DNA methylation is an important epigenetic modification that is essential for various developmental processes through regulating gene expression, genomic imprinting, and epigenetic inheritance1,2,3,4,5. Mammalian genomic DNA methylation is established during embryogenesis by de novo DNA methyltransferases, DNMT3A and DNMT3B6,7,8, and the methylation patterns vary with developmental stages and cell types9,10,11,12. DNA methyltransferase 3-like protein (DNMT3L) is a catalytically inactive paralogue of DNMT3 enzymes, which stimulates the enzymatic activity of Dnmt3a13. Recent studies have established a connection between DNA methylation and histone modifications, and revealed a histone-guided mechanism for the establishment of DNA methylation14. The ATRX–DNMT3–DNMT3L (ADD) domain of Dnmt3a recognizes unmethylated histone H3 (H3K4me0)15,16,17. The histone H3 tail stimulates the enzymatic activity of Dnmt3a in vitro17,18, whereas the molecular mechanism remains elusive. Here we show that DNMT3A exists in an autoinhibitory form and that the histone H3 tail stimulates its activity in a DNMT3L-independent manner. We determine the crystal structures of DNMT3A–DNMT3L (autoinhibitory form) and DNMT3A–DNMT3L-H3 (active form) complexes at 3.82 and 2.90 Å resolution, respectively. Structural and biochemical analyses indicate that the ADD domain of DNMT3A interacts with and inhibits enzymatic activity of the catalytic domain (CD) through blocking its DNA-binding affinity. Histone H3 (but not H3K4me3) disrupts ADD–CD interaction, induces a large movement of the ADD domain, and thus releases the autoinhibition of DNMT3A. The finding adds another layer of regulation of DNA methylation to ensure that the enzyme is mainly activated at proper targeting loci when unmethylated H3K4 is present, and strongly supports a negative correlation between H3K4me3 and DNA methylation across the mammalian genome9,10,19,20. Our study provides a new insight into an unexpected autoinhibition and histone H3-induced activation of the de novo DNA methyltransferase after its initial genomic positioning.
This is a preview of subscription content
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Get time limited or full article access on ReadCube.
All prices are NET prices.
Jaenisch, R. & Bird, A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nature Genet. 33 (suppl.). 245–254 (2003)
Smith, Z. D. & Meissner, A. DNA methylation: roles in mammalian development. Nature Rev. Genet. 14, 204–220 (2013)
Law, J. A. & Jacobsen, S. E. Establishing, maintaining and modifying DNA methylation patterns in plants and animals. Nature Rev. Genet. 11, 204–220 (2010)
Guibert, S., Forne, T. & Weber, M. Dynamic regulation of DNA methylation during mammalian development. Epigenomics 1, 81–98 (2009)
Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev. 16, 6–21 (2002)
Okano, M., Xie, S. & Li, E. Cloning and characterization of a family of novel mammalian DNA (cytosine-5) methyltransferases. Nature Genet. 19, 219–220 (1998)
Okano, M., Bell, D. W., Haber, D. A. & Li, E. DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99, 247–257 (1999)
Bestor, T. H. The DNA methyltransferases of mammals. Hum. Mol. Genet. 9, 2395–2402 (2000)
Laurent, L. et al. Dynamic changes in the human methylome during differentiation. Genome Res. 20, 320–331 (2010)
Lister, R. et al. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462, 315–322 (2009)
Bird, A. P. & Taggart, M. H. Variable patterns of total DNA and rDNA methylation in animals. Nucleic Acids Res. 8, 1485–1497 (1980)
