The N-terminus of histone H3 is required for de ... - Semantic Scholar

The N-terminus of histone H3 is required for de novo DNA methylation in chromatin Jia-Lei Hua,b, Bo O. Zhoua,b, Run-Rui Zhanga,b, Kang-Ling Zhangc, Jin-Qiu Zhoua,1, and Guo-Liang Xua,1 aThe State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 320 Yueyang Road, Shanghai 200031, China; bThe Graduate School, Chinese Academy of Sciences, 320 Yueyang Road, Shanghai 200031, China; and cDepartment of Biochemistry, School of Medicine, Loma Linda University, Loma Linda, CA 92350

DNA methylation and histone modification are two major epigenetic pathways that interplay to regulate transcriptional activity and other genome functions. Dnmt3L is a regulatory factor for the de novo DNA methyltransferases Dnmt3a and Dnmt3b. Although recent biochemical studies have revealed that Dnmt3L binds to the tail of histone H3 with unmethylated lysine 4 in vitro, the requirement of chromatin components for DNA methylation has not been examined, and functional evidence for the connection of histone tails to DNA methylation is still lacking. Here, we used the budding yeast Saccharomyces cerevisiae as a model system to investigate the chromatin determinants of DNA methylation through ectopic expression of murine Dnmt3a and Dnmt3L. We found that the N terminus of histone H3 tail is required for de novo methylation, while the central part encompassing lysines 9 and 27, as well as the H4 tail are dispensable. DNA methylation occurs predominantly in heterochromatin regions lacking H3K4 methylation. In mutant strains depleted of H3K4 methylation, the DNA methylation level increased 5-fold. The methylation activity of Dnmt3a largely depends on the Dnmt3L’s PHD domain recognizing the histone H3 tail with unmethylated lysine 4. Functional analysis of Dnmt3L in mouse ES cells confirmed that the chromatin-recognition ability of Dnmt3L’s PHD domain is indeed required for efficient methylation at the promoter of the endogenous Dnmt3L gene. These findings establish the N terminus of histone H3 tail with an unmethylated lysine 4 as a chromatin determinant for DNA methylation. Dnmt3a 兩 Dnmt3L 兩 H3K4 methylation 兩 PHD domain 兩 yeast Saccharomyces cerevisiae 兩 Dnmt3L promoter

T

he genomic DNA in mammalian cells is not only bound with basic histone proteins to form chromatin, but also covalently methylated at cytosine bases in specific regions. DNA cytosine methylation together with histone modifications in chromatin present profound epigenetic platforms for the regulation of gene expression and genome functions in various developmental and disease processes (1). A fundamental question concerns how DNA methylation occurs in the first place. In mammals, de novo methyltransferases Dnmt3a and 3b function to establish the methylation patterns in the genome, and the maintenance enzyme Dnmt1 acts to replicate the methylation patterns in cell divisions (2). Cross-talk between DNA methylation and histone modification exists in various contexts of transcriptional regulation and chromatin functions (3). There is some evidence that chromatin cues might be a determinant of DNA methylation. In the filamentous fungus Neurospora crassa, either the replacement of histone H3 lysine 9 (H3K9) with other amino acids, or deletion of the sole H3K9 methyltransferase DIM5 results in the loss of DNA methylation (4). In mouse ES cells, deficiency in the histone H3K9 methyltransferases Suv39h1- Suv39h2 is associated with hypomethylation at a subset of repeat elements (5). These correlative observations fail to establish a clear link between histone modifications and DNA methylation, partially because deletion of a histone methyltransferase or loss of a histone modification can have various indirect effects, including an influence on the www.pnas.org兾cgi兾doi兾10.1073兾pnas.0905767106

