The first line of evidence comes from nature rather than the laboratory. A wild plant population living on a drought-stricken Atlantic island showed genome-wide shifts in DNA methylation that could not be explained by sequence changes at the methylated sites themselves. When researchers traced the genetic basis of these altered methylation landscapes, they arrived repeatedly at the same missing stretch of DNA, a 2.7 kilobase deletion. The second line of evidence comes from a ten-generation greenhouse experiment, in which plants were propagated under controlled conditions while their methylomes were tracked generation by generation. Remarkably, this experiment converged on the same deletion. Two independent approaches, one observational and one experimental, pointed at the same genetic determinant.

To appreciate why this matters, it helps to recall how methylation memory is maintained. In flowering plants, cytosine methylation occurs in three sequence contexts: CG, CHG and CHH, where H stands for any base other than G. Each context is maintained by a dedicated molecular machinery. CG methylation, the most abundant and most stable form, is copied after DNA replication by the methyltransferase MET1, a homolog of the mammalian enzyme DNMT1. CHG methylation is reinforced through a self-reinforcing loop between the H3K9 histone methyltransferase CMT3 and chromomethylase activity. CHH methylation, the most dynamic of the three, depends on guided de novo methylation by small interfering RNAs through the RNA-directed DNA methylation pathway. Landmark studies, including the comprehensive methylome maps produced by Kawakatsu and colleagues in 2016 and the mechanistic framework laid out by Law and Jacobsen in 2010, established how faithfully these systems copy marks and what happens when key components are lost.

Fidelity, however, is not only about the methyltransferases that write the marks. It also depends on proteins that give the maintenance machinery access to the DNA. The chromatin remodeler DDM1, for example, was shown by Stroud and colleagues in 2013 to be required for methylation of transposable elements buried in nucleosomes; without DDM1, heterochromatin becomes inaccessible and methylation is progressively lost. Similarly, work on the VIM family of proteins and on factors such as those characterized by Woo and colleagues in 2007 revealed that mutations in single loci can cause genome-wide collapses or rearrangements of methylation patterns. Bostick and colleagues demonstrated in 2007 how the antagonistic interplay between methyltransferases and demethylases shapes the steady-state methylation landscape. In other words, the stability of the epigenome is a genetically encoded trait, and mutations in specific genes can tune it up or down.

The new studies extend this principle in an evolutionary direction. Rather than asking what happens when a maintenance gene is knocked out in a laboratory strain, they asked whether natural genetic variation, including structural variation such as deletions, modulates the stability of methylation inheritance in real populations. The answer appears to be yes. The 2.7 kilobase deletion acts as a genetic determinant of genome-wide methylation levels and, more importantly, of the fidelity with which those levels are transmitted across generations. Plants carrying the deletion do not simply have different methylation patterns; they have methylation patterns that are less reliably copied, so that epigenetic variation accumulates faster among their descendants.

This distinction between methylation state and methylation fidelity is the conceptual heart of the work. Most epigenomic surveys compare the methylation patterns of different individuals and ask where they differ. The new studies instead ask how much those patterns change over time within lineages, and whether the rate of change is itself heritable and genetically determined. The ten-generation greenhouse experiment was designed precisely to measure this drift. By propagating lines for ten generations and profiling their methylomes, the researchers could distinguish random epimutational drift from directed change, and could test whether the rate of drift depended on genotype. The convergence of this experiment with the island population study on the same deletion provides unusually strong evidence that the effect is real and biologically meaningful.

Why would a plant population on a drought-stricken island carry such a deletion in the first place? The authors of the accompanying commentary frame the answer as a provocative hypothesis: the potential to change can be an adaptation in itself. In a fluctuating or stressful environment, a genotype that generates more heritable epigenetic variation might explore a wider range of phenotypic states without waiting for slow sequence evolution. Transposable elements, which are the primary targets of methylation, can alter gene regulation when their methylation is lifted, and altered methylation near genes can change expression in ways that sometimes prove beneficial. A loosened methylation memory would increase the supply of such heritable variation, effectively raising the mutation rate of the epigenome. Whether this constitutes adaptation in the strict sense, or merely a byproduct of genetic change under stress, remains an open question that the new studies raise but cannot fully resolve.

The findings also carry practical implications for agriculture. Crop improvement increasingly looks to epigenetic variation as a source of novel traits, particularly for responses to drought, heat and salinity. Epigenetic breeding strategies assume that induced methylation changes can be selected and stably inherited. If the fidelity of methylation inheritance varies across genetic backgrounds, and if that variation can be mapped to specific loci such as the deletion identified here, breeders could in principle choose backgrounds that either stabilize desirable epigenetic states or, conversely, maximize the generation of new epigenetic variation. The genetic determinants of methylation fidelity thus become a breeding target in their own right, sitting at the interface of classical genetics and epigenomics.

There are also broader lessons for how biologists think about heredity. The traditional view treats DNA sequence as the sole substrate of inheritance, with methylation as a modifiable layer on top. The new work blurs that boundary: a sequence change, the deletion, controls the stability of the layer above it, and the resulting epigenetic drift may feed back into phenotypic evolution. Heredity, in this picture, is a multi-layered system in which genetic and epigenetic variation are coupled by defined molecular mechanisms. The 2.7 kilobase deletion is a concrete example of such a coupling, and its discovery in both a natural population and a controlled experiment suggests that similar couplings may be widespread and previously overlooked.

Much remains to be learned. The precise molecular function of the deleted region, the full set of loci whose methylation becomes unstable in its presence, and the fitness consequences of loosened methylation memory in the wild are all questions for future work. But the convergence of a drought-stricken island population and a decade-spanning greenhouse experiment on the same missing fragment of DNA is a vivid demonstration that epigenetic inheritance is not a fixed property of the genome. It is a trait under genetic control, variable among individuals, and potentially shaped by natural selection. In the evolving story of how organisms pass information to their offspring, methylation memory now appears to be not just a record of the past, but a dial whose sensitivity evolution can tune.

Subject of Research: Genetic determinants of DNA methylation fidelity and epigenetic inheritance in plants

Article Title: A deletion loosens methylation memory

Article References: Becker, C. (2026). A deletion loosens methylation memory. Nature Plants. https://doi.org/10.1038/s41477-026-02403-8

Image Credits: AI Generated

DOI: 10.1038/s41477-026-02403-8

Keywords: epigenetics, DNA methylation, epigenetic inheritance, plants, deletion, methylation fidelity, transposable elements, adaptation, epigenomics, plant evolution, Nature Plants, chromatin