What was built
Zhang, Hu, Xu, and colleagues at Sichuan University developed a chemically defined, serum-free medium for reconstituting mouse tooth germs in vitro (Zhang et al. 2024, Advanced Science, Experimental Section). They used a dual-fluorescence reporter mouse, Pitx2-copGFP and Msx1-tdTomato, to track epithelial and mesenchymal cells from embryonic day E13.5 through postnatal day PN1. Dispersed dental epithelial and mesenchymal cells were aggregated, coated in collagen, and cultured with 2% Matrigel in a medium based on Glasgow’s MEM, Knock-Out Serum Replacement, and small-molecule supplements. The reconstituted organoids, which the authors call toothoids, formed epithelial foci surrounded by dental mesenchyme and could be maintained in vitro for weeks.
After 10 days in vitro, the toothoids were transplanted under the kidney capsule for 2 weeks. They developed tooth-like structures containing dental pulp, odontoblasts, pre-dentin/dentin, enamel, and Periostin-positive periodontal-ligament-like tissue, as shown by hematoxylin and eosin staining and markers including Nestin, amelogenin, ameloblastin, Osterix/Sp7, Periostin, and CD34 (Zhang et al. 2024, Results; Figure 1F,G). E14.5 molar cells produced the highest number of tooth-like structures per toothoid, while E12.5 dissociated cells could not self-organize and PN1 cells lost the capability (Zhang et al. 2024, Results; Figure 1H,I). Incisor cells also reconstituted toothoids, but with fewer teeth per organoid and larger individual structures than molars (Zhang et al. 2024, Supporting Information).
What the model revealed
The toothoids recapitulated developmental timing. Sequential RT-qPCR and immunostaining showed induction of stage-specific markers such as Hand1/2, Fgf8, C1qtnf3, Krt13, Dmp1, Amelx, Enam, and Mmp20, and the formation of enamel knots marked by Fgf4 and LEF1 (Zhang et al. 2024, Results; Figure 2D-F). They also responded to chemical perturbations of TGF-beta and Wnt signaling in ways consistent with prior genetic studies: SB431542 and XAV939 reduced tooth formation and increased epidermalization, while CHIR99021 produced multiple dentin-pulp and enamel-like structures (Zhang et al. 2024, Results; Figure 2H,I).
The key mechanistic finding was that reconstitution did not reset the developmental clock. When cells were isolated from E16.5 or E17.5 tooth germs, the resulting toothoids expressed markers appropriate to those later stages and failed to re-express earlier markers such as Hand1/2, Fgf8, or Sdc1 in the papilla region. E14.5 toothoids likewise skipped markers from before E14.5 (Zhang et al. 2024, Results; Figure 3). This means a reconstituted tooth germ resumes development from the stage at which its cells were harvested, not from an embryonic default.
For the earliest stage, isolated E10.5 first branchial arches lost tooth-inductive potential in the defined medium within 10 days. Adding Activin A together with the Hedgehog/Smoothened agonist SAG preserved the ability to form tooth-like structures, whereas combinations that replaced SAG with the Wnt activator CHIR99021 did not (Zhang et al. 2024, Results; Figure 4B,C).
Finally, enamel deposition required an external push. In extended in vitro culture, toothoids formed ameloblasts and odontoblasts but only a thin pre-dentin layer. Moving them to an air-liquid interface did not trigger enamel; adding fetal bovine serum did, and the active ingredient appeared to be a serum growth factor rather than albumin. Inhibition of BMP signaling with LDN193189 or K02288 blocked enamel formation, while the BMP activator SJ000291942 induced enamel under serum-free conditions without changing ameloblast number or amelogenin/ameloblastin expression (Zhang et al. 2024, Results; Figure 5B-I). Scanning electron microscopy showed enamel rods and interrod structures resembling postnatal mouse molar enamel (Zhang et al. 2024, Results; Figure 5D,H).
What it means for the third dentition routes
For the bioengineered tooth germ program, the paper moves the field beyond the classic fetal-bovine-serum reconstitution recipe. A defined medium makes it possible to ask which factors are necessary and sufficient at each step, which is a prerequisite for scaling the approach to human cells or induced pluripotent stem cells. The finding that reconstitution preserves the original developmental stage also has practical consequences: it suggests that starting cell stage must be matched to the desired developmental window, rather than assuming any embryonic dental cells can be reset.
The enamel result is also relevant to the epithelial organoids program. Prior dental epithelial organoids produced ameloblast-like cells that expressed matrix proteins but did not deposit organized enamel. Zhang et al. show that BMP activation can trigger mineralized enamel deposition in a full toothoid context, which frames enamel induction as a signaling problem rather than a lineage-specification problem.
What it does not show
The work is in mouse embryonic cells. It does not demonstrate human toothoid formation, derivation from pluripotent stem cells, or transplantation into a functional jaw. Root development and eruption were not achieved in vitro; toothoids required kidney-capsule transplantation to mature. The defined medium still contains 2% Matrigel, a complex basement-membrane extract, so the system is not fully synthetic. And while enamel formed under BMP activation, the long-term hardness, thickness, and integration of that enamel remain uncharacterized.
Where we differ from the coverage
We found no press coverage of this paper. The result should not be read as “teeth grown in a dish” or as a near-term alternative to implants. It is a mouse developmental-biology tool that clarifies culture requirements for reconstituted tooth germs.
Provenance: grounded in the open-access full text of Zhang et al. 2024, Advanced Science 12(3):e2404345, DOI 10.1002/advs.202404345. Method and sourcing standard at /method/.