How teeth work, what the regrowth breakthrough actually showed, what repair can do today, what the future honestly holds — and, in our virtual lab, a reverse-engineered account of how you would grow a tooth back, step by step, with the tests that could kill each route. Every claim is graded from primary sources. No medical advice, ever.
PROVEN established in a peer-reviewed primary source or registry, population stated DOCUMENTED stated by a named primary actor or shown in a primary study, not yet replicated / peer-reviewed at scale UNPROVEN plausible, no adequate evidence yet MISIDENTIFIED a real thing described as a different thing FABRICATED invented, no source in the record
Enamel (the crown's shell, ~96% hydroxyapatite mineral), dentin (the living body), pulp (nerve, blood vessels, stem cells), and cementum — held in jaw bone by the periodontal ligament, a suspensory sling, not a fusion. An implant has none of this; a tooth is an organ, not a peg.
Enamel is built by ameloblasts — and ameloblasts are lost at eruption. Mature enamel has no cells, no blood supply, no repair crew. Lost enamel bulk cannot biologically regrow. Mineral can be put back into an early, still-standing lesion (remineralisation) — that is a different, smaller claim.
Odontoblasts survive for the life of a healthy tooth, adding secondary dentin lifelong and tertiary dentin after injury. The pulp holds dental pulp stem cells (DPSCs) that can form a dentin–pulp-like complex — proven in transplant experiments since Gronthos 2000.
A tooth develops by reciprocal signalling between epithelium and mesenchyme — BMP, Wnt, Shh and FGF, used again and again through bud, cap and bell stages. Root shape comes later, from Hertwig's epithelial sheath. Break the conversation and you get a missing, small, or extra tooth.
Humans are diphyodont: 20 baby teeth, then 32 permanent teeth. Sharks replace teeth continuously; rodents grow incisors forever; we do neither — our replacement program runs once.
A second successional lamina can bud off a developing permanent tooth and normally dies back. In a CT case series, about a third of premolar extra teeth were classified as coming from that rudimentary third dentition (Kiso 2019). That documents a remnant program — not a complete spare set waiting in every jaw. "Everyone has a third set ready" is MISIDENTIFIED.
USAG-1 (gene name SOSTDC1) is one of the body's own brakes on tooth development: it dampens BMP and Wnt signalling, the same signals that decide whether a young tooth bud keeps growing or regresses. Kyoto researcher Katsu Takahashi's group asked the reverse-engineering question first: if the brake helps decide tooth number, what happens if you release it?
| Claim | Grade | What the record shows |
|---|---|---|
| Mice lacking USAG-1 grow extra teeth | PROVEN | Kassai et al., Science 2005 — ectodin/USAG-1 knockout: enlarged enamel knots, extra teeth. The brake is real. |
| A single dose grew a whole tooth | PROVEN — in mice | Murashima-Suginami et al., Science Advances 2021: a BMP-selective anti-USAG-1 antibody, one systemic dose, generated a whole tooth in mice and rescued congenital tooth agenesis across several genetic models. |
| It worked in ferrets too | PROVEN — with conditions | Ferrets are diphyodont like us, which is why they matter. But the animals were young, the dose was 5× concentration, given three times, with immunosuppression — and the result was an extra front tooth. It was not an adult replacing a lost tooth. |
| The drug grew teeth in humans | FABRICATED | No human tooth growth has been reported in any paper, registry posting, or company announcement in the record. It did not happen — yet. |
| Phase 1 human trial (TRG035) | DOCUMENTED | Toregem BioPharma's TRG035, Kyoto University Hospital, from Oct 2024 (registry jRCT2051240154): 30 healthy men, 30–65, each missing at least one molar, single IV dose, randomised, double-blind, placebo-controlled. Primary endpoint: safety only. Toregem states the trial completed with safety confirmed (Nov 2025, FDA pre-IND announcement) — a company statement; peer-reviewed data have not been published. |
| Next trial: children born missing teeth | DOCUMENTED | Phase IIa, n=24 children with severe congenital partial anodontia, dose-finding. Japan's PMDA completed its trial-notification investigation in Aug 2026 per Toregem; hospital ethics review comes next. Still no trial, anywhere in the record, in adults who lost teeth. |
| Available by 2030 | UNPROVEN | 2030 is the developer's stated aim for a first approval, in the congenital indication, if Phase 2 and Phase 3 succeed. It is an aspiration, not a schedule. |
Open risks, on the record: USAG-1 also regulates kidney-protective BMP-7 (Yanagita 2004/2006), so systemic blocking needs kidney monitoring; a non-selective antibody that also disturbs Wnt was part of why the team engineered BMP selectivity — of five early antibodies, three harmed birth/survival in mice; and nothing yet answers control of a new tooth's number, position, shape, root, or eruption path in a person.