maintenance methylation. For example, recent studies showed that G9a exerts its effect on DNA methylation independently of its enzymatic activity (6–8), and depletion of this protein might have an impact on the maintenance function of Dnmt1 (9, 10). While a repressive mark such as H3K9 methylation can presumably trigger DNA methylation, the active chromatin mark H3K4 methylation has been implicated as a repulsive signal (11). Dnmt3L, the regulatory protein that forms a complex with Dnmt3a and Dnmt3b (12), binds specifically to the unmethylated H3 histone tail (11). This biochemical property of Dnmt3L leads to a model that the unmethylated state of H3 lysine 4 provides a chromatin clue for region-specific methylation by Dnmt3a, whereas other sequences are masked from this modification due to their association with repulsive histone marks (11). Although this model is consistent with the notion that genomic methylation is heavily associated with heterochromatin that is depleted of histone H3K4 methylation (13, 14), it lacks support from direct functional evidence in vivo. In this study, we exploited model organism, budding yeast Saccharomyces cerevisiae, to study DNA methylation. As an eukaryote, S. cerevisiae has conserved histones, but fewer types of histone modifications than other higher organisms. Most importantly, DNA methylation is absent from the yeast genome. This allows for sensitive detection of de novo methylation if mammalian DNA methyltransferases are functional in this heterologous host. In particular, we sought to address the chromatin determinants of de novo DNA methylation by the Dnmt3aDnmt3L complex, including the relationship between H3K4 methylation and DNA methylation. Our studies have revealed a requirement of the histone H3 tail for DNA methylation. The methylation function of Dnmt3a depends largely on the ability of Dnmt3L to recognize unmethylated H3K4 in target chromatin regions. These findings are supported by a functional assay of Dnmt3L in differentiating mouse ES cells. Results DNA Methylation by Murine Methyltransferases Expressed In Yeast.

Because yeast cells lack DNA cytosine methylation, we first examined whether murine DNA methyltransferases Dnmt3a and its positive regulator Dnmnt3L would function in this heterologous host. The ectopic expression of HA-tagged Dnmt3a and Flag-tagged Dnmt3L under the control of GAL1 promoter was induced with galactose (Fig. 1A). Cytosine methylation in the host genomic DNA was assessed with HPLC analysis. In DNA purified from the strain expressing Dnmt3a, a small peak was evident at the position exactly corresponding to the 5-meC Author contributions: J.-L.H., J.-Q.Z., and G.-L.X. designed research; J.-L.H. and R.-R.Z. performed research; B.O.Z., K.-L.Z., and J.-Q.Z. contributed new reagents/analytic tools; J.-L.H. and G.-L.X. analyzed data; and J.-L.H., J.-Q.Z., and G.-L.X. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1To whom correspondence may be addressed. E-mail: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0905767106/DCSupplemental.

PNAS 兩 December 29, 2009 兩 vol. 106 兩 no. 52 兩 22187–22192

BIOCHEMISTRY

Edited by Arthur D. Riggs, Beckman Research Institute of the City of Hope, Duarte, CA, and approved November 3, 2009 (received for review May 28, 2009)

Fig. 1. Yeast genomic DNA methylation by coexpressed Dnmt3a-Dnmt3L. (A) Western detection of HA-tagged Dnmt3a and Flag-tagged Dnmt3L expressed in different yeast strains upon galactose induction. Each strain was transformed with one or both of the expression constructs (⫹) or empty vectors (⫺) as indicated on the Top. (B) HPLC measurement (see Materials and Methods) of the methylcytosine content in genomic DNA isolated from the various strains. The bar graph shows the average percentages of methylcytosines in total cytosines and standard deviation among three clones for each strain.

standard (Fig. S1). This peak represents 0.07% of the total cytosine amount (Fig. 1B, lane 3). The accuracy of HPLC analysis was confirmed by mass spectrometry using the same DNA samples (data not shown). Moreover, no methylation was detected when Dnmt3a was absent (Fig. 1B, lanes 1 and 2). Notably, when Dnmt3a was expressed with Dnmt3L, the methylation level increased 9-fold to 0.6% (Fig. 1B). While the methylation targets did not show obvious sequence specificity in the flanking nucleotides, CpG was preferred ⬇10-fold over CpA (Fig. S2 and Table S1). This is consistent with the genomic methylation feature reported for mouse ES cells (15). Taken together, our data demonstrate the functionality of mammalian Dnmt3a and Dnmt3L for de novo methylation in the yeast system that can mimic the mammalian methylation system. Histone H3 N-Terminal Residues Are Crucial for DNA Methylation While the H4 Tail Is Dispensable. Genomic DNA is complexed with

