Fig. 4: Suppressing photorespiration alters DNA methylation patterns in the wild type and mitigates methylome and transcriptome defects in mthfd1-1.

a, Hierarchically clustered heat map of mean scaled mCG ratios (z scores) of all DMRs (rows) compared to the WT under cCO₂ (left); distributions of mCG ratios in four main clusters, B1 (n = 9,019), B2 (n = 13,430), B3 (n = 2,499) and B4 (n = 76) (centre); and per cent overlaps of DMRs with categorized DNA-methylated regions per cluster (right). mCG ratios were computed from per-site C/T counts pooled across two biological replicates for each genotype and CO₂ condition. The box plots indicate the medians (centre lines), IQRs (boxes) and 1.5 × IQR (whiskers); the violin width reflects the kernel density. b, Complete-linkage clustering dendrogram (Euclidean distance) of genotypes and CO₂ conditions based on mCG, shown above distributions of mCG ratios in union DMRs (covered across samples) from clusters A2 (n = 2,374) and B1 (n = 5,277). The cluster labels follow a and Fig. 1d. DMR mCG ratios were computed from per-site C/T counts pooled across two biological replicates for each genotype and CO₂ condition. The box plots indicate the medians (centre lines), IQRs (boxes) and 1.5 × IQR (whiskers); the violin width reflects the kernel density. n indicates the number of DMRs. c, DMR overlaps between CMT2-dependent methylation (A1 and B1) and gbM (A2 and B2) enriched clusters. d, Volcano plot of TE expression in m under hCO₂ compared to m under cCO₂ (n = 3). The top eight TEs ranked by q value are indicated by subfamily names. e, Heat map of mean (n = 3) expression changes relative to the WT under cCO₂ of mthfd1-1 DEGs encoding enzymes involved in FOCM, photorespiration and other related metabolic pathways (see row annotation; SMM, S-methylmethionine cycle). The rows were hierarchically clustered by scaled normalized transcript levels.