Extended Data Fig. 3: RdDM is dispensable for the recovery of H3K9me2 and CH methylation.

a, Recovery of CHG methylation in F9 progeny from a cross between ddm1 and wild-type (x-axis; reanalyzed from data⁶⁶), and F1 (cc x sss) (y-axis). Left, TE genes of ddm1-haplotype (that is TE genes localized on the ddm1-derived chromosome segments for both alleles in the line epiRIL98) were analyzed (n=1178) after exclusion of TE genes with small CH methylation difference in the mutants (CHG<0.1 and CHH<0.05). Middle, TE genes with low levels of matching 24-nt siRNA (RPKM < 2) in ddm1 (reanalyzed from data⁶⁴, n=461). Right, TE genes with low levels of matching 24-nt siRNA (RPKM < 2) in sss mutants (reanalyzed from data⁸, n=144). b, Recovery of CHH methylation as shown in a. c, d, The CH methylation recovery from the ddm1-epiallele was associated with CG methylation recovery. Recovery of CH methylation were compared to CG methylation level for CHG sites (c) and CHH sites (d). R represents the pearson’s correlation coefficient between two methylation recovery in the TE genes. e, Comparison of H3K9me2 levels in WT, cc and sss mutants, and their F1 progeny in DRM2 and drm2 backgrounds, based on western blot analysis. Black bars show the positions of molecular weight marker of the size 17kDa. Source data are provided as a Source Data file. f, CHG remethylation in F1 (cc x sss) in wild-type, drm2 and rdr1 rdr2 rdr6 (rdr126) mutant background estimated by digestion with Pst I, an enzyme sensitive to methylation at CHG sites, followed by quantitative-polymerase chain reaction analysis. The experimental details are described in the Methods section. Even in the drm2 or rdr126 background, CHG methylation lost in cc or sss mutations recovers in their F1 progeny to levels comparable to that in the control wild-type plant.