The discovery of a modified version of the base adenine, known as N6-methyladenine, in mouse DNA puts paid to the theory that cytosine derivatives are the only modified bases in mammals. See Article p.329
The DNA of most organisms is composed of four standard bases and a small set of modified bases that are produced enzymatically from these four after DNA replication. One modified base, N6-methyladenine (N6mA), is prevalent in prokaryotes (bacteria and archaea), but whether it is found in mammals has remained unclear. In this issue, Wu et al.1 (page 329) report the existence of N6mA in mouse stem cells. This exciting discovery is enhanced by the identification of an enzyme that removes methyl groups from N6mA, and by the finding that the modification is enriched in certain regulatory DNA sequences — data that together provide clues to N6mA's possible function in mammalian genomes.
The most abundant modified base in the mammalian genome is 5-methylcytosine (5mC), which regulates the expression of genes and of DNA regions called retrotransposons, which, through reverse transcription to produce an RNA intermediate, can move around the genome, disrupting gene regulation2,3. Enzymes called TETs can oxidize 5mC to form 5-hydroxymethylcytosine and other derivatives4. In prokaryotic genomes, N4-methylcytosine and N6mA are also widespread, and are thought to modulate DNA replication and repair, among other roles5. Moreover, N6mA is found in certain unicellular eukaryotic (nucleus-bearing) organisms6.
It was long thought that N6mA was absent from the DNA of most multicellular organisms. However, this view was biased by a lack of sufficiently sensitive detection methods, which prevented in-depth analysis of rare modified bases. In 2015, three studies characterized eukaryotic N6mA in detail — in green algae7 (Chlamydomonas reinhardtii), in nematode worms8 (Caenorhabditis elegans) and in fruit flies9 (Drosophila melanogaster). These reports revealed that the prevalence of the modification varied widely between eukaryotic species, and found evidence that, in these three organisms, N6mA is associated with active transcription.
Wu et al. looked for N6mA in the DNA of mouse embryonic stem (ES) cells. Using mass spectrometry, they found that N6mA represented only 6–7 bases per million adenines genome-wide. N6mA was enriched approximately fourfold in genomic regions associated with a rare histone protein, H2AX, although the reasons for this association remain unclear.
The authors next focused on DNA regions that are bound to this atypical histone protein, which is often associated with genomic regions involved in cell regulation and development. Analysing only these selected sequences enabled the researchers to make use of a sophisticated technique called single-molecule real-time (SMRT) sequencing, which detects the different kinetics with which a polymerase enzyme replicates modified bases compared with standard ones. SMRT sequencing has been used to map N6mA in several prokaryotic species10, and Wu et al. used the technique to identify consensus DNA sequences at which N6mA arises in mouse cells.
Demethylase enzymes remove methyl groups from DNA, RNA or proteins. A few N6mA demethylases of the ALKB protein family have been shown to remove methyl groups from N6mA in RNA — for example, to regulate messenger RNA11. In mammals, this protein family has nine members12. Wu et al. found that deletion of one member, ALKBH1, in mouse ES cells led to a high accumulation of N6mA in the genome, and that ALKBH1 could remove methyl groups from N6mA in DNA in vitro.
It is possible that other ALKBH proteins also possess DNA N6mA demethylase activity, because ALKBH4, 6 and 7 have no clearly defined substrate. Curiously, ALKBH1 worked most effectively on single-stranded DNA in vitro, raising the question of whether the enzyme preferentially operates during transcription or DNA replication in vivo, when DNA is transiently single stranded. Of note, the C. elegans N6mA demethylase NMAD-1, which is more closely related to ALKBH4 than to any other mammalian protein, does act on double-stranded DNA8. Thus, ALKBH4 might also be a mammalian N6mA demethylase, perhaps acting at a different stage of the cell cycle from ALKBH1.
Wu et al. deleted the Alkbh1 gene in ES cells, and found that N6mA accumulated on the X chromosome in these cells. This accumulation occurred in the regulatory regions of evolutionarily young LINE-1 retrotransposon sequences (those that moved into position relatively recently), suggesting that N6mA may help to control these genomic parasites. Indeed, the presence of N6mA correlated with inhibited expression of both LINE-1 elements and their adjacent genes (Fig. 1). This finding raises the possibility that N6mA has a role in mammalian X inactivation, a phenomenon that compensates for the fact that females have two X chromosomes, compared with males' one.
The gene-silencing role of N6mA in mouse ES cells contrasts with its presumed role in activating gene transcription in simpler eukaryotes7,8,9. A recent study independently detected N6mA in frogs and in mouse kidney DNA13, finding that the modified base was relatively depleted in protein-coding sequences. However, the effect of this depletion on transcription has yet to be determined.
Wu and colleagues' confirmation of N6mA in mammalian genomes raises many questions. For example, why does N6-adenine methylation occur in such seemingly different patterns and DNA sequences in different eukaryotic species? Is the modification differentially distributed across tissues or embryonic stages? An understanding of this variability might indicate whether there is a unifying role for N6mA in eukaryotes.
The enzymes that add methyl groups to adenine in mammalian DNA remain to be defined, as do the 'reader' proteins that detect genomic N6mA. Candidate reader proteins that can translate the N6mA code might be found among the several mammalian proteins that have domains predicted to interact with N6mA in DNA14.
ALKB-like proteins are currently the only known mammalian N6mA demethylases. In fruit flies, a relative of the mammalian TET 5mC oxidase acts as an N6mA demethylase9. However, the existence of 5mC in these organisms is still controversial15, and the fruit-fly protein is structurally different from ALKB proteins. Thus, it remains unclear whether TET 5mC oxidase might act as an N6mA demethylase in mammals.
The levels of N6mA in multicellular organisms are exceptionally low. So, to begin to answer these many questions, the field must await further technological developments that facilitate the study of rare modifications. Such an ability would surely lift this blossoming research area to the next level.
About this article
Nature Structural & Molecular Biology (2017)