What the study is

Zhang, Cao, Li, and colleagues report a preprint (posted 2026-05-01, doi:10.64898/2026.04.29.721551, not peer reviewed) identifying a Foxp4-positive skeletal stem cell population in the mandible, which they call mdSSCs. The core claim has two parts. First, lineage tracing in mice shows these cells build Meckel’s cartilage, mandibular bone and molar tooth germ mesenchyme during development, and keep contributing to bone, pulp and periodontal tissues in adults. Second, when purified by flow cytometry and transplanted on top of nasal sinus epithelium, these cells form ectopic teeth, including teeth assembled from adult human jaw bone cells in immunodeficient mice. No dental epithelial cells are transplanted; the epithelial partner is the host nasal sinus epithelium, which the authors argue is functionally similar to odontogenic epithelium.

What the mouse work showed

Using a Foxp4-CreERT2 knock-in crossed to a Rosa26-tdTomato reporter, the group administered tamoxifen at E11.5 and found labeled cells condensing at E12.5 into both Meckel’s cartilage and molar tooth germ mesenchyme; by E19.5 they had produced Col2a1-positive chondrocytes, Sp7-positive osteoprogenitors and osteoblasts, odontoblasts and dental follicle. Very few labeled cells appeared in the maxilla at E12.5, and maxillary contribution was much lower than mandibular. Single-cell RNA sequencing of sorted Foxp4-lineage cells from E12.5 mandibles (12,584 cells from 5 embryos, 7 clusters) placed the mdSSC cluster at 29% of cells, expressing the neural crest marker Meis2, sitting at the origin of Monocle3 chondrogenic, osteogenic and odontogenic trajectories, and acting as the strongest signaling hub in CellChat analysis. Postnatally, labeled cells appeared in endosteal, periosteal, pulp and periodontal compartments. In a distraction osteogenesis fracture model, they formed Col2a1-positive chondrocytes in the callus at 10 days post injury and Sp7-positive osteoprogenitors and osteocytes from 17 to 45 days. Genetic ablation via Rosa26-DTA (tamoxifen at E15.5) produced shorter mandibles and impaired molar tooth germ formation at E19.5 without a significant maxillary change (n=6 per genotype), and in adults ablation before surgery lowered callus bone volume ratio at 24 days post injury.

The transplantation result

The team prospectively isolated mouse mdSSCs as CD45-, Ter119-, CD31-, Thy1-, 6C3-, CD200+, CD105- cells, splitting tdTomato-positive from tdTomato-negative fractions, and confirmed higher colony-forming activity and trilineage differentiation in vitro. Under the renal capsule, embryonic mdSSCs formed bone and cartilage in 4 weeks while control cells formed fibrous tissue (n=10 mice per group, 5 independent experiments). The tooth result comes from a nasal sinus transplantation model: mdSSCs placed on the nasal sinus epithelium formed ectopic teeth with multiple cusps within 4 weeks, with pulp, ameloblasts and odontoblasts producing enamel and dentin, resembling native mouse tooth histology. E15.5 to E16.5 mdSSCs also formed a cusped crown with pulp at 4 weeks and roots with a visible apex at 8 weeks, and adult 12-week-old mouse mdSSCs formed multi-cusp teeth (quantification n=16 mice per group, 5 independent experiments).

For the human arm, the authors reanalyzed a published single-cell dataset of mandibulectomy bone fragments (GSE289393; 9,327 cells, 6 clusters) and highlighted a CADM1-positive, PDPN-positive cluster enriched for FOXP4, SOX9, GLI3 and TWIST1 regulons. From crushed mandible fragments of patients aged 24 to 29 (gating workflow from a 34-year-old donor), they sorted CD45-, CD31-, CD235a-, CADM1+, PDPN+ cells, confirmed FOXP4 enrichment by qPCR, and showed higher colony formation and trilineage differentiation. After expansion and sphere culture, these cells were xenotransplanted into NCG immunodeficient mice. Under the renal capsule they formed bone and cartilage while controls formed fibrous tissue (n=10 or 16 recipient mice per group, 5 independent experiments). In the nasal sinus they formed an ectopic human tooth at 4 weeks, donor origin confirmed by human vimentin staining (quantification n=10 or 15 recipient mice per group, 5 independent experiments). At 8 weeks the regenerated tooth showed an elongated root with a closed apex and a clear crown-root boundary, which the authors compare morphologically to a 14-year-old human premolar. Knocking down FOXP4 in these cells abolished in vivo tooth regeneration (n=10 recipient mice per group, 3 independent experiments).

