Extended Data Figure 3: Functional validation using a pooled shRNA screen.

This figure relates to Fig. 3 in the main text. a, Detailed outline of the pooled shRNA screen. Each stage (NE, ERG and MRG) was infected with an optimized virus titre aiming for an average of one shRNA integration per cell. Immediately after infection, cells were subjected to puromycin (puro) selection and bulk population material was collected 24 h after infection and before efficient shRNA knockdown. Five days after infection and selection, cells were FACS-sorted for HES5–GFP and both GFP⁺ and GFP⁻ cells were collected for analysis. Subsequently, gDNA was extracted and all integrated shRNAs were amplified by PCR for each population separately. The resulting material was then used to construct libraries for next-generation sequencing to count the number of shRNA integrations for each shRNA in each cell population. b, Overlap of genes identified to facilitate HES5⁺ cell maintenance, progression or proliferation determined by genes with at least two shRNAs significantly (q ≤ 0.05) over-represented in the HES5⁺ population with respect to the 24-h or HES5⁻ control. c, Regulator predictions based on differential gene expression. Performance is measured as percentage of the top 20 differentially expressed factors for each stage for those the transcription factors included in the shRNA library. d, Regulator predictions based on TERA ranking for H3K4me3, H3K4me1, H3K27ac or DNAme. Performance is measured as percentage of the top 20 predicted activating or repressive motifs for each stage map** to a transcription factor included in the shRNA library. e, Detailed heat map showing the top 30 predicted motifs and corresponding transcription factors differentially active between consecutive differentiation stages based on the combined TERA scores for H3K27ac, H3K4me3, H3K4me1 and DNAme. In addition, knockdown results as depletion scores (green/red heat map) obtained at each stage are shown on the right. f, Heat map showing the pairwise Pearson correlation coefficient (PCC) of the log₂ read-count normalized shRNA libraries across all conditions and technical replicates. g, Individual validation for shRNAs against OTX2 and PAX6 at the NE stage, which showed no effect in our pooled screening approach at any stage. Shown are qPCR levels for OTX2 or PAX6, HES5 and puromycin relative to HPRT. Each gene was measured in an independent knockdown experiment for a pool of the five shRNAs against PAX6 (blue), OTX2 (green), lacZ (orange) as well as the uninfected control (red).