Potency biomarker signature genes from multiparametric osteogenesis assays: Will cGMP human bone marrow mesenchymal stromal cells make bone?

In skeletal regeneration approaches using human bone marrow derived mesenchymal stromal cells (hBM-MSC), functional evaluation before implantation has traditionally used biomarkers identified using fetal bovine serum-based osteogenic induction media and time courses of at least two weeks. However, emerging pre-clinical evidence indicates donordependent discrepancies between these ex vivo measurements and the ability to form bone, calling for improved tests. Therefore, we adopted a multiparametric approach aiming to generate an osteogenic potency assay with improved correlation. hBM-MSC populations from six donors, each expanded under clinical-grade (cGMP) conditions, showed heterogeneity for ex vivo growth response, mineralization and bone-forming ability in a murine xenograft assay. A subset of literature-based biomarker genes was reproducibly upregulated to a significant extent across all populations as cells responded to two different osteogenic induction media. These 12 biomarkers were also measurable in a one-week assay, befitting clinical cell expansion time frames and cGMP growth conditions. They were selected for further challenge using a combinatorial approach aimed at determining ex vivo and in vivo consistency. We identified five globally relevant osteogenic signature genes, notably TGF-β1 pathway interactors; ALPL, COL1A2, DCN, ELN and RUNX2. Used in agglomerative cluster analysis, they correctly grouped the bone-forming cell populations as distinct. Although donor #6 cells were correlation slope outliers, they contrastingly formed bone without showing ex vivo mineralization. Mathematical expression level normalization of the most discrepantly upregulated signature gene COL1A2, sufficed to cluster donor #6 with the bone-forming classification. Moreover, attenuating factors causing genuine COL1A2 gene down-regulation, restored ex vivo mineralization. This suggested that the signature gene had an osteogenically influential role; nonetheless no single biomarker was fully deterministic whereas all five signature genes together led to accurate cluster analysis. We show proof of principle for an osteogenic potency assay providing early characterization of primary cGMP-hBM-MSC cultures according to their donor-specific bone-forming potential.