Ehrlich, M. et al. Amount and distribution of 5-methylcytosine in human DNA from different types of tissues of cells. Nucleic Acids Res. 10, 2709–2721 (1982)
Chedin, F., Lieber, M. R. & Hsieh, C. L. The DNA methyltransferase-like protein DNMT3L stimulates de novo methylation by Dnmt3a. Proc. Natl Acad. Sci. USA 99, 16916–16921 (2002)
Cedar, H. & Bergman, Y. Linking DNA methylation and histone modification: patterns and paradigms. Nature Rev. Genet. 10, 295–304 (2009)
Ooi, S. K. et al. DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature 448, 714–717 (2007)
Otani, J. et al. Structural basis for recognition of H3K4 methylation status by the DNA methyltransferase 3A ATRX–DNMT3–DNMT3L domain. EMBO Rep. 10, 1235–1241 (2009)
Zhang, Y. et al. Chromatin methylation activity of Dnmt3a and Dnmt3a/3L is guided by interaction of the ADD domain with the histone H3 tail. Nucleic Acids Res. 38, 4246–4253 (2010)
Li, B. Z. et al. Histone tails regulate DNA methylation by allosterically activating de novo methyltransferase. Cell Res. 21, 1172–1181 (2011)
Weber, M. et al. Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the human genome. Nature Genet. 39, 457–466 (2007)
Meissner, A. et al. Genome-scale DNA methylation maps of pluripotent and differentiated cells. Nature 454, 766–770 (2008)
Chen, T., Ueda, Y., Xie, S. & Li, E. A novel Dnmt3a isoform produced from an alternative promoter localizes to euchromatin and its expression correlates with active de novo methylation. J. Biol. Chem. 277, 38746–38754 (2002)
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 449, 248–251 (2007)
Klimasauskas, S., Kumar, S., Roberts, R. J. & Cheng, X. HhaI methyltransferase flips its target base out of the DNA helix. Cell 76, 357–369 (1994)
Rajavelu, A., Jurkowska, R. Z., Fritz, J. & Jeltsch, A. Function and disruption of DNA methyltransferase 3a cooperative DNA binding and nucleoprotein filament formation. Nucleic Acids Res. 40, 569–580 (2012)
Shi, P. et al. Site-specific protein backbone and side-chain NMR chemical shift and relaxation analysis of human vinexin SH3 domain using a genetically encoded 15N/19F-labeled unnatural amino acid. Biochem. Biophys. Res. Commun. 402, 461–466 (2010)
Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 (1997)
McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Cryst. 40, 658–674 (2007)
Adams, P. D. et al. PHENIX: building new software for automated crystallographic structure determination. Acta Crystallogr. D 58, 1948–1954 (2002)
Vagin, A. & Teplyakov, A. MOLREP: an automated program for molecular replacement. J. Appl. Cryst. 30, 1022–1025 (1997)
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D 60, 2126–2132 (2004)
Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26, 283–291 (1993)
DeLano, W. L. The PyMOL Molecular Graphics System. http://www.pymol.org (2002)
Roth, M. & Jeltsch, A. Biotin-avidin microplate assay for the quantitative analysis of enzymatic methylation of DNA by DNA methyltransferases. Biol. Chem. 381, 269–272 (2000)
Luger, K., Rechsteiner, T. J. & Richmond, T. J. Expression and purification of recombinant histones and nucleosome reconstitution. Methods Mol. Biol. 119, 1–16 (1999)
Simon, M. D. et al. The site-specific installation of methyl-lysine analogs into recombinant histones. Cell 128, 1003–1012 (2007)
Shi, P. et al. Site-specific (1)(9)F NMR chemical shift and side chain relaxation analysis of a membrane protein labeled with an unnatural amino acid. Protein Sci. 20, 224–228 (2011)
Tang, G. et al. EMAN2: an extensible image processing suite for electron microscopy. J. Struct. Biol. 157, 38–46 (2007)
