GE and Science Prize

Mechanisms of Transgenerational DNA Methylation Inheritance

Felipe Karam Teixeira

Nowadays, one can find many interpretations of the term "epigenetics" in the literature (1). In its most accepted definition, epigenetics refers to the study of heritable changes in gene expression that do not involve changes in the DNA sequence. This concept implies that, once established, a new epigenetic state can be stably propagated through mitosis or meiosis independently of the inducible signal, yet can still revert to its original state. Most important, these different states, known as epialleles or epimutations, can lead to phenotypic variation.

DNA methylation, which refers to the addition of methyl groups to DNA at specific nucleotides, is a classic epigenetic mark that plays key roles in the control of genome activity in plants and mammals. Indeed, multiple lines of evidence indicate that this modification is critical for the silencing of transposable elements (TEs)—mobile DNA modules that can jump within the genome—and other repeats, as well as for the epigenetic regulation of some genes, notably those that are mono-allelic, expressed in a parent-of-origin–specific manner (2). Therefore, DNA methylation must be tightly controlled. In plants, however, genetically induced loss of DNA methylation can be stably transmitted through meiosis independently of the inducing signal, leading to epimutations (3, 4). This and other findings led to the prevalent view, when I started my thesis, that DNA methylation cannot be restored once it has been severely compromised (2).

During my thesis work, I focused on the analysis of the inheritance patterns of DNA methylation in the model plant Arabidopsis thaliana. To address the transgenerational stability of DNA hypomethylation and its impact on heritable quantitative phenotypic traits, my Ph.D. supervisor and his collaborators had created a population of "epigenetic recombinant inbred lines" (epiRILs) (5). These were derived from two nearly isogenic parental lines in terms of genome sequence, but with contrasted DNA methylation patterns as a result of one parental line carrying a mutation in the adenosine triphosphatase chromatin remodeler gene DDM1. Mutations in this gene lead to a severe loss (>70%) of DNA methylation overall (4). Using these epiRILs and probing a small set of hypomethylated loci, I first confirmed that ddm1-induced loss of DNA methylation can be stably inherited at a fraction of loci across the Arabidopsis genome, independently of the inducing mutation and for multiple generations. However, I also found that approximately as many loci systematically regained wild-type DNA methylation in the same epiRILs (5). These observations were indicative of a previously unknown targeted DNA remethylation process that challenged the prevalent view on the transgenerational stability of DNA hypomethylation in plants.

To further investigate the DNA remethylation process, I conducted a systematic analysis using a large number of repeat elements. Our findings revealed the existence of an efficient RNA interference (RNAi)–based mechanism that protects numerous sequences across the Arabidopsis genome against irremediable loss of DNA methylation. Using molecular analysis, I have demonstrated that this process faithfully restores wild-type methylation over target sequences. Furthermore, I could show that DNA remethylation is progressive over several generations, takes up to five generations to be completed, and, in the case of reactivated transposable elements, is also associated with their resilencing. Deep sequencing of endogenous small RNA populations combined with bioinformatic analysis indicated that DNA remethylation is specific to the subset of heavily methylated loci that are associated with an abundance of small interfering RNAs (siRNAs). Indeed, genetic analysis revealed an essential role for the RNAi machinery, which is responsible for the production of siRNAs, in DNA remethylation (6). Studies carried out on another population of Arabidopsis epiRILs, obtained this time using a mutation in the maintenance DNA methyltransferase gene MET1, have largely confirmed our findings (7).

Finally, I was able to show that DNA remethylation is mainly confined to the reproductive stage of the plant life cycle (6, 8), where the RNAi machinery is likely to be most active (9, 10). In this respect, DNA remethylation parallels a series of other RNAi-dependent processes that have recently been described Drosophila and mammals, the function of which are to ensure the silencing of TEs and other repeats in germ cells and the embryo (11).

Collectively, our findings indicated that DNA methylation patterns tend to be propagated across generations in plants, rather than re-established anew at each generation as in mammals (8, 12). Indeed, genome-wide DNA methylation dynamics during reproductive stages in plant appears to be mostly restricted to accessory cells that do not contribute directly to the embryonic lineage (13, 14). However, we have shown that some DNA methylation defects can nonetheless be efficiently corrected over generations in plants, through a previously unknown RNAi-dependent process (Fig. 1) (6). Given its specificity, this process would endow plants with a unique ability to preserve the structural and functional integrity of their genomes while increasing adaptive opportunities through the creation of epialleles with variable transgenerational stability (8).

Fig. 1.
A model for the progressive restoration of DNA methylation over multiple generations. RNAi is largely dispensable for the maintenance of DNA methylation at most repeat elements. However, RNAi is critical for restoring wild-type DNA methylation after severe loss. This restoration is progressive across several generations because RNAi-based mechanisms can only enforce limited de novo DNA methylation during each reproductive cycle. Newly reached DNA methylation levels would be maintained during vegetative growth mainly by other pathways and would provide a starting point for further DNA methylation during the next reproductive phase. [The figure is modified from figure 2C of Teixeira and Colot (8).]


Importantly, phenotypic and quantitative genetic analyses conducted in parallel by colleagues and collaborators revealed that the epiRILs present variation and high heritability for two complex traits, namely flowering time and plant height. Although I have shown that mobilization of TEs also occurs in the epiRIL population—and therefore must be accounted as a source of variation—their results provided evidence that stably inherited epigenetic changes contribute significantly to heritable phenotypic variation in complex traits (5). Two long-standing and fascinating challenges for the future will be: (i) to explore the extent to which epigenetic variation does contribute to heritable phenotypic variation, either directly or indirectly through DNA sequence changes, such as those caused by TE mobilization, and (ii) to determine the impact of environmental perturbations on the mechanisms of transgenerational epigenetic inheritance.


  • 1. A. Bird, Nature 447, 396 (2007).
  • 2. R. K. Slotkin, R. Martienssen, Nat. Rev. Genet. 8, 272 (2007).
  • 3. T. Kakutani, K. Munakata, E. J. Richards, H. Hirochika, Genetics 151, 831 (1999).
  • 4. A. Vongs, T. Kakutani, R. A. Martienssen, E. J. Richards, Science 260, 1926 (1993).
  • 5. F. Johannes et al., PLoS Genet. 5, e1000530 (2009).
  • 6. F. K. Teixeira et al., Science 323, 1600 (2009).
  • 7. J. Reinders et al., Genes Dev. 23, 939 (2009).
  • 8. F. K. Teixeira, V. Colot, Heredity 105, 14 (2010).
  • 9. R. A. Mosher et al., Nature 460, 283 (2009).
  • 10. R. K. Slotkin et al., Cell 136, 461 (2009).
  • 11. D. Bourc'his, O. Voinnet, Science 330, 617 (2010).
  • 12. W. Reik, Nature 447, 425 (2007).
  • 13. M. Gehring, K. L. Bubb, S. Henikoff, Science 324, 1447 (2009).
  • 14. T.-F. Hsieh et al., Science 324, 1451 (2009).