Files
Abstract
As one of the major chromatin modifications, DNA methylation plays an important role in chromosome stability, transposable element (TE) deactivation and gene expression regulation. Although the majority of DNA methylation is occupied on repeats and TEs in heterochromatic regions, many coding genes also contain DNA methylation. In flowering plants, many genes contain exclusively CG methylation in the transcribed region, which is referred to as gene body DNA methylation (gbM). gbM is prevalent throughout angiosperms and is a major driver of methylation variations. Curiously, the loss of gbM was found in E. salsugineum, a member from the Brassicaceae family, and it was accompanied by a loss of the DNA methyltransferase CHROMOMETHYLASE 3 (CMT3). Those findings supported a hypothesis that CMT3 is required for the origination of gbM.My thesis was focused on investigating the mechanistic origin of gbM. Firstly, we tested what happens when the CMT3 gene of A. thaliana was transferred into E. salsugineum. I showed that de novo CHG methylation accumulated on genes over generations and was subsequently followed by the establishment of CG methylation. Additionally, by tracing CG methylation changes for six generations, I showed that the de novo CG methylation that accumulated in genes could be inherited even following silencing of the CMT3 transgene.
Next, I investigated the methylomes of 725 A. thaliana accessions. I found that DNA methylation is highly variable on both gene and heterochromatic regions among accessions. The number of genes possessing DNA methylation is inversely correlated with the abundance of methylation in heterochromatin. I trained a machine learning model that could utilize genic features to accurately predict gbM status, with gene length and CWG frequency being the most important predictors. Importantly, I found that the most conserved gbM genes were more likely to shift to transposon-like methylation, which affects transcription and could lead to phenotypic consequences.
Finally, I analyzed the methylomes for a variety of mutant lines to test a model of how heterochromatin methylation affects gbM. The results showed that ectopic hypermethylation on genes is due to feedback regulation of pathways involved in maintaining DNA methylation at heterochromatin. Using methylomes of ddm1 epigenetic recombinant inbred lines (epiRILs), I showed that hypomethylated chromosomes lead to ectopic DNA methylation on genes. Epiallelic variation in euchromatin is associated with QTL primarily located in heterochromatin.
Together, those studies investigated DNA methylomes for a variety of experimental mutant lines and natural populations, providing mechanistic investigation into spontaneous epiallele formation, which is associated with feedback regulation of pathways involved in maintenance of DNA methylation in heterochromatin. These findings expand our knowledge on the origination, distribution, and implication of spontaneous epialleles.