5mC-mediated epigenetic inheritance in animals

Some sort of dogma about DNA methylation research posits that environmental changes might induce epigenetic changes, and these will be inherited in the next generation causing transcriptional adaptation. While this has been proposed many times, it has not been demonstrated conclusively. Most importantly, even if methylation states would be inheritable, their actual effect on transcription was unclear. At least in invertebrates, the role of gene body methylation was debated, but not well understood.

In an attempt to tackle this conundrum, we turned some years ago to a new model system in the lab, the sea anemone Nematostella vectensis. We chose it as it has a smallish well characterised genome, it is easy to rear, and experimental manipulation is possible. Also, it has canonical invertebrate gene body methylation patterns, which remain stable along development. By using a combination of DNMT1 disruption treatments, we obtained hypomethylated individuals, that could progress developmentally and even produce fertile gametes, all while keeping an hypomethylated state. By profiling transcription in these individuals, we found that a lot of the proposed models about gene body methylation are likely to be wrong, as we didn't see major effects on alternative splicing, disregulation of the methylated genes or unstable transcription. Instead, we found that spurious transcriptional start sites and regulatory elements emerged from the gene bodies that were previously methylated, many of these actually derived from Transposable Elements. This role is in fact consistent with what is known about gene body methylation in mammals, so it is probably the ancestral state for animal DNA methylation.

Because the hypomethylated individuals could produce gametes, we then looked at the inheritability of these aberrant methylation states. While DNA methylation came back to some degree in the germ line, fertilisation did not prompt any major wave of epigenetic reprogramming, and "epi-mutants" were passively inherited in the next generation. This story is now published in Nature Ecology and Evolution.

The starlet sea anemone, Nematostella vectensis. Picture taken by Karmannye Chaudhary from a lab specimen.

We think this is helpful for the field, as it confirms that variable DNA methylation can be transmitted across generations in animals, at least invertebrates. This is very similar to what is known for plants. However, it does not fit the simplistic "Lamarckian" narrative. Gene body methylation prevents Transposable Elements modulating gene expression, and the outcome of this is hard to predict, sometimes it might be beneficial for the host, perhaps not have much effect or even be detrimental. So the "environmental epigenetic adaptation" hypothesis so frequently proposed requires a revision. At least when talking about DNA methylation.

Beyond 5mC: 6mA Joins the Epigenetic Cast

While we have traditionally focused on 5-methylcytosine DNA (5mC) methylation in the laboratory, we started exploring 6-methyladenine (6mA) some years ago with Pedro Romero Charria, the lab's second PhD student. The field was a bit messy since there have been high profile reports saying 6mA is present in all sorts of model systems, including mammals, plants or flies. However, this was also heavily contested. We decided to check for ourselves, since the scattered reports from ciliates, early diverging fungi and the green algae Chlamydomonas suggested that 6mA was probably more widespread. In a new publication in Nature Genetics, we trace the evolution of 6mA to the Last Eukaryotic Common Ancestor, and find that in contrast to 5mC, its patterns are surprisingly conserved throughout evolution. Still, 6mA and its main enzymatic complement, the AMT1 methyltransferase complex, have been lost in many branches of the tree of life, including all complex multicellular lineages (animals, plants, fungi, brown and red algae). Importantly, 6mA is a integral part of the ancestral eukaryotic chromatin, as it is inheritable (deposited in symmetric ApT dinucleotides) and co-localises with H3K4me3. In this regard, multicellular lineages are simpler than their unicellular ancestors, which is a fascinating question we want to explore.

We also got to write a Research Briefing summarising the findings and telling a bit of the behind the paper story, and this image made by our collaborator Omaya Dudin made it to the cover of the journal.

Based on the discoveries of this paper, we got granted a Wellcome Trust Discovery Award to keep exploring 6mA evolution. There are many mysteries that still surround this mark, so if you're curious and want to join our quest or collaborate, feel free to reach out!

Methylation goes viral: from curiosity to pattern?

After discovering that DNA methylation plays a key role in silencing hundreds of viral insertions in the protist Amoebidium, we began to question whether this was a rare exception or a widespread but overlooked phenomenon across eukaryotes. To explore this, we sought to expand our observations to a broader range of lineages. The challenge, however, lies in identifying species that have retained functional 5mC machinery, as the loss of DNA methylation is surprisingly common across the eukaryotic tree.

We focused on three species from distinct eukaryotic supergroups: Acanthamoeba castellanii (Amoebozoa), Naegleria gruberi (Heterolobosea), and Cyanophora paradoxa (Glaucophyta). In all three, we found that 5mC acts as a repressive mark, primarily targeting transposable elements and transcriptionally silent genes. As we extended our analysis to additional lineages, we repeatedly observed that viral insertions tend to localise within methylated, transcriptionally repressed regions of the genome. This pattern supports a broader model in which epigenetic silencing contributes to the genomic containment of potentially harmful, horizontally acquired DNA. This has been published in MBE:

https://academic.oup.com/mbe/advance-article/doi/10.1093/molbev/msaf176/8213644

Figure showing enrichment of genes of viral origin in the methylated fraction of the genome across a diverse set of eukaryotes.

Focusing on Acanthamoeba, we collaborated with John Archibald’s group at Dalhousie University. By comparing two high-quality, chromosome-scale assemblies (Neff and C3 strains), we observed extensive macrosynteny between the genomes, yet found that strain-specific regions were consistently enriched for giant virus insertions. These insertions derived from nuclear-replicating viruses, so we hypothesize they are the kind most likely to integrate accidentally into the host genome. This work can be found in BMC Biology:

https://bmcbiol.biomedcentral.com/articles/10.1186/s12915-025-02280-1