Extended Data Figure 6: Global transcriptomic and epigenomic analysis.

a, Hierarchical clustering of microarray gene expression data from ESCs, EpiSCs, rsEpiSCs and in vivo isolated E5.75 and E6.5 epiblasts. b, Two-way scatter plot of gene expression data from RNA-seq of EpiSCs and rsEpiSCs. Black lines indicate fourfold cut-off in expression level difference. Pearson correlation coefficient (r) between samples is shown at the upper right corner. c, Top five Gene Ontology (GO) terms enriched in the set of genes that are differentially expressed by at least fourfold (either up or down) between rsEpiSCs and EpiSCs. d, Average H3K27me3 signal at Polycomb target genes in rsEpiSCs (purple) and EpiSCs (green). e, Plots of DNA methylation and histone methylation (H3K4me3 and H3K27me3) signals around the transcription start sites of representative classes of genes. Examples given include pluripotent genes (Sall4, Klf4 and Lin28a), neuronal-related genes (Sox2, Gbx2 and Sox1) and cell-membrane-related genes (Cldn6, Cldn3 and Cdh1). f, Global cytosine methylation at CG sites (mCG) levels of EpiSCs and rsEpiSCs (top left). The numbers of hyper- and hypo-DMRs discovered in rsEpiSCs (top right). The numbers of promoter-associated (transcription start site ± 2.5 kb), distal (>10 kb from transcription start site) and proximal (transcription start site ± 2.5 to 10 kb) rsEpiSCs hyper- and hypo-DMRs (bottom). g, Positive correlation between the amount of gene body non-CG DNA methylation and the level of gene expression. h, Gene Ontology biological process and molecular function term enrichment for genomic regions associated with rsEpiSCs hyper-DMRs. i, PC1–PC2 plane from PCA analysis of transcriptome comparison between samples from this study (rsEpiSCs (circled with red line) and EpiSCs (circled with yellow line)) and a published data set⁶ (in vivo epiblast isolated from different developmental stages (CAV, cavity; PS, pre-streak; LMS, late mid-streak; LS, late streak; OB, no bud; EB, early bud; LB, late bud) and EpiSCs (circled with thick blue line)). The green arrow through the in vivo samples delineates a progressing ‘developmental axis’. j, Hierarchical clustering of rsEpiSCs (red), EpiSCs (both from this study (yellow) and ref. 6 (blue)) and epiblasts of CAV, PS, LMS, LS, OB, EB and LB stages (green) using data from all annotated probes. k, Relative expression between rsEpiSCs and in vivo late-streak-stage epiblast (ref. 6) of genes characteristic of anterior mesendoderm (AME), anterior definitive endoderm (ADE), anterior primitive streak (APrS), whole primitive streak (PrS) and posterior primitive streak (PPrS). i, Primordial germ cell induction from Blimp1–YFP mESCs and rsEpiSCs. Left, before induction, both mESCs and rsEpiSCs were found negative for YFP; successful induction was observed with mESCs, as indicated by a positive YFP signal in cell aggregates, but not with rsEpiSCs. Right, PGC induction efficiency was compared between Blimp1–YFP mESCs and rsEpiSCs. Error bars indicate s.d. (n = 3, independent experiments); t-test, **P < 0.01, *P < 0.05. CAV, cavity; PS, pre-streak; ES, early streak; MS, mid-streak; LMS, late mid-streak; LS, late streak; OB/EB, no bud/early bud; LB, late bud.