Feil
R.
Epigenetic deregulation of genomic imprinting in human disorders and following assisted reproduction
Birth Defects Res. C Embryo Today
 , 
2005
, vol. 
75
 (pg. 
81
-
97
)
39
Sharif
J.
Muto
M.
Takebayashi
S.
Suetake
I.
Iwamatsu
A.
Endo
T.A.
Shinga
J.
Mizutani-Koseki
Y.
Toyoda
T.
Okamura
K.
, et al.  . 
The SRA protein Np95 mediates epigenetic inheritance by recruiting Dnmt1 to methylated DNA
Nature
 , 
2007
, vol. 
450
 (pg. 
908
-
912
)
40
Karagianni
P.
Amazit
L.
Qin
J.
Wong
J.
ICBP90, a novel methyl K9 H3 binding protein linking protein ubiquitination with heterochromatin formation
Mol. Cell. Biol.
 , 
2008
, vol. 
28
 (pg. 
705
-
717
)
41
Wu
M.Y.
Tsai
T.F.
Beaudet
A.L.
Deficiency of Rbbp1/Arid4a and Rbbp1l1/Arid4b alters epigenetic modifications and suppresses an imprinting defect in the PWS/AS domain
Genes Dev.
 , 
2006
, vol. 
20
 (pg. 
2859
-
2870
)
42
Wagschal
A.
Delaval
K.
Pannetier
M.
Arnaud
P.
Feil
R.
Chromatin immunoprecipitation (ChIP) on unfixed chromatin from cells and tissues to analyze histone modifications
2007
 
CSH Protocols, 10.1101/pdb.prot4767