histones, which are highly conserved among most species. To define the role of histone tails in DNA methylation, we generated a series of yeast strains (Table S2) that contain deletion and point mutations in the histone H3 and H4 genes. The mutated histone H3 and H4 gene were cloned into a yeast CEN plasmid. The mutant constructs were transformed into yeast strain JLH511 (Table S2), which lacks the chromosomal copies of the H3/H4 genes, but harbors a plasmid carrying the wild-type H3 or H4 allele under the control of their endogenous promoters. The preexisting plasmid was subsequently removed by negative selection using 5-fluoroorotic acid (5-FOA). The resulting strains expressing a histone mutant each were then transformed with Dnmt3a and Dnmt3L expression constructs. Western blot analysis of yeast cell extracts confirmed the expression of Flag-tagged Dnmt3a and Dnmt3L (Fig. 2A). Their low expression levels in strains bearing H3 R2A and H4 ⌬1–12 are presumably due to the direct effect of histone mutation on the GAL1 promoter as reported (16). While DNA methylation reached approximately 0.5% in the strain expressing the wild-type histone H3 from the episomal plasmid (a level similar to the YPH499 strain with a chromosomal H3 gene, Fig. 1B), truncation of its N-terminal tail by as few as four amino acid residues abolished DNA methylation by Dnmt3a-Dnmt3L (Fig. 2 A and B, lanes 1–4). In contrast, deletion of the central part (⌬13–28) of histone H3 N-terminal tail provoked a 2-fold increase of the methylation level. As arginine 2 (R2) and lysine 4 (K4) are the two key residues of the H3 tail terminus subjected to posttranslational methylation, we respectively made substitutions for these two positions. While substitution of K4 for arginine (K4R) had a negative effect on DNA methylation, the replacement of R2 with alanine (R2A) appeared to enhance the methylation activity, given the drastic 22188 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0905767106

Fig. 2. Requirement of the N-terminal of histone H3 tail for DNA methylation. (A) Western detection of Flag-tagged Dnmt3a and Dnmt3L proteins in yeast strains carrying mutant histone H3 or H4. (B) Methylcytosine contents in histone mutant strains coexpressing Dnmt3a and Dnmt3L. For each histone mutant, three individual clones were analyzed. (C) MNase digestion assay of chromatin structure in strains with wild-type and mutant histones. Permeabilized nuclei were isolated from cells grown at logarithmic phase, and treated with MNase at increasing concentrations (0, 50, 100, 200, and 400 U/mL).

reduction in Dnmt3a and Dnmt3L protein levels in the strain harboring R2A mutation (Fig. 2 A and B). The stimulatory effect exhibited by the H3 (⌬13–28 and R2A) mutations might be mediated by reduced H3K4 methylation [Fig. S3 and (17)]. In contrast to the essential role of H3 N terminus, deletion of the N-terminal tail of histone H4 caused no obvious change in DNA methylation (Fig. 2B, lanes 8 and 9). The slight decrease in H4 ⌬1–12 strain was probably due to the lower protein expression of Dnmt3a and Dnmt3L (Fig. 2 A, lane 8). These results indicate an essential role of the histone H3 tail and the crucial function of a few N-terminal residues for DNA methylation. To exclude the possibility that histone H3 mutants affect DNA methylation via their effects on chromatin assembly, we conducted the micrococcal nuclease (MNase) digestion assay in three deletion strains H3 ⌬1–4, H4 ⌬1–12, and H4 ⌬12–20. Comparisons of the MNase digestion profiles with the wild-type strain showed little alteration (Fig. 2C). Thus, it is unlikely that the histone tail mutations affected DNA methylation by causing a radical change in chromatin configuration. Elimination of Histone H3K4 Methylation Increases DNA Methylation in Yeast. Because the very end of the H3 tail was found to be crucial

for genomic methylation, we next examined whether the lysine 4 methylation status would influence the activity of Dnmt3aDnmt3L. The two proteins were therefore expressed in strains deleted for individual components of the COMPASS complex, the yeast analogue of mammalian MLL complex that catalyzes H3K4 methylation. The integrity and stability of COMPASS complex requires the heterodimer of Swd1/Swd3 (WD40 repeat protein), the catalytic subunit Set1, and the plant homeodomain finger protein Spp1 (18). As expected, deletion of SET1, SWD1, and SWD3 Hu et al.

H3K4 methylation signal was enriched at the coding regions of the actively transcribed genes YPR113W, YLR340W, and YBR119W relative to the silent sites YPL017C, the silent mating-type locus (HMR) and the left telomeric region of chromosome IV (TEL04L), but was absent in the set1⌬ strain (Fig. 4B). The corresponding DNA methylation status in these regions was determined by both restriction assay of the PCR fragments amplified from bisulfite-treated genomic DNA and sequencing analysis of cloned PCR fragments. Both experiments revealed that the coexpressed Dnmt3a and Dnmt3L could methylate sequences in heterochromatin while conferred no methylation at all at the euchromatic sequences (Fig. 4 C and D). Most strikingly, DNA methylation became detectable in the euchromatin when H3K4 methylation was abolished in the set1⌬ strain. These observations clearly indicate that de novo methylation by Dnmt3a-Dnmt3L occurs at sequences associated with unmethylated H3K4, but was prevented at other sequences by the presence of H3K4 methylation.

