Identification of differentiation-stage specific markers that define the ex vivo osteoblastic phenotype

The phenotype of osteoblastic (OB) cells in culture is currently defined using a limited number of markers of low sensitivity and specificity. For the clinical use of human skeletal (stromal, mesenchymal) stem cells (hMSC) in therapy, there is also a need to identify a set of gene markers that predict in vivo bone forming capacity. Thus, we used RNA sequencing to examine changes in expression for a set of skeletally-related genes across 8 time points between 0 and 12 days of ex vivo OB differentiation of hMSC. We identified 123 genes showing significant temporal expression change. Hierarchical clustering and Pearson's correlation generated 4 groups of genes: early stage differentiation genes (peak expression: 0-24 h, n = 28) which were enriched for extracellular matrix organisation, e.g. COL1A1, LOX, and SERPINH1; middle stage differentiating genes (peak expression days: 3 and 6, n = 20) which were enriched for extracellular matrix/skeletal system development e.g. BMP4, CYP24A1, and TGFBR2; and late stage differentiation genes (peak expression days: 9 and 12, n = 27) which were enriched for bone development/osteoblast differentiation, e.g. BMP2 and IGF2. In addition, we identified 13 genes with bimodal temporal expression (2 peaks of expression: days 0 and 12) including VEGFA, PDGFA and FGF2. We examined the specificity of the 123 genes' expression in skeletal tissues and thus propose a set of ex vivo differentiation-stage-specific markers (n = 21). In an independent analysis, we identified a subset of genes (n = 20, e.g. ELN, COL11A1, BMP4) to predict the bone forming capacity of hMSC and another set (n = 20, e.g. IGF2, TGFB2, SMAD3) associated with the ex vivo phenotype of hMSC obtained from osteoporotic patients.