Fig. 3

Hypermethylation in the mature embryo correspond to hypomethylation in the endosperm. a Representation of a 740-kb genomic region (top) and a zoom-in of 86 kb (bottom) showing genes (in blue) and TEs (in orange) as well as DMRs between endosperm [6] vs. four-day-old seedlings and mature embryos vs. four-day-old seedlings with CG = red, CHG = blue, and CHH = green. DNA hypomethylation is shown as downward bars, hypermethylation as upward bars in the respective comparisons. siRNA abundance for up to 50 reads is shown for 24 nt size (red) and 21–22 nt (green). b Venn diagram representing the overlap between TE annotations affected by CHH-hypermethylation DMRs in the mature embryo with CG- as well as CHG-hypomethylation DMRs in the endosperm. RF = representation factor; RF > 1 = overlap higher than random, RF < 1 = overlap lower than random, with p value < 1.0xe^–30. c Heat map showing absolute values of methylation for CG, CHG, and CHH contexts at five developmental stages/tissues based on the CHH DMRs detected between mature embryos and early embryos as well as four-day-old and ten-day-old seedlings (27,528 DMRs in total). Rows were sorted by complete linkage hierarchical clustering with Manhattan distance as a distance measure using the CG methylation values and aligning the other contexts to these coordinates. d Box plots showing the distribution of seed-derived 24-nt siRNA [27] abundance over endosperm vs. four-day-old seedlings CG hypomethylation DMRs (Hypo CG-DMRs endosperm/seedling), CHH hyper-methylation DMRs in the mature embryo vs. four-day-old seedlings (Hyper CHH-DMRs embryo/seedling) and the overlap of both DMRs (Overlap of DMRs). siRNA reads of 24 nt were counted over each feature and converted to count per million. Student’s t-test in comparison of siRNA abundance of CG hypomethylation endosperm vs. four-day-old seedlings; * = p value < 0.05; *** = p value < 0.001