Fig. 3. Increased DNA methylation in yeast depleted of histone H3 lysine K4 methylation. (A) Western analysis for H3K4 mono-, di-, and trimethylation in strains of set1⌬, swd1⌬, swd3⌬, and spp1⌬. Blotting with antibodies against unmodified H3 and actin served as loading controls. (B) Methylcytosine contents in various strains expressing Dnmt3a alone or together with Dnmt3L. (C) Western analysis for H3K4 mono-, di-, and trimethylation in set1⌬ strains expressing reintroduced WT and mutant Set1 protein. (D) Methylcytosine contents in set1⌬ strains expressing reintroduced wild-type and mutant Set1 protein.

abolished mono-, di-, and trimethylation at H3K4, whereas deletion of SPP1 had no effect except for a decrease in trimethylation (Fig. 3A). Strikingly, the methylation activity of Dnmt3a-Dnmt3L increased greatly in strains deficient in H3K4 methylation. The 5mC content reached 1.7–2.0% in the strains of set1⌬, swd1⌬, and swd3⌬, a 3-fold increase above the wild-type strain, while only a modest increase was seen in the spp1⌬ strain (Fig. 3B). Yet the expression level of exogenous Dnmt3s in COMPASS mutant strains was lower than that in the wild-type strain (Fig. S4), very likely due to the loss of active histone mark H3K4 methylation in these strains. The elevated DNA methylation led to a slow growth phenotype (Fig. S5). Restoration of H3K4 methylation by ectopic expression of wild-type Set1 protein in the set1⌬ mutant strain (Fig. 3C) reduced DNA methylation approximately to the level seen in the wild-type strain (Fig. 3D, compare samples 6–8). Reintroduction of the catalytically inactive mutants (Set1p-N1016Q or -C1068A) (19, 20) into set1⌬ cells did not increase histone methylation (Fig. 3C, lanes 9 and 10), and had no effect on DNA methylation (Fig. 3D). This rescue experiment suggests that elevated DNA methylation in the mutant strains is not attributable to a casual disruption of any unrelated COMPASS-associated functions (21). Therefore, it is likely that DNA methylation directly links to the unmethylated state of H3K4. DNA Methylation Occurs in Genomic Regions Devoid of Histone H3K4 Methylation. Based on the above observations with the COM-

PASS mutants, we postulated that de novo DNA methylation might occur in yeast genomic regions with H3K4 hypomethylation. To test this possibility, we analyzed the distribution of DNA methylation at three genomic loci of heterochromatin as well as three loci of euchromatin (Fig. 4A). Chromatin-immunoprecipation assay confirmed the reported histone methylation status at these examined sites (17). The Hu et al.

unmethylated lysine 4 of histone H3 via its N-terminal PHD domain (11). Dnmt3L also forms a complex with Dnmt3a via its C-terminal domain (22) and stimulates the methylation activity (23–26). We therefore assessed the significance of these dual functions of Dnmt3L for Dnmt3a-mediated de novo methylation in yeast. We generated a deletion mutant of the Dnmt3L PHD domain, as well as two point mutants D124A and I141W, equivalent to D90A and I107W of the human DNMT3L, which specifically disrupts the interaction with the H3 tail (11). Another mutant we constructed was F297A in the C-terminal domain of Dnmt3L that is unable to stimulate the Dnmt3a activity in vitro (22), but can still efficiently interact with Dnmt3a (27). These Dnmt3L mutants were coexpressed with Dnmt3a at a similar level in yeast cells (Fig. 5A). The functional deficiency of the Dnm3L PHD mutants was confirmed by a peptide pull-down assay using the protein purified from yeast. While the wild-type Dnmt3L could bind to the K4-unmethylated H3 peptide irrespective of arginine methylation at R2, all three PHD mutants lost this binding ability (Fig. 5B). HPLC analysis of the genomic DNA revealed that de novo methylation was reduced significantly in cells with PHD-mutated Dnmt3L (Fig. 5C, lanes 2–4). In contrast, the C-terminal point mutation F297A that weakens the stimulatory effect on Dnmt3a in vitro did not show any negative effect on de novo methylation in vivo (Fig. 5C, lane 5). These results suggest that the function of Dnmt3L in recognizing unmethylated-K4 H3 histone might be of primary importance for promoting DNA methylation by Dnmt3a in vivo. De Novo Methylation at the Dnmt3L Promoter in Differentiating ES Cells Is Compromised By Dnmt3L PHD Domain Mutations Affecting the Binding of Unmethylated H3K4 Histone Tails. We have shown above

