Single cell lineage analysis reveals cell fate determination events during directed β-cell differentiation

In vitro differentiation of human ES cells into insulin-producing β-like cells offers new opportunities for pancreatic development modeling and potential diabetes therapy. However, the precise molecular events associated with this multi-stage process remain unclear. Here, we generated 95,308 single cell transcriptome data encompassing the entire differentiation process, and reconstructed a tree delineating the fate choices of all major cell populations in both endocrine and non-endocrine lineages. Most detectable genes are dynamically regulated during differentiation, and stage-specific transcription factors create the time-dependent enhancer landscapes. Interestingly, we found that many diabetes/obesity risk genes are only transiently expressed during differentiation. One example is TCF7L2, which peaked during the pancreatic progenitor stages driven by an enhancer located within a diabetes GWAS locus. From the lineage analysis, we found that the NOTCH signaling-downstream gene HES1 is one of the “switch genes” associated with the choice of non-endocrine cell fate; this led to an improved differentiation protocol for better yield of endocrine cells by adjusting the timing of NOTCH inhibition. Additionally, we discovered that ROCKII inhibitor also promotes endocrine differentiation by suppressing non-endocrine lineage. Taken together, our comprehensive single cell lineage analyses provided a valuable resource for the study of pancreatic development and disease etiology. Single cell sequencing (Drop-seq) for In vitro beta cell differentiation