Which factors shape the rice DNA methylome?

Charting the epigenome is of key importance if we want to understand and engineer the genome of crop species. DNA methylation—the addition of a methyl group to cytosine—is a faithfully inherited epigenetic modification marking primarily repressed regions of the genome such as transposable elements (TEs). Loss of methylated cytosines leads to undesired reactivation of transcription and therefore genomic instability, which has dramatic consequences for the development of plants and animals.

In plants, cytosine methylation is commonly found in three different sequence contexts: CG, CHG, and CHH (where H denotes A, T, or C). Several methyltransferases establish and maintain genome-wide DNA methylation patterns. In Arabidopsis thaliana, CG and CHG methylation are maintained by METHYLTRANSFERASE1 (MET1) and CHROMOMETHYLTRANSFERASE3 (CMT3), respectively (reviewed in Law and Jacobsen, 2010; Zhang et al., 2018). CHH methylation is established de novo by DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) and CMT2. The relationship between methyltransferases and their deposited mark is complex. Certain genomic regions bear methylation in all three contexts while others are marked only by one or two. Furthermore, the loss of methylation in one sequence context does not imply the loss of the two other types. It appears thus critical to understand the contribution of each methyltransferase to the DNA methylome.

The rice (Oryza sativa) genome contains two MET, three CMT, and four DRM genes. So far, their role in establishing and maintaining DNA methylation remains not fully understood. In this issue of The Plant Cell, Hu et al. (2021) combined genetics and genomics to assess the contribution and synergies between all rice methyltransferases. To generate mutant plants for each individual and for multiple methyltransferases, the authors used a multiplexed CRISPR-Cas9 system that enabled them to obtain homozygous deletions/insertions for single and multiple genes in the first transgenic generation (see Figure).

Figure.

Visible and molecular phenotypes of rice DNA methyltransferase mutants. Top: Rice plants deficient for all different DNA methyltransferases were obtained by CRISPR-Cas9. Bottom: DNA methylation profiles at single-base pair resolution were generated by whole-genome bisulfite sequencing. CHG and CHH profiles of the obtained knock-out mutants are shown here. Adapted from Hu et al. (2021), Figures 1 and 2.

Using whole-genome bisulfite sequencing, the authors determined the target cytosines of each methyltransferase at single base-pair resolution (see Figure). They found that CMT2, CMT3a, and DRM2 are the three major players in non-CG methylation in rice. The majority of CHH methylation was eliminated in the drm2^−/− mutant and DRM2-dependent CHH differentially methylated regions (DMRs) did not show loss of methylation when CMT2 and/or CMT3a were mutated. In addition, loss of CMT2 led to decreased CHH methylation at DMR2-independent sites. CMT3a is the main CHG methyltransferase and contributes most to repressing TEs. Loss of CMT3a also affected CHH methylation, mostly in a DRM2-independent manner. Similarly, drm2^−/− plants displayed reduced CHG methylation. However, these DRM2 CHG DMRs were both CMT3a-dependent and -independent. Together, these observations nicely illustrate the complex and site-dependent crosstalk between CHG and CHH methylation.

Some residual non-CG methylation remained in the triple drm2 cmt2 cmt3a mutant (dcc), which was lost in the drm2 drm3 cmt2 cmt3a cmt3b quintuple mutant. The corresponding genomic regions displayed heterochromatic features such as high CG content, TE density, and histone H3 lysine 9 dimethylation but low levels of 24-nt siRNAs. In particular, CMT3b seemed to have evolved to maintain non-CG methylation regions with high cytosine density. Interestingly, such GC-rich clusters of non-CG methylated sites seem to be unique to the rice genome.

CMT2 and CMT3 in rice and Arabidopsis seem to have evolved slightly different functions. Indeed, loss of rice CMT2 did not affect CHG methylation while in Arabidopsis, CMT2 contributes to maintain both CHH and CHG methylation (Stroud et al., 2014). Furthermore, rice CMT3 CHH DMRs are independent of DRM2 while Arabidopsis CMT3 CHH DMRs are a subset of CMT2 DMRs.

In summary, Hu and colleagues charted in detail the DNA methylome of rice, providing a key resource and stepping stone for the future epigenomic engineering of this important crop species.