that Dnmt3L promotes Dnmt3a methylation activity through the binding of its PHD domain to histone H3 in yeast. To evaluate the functional significance of the PHD domain in vivo, we tested the Dnmt3L mutants (D124A, I141W, and ⌬PHD) in mouse ES cells deleted of the endogenous Dnmt3L gene and examined whether the mutations would affect de novo methylation at the Dnmt3L promoter. Dnmt3L has been shown to participate in the methylation of its own promoter during embryonic development and in vitro ES cell differentiation (28). We selected stable cell lines in which the exogenous mutant and wild-type Dnmt3L proteins were expressed at a level close to the endogenous one of wild-type ES cells (Fig. 6A). The cells were cultured to form embryoid bodies (EBs) in which they underwent spontaneous differentiation, a process mimicking mouse early embryonic development. Both COBRA and bisulfate sequencing analysis were performed to examine the methylation state at the Dnmt3L PNAS 兩 December 29, 2009 兩 vol. 106 兩 no. 52 兩 22189

BIOCHEMISTRY

DNA Methylation Depends on the Ability of Dnmt3L to Bind Histone H3 with Unmethylated K4. Dnmt3L interacts specifically with

Fig. 4. Distribution of DNA methylation at genomic sequences associated with unmethylated H3K4. (A) Schematic diagrams of six yeast genomic loci chosen for DNA methylation analysis. Regions for chromatin-immunoprecipitation (ChIP) assay and bisulfite-based DNA methylation analysis are indicated. (B) ChIP with anti-H3K4 trimethylation antibody to examine the histone methylation state in the regions chosen for analysis. (C) COBRA DNA methylation analysis of six representative loci in wild-type and set1⌬ strains. TaqI digestion of a PCR fragment into smaller fragments reflects full or partial methylation (me) in the region analyzed. Resistance to the digestion represents an unmethylated (un) state. (D) Bisulfite sequencing profiles of DNA methylation of four representative loci. CpG sequences are shown with filled (methylated) and open (unmethylated) circles.

promoter (Fig. 6 B and C). The transgenic wild-type Dnmt3L appeared to be more proficient than the endogenous protein (51.4 vs. 36.1%), probably due to its unattenuated expression during differentiation for an ectopic CAG promoter. Most notably, the Dnmt3L mutants conferred at most only half of the methylation seen in the wild-type transgenic counterpart. This drastic reduction in the methylation efficiency actually represented a nullifying effect on the Dnmt3L function because the maximal methylation level (25.7%) in cells expressing the mutants was close to the level (23.6%) in cells deleted of Dnmt3L. These data suggest that the function of Dnmt3L in de novo methylation might fully rely on its binding to histone H3 tails. Discussion We describe here the use of budding yeast S. cerevisiae as a model system to study the mechanism of de novo DNA methylation by mammalian methyltransferases. The histone H3 tail terminus of seven residues was shown to be critical in de novo methylation by providing a binding site for Dnm3L in chromatin. De novo 22190 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0905767106

methylation appeared predominantly in heterochromatin containing hypomethylated lysine 4. Methylation of this residue prevented the associated genomic DNA from being methylated. In support of the role of histone H3 tail in DNA methylation, the ability of Dnmt3L to interact with the terminal region of the H3 tail was shown to be essential for promoting Dnmt3a methylation activity at the Dnmt3L promoter in mouse embryonic stem cells. DNA de novo methyltransferases are intimately associated with chromatin (11, 29, 30). However, the study of the chromatin requirement for DNA methylation in mammalian cells has been hindered by at least three aspects. First, mutational analysis of histones in mammals is impossible because histones are encoded by hundreds of gene copies. Second, histone modification is very complex in mammals, for example, more than 10 different enzymes could act on the methylation of H3 at lysine 4 (31). Finally, mammalian chromatin contains the preexisting DNA methylation. The use of budding yeast largely circumvents these problems. Yeast host strains deleted for both copies of chromosomal H3 or H4 genes can express mutant histones from an Hu et al.

