Single-cell multi-omics, spatial transcriptomics and systematic perturbation decode circuitry of neural crest fate decisions

Cranial neural crest (NC) cells, which can migrate, adopt multiple fates, and form most of the craniofacial skeleton, are an excellent model for studying cell fate decisions. Using time-resolved single-cell multi-omics, spatial transcriptomics, and systematic Perturb-seq, we fully deciphered zebrafish cranial NC programs, including 23 cell states and three spatial trajectories, reconstructed and tested the complete gene regulatory network (GRN). Our GRN model, combined with a novel velocity-embedded simulation method, accurately predicted functions of all major regulons, with over a 3-fold increase in correlation between in vivo and in silico perturbations. Using our new approach based on regulatory synchronization, we discovered a post-epithelial-mesenchymal-transition endothelial-like program crucial for migration, identified motif coordinators for dual-fate priming, and quantified lineage-specific cooperative transcription factor functions. This study provides a comprehensive and validated NC regulatory landscape with unprecedented resolution, offering general regulatory models for cell fate decisions in vertebrates.