What the review argues
Xiaoqing Li, Nan Jiang and four colleagues have published a comparative synthesis of how vertebrates replace teeth, built around one organizing claim: the difference between a shark that renews teeth all its life and a human who gets one replacement set is not whether a successional dental lamina (SDL) exists, but how permissive its local control is (Fundamental Research, 2026). They call this the SDL rheostat model. The review is a synthesis of published work across sharks, cichlids, snakes, geckos, mice, ferrets and miniature pigs, plus human clinical series; it reports no new experiments of its own.
The suppressed structure
The review defines a rudimentary SDL (RSDL), an epithelial offshoot that forms and then regresses without producing a successor tooth. In mice, the incisor RSDL arises around embryonic day 13, peaks by E15, and is gone by postnatal day 5 through apoptosis; the molar RSDL appears as a lingual thickening of the first molar at E16.5 and regresses by PN12, driven more by failed proliferation than by cell death. In diphyodont mammals the picture is more dynamic. In miniature pigs, the authors describe a mechanosensitive switch: the developing deciduous tooth compresses the intervening dental follicle progenitor cells at 3 to 20 kPa, and that compression runs through an integrin beta1-ERK1-RUNX2 axis that suppresses epithelial Wnt signaling and holds the SDL quiescent; eruption relieves the stress and releases the brake. Regression itself is attributed to a combination of epithelial-mesenchymal transition, localized apoptosis, reduced proliferation and cell migration out of the lamina, with non-canonical Wnt receptor Fzd6 marking regressing regions in the pig.
What revives it
The same structure can be pushed back into tooth formation, and this is where the piece connects to the anti-USAG-1 / TRG-035 program. The review restates the originator work: USAG-1 is a BMP antagonist that also modulates Wnt signaling; in USAG-1 knockout mice the incisor RSDL survives instead of dying and supernumerary teeth form, and systemic anti-USAG-1 antibody in diphyodont ferrets produced functional third-generation incisors in a dose-dependent manner. It also catalogues related revivals in mice: genetically stabilized beta-catenin in Sox2-positive dental epithelium generates supernumerary teeth through repeated enamel knots, transient Shh inhibition reactivates the RSDL through FGF-Wnt crosstalk, and Osr2-deficient mice form lingual supernumerary teeth via expanded Bmp4 signaling, an effect that requires Msx1 and disappears in double mutants. Removing the neighboring molar germ lets an isolated mouse RSDL form tooth-like structures in vitro, which the authors read as evidence that the predecessor tooth itself is part of the inhibitory circuit.
The human claim and its boundary
For humans, the review leans on two lines of argument and states both cautiously. Morphologically, it cites Ooe’s classic embryological proposal of a third-dentition anlage, lingual to the permanent germs, reportedly visible in all examined specimens and gone by permanent tooth eruption. Clinically, it cites a CBCT-based retrospective study that put the prevalence of late-developing supernumerary teeth (LDST) at 0.14 percent, found about three quarters of LDSTs met the clinical criteria for a third-dentition origin, and measured a developmental delay of 6.48 to 10.45 years behind the adjacent permanent teeth; a separate 78-case series put the fraction meeting third-dentition criteria at one third, strongest for multiple and premolar-region LDST. The authors are explicit that no molecular or lineage-tracing evidence currently demonstrates the existence, persistence or odontogenic competence of a third dentition in humans, and they frame it as a hypothesis for explaining LDST, not an established structure. Their translational proposal is a protect-and-activate sequence: transient inhibitors of regulated cell death to preserve the progenitor pool, followed by pathway activation, with large-animal models and safety work first, precisely because Wnt, Shh, BMP and USAG-1 are pleiotropic and oncogenic risk is unresolved.
Where we differ from the coverage
Coverage of tooth-regeneration drugs tends to present the human third dentition as found biology waiting for a trigger. This review, which is broadly favorable to that research direction, does not: it calls the human structure hypothetical, notes that revitalization evidence is predominantly morphological rather than functional, and states that efficacy and safety in humans remain unverified. That is the honest current boundary of the field record as well: one antibody program in Phase I, and the strongest animal result, third-generation incisors in ferrets, is the same preclinical lineage the program page already tracks.
Provenance: grounded in the full text of the open-access review (CC-BY, Fundamental Research, DOI 10.1016/j.fmre.2026.09.014, published online September 2026); every number above was checked against that text, and the anti-USAG-1 ferret result traces to the Murashima-Suginami et al. 2021 lineage already on the program page. Method and sourcing standard at /method/.