episomal vector. In addition, yeast histone modifications are relatively simple. For example, lysine methylation only occurs at K4, K36, and K79 on H3, and repressive marks such as H3K9 methylation and H3K27 methylation commonly found in other eukaryotes are absent. More importantly, yeast cells do not have any endogenous DNA methylation and thus provides a clean substrate in vivo for the functional study of exogenously expressed mammalian methyltransferases. De novo DNA methylation is one of the events in a multistep gene silencing process in which certain chromatin modifications, including histone deacetylation also take place (32). The extent of the requirement for the action of de novo methyltransferases is poorly understood in mammals. Repressive marks generated by locally recruited chromatin-modifying enzymes are thought to serve as signals for DNA methylation (1, 33). In this connection, deficiency in histone H3K9 methyltransferases Suv39h1 and Suv39h2 causes loss of methylation at satellite repeats in mouse ES cells (5). Another histone H3K9 methyltransferase, G9a has been implicated in methylation of the Oct4 promoter (32). Despite the essential role of H3K9 methylation clearly demonstrated for DNA methylation in fungi and plants (4, 34), we are concerning whether mammalian DNA methylation machinery might use different chromatin clues given the fact that the regulatory domains of mammalian de novo methyltransferases show no conservation with the fungal and plant enzymes. Our observation of robust methylation activity of Dnmt3a-Dnmt3L in yeast, which is free of repressive marks, demonstrates that the mammalian enzymes can function independently of histone K9 methylation. Significant methylation occurred even when the region of H3 tail encompassing K9 was deleted (Fig. S6). This is in a sharp contrast to the finding in Neurospora in which lysine 9 and its trimethylation are absolutely required for DNA methylation (4). Instead, we found that the absence of an active mark (H3K4 methylation) on the H3 tail signals DNA methylation, in favor of the model suggested by Ooi (11). Although we cannot exclude the possible role of a repressive mark in enhancing DNA methylation in the presence of a mediating factor such as HP1, it is clear that the mammalian methyltransferase complex is able to interact directly with an unmodified H3 histone tail terminus to confer de novo methylation on nucleosomal DNA in vivo. The maximal amount of 5mC generated in the yeast genome upon induction of Dnmt3a and Dnmt3L is 2%, which is below the Hu et al.

BIOCHEMISTRY

Fig. 5. Functional dependence of Dnmt3L on binding to unmethylated histone tails. (A) Western detection of Flag-tagged Dnmt3L PHD-mutants coexpressed with HA-Dnmt3a in yeast. (B) Peptide binding assay for the Dnmt3L PHD mutant proteins purified from yeast. Note that only the wildtype Dnmt3L control could bind to the H3 peptide that lacked modification at lysine 4. WDR5 (WD repeat domain 5) known to bind H3 terminus with unmethylated arginine 2 (40) was used as a control. (C) Requirement of Dnmt3L’s PHD function to promote the Dnmt3a methylation activity.

Fig. 6. Function of Dnmt3L’s PHD in de novo methylation at the Dnmt3L promoter in differentiating mouse embryonic stem cells. (A) Western detection of the Dnmt3L PHD mutant proteins expressed in Dnmt3L-knockout ES cells. Flag-tagged Dnmt3L wild-type and mutants with substitutions D124A, I141W and PHD deletion were introduced into Dnmt3L-knockout ES cells to generate stable cell lines. Wild-type (WT) and Dnmt3L knockout (KO) ES cell lines were used as controls. GAPDH was used for loading normalization. (B) COBRA analysis of the methylation level at the Dnmt3L promoter upon in vitro differentiation of ES cells. PCR fragments amplified from bisulfite-treated genomic DNA were digested with TaqI. The relative amount of DNA sensitive and resistant to the digestion reflects methylated (me) and unmethylated (un) states at the TaqI sites examined. (C) Bisulfite sequencing profiles of DNA methylation of the Dnmt3L promoter in differentiated ES cells. CpG sequences are shown with filled (methylated) and open (unmethylated) circles. The numbers above the profiles show the percentages of methylated CpGs. Arrows below indicate the CpG dinucleotides within the TaqI recognition sequence.

3–5% present in mammalian cells. Of note, this level was achieved when H3K4 methylation was completely erased in the COMPASS mutant strains. In addition to the lack of maintenance-methylation PNAS 兩 December 29, 2009 兩 vol. 106 兩 no. 52 兩 22191

machinery that can propagate methylation patterns in fast-dividing yeast cells, the differences in the yeast epigenome might also limit the activity of the mammalian methyltransferase. Chromatin modifications in the budding yeast differ substantially in type and extent from those of mammals. While most repressive marks including H3K9 and H3K27 methylation and silencing factors such as heterochromatin proteins (HP1) common in higher organisms are missing, the yeast chromatin is rich in active marks such as acetylation. Conceivably these active marks may mask the yeast genome from de novo methylation in a similar manner to H3K4 methylation. Nevertheless it is due to the presence of distinctive histone modifications, silencing factors as well as histone variants that the yeast presents a unique system for the study of mammalian DNA methylation. Materials and Methods

