The maintenance of pluripotency requires coordinated expression of a set of essential genes. Using our recently established haploid human pluripotent stem cells (hPSCs), we generated a genome-wide loss-of-function library targeting 18,166 protein-coding genes to define the essential genes in hPSCs. With this we could allude to an intrinsic bias of essentiality across cellular compartments, uncover two opposing roles for tumour suppressor genes and link autosomal-recessive disorders with growth-retardation phenotypes to early embryogenesis. hPSC-enriched essential genes mainly encode transcription factors and proteins related to cell-cycle and DNA-repair, revealing that a quarter of the nuclear factors are essential for normal growth. Our screen also led to the identification of growth-restricting genes whose loss of function provides a growth advantage to hPSCs, highlighting the role of the P53–mTOR pathway in this context. Overall, we have constructed an atlas of essential and growth-restricting genes in hPSCs, revealing key aspects of cellular essentiality and providing a reference for future studies on human pluripotency.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors thank E. Meshorer and all members of The Azrieli Center for Stem Cells and Genetic Research for their input and critical reading of the manuscript. The authors also thank O. Yanuka, T. Golan-Lev and A. Petcho for assistance with tissue culture. This work was partially supported by the US–Israel Binational Science Foundation (grant no. 2015089), by the Israel Science Foundation (grant no. 494/17) and by the Azrieli Foundation. A.Y. is supported by the Lady Davis Postdoctoral Fellowship. I.S. is supported by the Adams Fellowship Program of the Israel Academy of Sciences and Humanities, and N.B. is the Herbert Cohn Chair in Cancer Research.
Integrated supplementary information
a, Analysis of the abundance of sgRNAs within the haploid hESC population at multiple time points after the co-delivery of sgRNAs and Cas9. Shown is a correlation matrix for all time points with correlation coefficients on the right diagonal and the scatter plots on the left diagonal of the matrix (n=181,131 gRNAs). b, CRISPR score formula. c, Scatter plot of CRISPR scores between two replicate experiments (n=18,166 genes, r = Pearson’s correlation coefficient). d, Chromosomal distribution of growth-restricting genes. Shown are growth-restricting genes (blue lines) and all other genes targeted in the library (grey lines).
a, Venn diagram showing the overlap of cell-essential genes identified in the current study in hESCs and in near-haploid KBM7 leukemic cells by Wang et al. 13 and by Blomen et al.14 and in several immortalized and cancer lines from different tissues of origin by Hart et al.16 (n=15,899 genes). b, PCA plot demonstrating the separation of genes in different studies. While the current study and Wang et al. study13 used the same sgRNA library, Blomen et al.14 used a transposon-mediated gene-trap mutant library and Hart et al.16 used two independent sgRNA libraries. c, Analysis of first percentile CRISPR scores across autosomes of near-haploid KBM7 cells (in which chromosome 8 is diploid), three diploid cell lines used in Wang et al. study6 and haploid hESCs from the current study. Chromosome 8 is highlighted in red. Dashed lines indicate the threshold corresponding to p-value of 0.025 on either side of the distribution of first percentile CRISPR scores (two-tailed Z-test, n=22 chromosomes, *P=0.02). d, -log10 P-values for chromosome 8 in the distribution of the first percentile CRISPR scores across autosomes in different cell lines. In near-haploid KBM7 cells, the first percentile CRISPR score for the diploid chromosome 8 is significantly different from those of the other autosomes (marked by asterisk), suggesting more efficient loss-of-function analysis in haploid chromosomes (two-tailed Z-test, n=22 chromosomes, *P=0.02).
Supplementary Figure 3 Analysis of cell-essential genes for their cellular compartmentalization, disease association and tumor formation.
a, Pie charts for each cellular compartment in hESCs demonstrating both enriched and depleted genes. b, Fraction of genes that are essential and expressed in hESCs within each cellular compartment c, Fraction of essential genes within the total number of genes in each cellular compartment in KBM7 cells. d, Distribution of CRISPR scores of genes associated with AR human disorders with a growth retardation phenotype in hESCs (green curve) and in leukemic KBM7 cells (purple curve). e-f, Volcano plots representing Q-value and CRISPR score of the canonical oncogenes (e) and tumor suppressor genes (f) that did not show significant values (see Fig. 2e and f, KS-test, n=20 gRNAs, two biological replicates of 10 independent gRNAs per gene). g, Heatmaps showing the comparison of CRISPR scores of three groups of genes between hESCs and the cancer cell lines: Growth-restricting tumor suppressors in hESCs that lost this feature of growth-restriction in cancer cell lines (left heatmap), essential oncogenes in hESCs that lost their essentiality in cancer cell lines (middle heatmap) and nonessential oncogenes in hESCs that gained essentiality in cancer cell lines (right heatmap). Light blue and light red boxes designate significant FDR values (shown below the CRISPR score values, positive and negative values indicate enriched and depleted gRNAs, respectively) for growth-restricting and essential genes, respectively (KS-test, n=20 gRNAs).
