What was tested
GH12 is a synthetic 12-amino acid cationic, amphipathic peptide (sequence GLLWHLLHHLLH) designed de novo against cariogenic bacteria by Wang and colleagues in 2017 (Archives of Oral Biology, DOI 10.1016/j.archoralbio.2017.03.017); it inhibits eight major cariogenic bacteria at 4 to 8 µg/mL minimum inhibitory concentration and was seen by electron microscopy to lyse bacterial membranes. Follow-up work reported activity against endodontic biofilms and Enterococcus faecalis when used as a root canal irrigant (Li et al. 2020, International Endodontic Journal, DOI 10.1111/iej.13296). The open question for regenerative endodontics is whether a peptide that perforates membranes is safe for the pulp cells a scaffold is meant to host. Reiniche, Faure, Leveque, Orelle, Ziverec, Farges, Virard, Malcor, Bekhouche, and Ducret, from Universite Claude Bernard Lyon 1, the CNRS, and the Hospices Civils de Lyon, address that compatibility question in Journal of Dentistry (volume 163, article 106105, published 11 September 2025). The study belongs to the pulp-dentin repair program (/programs/pulp-dentin-repair/): everything in it is in vitro.
A predicted pore, seen in liposomes
The authors modeled GH12 assemblies with AlphaFold 3 and Boltz-2, testing 2 to 7 monomers; both tools converged on trimeric and tetrameric arrangements as the most stable. The authors themselves flag the AlphaFold confidence scores as low (pTM and ipTM below 0.5) and treat the structures as indicative rather than proof. In liposomes built from E. coli lipid extract and loaded with self-quenched carboxyfluorescein, 40 µg/mL GH12 produced a time-dependent dye signal within about 10 minutes, consistent with rapid membrane permeabilization, with the caveat that GH12 also quenches the dye directly, which the authors note complicates the readout (same source, Methods and Results).
On plastic, cells take the hit
Human dental pulp mesenchymal stem cells were isolated from extracted third molars of donors aged 12 to 17 and treated at 0, 10, 25, or 50 µg/mL for 2, 4, or 6 hours, with three donors per assay. Viability fell significantly only at 50 µg/mL, and modestly: at 6 hours, 99.17 percent of cells were alive in control versus 86.70 percent with GH12. Lysosomal staining, read as a cellular stress response, rose significantly at 50 µg/mL at all three time points (by 22.2, 35.8, and 24.3 percent at 2, 4, and 6 hours) and at 25 µg/mL at 4 and 6 hours, with the strongest signal in small, rounded cells likely to be dying.
In fibrin, no detectable harm
The same cell type, embedded at 22,000 cells per µL in a fibrin hydrogel (3 mg/mL fibrinogen, 4 U/mL thrombin) with 50 µg/mL GH12, showed 92 percent viability after 48 hours, not significantly different from the GH12-free control (p = 0.18). A supplementary agar-diffusion assay found the peptide retained antibacterial activity inside the gel, producing inhibition zones on Staphylococcus aureus and E. faecalis lawns with a clindamycin-loaded gel as positive control. The authors credit interactions between the peptide and the fibrin matrix, which can bind and partially retain it, acting as a local reservoir and limiting free concentration around cells.
Why the route should care
Residual bacteria are a leading reason regenerative endodontic procedures produce fibrosis or mineralized scar instead of pulp, so an antimicrobial built into the scaffold addresses a real failure mode for the route. The numbers sketch a working window: antibacterial MICs of 4 to 8 µg/mL against cariogenic bacteria sit well below the 50 µg/mL that stressed cells in open culture, and the fibrin matrix lowered effective exposure further. The contrast also underlines how harsh 2D plastic is as a pulp stand-in: the authors note that 256 µg/mL was reported as non-cytotoxic to human gingival fibroblasts after short exposure, four times the concentration that stressed pulp cells here.
What it does not show
One cell type, three donors per assay, short exposure windows, no endothelial or neuronal cells, and no migration, differentiation, or mineralization readouts, in a single fibrin formulation. The structural models are moderate-confidence predictions, and the liposome assay is partly confounded by dye quenching. This is an in vitro biocompatibility screen, not a pulp-capping model and not an animal study; whether a 92 percent viability readout at 48 hours translates into intact pulp regeneration is exactly the work this paper does not do.
Where we differ from the coverage
We found no press or popular coverage of this paper to differ from. The paper’s own clinical-relevance statement, which frames the work as paving the way toward clinical translation, outruns a three-donor in vitro study. The defensible result is narrower: a pore-forming peptide that is measurably harsh to pulp stem cells in open culture, and compatible with the same cells when delivered from a fibrin hydrogel at the tested concentration.
Provenance: grounded in the full author manuscript of Reiniche et al. (2025), Journal of Dentistry 163:106105, archived openly under CC BY at HAL (hal-05343278) and read in full; background citations checked against their own records (DOI 10.1016/j.archoralbio.2017.03.017, DOI 10.1111/iej.13296). Method and sourcing standard at /method/.