Chromatin Immunoprecipitation Assay. Chromatin immunoprecipitation (ChIP) assays were performed as described in ref. 36, with slight modifications. Briefly, yeast cells were grown at an OD600 of approximately 1.2 and cross-linked with 1% formaldehyde. Cells were lysed with glass beads and sonicated to shear the chromatin to fragment sizes of 250- to 500-bp. Cross-linked chromatin fragments were immunoprecipitated with rabbit IgG (Sigma) or Abs against H3 (Upstate) or H3K4me3 (Abcam) bound to Protein A Sepharose beads (GE Healthcare). After the cross-links were reversed, the DNA fragments were extracted for PCR assay. The ChIP primers are listed in Table S3. Establishment of Stable ES Cell Lines and In Vitro Differentiation. Dnmt3L⫺/⫺ ES cells (37) were transfected with pCAG-IRESblast (38) expression constructs encoding Flag-tagged Dnmt3L. Blasticidin selection for stable cell lines was performed as described in ref. 38. At least 15 drug-resistant colonies were picked and expanded to assay for recombinant protein expression. In vitro differentiation was performed essentially as described in ref. 39.

DNA Methylation Analysis. HPLC experiments were performed as described in ref 35. with slight modifications. Details about the procedure for HPLC and bisulfite methylation analysis are included in the SI Text.

ACKNOWLEDGMENTS. We thank Randall H. Morse for plasmids and yeast strains, Li Shen for help with HPLC, Jun Yan for bioinformatics, and C. P. Walsh for critical review of the manuscript. This work was supported by the Ministry of Science and Technology of China Grants 2005CB522400 and 2006CB943900 and the National Science Foundation of China and Chinese Academy of Sciences Grant KSCX2-YWN-019 (to G.-L.X.) and Ministry of Science and Technology of China Grant 2005CB522402 and National Science Foundation of China Grant 90919027 (to J.-Q.Z.).

1. Jaenisch R, Bird A (2003) Epigenetic regulation of gene expression: How the genome integrates intrinsic and environmental signals. Nat Gen 33:245–254. 2. Bestor TH (2000) The DNA methyltransferases of mammals. Hum Mol Gen 9:2395–2402. 3. Li E (2002) Chromatin modification and epigenetic reprogramming in mammalian development. Nat Rev Genet 3:662– 673. 4. Tamaru H, Selker EU (2001) A histone H3 methyltransferase controls DNA methylation in Neurospora crassa. Nature 414:277–283. 5. Lehnertz B, et al. (2003) Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin. Curr Biol 13:1192–1200. 6. Dong KB, et al. (2008) DNA methylation in ES cells requires the lysine methyltransferase G9a but not its catalytic activity. EMBO J 27:2691–2701. 7. Tachibana M, Matsumura Y, Fukuda M, Kimura H, Shinkai Y (2008) G9a/GLP complexes independently mediate H3K9 and DNA methylation to silence transcription. EMBO J 27:2681–2690. 8. Epsztejn-Litman S, et al. (2008) De novo DNA methylation promoted by G9a prevents reprogramming of embryonically silenced genes. Nat Struct Mol Bio 15:1176 –1183. 9. Esteve PO, et al. (2006) Direct interaction between DNMT1 and G9a coordinates DNA and histone methylation during replication. Genes Dev 20:3089 –3103. 10. Sharif J, et al. (2007) The SRA protein Np95 mediates epigenetic inheritance by recruiting Dnmt1 to methylated DNA. Nature 450:908 –912. 11. Ooi SK, et al. (2007) DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature 448:714 –717. 12. Cheng X, Blumenthal RM (2008) Mammalian DNA methyltransferases: A structural perspective. Structure 16:341–350. 13. Mohn F, et al. (2008) Lineage-specific polycomb targets and de novo DNA methylation define restriction and potential of neuronal progenitors. Mol Cell 30:755–766. 14. Meissner A, et al. (2008) Genome-scale DNA methylation maps of pluripotent and differentiated cells. Nature 454:766 –770. 15. Ramsahoye BH, et al. (2000) Non-CpG methylation is prevalent in embryonic stem cells and may be mediated by DNA methyltransferase 3a. Proc Natl Acad Sci USA 97:5237– 5242. 16. Durrin LK, Mann RK, Kayne PS, Grunstein M (1991) Yeast histone H4 N-terminal sequence is required for promoter activation in vivo. Cell 65:1023–1031. 17. Kirmizis A, et al. (2007) Arginine methylation at histone H3R2 controls deposition of H3K4 trimethylation. Nature 449:928 –932. 18. Dehe PM, et al. (2006) Protein interactions within the Set1 complex and their roles in the regulation of histone 3 lysine 4 methylation. J Biol Chem 281:35404 –35412. 19. Santos-Rosa H, et al. (2003) Methylation of histone H3 K4 mediates association of the Isw1p ATPase with chromatin. Mol Cell 12:1325–1332. 20. Santos-Rosa H, Bannister AJ, Dehe PM, Geli V, Kouzarides T (2004) Methylation of H3 lysine 4 at euchromatin promotes Sir3p association with heterochromatin. J Biol Chem 279:47506 – 47512. 21. Nislow C, Ray E, Pillus L (1997) SET1, a yeast member of the trithorax family, functions in transcriptional silencing and diverse cellular processes. Mol Biol Cell 8:2421–2436. 22. Jia D, Jurkowska RZ, Zhang X, Jeltsch A, Cheng X (2007) Structure of Dnmt3a bound to Dnmt3L suggests a model for de novo DNA methylation. Nature 449:248 –251.

