What they built
Tao et al. dispersed lithium-calcium-silicate (Li2Ca2Si2O7, LCS) particles, synthesized by a sol-gel route, milled, and sieved, with a measured particle size below 50 micrometers, into gelatin methacryloyl (GelMA) with a lithium phenyl phosphinate photoinitiator. The mix is injectable through a 25-gauge needle and crosslinks under blue light in 5 to 40 seconds, a fast set compared with the hours-long curing of conventional calcium-silicate caps such as MTA or iRoot BP (Tao et al. 2025, Discussion). The composite mineralizes on its own: after one to three days in simulated oral fluid, the LCS-loaded groups show hydroxyapatite by X-ray diffraction, while plain GelMA does not. The dose matters. Higher LCS load raises the pH of the soak solution (7.40 for GelMA, 9.87 for 15% LCS at 21 days) and slows degradation, so the working formulation settled on 5% LCS, which performed best across the cell assays, between inactive plain GelMA and inhibitory 15% loadings (Tao et al. 2025, Results and Discussion).
In vitro: Schwann cells and pulp cells move together
The group tested the hydrogels against rat Schwann cells and human dental pulp stem cells (DPSCs). At 5% LCS, Schwann cells migrated more and expressed more neurotrophic and myelination-associated genes (GDNF, PMP22, NGF, BDNF) and proteins (S100, GDNF); at 15% LCS that effect was no longer seen (Tao et al. 2025, Results). The same 5% formulation slightly promoted DPSC proliferation, and promoted migration and odontogenic differentiation, with dentin-matrix proteins DMP-1 and DSPP and mineral nodules all elevated. For pulp cells the higher dose went further and hurt: 15% LCS cut proliferation and migration below plain GelMA. To test whether the neural side feeds the odontogenic side, the authors conditioned medium on Schwann cells cultured with the hydrogels and applied it to DPSCs: medium from the GelMA-5LCS group outperformed controls for DPSC migration and for DMP-1, DSPP, and OPN expression (Tao et al. 2025, Results). This is an indirect, conditioned-medium readout of neuro-odontogenic coupling, not a direct demonstration, and the factors responsible were not identified.
In vivo: rat molar defects, six weeks
For the animal test, the team drilled 0.8 mm dentin-pulp defects into the molars of male SD rats (8 weeks old), filled them in situ with hydrogel, crosslinked with blue light for 40 seconds, and sealed the cavity; after six weeks they harvested the jaws (Tao et al. 2025, Materials and Methods). Five groups were compared: normal teeth, blank defects, GelMA, GelMA-5LCS, and iRoot BP, a commercial calcium-silicate repair cap used as the benchmark. By micro-CT, the GelMA-5LCS group showed higher bone-volume fraction, mineral density, and trabecular thickness in the defect than blank and GelMA groups, similar to the iRoot BP benchmark (n = 4). Hematoxylin and eosin staining showed pulp-like tissue in the GelMA-5LCS defects that looked closer to normal than in the other repair groups, and dentin markers DMP-1 and DSPP were elevated, with no significant difference from iRoot BP (n = 5). The distinguishing result is neural: neurofilament staining showed more nerve-fiber ingrowth into the defect area of the GelMA-5LCS group than into blank or GelMA defects (n = 5). The comparison does not include the iRoot BP benchmark, and normal pulp had the densest fibers.
Why innervation is the point
Most pulp-repair biomaterials are scored on hard-tissue endpoints: dentin-bridge area, mineral density, occluded versus open defects. Nerves rarely get measured, even though the one controlled human result in this program, the Xuan et al. autologous pulp-stem-cell trial (Science Translational Medicine, 2018; see the pulp-dentin-repair program record), regenerated pulp that contained both vessels and nerves. Tao et al. make innervation a headline readout and show a material-only route to it: no cells are transplanted, just a lithium-, calcium-, and silicon-releasing ceramic in a gelatin gel. Their mechanistic suggestion is that lithium and silicon drive Schwann-cell neurotrophic output (they cite lithium’s effect on beta-catenin signaling), which then supports pulp-cell differentiation. Readers of this program will recognize the echo of the tideglusib result in the program record (Neves et al., 2017): tideglusib targets GSK-3, which lithium is reported elsewhere to inhibit, though this paper does not test either pathway directly.
What it does not show
The study is preclinical and short. Six weeks in a rat molar says nothing about long-term survival, eruption, or function of the repaired tooth. The innervation result is anatomical: neurofilament density, not a functional recovery of sensation or dentin sensitivity. The micro-CT mineralized tissue in a drilled defect is not proven to be organized tubular dentin, and no mechanical testing of the regenerated tissue was done. The defects are sterile drill injuries, so the antibacterial and anti-inflammatory performance that decides real pulp-capping outcomes was not tested; the authors list this themselves. Whether this material outperforms commercial caps on innervation is simply unknown: no benchmark group was stained for neurofilament. The Schwann-cell to DPSC coupling was shown only with conditioned medium, and the specific signaling molecules remain unidentified. Sample sizes are small (n = 4 for micro-CT, n = 5 for histology). Finally, the dose window is a practical constraint: at 10 to 15% LCS the soak pH climbs above 9 and pulp-cell behavior suffers, so the material’s margin between bioactive and caustic is not wide. The study shows a workable formulation and a fresh endpoint in one animal model, not a clinically validated material.
Provenance: grounded in the open-access full text of Tao X, Zhang H, Mei P, Huang J, Fang B, Huan Z, Wu C. An injectable bioceramics-containing composite hydrogel promoting innervation for pulp-dentin complex repair. Int J Oral Sci. 2025;17:66. doi:10.1038/s41368-025-00398-0, retrieved from PMC12485191 and read in full. Method and sourcing standard at /method/.