Charting the emergent organotypic landscape of the mammalian gut endoderm at single-cell resolution

Abstract

words)To comprehensively delineate the ontogeny of an organ system, we generated 112,217 singlecell transcriptomes representing all endoderm populations within the mouse embryo until midgestation. We employed graph-based approaches to model differentiating cells for spatiotemporal characterization of developmental trajectories. Our analysis reveals the detailed architecture of the emergence of the first (primitive or extra-embryonic) endodermal population and pluripotent epiblast. We uncover an unappreciated relationship between descendants of these lineages, before the onset of gastrulation, suggesting that mixing of extra-embryonic and embryonic endoderm cells occurs more than once during mammalian development. We map the trajectories of endoderm cells as they acquire embryonic versus extra-embryonic fates, and their spatial convergence within the gut endoderm; revealing them to be globally similar but retaining aspects of their lineage history. We observe the regionalized localization of cells along the forming gut tube, reflecting their extra-embryonic or embryonic origin, and their coordinate patterning into organ-specific territories along the anterior-posterior axis.

Concepts and limitations for learning developmental trajectories from single cell genomics

“…This will play an important role for developmental research as, by integrating multiple data setspotentially at some point generating a cell atlas and a road map of development (Regev et al, 2017)more robust estimates of cell states can be made. Recent work on mouse development have highlighted the potential to obtain such a global view of developmental processes from single cell sequencing data (Cao et al, 2019;Pijuan-Sala et al, 2019;Nowotschin et al, 2018). Moreover, with neural network models for estimating a latent state slowly coming to fruition in single cell omics (Eraslan et al, 2019;Lopez et al, 2018), we can potentially combine data sets using dimension-reduced latent spaces, which represent biologically relevant features.…”

Lineage reconstruction from clonal correlations

“…Multiple Paths in Development. There are several well-documented cases where cells violate a strict tree-like developmental hierarchy (18,28,29). We cannot yet provide a general account of how barcode statistics are determined for arbitrary differentiation topologies, but motivated by the observation that cross-tree transitions are usually sparsely superimposed on otherwise tree-like processes we can ask how a single cross-tree transition would affect conformal symmetries on the tree.…”