Direct 3D isolation in gelatin methacryloyl microbeads versus conventional 2D isolation and culture: a trade-off between mesenchymal stromal cell purity and niche-associated transcriptional profiles

Bone-marrow mesenchymal stromal cell (MSC) isolation typically couples plastic-adherence selection to planar expansion. Therefore, the recovered cells may reflect culture adaptation as much as native stromal biology. Direct isolation within a three-dimensional matrix could reduce planar selection, but its effects on cellular recovery, composition and MSC transcriptional profiles remain insufficiently resolved. We therefore developed a gelatin methacryloyl (GelMA) workflow progressing from bulk hydrogels to medium-surrounded GelMA microbeads (GMBs) and compared 3D-derived cells with conventional 2D cultures. Qualitative live/dead staining with calcein acetoxymethyl ester (CAM) and propidium iodide (PI) and 5-ethynyl-2′-deoxyuridine (EdU) imaging established cellular activity in both GelMA formats. In bulk GelMA cultures from four human donors, CD73-, CD90- and CD105-associated flow-cytometric readouts did not increase from day 7 to day 21. At day 21, the combined marker readout was numerically higher in donor-matched 2D cultures than in bulk 3D cultures, which retained heterogeneous marker-defined populations. GMBs enabled cell recovery and flow-cytometric characterization and motivated an exploratory one-donor single-cell RNA sequencing (scRNA-seq) analysis. The final dataset contained 37,784 cells across four samples and 23 annotated populations. MSC-focused re-embedding resolved 13,175 cells into two 3D and four 2D clusters. A pooled comparison identified 4,840 differentially expressed genes (DEGs, false discovery rate [FDR]<0.01, |log2FC|≥1), indicating a broad method-associated expression shift. The 3D C1 cluster combined progenitor-, extracellular-matrix-, adhesion-, cytokine-, hypoxia/glycolysis- and predicted-interaction-associated programmes, whereas 2D clusters showed mitochondrial, cytoskeletal/mechanoadaptive and proliferative programmes. Thus, 2D culture yielded a more strongly represented and compositionally purer MSC compartment, while 3D isolation contained rare MSC clusters with niche-associated transcriptional profiles within a heterogeneous marrow environment. These findings support a testable native-associated tendency for 3D, but donor replication, matched recovery and direct native-reference mapping are required before inferring greater in-vivo similarity or functional superiority.