Ludtke, S. J., Baldwin, P. R. & Chiu, W. EMAN: semiautomated software for high-resolution single-particle reconstructions. J. Struct. Biol. 128, 82–97 (1999)
Ludtke, S. J., Jakana, J., Song, J. L., Chuang, D. T. & Chiu, W. A 11.5 Å single particle reconstruction of GroEL using EMAN. JMB (2001)
van Heel, M., Harauz, G., Orlova, E. V., Schmidt, R. & Schatz, M. A new generation of the IMAGIC image processing system. J. Struct. Biol. 116, 17–24 (1996)
Cong, Y. & Ludtke, S. J. Single particle analysis at high resolution. Methods Enzymol. 482, 211–235 (2010)
Cong, Y. et al. 4.0-Å resolution cryo-EM structure of the mammalian chaperonin TRiC/CCT reveals its unique subunit arrangement. Proc. Natl Acad. Sci. USA 107, 4967–4972 (2010)
Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004)
We thank staff of beamline BL17U at Shanghai Synchrotron Radiation Facility, China, for their assistance in data collection, and H. Wang for help on electron microscopy analyses. We thank staff of the Biomedical Core Facility, Fudan University, for their help on biochemical analyses, and A. D. Riggs for providing the complementary DNAs of DNMT3A and DNMT3L. This work was supported by grants from the National Basic Research Program of China (2011CB965300, 2009CB918600, 2013CB910401), the National Science & Technology Major Project ‘Key New Drug Creation and Manufacturing Program’ of China (2014ZX09507-002, 2011ZX09506-001), the National Natural Science Foundation of China (31270779, 91419301, 31030019, U1432242, 31270771, 31222016, 31300685, U1332138), the Basic Research Project of Shanghai Science and Technology Commission (12JC1402700, 13JC1406300), the Fok Ying Tung Education Foundation (20090071220012), and the Chinese Academy of Sciences Pilot Strategic Science and Technology Projects B (numbers XDB08030201, XDB08030302). Y.C. is a scholar of the Hundred Talents Program of the Chinese Academy of Sciences.
The authors declare no competing financial interests.
Extended data figures and tables
a, In vitro DNA methyltransferase activities of DNMT3A2 (purified from insect cells) in the absence or presence of CDNMT3L. The assays were performed in the presence or absence of histone H3 peptide (residues 1–12). Note that CDNMT3L could enhance the activity of DNMT3A2 by a factor of 2–3, which is consistent with previous study22. However, histone H3-mediated activation of DNMT3A is independent of the existence of CDNMT3L. b, In vitro DNA methyltransferase activities of DNMT3A2 (purified from insect cells) or DNMT3A2–CDNMT3L (purified from bacteria) in the presence or absence of histone H3 peptides. c, Enzymatic activities of DNMT3A2 (purified from insect cells) or DNMT3A2–CDNMT3L (purified from bacteria) using reconstituted nucleosomes as substrates. Nucleosomes containing unmodified histone H3 or H3KC4me3 were subject to SDS–PAGE and visualized using specific antibodies. d, e, Enzymatic activities of various N-terminal deletions of DNMT3A2 in the absence (d) or presence (e) of CDNMT3L. Corresponding relative activities are indicated at the bottom of each figure. CPM, counts per minute. Error bars, s.d. for triplicate experiments. The ADD–CD or CD protein purified from bacteria was not stable in solution and tended to precipitate out, which may have resulted in their lower activities under our experimental conditions (compared with DNMT3A2 purified from insect cells). Because CDNMT3L could stabilize DNMT3A and had no effect on histone H3-mediated activation, protein complexes ADD–CD–CDNMT3L and CD–CDNMT3L were used in the following studies if not specified.
