A single-cell CRISPR screen defines a gene regulatory network governing human pluripotency in primed and naïve cells

Human pluripotent stem cells (hPSCs) can give rise to the three embryonic germ layers and possess a limitless propagation potential. We have previously defined the essentialome of hPSCs through a genome-wide CRISPR screen, but the function of each essential gene is still obscure. Here, we uncovered the role of pluripotent-specific transcription factors (TFs) that are essential in hPSCs, using a pooled CRISPR screen combined with single-cell transcriptomics. We show that most of these TFs form a remarkably interconnected gene regulatory network (GRN), controlling multiple aspects of pluripotency, spanning regulation of differentiation, self-renewal, cell survival, and expression of transposable elements. Interestingly, various TFs function as positive regulators of pluripotency, serving as gatekeepers from differentiation, and preventing transition into different cell lineages in a TF-specific manner. Surprisingly, other TFs inhibit the pluripotency network, potentially balancing its maintenance with responsiveness to differentiation cues, which we could validate through individual gene knockouts. Finally, by perturbing the TFs network in naïve cells, we reveal both shared and distinct roles for the GRN between the two pluripotency states. In summary, this analysis enabled the construction of an extended network for human pluripotency. This study sheds new light on principles in early human development and can hopefully be used to enhance the use of hPSCs for disease modelling or cell therapy. Overall design: For the primed perturb-seq experiment, Cas9-EGFP hPSCs cells were dissociated using TrypLE Select (Thermo Fisher Scientific), centrifuged, and resuspended in mTeSR1 medium (STEMCELL Technologies) supplemented with 10 µM ROCK inhibitor (Y27632) and 8 µg/ml polybrene (Sigma). A total 3.6X106 cells were transduced in suspension with the virus library at a multiplicity of infection (MOI) of 0.3 and the cells were plated on 3 Matrigel coated 6-well plates (Corning). 24h after transduction, the virus containing medium was replaced with puromycin-containing mTeSR1 medium (0.3 mg ml-1, Sigma). The medium was subsequently replaced daily with puromycin-containing medium until day 6 post transduction, when cells were harvested for 10X genomics. For harvesting, cells were dissociated using TrypLE Select (Thermo Fisher Scientific), and after centrifugation were re-suspended in PBS containing 0.002% Trypan Blue to stain dead cells. Cells were again centrifuged and resuspended in PBS containing 10% KSR and 10 µM ROCK inhibitor Y-27632, filtered through a 70 µm cell strainer (Corning) and sorted using BD FACSAria III (BD Biosciences). Specifically, cells which were GFP+ (positive for Cas9-EGFP), BFP+ (positive for the gRNA library) and far-red negative (Trypan Blue+ cells emit far red signal, enabling the isolation of live cells) were sorted with 488 nm, 405 nm, and 650 nm lasers, and subjected to scRNA-seq using 10X genomics. For the naïve condition, the experimental design of the experiment was adjusted as naïve cells under the 5i/L/A condition proved to be highly resistant to lenti-viral infection. Specifically, primed hPSCs were first infected with the viral library as mentioned above and kept under puromycin selection for 7 days. Then, naïve conversion was performed as mentioned above. On day 18 of the conversion, naïve cells were dissociated using TrypLE Select (Thermo Fisher Scientific) and split into 6 6-well plates with a density of 1.5X105 cells/well. After 24h, the cells were transfected with a pSpCas9(BB)-2A-GFP (PX458) construct (a gift from Feng Zhang; addgene plasmid #48138) in the presence of Lipofectamine Stem Transfection Reagent (Invitrogen; STEM00003). 5 days after the transfection, cells were sorted based on the same criteria as in the primed experiment and subjected to scRNA-seq using 10X genomics.