23. Suetake I, Shinozaki F, Miyagawa J, Takeshima H, Tajima S (2004) DNMT3L stimulates the DNA methylation activity of Dnmt3a and Dnmt3b through a direct interaction. J Biol Chem 279:27816 –27823. 24. Gowher H, Liebert K, Hermann A, Xu G, Jeltsch A (2005) Mechanism of stimulation of catalytic activity of Dnmt3A and Dnmt3B DNA-(cytosine-C5)-methyltransferases by Dnmt3L. J Biol Chem 280:13341–13348. 25. Chen ZX, Mann JR, Hsieh CL, Riggs AD, Chedin F (2005) Physical and functional interactions between the human DNMT3L protein and members of the de novo methyltransferase family. J Cell Biochem 95:902–917. 26. Chedin F, Lieber MR, Hsieh CL (2002) The DNA methyltransferase-like protein DNMT3L stimulates de novo methylation by Dnmt3a. Proc Natl Acad Sci USA 99:16916 –16921. 27. Jurkowska RZ, et al. (2008) Formation of nucleoprotein filaments by mammalian DNA methyltransferase Dnmt3a in complex with regulator Dnmt3L. Nucleic Acids Res 36:6656 – 6663. 28. Hu YG, et al. (2008) Regulation of DNA methylation activity through Dnmt3L promoter methylation by Dnmt3 enzymes in embryonic development. Hum Mol Gen 17:2654 – 2664. 29. Ge YZ, et al. (2004) Chromatin targeting of de novo DNA methyltransferases by the PWWP domain. J Biol Chem 279:25447–25454. 30. Li JY, et al. (2007) Synergistic function of DNA methyltransferases Dnmt3a and Dnmt3b in the methylation of Oct4 and Nanog. Mol Cell Biol 27:8748 – 8759. 31. Ruthenburg AJ, Allis CD, Wysocka J (2007) Methylation of lysine 4 on histone H3: Intricacy of writing and reading a single epigenetic mark. Mol Cell 25:15–30. 32. Feldman N, et al. (2006) G9a-mediated irreversible epigenetic inactivation of Oct-3/4 during early embryogenesis. Nat Cell Bio 8:188 –194. 33. Bird A (2002) DNA methylation patterns and epigenetic memory. Genes Dev 16:6 –21. 34. Jackson JP, Lindroth AM, Cao X, Jacobsen SE (2002) Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416:556 –560. 35. Ramsahoye BH (2002) Measurement of genome wide DNA methylation by reversedphase high-performance liquid chromatography. Methods 27:156 –161. 36. Hecht A, Grunstein M (1999) Mapping DNA interaction sites of chromosomal proteins using immunoprecipitation and polymerase chain reaction. Methods Enzymol 304:399 – 414. 37. Hata K, Okano M, Lei H, Li E (2002) Dnmt3L cooperates with the Dnmt3 family of de novo DNA methyltransferases to establish maternal imprints in mice. Development 129:1983–1993. 38. Chen T, Ueda Y, Dodge JE, Wang Z, Li E (2003) Establishment and maintenance of genomic methylation patterns in mouse embryonic stem cells by Dnmt3a and Dnmt3b. Mol Cell Bio 23:5594 –5605. 39. Gu P, Le Menuet D, Chung AC, Cooney AJ (2006) Differential recruitment of methylated CpG binding domains by the orphan receptor GCNF initiates the repression and silencing of Oct4 expression. Mol Cell Bio 26:9471–9483. 40. Iberg AN, et al. (2008) Arginine methylation of the histone H3 tail impedes effector binding. J Biol Chem 283:3006 –3010.

Yeast Strains and Plasmid Constructions. The yeast strains used in this study are listed in Table S2. Details for constructs are presented in the SI Text and the primers used are shown in Table S3.

22192 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0905767106

Hu et al.