Supplementary Figure 8: Reduced gene expression in hypermethylated DMRs.

a,b, Reproducibility of RNA-seq between samples. a, Scatterplots with pairwise comparisons for control (C1, C2) and patient (P1, P2) fibroblast RNA-seq data sets. Each dot represents the expression level for a gene (n = 11,963 genes) for both samples, measured in log₂ CPM. b, Scatterplots with pairwise comparisons for wild-type (WT1, WT2, WT3) and Dnmt3a^W326R/W326R (hom1, hom2, hom3) 9-d NPC differentiation RNA-seq data sets. Each dot represents the expression level for a gene (n = 13,022 genes) for both samples, measured in log₂ CPM. CPM, counts per million; R, Pearson correlation coefficient. c, Reduced expression levels for genes associated with hyper-DMRs are evident in fibroblasts. Plotted, log₂ CPM ratios, box, 25th–75th percentile; whiskers, 1.5× interquartile range from box; center line, median. P value, two-sided Wilcoxon rank-sum test, All genes (n = 11,495) with coverage in beadarrays versus genes associated with hypo-DMRs (n = 286) or hyper-DMRs (n = 435). d, Model linking transcriptional changes in stem/progenitor cells due to DNMT3A mutations with organ/organism size. Gain- and loss-of-function mutations in DNMT3A are proposed to impact on stem cell fate dynamics. Owing to increased expression of multipotency genes and a decrease in differentiation/neurogenic gene expression, loss of DNMT3A promotes self-renewal of stem/progenitor cells at the expense of differentiation^31,39. This increase in stem cell/progenitor pool size would be expected to result in increased final cell number generated during development, that is, overgrowth. Gain-of-function DNMT3A mutations in the PWWP domain cause an opposite transcriptional bias and therefore might be expected to accelerate differentiation at the expense of self-renewal, depleting progenitor pools and resulting in reduced final cell number and brain/organism size. ~46 rounds of cell division are required to generate the ~10¹⁴ cells that make up the human body; therefore, small effects over an extended timeframe in cell fate choice (progenitor versus differentiation) could account for substantial differences in final size.