Scientists Grow Human Jawbone Organoids from Stem Cells in 3D Culture
Kyoto researchers created jawbone-like organoids from human iPSCs, recapitulating mandibular development and modeling fragile bone disease.
Summary
Researchers at Kyoto University developed a method to grow jawbone-like organoids from human induced pluripotent stem cells (iPSCs) using a stepwise 3D culture system. By mimicking the embryonic sequence — from neural crest cells through mandibular ectomesenchyme — they produced organoids containing osteoblasts, network-forming osteocytes, and self-mineralized bone matrix. The organoids promoted bone repair when transplanted into jaw defects in animal models, and replicated bone fragility seen in osteogenesis imperfecta patients using disease-specific iPSCs. This is the first system to faithfully recapitulate jaw-specific bone development in a 3D human tissue model, opening doors for studying jaw diseases, testing therapies, and eventually rebuilding bone lost to infection, trauma, or cancer.
Detailed Summary
Jawbones are among the most clinically vulnerable skeletal structures in the body, subject to irreversible damage from infections, trauma, tumors, and congenital defects. Yet no faithful human jawbone model existed for studying the biology or testing treatments — until now. Researchers at Kyoto University's Center for iPS Cell Research and Application (CiRA) developed a complete 3D protocol to generate jawbone-like organoids from human iPSCs, published in Nature Biomedical Engineering. The work addresses a fundamental challenge: jawbones arise not from mesoderm like most bones, but from cranial neural crest cell (NCC)-derived ectomesenchyme in the first pharyngeal arch (PA1), requiring highly specific developmental signals to recapitulate.
The protocol begins by aggregating dissociated iPSCs in 96-well ultra-low-adhesion plates with ROCK inhibitor Y-27632, then sequentially treating aggregates with BMP4 (10 ng/ml, 1 day) followed by TGF-β inhibitor SB431542 and GSK3β inhibitor CHIR99021 under shaking conditions. This reliably generated HOX-negative, SOX10+CD271+TFAP2A+ neural crest cells with 94.9 ± 1.9% CD271-high efficiency by day 5, confirmed across six independent iPSC lines. Critically, these NCCs lacked HOX gene expression (HOXA1, HOXA2, HOXB2, HOXA3), matching the identity of midbrain–anterior hindbrain NCCs that naturally give rise to PA1 ectomesenchyme in the embryo.
To drive NCCs toward mandibular ectomesenchyme (mdEM) identity, the team tested combinations of FGF8, endothelin-1 (EDN1), and BMP4 — signals normally provided by pharyngeal arch epithelium. The combined FEDB regimen (FGF8 + EDN1 + BMP4) most effectively downregulated the NCC marker SOX10 while upregulating mdEM markers including DLX1, DLX2, DLX5, DLX3, GSC, HAND2, TWIST1, and PRRX1. Remarkably, the 3D aggregates developed a proximal-to-distal patterning gradient from center outward — mirroring in vivo mandibular development — with DLX2 expressed broadly and HAND2 restricted to outer (distal) regions. Addition of pharyngeal epithelial signals further induced mandibular prominence-specific regional patterning, including Runx2 and Sp7 expression indicative of osteoprogenitor commitment.
Under osteogenic culture conditions, the mdEM aggregates formed jawbone-like organoids with histologically confirmed osteoblasts secreting type-I collagen-rich extracellular matrix, osteocytes embedded within self-produced mineralized bone matrix, and importantly, three-dimensional dendritic osteocyte networks — a feature extremely difficult to recapitulate in 2D culture. Calcium deposition and mineralization were confirmed by Alizarin Red staining. When transplanted into jawbone defect models, the organoids promoted bone regeneration, demonstrating in vivo functional capacity.
The versatility of the platform was further demonstrated using patient-derived iPSCs from an osteogenesis imperfecta (OI) patient carrying a COL1A1/COL1A2 mutation. OI organoids recapitulated disease phenotypes including reduced and abnormally structured collagen matrix and impaired mineralization. Mutation-corrected iPSC lines partially rescued these phenotypes, validating the model for disease research and drug screening. The authors used xeno-free induction conditions throughout, which is a meaningful step toward eventual clinical translation. Limitations include the absence of vascularization and osteoclasts in the organoids, and the fact that in vivo transplantation experiments were conducted in immunocompromised animal models rather than humans.
Key Findings
- 3D induction achieved 94.9 ± 1.9% CD271-high HOX-negative neural crest cell efficiency from human iPSCs by day 5, confirmed across 6 independent iPSC lines
- FEDB signal combination (FGF8 + EDN1 + BMP4) reliably induced mandibular ectomesenchyme markers DLX5, DLX3, GSC, and HAND2 while suppressing NCC marker SOX10
- mdEM aggregates displayed center-to-periphery proximal-distal patterning mirroring embryonic mandibular development, with DLX2 broadly expressed and HAND2 restricted to outer/distal regions
- Jawbone-like organoids contained self-mineralized bone matrix with embedded osteoblasts and 3D dendritic osteocyte networks — a feature not achievable in previous 2D models
- Transplantation of organoids into jawbone defect animal models promoted bone regeneration in vivo
- OI patient-derived iPSC organoids recapitulated disease phenotypes including defective collagen matrix and impaired mineralization, partially rescued by mutation-corrected iPSC lines
- Protocol was executed entirely under xeno-free conditions, a key requirement for future clinical translation
Methodology
Human iPSCs (1231A3 and five additional lines including OI patient-derived) were aggregated in 96-well ultra-low-adhesion plates and subjected to sequential 3D induction using BMP4, SB431542, CHIR99021, and FEDB (FGF8 + EDN1 + BMP4) signals over multiple weeks. Characterization used flow cytometry, immunofluorescence, qPCR, RNA sequencing, and histology (Alizarin Red, type-I collagen staining). In vivo bone regeneration was tested by transplanting organoids into jawbone defect models in immunocompromised animals. Statistical comparisons used one-way ANOVA with Tukey's multiple comparisons test, with data expressed as mean ± s.d. from multiple biologically independent experiments.
Study Limitations
The organoids lack vascularization and osteoclasts, limiting their ability to model bone remodeling and homeostasis fully. In vivo transplantation experiments used immunocompromised animal models, which may not reflect the immune environment of human patients. The authors also note that complete recapitulation of the full mandibular ossification program — including endochondral ossification of proximal and distal ends — was not achieved in this iteration.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.