De Novo DNA Cytosine Methyltransferase Enzyme Dynamics
The DNA methyltransferases (DNMT) are one busy enzyme family! While they are responsible for establishing and maintaining the mammalian cytosine methylome, they aren’t one and the same, as they take on some diverse roles. The DNMTs can from complexes with more than one DNMT to accomplish additional functions and their individual functions can also be generalized. DNMT1 is considered to be the maintenance methyltransferase as it is primarily involved in actively maintaining DNA methylation by recognizing hemi-methylation and copying the pattern onto the opposite strand. DNMT3A and DNMT3B are de novo methyltransferases that are mainly involved in creating new methylation patterns – with a particular fondness for early embryonic events. DNMT3L is a catalytically inactive DNMT3 paralogue. It interacts with DNMT3A and DNMT3B to alter their activity, and is used in special situations such as genomic imprinting. Lastly is DNMT2, which has been renamed as TRDMT1, as it methylates aspartic acid in tRNA and not DNA. Further adding to the complexity of this family is the fact that the DNMT family members and paralogues also have unique isoforms that have different temporal and spatial regulation!
How exactly the DNMTs target specific genomic sequences is an area of active research. There are a number of models suggesting a role for histone modification, RNA guidance, accessory proteins, and/or DNA sequence specificity in determining what region a DNMT will target (see DNMT review)1. In order to investigate the mechanisms underlying de novo methylation, Baubec et al. examined the genomic binding of the major DNMT3A and DNMT3B isoforms in mouse embryonic stem cells. In their approach they used a biotin tag on DNMT3A and DNMT3B that allowed for ChIP assessment of DNMT3 paralogue binding to DNA. They found that both DNMT3A and DNMT3B have a broad genomic binding pattern that localizes to methylated CpG rich regions and avoids H3K4me-modified histones. This approach showed that both DNMT3A and DNMT3B bind similar regions, however, DNMT3B primarily binds actively transcribed gene bodies, whereas DNMT3A appears to avoid them. They also found that features of active transcription are present at DNMT3B binding sites, including mRNA expression, RNA polymerase II binding, and H3K6me3-modified histones.
Next, in order to understand how localization relates to enzyme activity, the team used DNMT triple KO’s and transfected cells with a de novo DNMT3 paralogue of choice (either DNMT3A2 or DNMT3B1). This allowed them to analyze DNMT function on fresh template, as the triple KO model lacks methylation to start with. Attesting to the quality of the experimental triple KO system, the group found that the transfected de novo DNMTs associated to the same locations in the DNA as they did in non-KO, endogenously methylated-cells. This indicates that recruitment is preferentially independent of pre-existing methylation. The team then measured the rate of methylation using high-pressure liquid chromatography coupled to mass spectrometry. The DNMT3A2-transfected line showed higher activity than DNMT3B1 transfected line, with 0.7% as opposed 0.18% methylation, occurring at all cytosines. The wild-type, non-KO lines showed 4.2%. These rates were reproduced not only across replicates, but were also confirmed by bisulfite sequencing – with DNMT3A methylating 7% and DNMT3B methylating 2.8% of cytosines in CpG dinucleotide contexts. De novo methylation by both paralogues did not occur in the core nucleosome unit, but rather in regions occupied by the linker histone (H1). These results indicate that de novo methylation is regulated by enzyme recruitment and nucleosome occupancy.
Finally, the team used CRISPR/Cas9 genome editing to deplete the H3K6 methlyransferase SET2D. The editing was done in the triple KO line that had been transfected with DNMT3B. It was found that methylation at CpG rich sites and linkers sites wasn’t changed, but that there was a significant depletion of methylation in the regions where H3K6me3 would be present in actively transcribed gene bodies. Their findings also showed that the N-terminal of DNMT3B contains a functional PWWP domain. PWWP is over 100 aa in length and contains a Proline-Tryptophan-Tryptophan-Proline motif that is used in nuclear proteins for protein interactions that govern growth and differentiation 2. Interestingly, they found that the PWWP domain is needed to bind to H3K6me3-modified nucleosomes. Thus, the team elucidated the mechanism of how SETD2 mediated H3K36me3 can target DNMT3B binding and subsequent de novo DNA methylation to transcribed gene bodies.
Ultimately, this body of research suggests that, in general, umethylated cytosine substrate, as well as context, and accessibility, determine de novo methylation outside of positioned nucleosomes and/or active regulatory elements. Furthermore it shows that DNMT3B’s targeting to gene bodies is dependent on H3K6me3 histone modification. The research also provides further evidence that DNA cytosine methylation is not simply a repressive mark but rather a context dependent regulator.
Baubec T, Colombo DF, Wirbelauer C, Schmidt J, Burger L, Krebs AR, Akalin A, & Schübeler D (2015). Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature, 520 (7546), 243-7 PMID: 25607372
1. Smith, Z., & Meissner, A. (2013). DNA methylation: roles in mammalian development Nature Reviews Genetics, 14 (3), 204-220 DOI: 10.1038/nrg3354
2. Qiu C, Sawada K, Zhang X, & Cheng X (2002). The PWWP domain of mammalian DNA methyltransferase Dnmt3b defines a new family of DNA-binding folds. Nature structural biology, 9 (3), 217-24 PMID: 11836534