Reconstitution of G9a/GLP complex carrying mutations in their SET-domains. (A) Amino-acid alignment of the SET-domains of mG9a-L (Genbank, 21832048) and mGLP (Genbank, 60115440). Conserved and similar residues (shown as +) are shown between the two sequences. Mutagenized amino acids are shown in grey boxes. (B) Table showing serial mutants of G9a and GLP. (C) In vitro histone methyltransferase activity of mutant SET-domains of G9a (left) and GLP (right). GST-fused carboxy terminus of mouse G9a (969-1263 of G9a-L) and that of mouse GLP (1002–1296) were incubated with recombinant histone H3/H4 and S-adenosyl-[methyl-14C]-L-methionine, and then separated by SDS–PAGE. Coomassie staining shows enzymes and substrates. Autoradiography shows 14C-methyl group incorporated into recombinant H3. Activities of mutant enzymes were quantified using image J (http://rsb.info.nih.gov/ij/) and those of GWT or LWT are normalized to 100%. (D) List of ES cell lines expressing mutant G9a/GLP complexes. Serial cell lines were established by stable introduction of the expression plasmids for full-length of G9a-L or GLP into the corresponding KO cells shown in Supplementary Figure S1. Expression levels of mutant G9a or GLP proteins are shown in Supplementary Figure S2. (E) The ability of serial mutant proteins to form heterodimers are numerically represented as G9a/GLP interaction index. Protein levels within anti-G9a or anti-GLP immune complexes shown in Supplementary Figure S3 were measured by image J software, and the ratio of GLP protein immunoprecipitated with anti-G9a to those immunoprecipitated with anti-GLP were calculated. The interaction index of wild-type (WT) ES cells (top left panel) is ∼100%, indicating that endogenous G9a and GLP form a stoichiometric (1:1) heteromer in wild-type cells. (F) H3K9 methylation status of serial mutant ES lines were measured by immunoblot analysis and numerically represented as H3K9me recovery (Raw data are presented in Supplementary Figure S4). The values of signal intensities for methylated H3K9 of GW and LW are normalized to 100%, and those of the corresponding KO cells are normalized to 0%. Data represents an average value of the H3K9 recovery of two clones established independently.

G9a/GLP complex formation induces DNA methylation of the Mage-a2 promoter and the 5′-upstream region of the Wfdc15a gene. (A) A schematic diagram of CpG positions and amplicon for ChIP analysis within Mage-a2 promoter is shown. DNA methylation status of eight CpG sequences (no. 1-8) within the promoter (0 to −287) and a CpG (no. 9) located at the 5′ end of the transcribed region were examined in this study. (B) Methylation differences among serial ES lines expressing mutant G9a/GLP complex at the Mage-a2 promoter. DNA methylation status of CpG positions nos. 1–9 shown in (A) was measured by bisulphite sequencing. Open circles indicate unmethylated cytosines and filled circles indicate methylated cytosines. (C) Schematic diagram of CpG positions and amplicon for ChIP analysis within the 5′-upstream region of Wfdc15a. (D, E) Methylation differences among serial ES lines expressing mutant G9a/GLP complex at the position no. 1–8 (D) and no. 9–11 (E) of CpG sequences on the 5′-upstream region of Wfdc15a.

H3K9 methylation suppresses G9a/GLP-target genes in DNA methyltransferase-deficient ES cells. (A) Northern blot analyses of Dnmts-KO ES cells are shown. Total RNA was separated and probed with radiolabelled cDNAs. (B) Expression levels for Mage-a (left) and Wfdc15a (right) are numerically represented. The values for G9a-KO cells are normalized to 100%. Each experiment was carried out twice. (C) Cytosine methylation status of Mage-a2 promoter and 5′-upstream region of Wfdc15a in Dnmt1-KO (left) and Dnmt3a+b DKO (right) ES cells. CpG positions are as described in Figure 3B and D. (D) Histone methylation patterns of the Mage-a2 promoter region (top) and 5′-upstream region of Wfdc15a (bottom) in Dnmts-KO ES cells were determined by ChIP analyses. H3K9me2 around the promoter is still maintained in Dnmt1-KO and Dnmt3a+b DKO ES cells. Experiments were carried out twice.

Schematic representation of G9a/GLP complex-mediated gene suppression. One candidate molecule links G9a/GLP to DNA methyltransferase is Wiz, as it binds preferentially G9a/GLP heteromer (Ueda et al, 2006). Genetic analyses indicate that orphan G9a (right) or GLP (left) is both incapable of catalysing H3K9 methylation and recruiting DNA methylation, accompanied with the reactivation of the target genes (see panels of G9a-KO and GLP-KO). Mutant complex incapable of H3K9 methylation, such as GM4/LWT or GM7/LWT, can still suppress transcription via recruiting DNA methyltransferase(s) (see panels of inactive G9a/GLP). This suppression is cancelled by 5′Aza-dC treatment. G9a/GLP complex can also suppress transcription even without DNA methylation, solely by depositing H3K9 methylation (see panel of Dnmts-KO).

Effect of G9a or GLP mutations on transcriptional suppression. (A) Northern blot analyses of Mage-a and Wfdc15a genes in ES cells expressing mutant G9a/GLP are shown. Total RNA collected from two independent clones were separated and probed with radiolabelled cDNAs. (B) Signal intensities for the transcripts of Mage-a (left) and Wfdc15a (right) were calculated by image J software and numerically represented. The value for G9a-KO cells was set at 100%. Data represents an average value between two independent clones. (C) Differences in the histone modification patterns on the Mage-a2 promoter region (top) and 5′-upstream region of Wfdc15a (bottom) in the serial ES cell lines were measured by chromatin immunoprecipitation (ChIP) analysis using anti-H3K9me1 and me2 (left) and H3K4me2 (right). Each experiment was performed twice.

Inhibition of DNA methyltransferase activity reactivates G9a/GLP-target gene expression independent of H3K9 methylation. (A) Serial ES lines were cultured with histone deacetylase inhibitor, tricostatin-A (TSA) or the DNA methyltransferase inhibitor, 5-Aza-2′-deoxycytidine (5AZ). Total RNA from treated cells was isolated and analysed through northern blotting using radiolabelled cDNA. (B) The effect of TSA on H3 acetylation was confirmed by immunoblot analysis with specific antibodies directed against acetyl histone H3. Ponceau staining represents equal loading of histones. (C) Reactivation of the transcripts for Mage-a (left) and Wfdc15a (right) is represented numerically. The values for G9a-KO cells without drug treatment are normalized to 100%. Experiments were performed twice.

Dnmt3a and 3b redundantly methylate CpG sequences on G9a/GLP-target genes. (A) Dnmt3 protein expression of several rescued ES lines on Dnmt3a+b DKO background, were analysed with mouse monoclonal antibody 64B1446, which recognizes carboxy terminus of both Dnmt3a and 3b (Chen et al, 2002). (B) Cytosine methylation status of Mage-a2 promoter and 5′-upstream region of Wfdc15a in Dnmt3-rescued line. CpG positions are as described in Figure 3B and D.