Extended Data Fig. 4: Direct targets of SOX17 in hPGCLCs and DE.

a, Cumulative distribution function plot showing the functional prediction of SOX17 and PRDM1. The ChIP peaks of SOX17/PRDM1 in hPGCLCs were assigned to genes with TSS within 100 kb of the peak summits. A regulatory potential score was calculated for each gene based on the distance between the peak summit and the TSS⁹⁶. The genes were then divided into upregulated, downregulated and unchanged according to their expression patterns upon SOX17 or PRDM1 overexpression. Cumulative distribution function plot was generated for each group with genes ranked by decreasing regulatory potential. A one-tailed Kolmogorov-Smirnov test was used to determine the statistical significance between the differentially expressed groups and the unchanged group. Note that upregulated genes (but not downregulated genes) upon SOX17 induction have a significantly higher tendency to be bound by SOX17. In contrast, genes downregulated upon PRDM1 overexpression tend to be bound by PRDM1. b, GO biological process terms that were enriched in SOX17 direct up targets (red dots in Fig. 3d). c, Expression heatmap of SOX17 direct up target genes. Shown are the genes which were (1) upregulated both by SOX17 alone (log2(fold change) >1 and adjusted p-value <0.05 between Dex-treated and non-treated 12h PreME aggregates); (2) upregulated by cytokines (log2(fold change) >2 and adjusted p-value <0.05 between day 2 hPGCLCs and PreME); and (3) downregulated in SOX17 KO (log2(fold change) >1 and adjusted p-value <0.05 between SOX17 KO day 2 aggregate and WT day 2 aggregate)²⁰. d, Top eight TFs with binding sites (ReMap2020) enriched in hPGCLC-specific and DE-specific SOX17 peaks. e, Genome browser snapshots showing the epigenetic landscape of DE-specific (CER1 and LEFTY2) and hPGCLC-specific (NANOS3 and PDPN) SOX17-bound gene targets. f, Heatmap showing expression of genes associated with the top enriched motifs (Fig. 3h) and the top enriched TF binding sites (in d) in hPGCLC-specific and DE-specific SOX17 peaks. g, UniProtKB Keywords that were enriched in PRDM1 direct down targets (blue dots in Fig. 3i).