BackgroundHuman bone marrow stromal cells (BMSCs) are an easily accessible and expandable progenitor population with the capacity to generate neural cell types in addition to mesoderm. Lineage tracing studies in transgenic animals have indicated Nestin + BMSCs to be descended from the truncal neural crest. Single-cell analysis provides a means to identify the developmental origin and identity of human BMSC-derived neural progenitors when lineage tracing remains infeasible. This is a prerequisite towards translational application.MethodsWe attained transcriptomic profiles of embryonic long bone, adult human bone marrow, cultured BMSCs and BMSC-derived neurospheres. Integrated scRNAseq analysis was supplemented by characterization of cells during culture expansion and following provision of growth factors and signalling agonists to bias lineage.ResultsReconstructed pseudotime upon the integrated dataset indicated distinct neural and osteogenic differentiation trajectories. The starting state towards the neural differentiation trajectory consisted of Nestin + /MKI67 + BMSCs, which could also be diverted towards the osteogenic trajectory via a branch point. Nestin + /PDGFRA + BMSCs responded to neurosphere culture conditions to generate a subpopulation of cells with a neuronal phenotype according to marker expression and gene ontogeny analysis that occupied the end state along the neural differentiation trajectory. Reconstructed pseudotime also revealed an upregulation of BMP4 expression during culture of BMSC-neurospheres. This provided the rationale for culture supplementation with the BMP signalling agonist SB4, which directed progenitors to upregulate Pax6 and downregulate Nestin.ConclusionsThis study suggested BMSCs originating from truncal neural crest to be the source of cells within long bone marrow possessing neural differentiation potential. Unravelling the transcriptomic dynamics of BMSC-derived neural progenitors promises to enhance differentiation efficiency and safety towards clinical application in cell therapy and disease modelling.