Grem2 marks a root-follicle stem cell that builds alveolar bone and periodontal ligament in mice

What they tested

Zhang, Yang, Bian, Xiong, Li, and Dong at Shanghai Jiao Tong University asked whether Grem2, a BMP antagonist linked to human tooth agenesis and taurodontism, labels a specific dental follicle stem cell (DFSC) subset and what that subset contributes to tooth root and alveolar bone development. The work sits in the bioengineered tooth germ program, because DFSC heterogeneity and lineage tracing are central to any protocol that tries to rebuild a tooth and its supporting periodontium. The study combines reanalysis of published single-cell RNA-seq datasets, a new Grem2-CreERT2 knock-in mouse line, and a conventional Grem2 knockout mouse model.

A distinct root-follicle progenitor

Using reanalyzed scRNA-seq data from mouse molar development (GSE189381) and FACS-sorted Gli1+ and PTHrP+ datasets (GSE237439, GSE313714, GSE120108), the authors found that Grem2 expression becomes restricted to the lateral, or “root,” follicle during postnatal stages (Zhang et al. 2026, Results). In these datasets, Grem2+ cells formed a subpopulation that was partially overlapping but transcriptionally distinct from PTHrP+ and Lepr+ dental follicle cells. Differential expression analysis highlighted Id2, Igfbp3, and H19 in Grem2+ cells, and gene-ontology terms pointed toward ossification, extracellular matrix organization, and osteoblast differentiation (Zhang et al. 2026, Results).

Lineage tracing with Grem2-CreERT2; Rosa26-tdTomato mice, induced at postnatal day 3.5 (PN3.5), showed tdTomato+ descendants in the periodontal ligament (PDL), interradicular alveolar bone, and cementoblasts. At PN28, after root formation was complete, Grem2+ descendants accounted for 59.09% ± 13.58% of PDL cells and 36.29% ± 7.69% of interradicular alveolar bone (Zhang et al. 2026, Results). Immunostaining showed overlap with Runx2, Osterix, osteocalcin, and periostin, consistent with differentiation into bone, cementum, and PDL lineages.

Stem-cell properties and persistence

Cultured Grem2+ cells harvested at PN5.5 made up approximately 80% of the total DFSC population and could be expanded clonally. They expressed CD90/Thy1, CD44, and CD29, were negative for CD34, Sca1, and CD31, and showed a CD44+CD31- flow-cytometry profile (Zhang et al. 2026, Results). Under chondrogenic, osteogenic, and adipogenic conditions, the cells expressed aggrecan, Runx2, and Perilipin-1, respectively, meeting the standard trilineage differentiation criterion for mesenchymal stem cells.

The cells also persisted beyond development. Tamoxifen-induced labeling at PN5 or PN14 produced tdTomato+ descendants at PN21 around root apices and in periodontal tissues. Labeling at PN30 to PN35 produced detectable signals at PN35 in the root apex and alveolar bone. In a six-week-old injury model, a defect in the cervical interradicular alveolar bone recruited more tdTomato+ cells to the repair site over three weeks, indicating an adult repair response (Zhang et al. 2026, Results).

What happens when Grem2 is removed

Conventional Grem2 knockout mice, generated by CRISPR-Cas9 deletion of exon 2, showed delayed root development and reduced interradicular alveolar bone volume at PN15, with the phenotype becoming more pronounced by PN90. By PN90 the knockout molars had shorter roots, increased bone density under the furcation, an enlarged pulp chamber, and a higher crown-to-root length ratio, resembling but more severe than the phenotype seen after Grem2+ cell ablation with Grem2-CreERT2; Rosa26-DTA mice (Zhang et al. 2026, Results). The interseptal alveolar bone between the first and second molars was less affected.

Mechanistically, Grem2 knockout reduced both osteogenic activity (lower ALP, Runx2, and Sp7) and osteoclastic activity (weaker TRAP staining), suggesting lower bone turnover. EdU label-retaining experiments showed fewer quiescent stem cells in Grem2-knockout mice at PN12.5, and EdU+ proliferating cells at PN15 were confined to the most apical interradicular bone rather than distributed across the cervical portion (Zhang et al. 2026, Results). CellChat analysis of the P3.5 scRNA-seq dataset identified IGF signaling as enriched in the root follicle. IGF2 expression, which colocalized with Grem2+ descendants at the root furcation in wild-type mice, was absent in the interradicular alveolar bone of Grem2-knockout mice. Conditional deletion of Igf2 in Grem2+ cells (Grem2-CreERT2; Igf2flox/-) partially recapitulated the molar phenotype, with shortened mesial roots and reduced furcation alveolar bone volume (Zhang et al. 2026, Results).

What it means for tooth germ engineering

The finding gives the bioengineered tooth germ program a more specific DFSC marker than the broader PTHrP+ or Lepr+ populations: Grem2+ cells sit in the root follicle, contribute to both cervical and basal interradicular bone, and remain active into adulthood. If the same population exists in human dental follicle tissue, it could become a target for building periodontal support around a bioengineered tooth. The BMP-antagonist role of Grem2 also adds to the evidence that local BMP gradients, not just absolute BMP levels, control whether DFSCs maintain stemness or differentiate.

What it does not show

This is a mouse developmental study. It does not demonstrate a human Grem2+ DFSC population, nor does it report isolation or differentiation of human cells. The lineage-tracing and knockout experiments were performed in conventional and conditional mouse models, and the injury model was a surgical defect in alveolar bone, not regeneration of an entire tooth-support apparatus. The IGF2 mechanism is supported by conditional genetics but has not been tested pharmacologically or in a repair outcome. Finally, because the paper is a preprint under review at International Journal of Oral Science, it has not yet completed peer review.

Provenance: grounded in Zhang Q, Yang B, Bian J, Xiong Z, Li Y, Dong C. Identification of Grem2+ Dental Follicle Stem Cells in Alveolar Bone Homeostasis and Molar Morphogenesis. Research Square. 2026. doi:10.21203/rs.3.rs-10692088/v1, verified against the full preprint text and metadata. Method and sourcing standard at /method/.