a, b, Two different views of the 2Fobserved − Fcalculated maps for CDNMT3L (a) and CD (b) domains in the ADD–CD–CDNMT3L structure. The maps were calculated at 3.82 Å and contoured at 1.5σ. Only main-chains are shown for simplicity. c, The 2Fobserved − Fcalculated maps for the ADD domain after refinement of the CD–CDNMT3L complex (top) and after refinement of the ADD–CD–CDNMT3L complex (bottom). The maps were calculated at 3.82 Å and contoured at 0.8σ. Main-chains from most residues, including residues 526–533 involved in the interaction with CD domain, fit well into the electron density. Some loop regions were not well covered by electron density, which is consistent with a high B factor (Extended Data Fig. 8a) of the ADD domain in the complex structure, supporting the dynamic feature of the ADD domain for regulating enzymatic activity of DNMT3A. d, Zn-anomalous difference map contoured at 3.5σ shows the positions of zinc cations in the ADD domain. e, Gel filtration profiles for standard proteins and the ADD–CD–CDNMT3L complex. The peak position corresponds to the dimer of ADD–CD–CDNMT3L with a molecular mass of about 140 kDa. f, Dimer formation of the ADD–CD–CDNMT3L complex in crystals. The dimer of ADD–CD–CDNMT3L complexes is mediated by CD–CD interaction in a two-fold crystallographic symmetry. Given the difficulty in tracing the conformation of the side chain in 3.82 Å resolution structure, we have not discussed the specific hydrogen bond or hydrophobic interaction within ADD–CD–CDNMT3L. Residues 832–846 of DNMT3A were not built in the model because they lacked electron density, which may have resulted from their flexibility in crystals.
a, Superimposition of human ADD–CD–CDNMT3L with mouse CDDnmt3a-CDnmt3L (lack of ADD domain, PDB accession number 2QRV)22 structures shown in ribbon representations. CD–CDNMT3L in two structures is well aligned with a root mean squared deviation of 1.28 Å for 723 Cα aligned. The function of DNMT3A–DNMT3L complex dimerization has been characterized in a previous study22. The functions and structures of the CD and CDNMT3L domains, and the CD–CD and CD–CDNMT3L interfaces, were not discussed in this work. b, Overall structure of ADD–CD–CDNMT3L with CD–CDNMT3L shown in electrostatic potential surface, and the ADD domain and linker shown in ribbon representation. The linker packs against a hydrophobic surface of the CD domain. c, d, Close-up view of linker-CD (c) and ADD–CD (d) interfaces with the electrostatic potential surface of the CD domain indicated. Critical residues are shown in stick representation. e, Superimposition of ADD–CD–CDNMT3L and DNMT3L-H3 structures (PDB accession number 2PVC)15 shown in ribbon representations in two different views. The CD domain and C-like domain of DNMT3L were aligned for comparison. Note that the extended loop of the ADD domain in ADD–CD–CDNMT3L overlaps with an α helix in the DNMT3L–H3 structure. DNMT3L is unlikely to adopt a similar conformation to that of ADD–CD–CDNMT3L because otherwise the ADD domain will have steric hindrance with the C-like domain of DNMT3L (dashed circle). According to the above analyses, the structure of the autoinhibitory form of DNMT3A could not be predicted on the basis of the DNMT3L structure because the overall structures of DNMT3A and DNMT3L are different.
Sequences of human DNMT3A (NP_072046), DNMT3B (NP_008823), DNMT3L (NP_787063), mouse Dnmt3a (NP_031898), zebrafish Dnmt3a (NP_001018150), and DNA methyltransferase from Haemophilus parahaemolyticus (WP_005706946) used in the alignment. Highly conserved and identical residues are highlighted with dark green background, and conserved residues are indicated with light green background. Secondary structural elements are coloured as in Fig. 1a and indicated above the sequences. Invisible residues in the structure of ADD–CD–CDNMT3L are indicated as dashed lines above the sequences. Residues involved in ADD–CD interactions in active form or autoinhibitory form are indicated as black stars and red triangles, respectively. Residues involved in H3–ADD interactions are indicated as blue squares.
a, b, GST pull-down assays with recombinant CD (residues 627–912) protein incubated with wild-type or mutant GST–ADD–linker proteins immobilized on glutathione resin. The bound proteins were analysed by SDS–PAGE and Coomassie blue staining. c, GST pull-down assays using wild-type or mutant of the CD domain. d, GST pull-down assays in the absence or presence of histone H3 peptide (H3K4me0 or H3K4me3). e, GST pull-down assays with the CD domain incubated with GST–ADD or GST–ADD–linker proteins immobilized on glutathione resin. f, Activities of wild-type and mutant ADD–CD. Residues 621–632 were replaced by a GS linker in the mutant proteins.
