Ceci n'est pas une DNMT: Recently discovered functions of DNMT2 and their relation to methyltransferase activity (Comment on DOI 10.1002/bies.201300088)


DNMT2 is an unfortunately named protein. Along with its paralogs DNMT1, DNMT3a, DNMT3b, and DNMT3L, it was named thus because of its similarity with bacterial DNA methyltransferases (DNMTs). However, DNMT2 has no measurable activity towards DNA in vitro. In addition, several eukaryotic organisms, including Schizosaccharomyces pombe and Drosophila melanogaster, contain DNMT2 as their sole DNMT-like protein; these organisms lack any detectable DNA methylation [1]. DNMT2, then, is not a DNMT.

A catalytic function of DNMT2 was brought to light in a groundbreaking paper from the Bestor lab, which showed that DNMT2 is actually a tRNA methyltransferase. Its preferred target is tRNAAsp, and this function is conserved between mammals, flies, and plants [2]. However, the flies, plants, and mice lacking DNMT2 had no obvious phenotypic defect. So, what might DNMT2 be important for?

This is the question addressed in the review by Schaefer and Durdevic [1]. The authors provide an excellent overview of the germane literature. Two main messages emerge, along with one hypothesis.

The first message is that the crucial function of tRNA methylation by DNMT2 is not apparent in standard conditions, but only after genetic sensitization (by removing another RNA methyltransferase), or under heat shock. In other words, the evolutionary forces that is selected for the retention of DNMT2 have no measurable effect when a model organism is grown in an optimal laboratory environment. The idea is well-known to many geneticists, but it bears repeating.

The second message is that DNMT2 is involved in genome protection. In its absence, (+) RNA viruses, and possibly also retrotransposons, are deregulated. It is unclear yet how this relates to the tRNA methyltransferase function of DNMT2.

The hypothesis, finally, is that DNMT2 could act catalytically as something other than a cytosine methyltransferase. It has been clearly shown that DNMTs can deaminate or even transfer aldehydes to cytosines under certain circumstances. DNMT2 derives from an ancestral DNMT gene, and could have maintained these capacities. This could possibly be important for its role in the defense against genome invaders.

What is left to discover? An exciting recent paper reported that DNMT2 is necessary for RNA-mediated paramutation in mouse [3]. Unexpectedly, we seem to have come full circle, and DNMT2, which was initially expected to be involved in epigenetic inheritance, is indeed involved in the heritability of phenotypes, even though they are based on RNA, and not DNA. Paramutation was initially discovered in maize, and has been extensively investigated in plants [4]. It will be interesting to determine whether DNMT2 is involved in this form of non-Mendelian inheritance in plants as well as mammals.

Another recent paper has highlighted a new and potentially important role of DNMT2. In stem cells of the male Drosophila germline, the chromatids of sex chromosomes are segregated nonrandomly during asymmetric cell divisions: one specific strand of DNA is preferentially retained in the stem cell, whereas the other strand is passed on to the more differentiated daughter cell [5]. This asymmetry is lost in the absence of DNMT2 [5]. It will be fascinating to find out whether this effect is mediated by RNA, and by the catalytic activity of DNMT2, and whether it relates to paramutation.