Fig. 1

Integrated scRNAseq analysis of embryonic and adult bone marrow datasets. A UMAP characterizing the distribution of cells within each individual dataset (embryonic long bone, fresh BMSCs, cultured BMSCs, BMSC-neurospheres) and upon integration. B Unsupervised clustering of the integrated dataset yielded 14 clusters (C1-C14). C The RNA velocity graph revealed C8 and C2 as starting states, C3 and C13 as end points. D Pseudotime reconstruction indicated C3 (neural differentiation) and C13 (osteogenesis) were two distinct differentiation end states. E C1 and C7 were branch points originating from C8, with the former expressing genes associated with neural differentiation and the latter expressing genes related to skeletal development. F Identification of genes with differential expression across clusters was achieved via Moran’s index calculations. Subsequently, these genes were organized into 27 distinct modules based on their expression patterns, and associated with their corresponding clusters. G Functional enrichment analysis upon gene sets revealed that module 17 (correlating with C3) exhibited processes related to neurodevelopmental potential, whilst module 21 (correlating with C13) exhibited processes related to skeletal muscle development. H Along the pseudotime trajectory of the integrated dataset, a continuous decrease in the mesenchymal marker CXCL12 was detected. NC-related markers (P75, Pax7, Snail2 and Twist2) also decreased whilst on the contrary, neurogenesis-related genes (NRN1, GAP43 and TUBB3/TUJ1) increased. I Joint density map visualizing the co-expression pattern of genes in the integrated dataset