a, Scatter plot demonstrating CRISPR scores of hESC-enriched transcription factors in hESCs and KBM7 cells; in green: essential genes in KBM7 cells and hESCs, in red: essential genes only in hESCs. b, Mean ± s.e.m. of gRNA reads/gene over time in culture for validated genes (SALL4, DSCC1, VRTN, SEPHS1) along with their expression (normalized FPKM from RNA sequencing data25) in hPSCs and 27 somatic cell types. Lymphocytes and fibroblasts are transformed cell lines. CRISPR scores and P-values appear at the bottom (KS-test, n=20 gRNAs). c, Relative transcript levels (analysed by qRT–PCR) of targeted hESC-essential genes by siRNA knockdown as compared to a non-targeting siRNA control in diploid hESCs. Cells were harvested 24 hours after transfection of siRNAs for SALL4, 48 hours for VRTN and 72 hours for DSCC1 and SEPHS1 (n=3 biological replicates, mean ± s.e.m values). d, Relative wild-type transcript levels (analysed by qRT–PCR) of targeted hESC-essential genes in sgRNA-knockout lines 3 days after the delivery of sgRNAs and Cas9. Control lines received only Cas9 in the absence of a sgRNA. Values represent the averages of two biological replicate experiments with three technical replicates for each. e, Cell viability assay in hESC and KBM7 sgRNA-knockout lines for PIK3CA and PDIA4, two genes expressed in both cell types, 4 days after the delivery of sgRNAs and Cas9 (n=3 biological replicates for KBM7 cells, n=6 biological replicates for hESC cells, mean ± s.e.m values). f, Relative wild-type transcript levels (analysed by qRT–PCR) of targeted genes in sgRNA-knockout lines of KBM7 cells, 3 days after the delivery of sgRNAs and Cas9. Values represent the average of three technical replicates. g, Relative wild-type transcript levels (analysed by qRT–PCR) of targeted hESC-essential genes in sgRNA-knockout lines 3 days after the delivery of sgRNAs and Cas9. Values represent the average of two biological replicate experiments with three technical replicates for each. When applicable, source data are provided in Supplementary Table 4.
Highlighted in red are significantly enriched growth-restricting genes that were found among the top 50 genes of this group in the haploid hESC screening. Enrichment fold-changes of TP53 target genes are shown on the right. Numbers indicate different P53 target pathways, red and black bars show high and low-ranking growth-restricting genes, respectively.
a-d, Growth curves of IGF1-treated (red) vs. control (blue) diploid hESCs grown in mTESR1 (a), conditioned medium with 12% KSR (standard) (b), conditioned medium with 6% KSR (c), conditioned medium with 3% KSR (d) (n=4 biologically independent samples). Mean ± s.e.m. are presented and unpaired two-tailed t-test was applied (PmTeSR1=0.01, P12% KSR=2.7e-4, P6% KSR=8.4e-4, P3% KSR=3.6e-4, *P < 0.05; ***P < 0.001). e-g, Gating strategy for flow cytometry analysis of apoptotic cells (e, Annexin-V staining), cell-cycle profile (f, Propidium Iodide staining) and TRA-1-60+ pluripotent cells (g, TRA-1-60-staining). A minimum of 30,000 cells were analysed for each sample. h-i, Western blots demonstrating phosphorylated levels of mTORC1 target, 4E-BP1 and mTORC2 target AKT (top panels) upon treatment of hESCs with rapamycin and Torin 1 for short- (h) and long-term (i) treatments. To enable incubation with three primary antibodies separately, nitrocellulose membrane was cut where the dashed lines were indicated. Quantification of the western blots are shown in the bottom panels (n=3 biological replicates, mean ± s.d. values). As shown, rapamycin inhibits the activity of mTORC1 but not that of mTORC2 in hESCs in both short- and long-term treatments, whereas Torin 1 inhibits both complexes. j and k, Growth curves of rapamycin- or Torin 1-treated diploid hESC lines, REX1-GFP WA09 (j) and CSES9 (k), over the course of four days (n=3 biological replicates, mean ± s.e.m values). Shown are the values normalized to the untreated controls of the corresponding time points. l, Gene ontology analysis for the significantly down-regulated transcripts after four days of rapamycin treatment in hESCs (CSES9, blue) and in human foreskin fibroblasts (pink) (n=3 biological replicates). When applicable, source data are provided in Supplementary Table 4.
Uncropped membranes used in the Supplementary Figure 6h-i. Sample IDs as follows: Gel1: 1) untreated (short term), biological replicate 1; 2) untreated (short term), biological replicate 2; 3) untreated (short term), biological replicate 3; 4) Rapamycin (1h), biological replicate 1; 5) Rapamycin (1h), biological replicate 2; 6) Rapamycin (1h), biological replicate 3; 7) Rapamycin (3h), biological replicate 1; 8) Rapamycin (3h), biological replicate 2; 9) Rapamycin (3h), biological replicate 3; 10) Torin 1 (1h), biological replicate 1; 11) Torin 1 (1h), biological replicate 2; 12) Torin 1 (1h), biological replicate 3. Gel2: 13) Untreated (long term), biological replicate 1; 14) Untreated (long term), biological replicate 2; 15) Untreated (long term), biological replicate 3; 16) Rapamycin (2 days), biological replicate 1; 17) Rapamycin (2 days), biological replicate 2; 18) Rapamycin (2 days), biological replicate 3.
Supplementary Figures 1–7 and Supplementary Table legends
CRISPR scores and significance values of all genes in the loss-of-function mutant library of haploid hESCs.
List of essential genes linked to autosomal recessive disorders with growth retardation phenotypes.
List of hESC-essentialome genes.
Statistics source data.