a, Enzymatic activities of wild-type and mutant ADD–CD–CDNMT3L. Residues on the missing loop (residues 831–846) were mutated for the in vitro DNA methyltransferase activity assay. Error bars, s.d. for triplicate experiments. Mutating above residues leads to loss of activity of ADD–CD–CDNMT3L, supporting their important role in catalysis or DNA recognition. The missing loop in the ADD–CD–CDNMT3L structure is equivalent to a DNA-binding loop in the HhaI–DNA structure. b, DNA has no effect on the interaction between histone H3 and DNMT3A. Left, isothermal titration calorimetry enthalpy plot for the binding of isolated ADD domain (in cell) to histone H3 peptide (residues 1–12, in syringe), with the estimated binding affinities (Kd) listed. Right, superimposed isothermal titration calorimetry enthalpy plots for the binding of ADD–CD–CDNMT3L (in cell) to histone H3 peptide (residues 1–12, in syringe) in the absence or presence of dsDNA. The estimated binding affinities (Kd) are listed. Histone H3 peptide has comparable binding affinity to the ADD domain alone (1.75 μM) and ADD–CD–CDNMT3L in autoinhibitory form (2.14 μM), and the addition of DNA was not able to enhance the binding affinity further. The presence of DNA led to a slight decrease in the binding affinity between histone H3 peptide and ADD–CD–CDNMT3L, which may have resulted from slight precipitation of the protein caused by the high concentration of DNA used for titration. c, Electrophoretic mobility-shift assays for DNMT3A proteins in the absence or presence of histone H3 peptide, with protein concentrations indicated. H3–ADD–CD represents a fusion protein with histone H3 (residues 1–20) at the N terminus of ADD–CD. The assays showed that CD–CDNMT3L strongly bound to the FAM-labelled DNA duplex, whereas the existence of the ADD domain markedly decreased DNA-binding affinity, which was partly restored by the addition of histone H3 peptide or largely restored by H3–ADD–CD fusion protein.
a, Ribbon representations of the overall structures of the ADD–CD–CDNMT3L in active (left) and autoinhibitory (right) forms. Histone H3 peptides are coloured in yellow. b, Structural comparison of human ADD–CD–CDNMT3L–H3 and mouse CDDnmt3a–CDnmt3L complexes. The compared structures are shown in ribbon representations. The ADD–CD–CDNMT3L complex structure (this study) is coloured as in Fig. 1d, and the CDDnmt3a-CDnmt3L complex structure22 is coloured in grey. Residues 611–620 and 833–846 of DNMT3A were not built in the model because they lacked electron density. c, LIGPLOT representation of the ADD–CD interactions in the ADD–CD–CDNMT3L-H3 structure. Carbon, oxygen, and nitrogen are shown as black, red, and blue balls, respectively. Hydrogen bonds are indicated as dashed lines, with lengths given in Å. d, Close-up view of the ADD–CD interface. Critical residues for the interactions are shown in stick representation, and hydrogen bonds are indicated as dashed lines. The C terminus (residues 903–911) of the CD domain and a loop region (residues 621–632) together form a flat patch for interaction with the ADD domain. Hydrogen bonds are formed between residues N551, N553, and R556 of the ADD domain and residues E629, C911, and E907 of the CD domain. Residues Y526, Y528, W601, and F609 of the ADD domain, V622 and P625 of the linker, and R803 and P904 of the CD domain are involved in hydrophobic interactions. e, Structural comparison of ADD–CD–H3 in ADD–CD–CDNMT3L–H3 (this study) and DNMT3L–H3 structures (PDB accession number 2PVC)15. Two compared structures are shown in ribbon representations with ADD domains (left) or catalytic domains (right) aligned, respectively. The DNMT3L–H3 structure is coloured in grey. When the ADD domains are superimposed, the catalytic domain moves with a longest distance of 19 Å. When the CD domains are superimposed, the ADD domain moves 6 Å. f, Close-up view of the H3–ADD interface. Critical residues for the interactions are shown in stick representation, and hydrogen bonds are indicated as dashed lines. The fashion of histone H3–ADD interaction is similar to that observed in the structure of the H3–ADD fusion protein16. g, Histone H3 peptide pull-down assay. Recombinant wild-type and mutant ADD–CD–CDNMT3L proteins were incubated with biotinylated histone H3 peptide (residues 1–21) and immobilized onto streptavidin sepharose beads. Bound proteins were subjected to SDS–PAGE and stained by Coomassie blue.