The primary-cilia mechanism arm

Gene ontology and KEGG analyses of the mdSSC clusters flagged response to mechanical stimulus at both embryonic and adult stages, cilium assembly and organization in adults, and calcium, Wnt and Hedgehog signaling. In embryonic bulk RNA sequencing, Ift140, an intraflagellar transport gene, was the most significantly upregulated in mdSSCs versus control cells. Ift140 silencing in vitro downregulated Wnt and Hedgehog target genes and impaired colony formation and trilineage differentiation. Conditional deletion of Ift140 in Foxp4-positive cells (Foxp4-CreER; Ift140 flox/flox) phenocopied ablation: mandibular but not maxillary shortening, delayed molar tooth germ development, fewer primary cilia and Foxp4-positive cells, and reduced Col2a1-positive chondrocytes and Runx2-positive osteoprogenitors at E19.5 (n=6 per genotype); adult knockout mice had lower callus bone volume ratio at 24 days post injury, and transplanting sphere-cultured mdSSCs into the fracture site accelerated repair in an Ift140-dependent manner. Foxp4 knockdown itself lowered Hedgehog targets Gli1 and Gli2 while modestly raising Wnt targets Dkk1 and Axin2, so the regulatory logic between Foxp4 and cilia-associated signaling is not a simple linear chain.

Boundary of the result

Every tooth in this paper is ectopic, grown in the nasal sinus of an immunodeficient mouse. None erupted into a socket, occluded, or was tested for innervation or function. The epithelial partner is host nasal sinus epithelium reprogrammed in situ, so the experiment does not show that these cells can build a tooth with oral epithelium in a human jaw, which is the clinically interesting configuration. The text never states what fraction of transplants actually formed a tooth, so formation efficiency is unknown. The human material came from a small number of mandibulectomy patients in their 20s and 30s; recipient mouse numbers are reported, but donor counts are not broken out, and donor heterogeneity is unaddressed. The resemblance of the 8-week construct to a 14-year-old premolar is a descriptive morphological comparison, not a functional match. The authors themselves list the critical gap: in situ (orthotopic) tooth regeneration in the native mandibular microenvironment has not been achieved, and formation efficiency and tooth shape remain unoptimized. And all of this sits in a preprint that has not yet passed peer review.

What it changes for the program

For the bioengineered tooth germ program, the interesting departure from the 2009 Ikeda result and the 2024 chemically defined reconstitution work is the sourcing: the program record says human cell sourcing and developmental control remain the major gaps, and this paper claims a prospectively isolated, phenotypically defined mesenchymal population from adult human jaw bone that recruits host epithelium to build a tooth-organized structure, persisting from embryo to adulthood in mice. If it replicates, that is a new candidate cell source for reconstitution protocols. It does not move the tier: it is one lab, mouse recipients, an ectopic site, and an unreviewed preprint, with no independent replication and no orthotopic test. The program’s standing bar, independent tooth formation in a large animal by a group with no commercial stake, is untouched.

Provenance: grounded in the full text of Zhang L, Cao D, Li X, Yu H, Wang J, Zhu Q, Zhang X, Chen C, Li G, Xu X, Xu X, Tao D, Gong X, Niu P, Wu X, Yu M, Yue R, Sun Y. Foxp4+ Mandibular Skeletal Stem Cells Orchestrate Bone/Tooth Development and Regeneration. Preprint, posted 2026-05-01. doi:10.64898/2026.04.29.721551, retrieved in full and read. Method and sourcing standard at /method/.