Fig. 3: GbM quantitatively affects gene expression.

a, GbM level and expression of AT5G53010 across accessions. Per cent expression variance explained (PVE) by gbM is indicated. Pearson’s correlation analysis was used to assess the association between the two variables. FPKM, fragments per kilobase of transcript per million mapped reads. b,c, Expression in met1 seedlings compared with WT of Bonferroni (α = 0.05) retained eQTL^gbM genes across 16 accessions (b) and across Col-0 tissues (leaf, seedling and inflorescence; c). Numbers of unique genes within each group are noted above the plots, means are indicated by ‘+’ and noted below the plots. Sample medians are shown by centre lines, and box edges represent the 25th and 75th percentiles. Whiskers extend to 1.5 times the interquartile range. P values were calculated using a two-tailed Student’s t-test to compare the indicated eQTL group with non-associated (NA) genes. d, Relationship between gbM levels of retained +eQTL^gbM and −eQTL^gbM genes in WT plants across 16 accessions and log₂ fold expression change in met1 compared with the WT of that accession. Genes were grouped by gbM levels. R and P values correspond to Pearson’s correlation. e, Relationship between the gene expression coefficient of variation (CV) across biological replicates in WT (blue) and met1 (purple) and WT gbM level across 16 accessions. Genes were grouped by gbM levels in WT. ρ and P values correspond to Spearman’s rank correlation coefficient. f, Relationship between the log₂ fold CV change in met1 compared with WT and the gbM level across 16 accessions. R and P values correspond to Pearson’s correlation. g,h, Expression in h1met1^+/− (g) and h1met1^+/+ (h) compared with h1 of Bonferroni (α = 0.05) retained (R) or lost (L) eQTL^gbM genes that are either demethylated (Demeth.) or keep methylation. Box plots as in b and c. P, two-tailed Student’s t-test.