a, Average B factors for domains of ADD–CD–CDNMT3L in the structures of ADD–CD–CDNMT3L and ADD–CD–CDNMT3L bound to H3 peptide. The average B factor of the ADD domain is higher than other domains in both structures, and is higher in autoinhibitory form (177.5 Å2) than that in active form (107.6 Å2). The results indicate that the ADD domain is more dynamic than other domains of the complex, especially in its autoinhibitory form. The observation further supports the idea that DNMT3A undergoes conformational changes on the ADD domain induced by histone H3. b, Two different views of the electron microscopy density maps of DNMT3A2–CDNMT3L (left) and DNMT3A2–CDNMT3L-H3 (right) processed to 24 Å and 20 Å resolution, respectively. The corresponding crystal structure was fitted into the electron microscopy density map for each state. The density is not fully occupied, which might because of the missing PWWP domain in the crystal structures. c, Typical negative stain CCD images of DNMT3A2–CDNMT3L (left) and DNMT3A2–CDNMT3L-H3 (right). Representative particles are highlighted by white boxes. d, Comparison of the two-dimensional projections (bottom) from the electron microscopy map with the corresponding reference-free two-dimensional class averages (top) reveals similar structural features. e, Position of residues F827 and F868 for 19F NMR measurements. Close-up view of the DNMT3A structure in autoinhibitory form with residues F827 and F868 indicated in stick representation. Residue F827 is located in loop L2 (for DNA binding) and close to the ADD domain. As a negative control, residue F868 is located close to the catalytic cavity and away from the ADD domain. Residue F868 is unlikely to undergo conformational change when the ADD domain dissociates from the CD domain. To detect conformational changes of DNMT3A in solution, residues F827 and F868 were substituted by 19F-labelled l-4-trifluoromethylphenylalanine (19F-tfmF) in ADD–CD. f, One-dimensional 19F NMR measurements were performed using ADD–CD with substitution of F827tfmF (left) or F868tfmF (right) in the absence or presence of H3K4me0 or H3K4me3 peptide. The chemical shift for each measurement is indicated.
This video illustrates the conformational changes of DNMT3A induced by histone H3 tail. DNMT3A exists in an autoinhibitory form, in which the ADD domain (green) binds to and inhibits the DNA-binding affinity of the CD domain (purple). Histone H3 (yellow) disrupts ADD-CD interaction, induces a large movement of the ADD domain, and releases the autoinhibition of DNMT3A. In the active form of DNMT3A, the ADD domain has no steric hindrance for DNA recognition by the CD domain. As pointed out in the main text, residues R790, R792, D529, and D531 of DNMT3A and H3K4 are shown as sticks. (MP4 5606 kb)
About this article
Cite this article
Guo, X., Wang, L., Li, J. et al. Structural insight into autoinhibition and histone H3-induced activation of DNMT3A. Nature 517, 640–644 (2015). https://doi.org/10.1038/nature13899
Methylation of recombinant mononucleosomes by DNMT3A demonstrates efficient linker DNA methylation and a role of H3K36me3
Communications Biology (2022)
The disordered N-terminal domain of DNMT3A recognizes H2AK119ub and is required for postnatal development
Nature Genetics (2022)
Nature Reviews Genetics (2022)
CD4 expression in effector T cells depends on DNA demethylation over a developmentally established stimulus-responsive element
Nature Communications (2022)
Macrophages use apoptotic cell-derived methionine and DNMT3A during efferocytosis to promote tissue resolution
Nature Metabolism (2022)