In vivo clonal lineage tracing combined with clone2vec machine learning analysis reveals that mouse embryonic development is organized by continuous, embryo-wide spatial gradients of clonal fate biases where anatomical position and clonal composition are mutually predictive through position-specific auxiliary transcriptional programs, and these axial patterning systems can be experimentally overridden by mosaic perturbations of extrinsic signaling pathways such as Hedgehog signaling.

Embryogenesis is commonly viewed through a tree model of cell differentiation, which fails to capture the spatiotemporal modulation of cell multipotency underlying morphogenesis. In this study we profile the multipotency landscape of the embryo, using in vivo single-cell clonal lineage tracing of mouse embryos traced from neurulation until mid-gestation, combined with a machine learning tool that categorizes individual clones into lineages based on shared transcriptional context. This revealed a previously unrecognized continuous, embryo-wide gradient of clonal fate biases, in which anatomical position and clonal composition are mutually predictive. Comparing clonal lineages revealed gene regulatory networks underlying the dynamic biasing of cells towards specific fates by spatial transcription factor programs. However, mosaic combinatorial perturbations targeting the Hedgehog pathway generated clones in which positional identity was mismatched with clonal composition, demonstrating that extrinsic signals can override the axial patterning system underlying clonal fate biases. Altogether, our work demonstrates an effective practical approach for dissecting mechanisms of lineage specification and has implications for stem cell engineering.