The DNMT3A PWWP domain is essential for the normal DNA methylation landscape in mouse somatic cells and oocytes

Author summary Epigenetic modifications of histones and DNA play important roles in gene regulation and embryonic development. It is increasingly evident that interplay between the modifications is important for shaping their distribution patterns. We herein examine the role of the PWWP domain of DNMT3A, the major DNA methyltransferase expressed in mammalian oocytes, in shaping the oocyte-specific DNA methylation landscape. The PWWP domain is known to recognize di- and tri-methylated lysine 36 of histone H3 (H3K36me2/3), the latter of which is a histone modification especially important for shaping the DNA methylation landscape in oocytes. We found that, despite the potential importance of the PWWP domain in H3K36me3-dependent DNA methylation, mouse DNMT3A protein with an amino acid substitution disrupting the H3K36me3 recognition has little impact on the DNA methylation of H3K36me3-marked regions. Instead, the mutated protein causes aberrant DNA hypermethylation in regions lacking H3K36me2/3, including H3K27me3-marked regions. Furthermore, mouse oocytes that only express the mutated protein can support normal development when fertilized with wild-type sperm. Our findings suggest that DNMT3A has a backup system to recognize H3K36me3 without a normal PWWP domain and that mouse embryos have the potential to tolerate or reprogram the aberrant DNA hypermethylation caused by the PWWP mutation. DNA methylation at CG sites is important for gene regulation and embryonic development. In mouse oocytes, de novo CG methylation requires preceding transcription-coupled histone mark H3K36me3 and is mediated by a DNA methyltransferase DNMT3A. DNMT3A has a PWWP domain, which recognizes H3K36me2/3, and heterozygous mutations in this domain, including D329A substitution, cause aberrant CG hypermethylation of regions marked by H3K27me3 in somatic cells, leading to a dwarfism phenotype. We herein demonstrate that D329A homozygous mice show greater CG hypermethylation and severer dwarfism. In oocytes, D329A substitution did not affect CG methylation of H3K36me2/3-marked regions, including maternally methylated imprinting control regions; rather, it caused aberrant hypermethylation in regions lacking H3K36me2/3, including H3K27me3-marked regions. Thus, the role of the PWWP domain in CG methylation seems similar in somatic cells and oocytes; however, there were cell-type-specific differences in affected regions. The major satellite repeat was also hypermethylated in mutant oocytes. Contrary to the CA hypomethylation in somatic cells, the mutation caused hypermethylation at CH sites, including CA sites. Surprisingly, oocytes expressing only the mutated protein could support embryonic and postnatal development. Our study reveals that the DNMT3A PWWP domain is important for suppressing aberrant CG hypermethylation in both somatic cells and oocytes but that D329A mutation has little impact on the developmental potential of oocytes.