Fig. 2

Analysis of “germline” epimutations in A. thalianamutation accumulation (MA) lines. a Three different MA pedigrees were analyzed. All three pedigrees were derived from a single Columbia (Col-0) inbred genotype. Two of the pedigrees were previously published (MA1_1, Becker et al. 2011; MA1_3, van der Graaf et al. 2015), and one pedigree (MA3) is new. These three MA pedigrees were chosen because they differ in their topologies, 5mC measurement strategies, and the temporal resolution of the 5mC samples. b Overview of the data: N is the total number of sequenced samples; Seq depth is the average sequence depths of the samples; # TP is the number of unique time-points (or generations) that are sampled; max. (Δt) is the maximum divergence time (in generations) in the pedigree. c Application of models Abnull, Abneutral, ABmm, and Abuu. The best fitting model is indicated for each MA pedigree, sequence context (CG, CHG, and CHH), and genomic feature (global, exons, promoters, TEs). d Shown are the fits of the best fitting models for each pedigree and context. e Schematic representation of transgenerationally stable CHH epimutations. The barplots indicate the density of stable CHH epimutations in lineages L2 and L8 of the MA3 pedigree. f CHH sites featuring stable epimutations tend to fall outside of sRNA clusters in lineages L2 and L8. g Analysis of cmt2 mutant and Col-0 wt from Stroud et al. [2] show loss of methylation in the mutant at the stable CHH epimutation sites, indicating that these loci are targeted by CMT2. e Compared to the whole genome (wg), stable CHH loci with stable epimutations are enriched for CWA trinucleotides, which is a preferred substrate for CMT2 binding