Identification of the central intermediate in the extra-embryonic to embryonic endoderm transition through single cell transcriptomics

An increasing number of studies suggests that embryonic development is a highly plastic process and that lineage specification can proceed via alternative routes not encompassed in canonically defined germ layers. Despite these recent revisions to fate maps, the differentiation of pluripotent stem cells (PSCs) to specific lineages tends to continue to focus on defined germ layer differentiation. Thus although recent studies suggest that the visceral organs originate from both embryonic and extra-embryonic lineages, PSC differentiation has focused on generating definitive and gut endoderm solely via an embryonic trajectory. To better resolve the lineage trajectories during development and during corresponding in vitro differentiation, we used MARS-seq to generate a single-cell RNA-seq datasets from mouse embryos expressing a Foxa2Venus reporter and in vitro embryonic stem cell (ESC) differentiation towards definitive endoderm. We coupled this data to a new simplified computational approach to compare cell types, both within our own transcriptomes and across different datasets. Based on this analysis, we captured the central intermediate in the transition between extra-embryonic and embryonic endoderm. When we assessed in vitro ESC differentiation in a number of different endoderm protocols, we found no evidence for this transient intermediate population during differentiation, only traditional embryo-like populations. We then directly tested the capacity of extra-embryonic endoderm to generate organ specific cell types, exploiting our previously defined cell culture system for naïve extra-embryonic endoderm (nEnd). We found that nEnd exhibited gene expression states that reflects early extra-embryonic identity and could be further differentiated to embryonic gut organoids. Taken together, this suggests self-renewing extra-embryonic nEnd represents a source of definitive organ cell types and suggests that in vitro differentiation should take full advantage of the recent observations about developmental plasticity. Overall design: The in Vivo study contains 41 samples from 4 developmental stages of Foxa2-Venus mouse embryos. There are 12 samples of E6.5, 8 samples of E7.5, 17 samples of E8.5, 4 samples of E9.5. The single cells were specifically sorted for Foxa2-Venus protein. The in Vitro study contains 34 samples from 10 experimental stages. The single cells were sorted for Gsc-GFP and Hhex-RedStar or E-cad and Cxcr4.