| Route | Stage | Grade |
|---|---|---|
| Fluoride remineralisation, fillings, crowns, root canal, implants, dentures | Standard care now | PROVEN — as sealing/replacement. Implants survive ~94.6% at 10+ years across 7,711 implants — but an implant is a prosthetic root: no pulp, no ligament, no enamel. |
| Early-lesion remineralisation boosters (P11-4 peptide, hydroxyapatite) | Small human trials | DOCUMENTED for early white-spot lesions. Rebuilding missing enamel or a crown: UNPROVEN. Selling the first as the second: MISIDENTIFIED. |
| Pulp regeneration with dental pulp stem cells | Human pilots | DOCUMENTED pilot-grade: Nakashima 2017 (n=5, dentin in 3/5, no control); Xuan 2018 (n=40 children, regained sensation and root growth vs standard care). |
| Regenerative endodontics (blood-clot scaffold, immature teeth) | In clinical guidelines | DOCUMENTED — but the tissue formed is often bone/cementum-like, not true pulp. Calling it true pulp regrowth: MISIDENTIFIED. |
| Tideglusib / GSK-3 inhibition for dentin repair | Mice only | DOCUMENTED in mice (~2× dentin, Neves & Sharpe 2017); human claim UNPROVEN. "Ends fillings" is MISIDENTIFIED. |
| Whole bioengineered tooth (organ-germ method) | Mice only | DOCUMENTED in mice: Ikeda/Tsuji, PNAS 2009 — an embryonic-cell tooth germ erupted, occluded, and responded to pain in an adult mouse jaw. In humans: UNPROVEN. The blocker is proven biology: adults have no dental epithelium left to make enamel with — iPSC-derived ameloblasts (2023) are an in-vitro start, not a tooth. |
| 3D-printed living teeth, laser regrowth, cell-free homing in people | Preclinical | UNPROVEN in humans (printed prosthetics themselves are proven — a different claim). |
No honest date exists for a bioengineered human tooth. Anyone giving you one is selling something.
Our lab works on paper and public data only — no wet lab, no patients. Every route below is a filed hypothesis, starts UNPROVEN, and carries the test that would kill it. A tooth is not a peg: a real replacement needs the right crown shape for its position, enamel, dentin, a living pulp fed through a root tip, roots, a ligament sling (fused to bone = failure), eruption into bite, and lifelong maintenance. Every successful tissue-forming system shares a triad: a competent cell, an instructive signal, and a vascularisable, correctly shaped space.
Mechanism: dormant lamina remnant at the site → USAG-1 holds its BMP signal below survival → BMP-selective antibody lifts it → the bud/cap/bell program re-runs → eruption. This is the Toregem route, and the only one in human trials.
Kill tests: no lamina remnant at most adult lost-tooth sites; Phase 2 in congenital children produces no teeth; or teeth form uncontrolled in number, position, or shape.
Mechanism: manufacture epithelium (iPSC-derived is the leading candidate) + mesenchyme (pulp stem cells or iPSC neural crest), reaggregate in a shaped scaffold, mature, transplant into the socket, erupt. The Tsuji mouse proof, rebuilt with adult-sourced cells.
Kill tests: no adult-accessible epithelial source achieves true ameloblast function; or germs make mineralised "tooth-like" structures without ordered enamel and ligament — tooth-like is not a tooth.
Mechanism: a scan-shaped degradable scaffold plus recruitment signals coaxes the patient's own cells to rebuild ligament, bone, and a dentin–pulp complex in place. Its known gap: it cannot make an enamel crown.
Kill test: human work keeps producing bone/ligament only — then this is socket preservation, valuable, but not regrowth.
Mechanism: stack the near-term wins — GSK-3 dentin repair over a living pulp, guided remineralisation of early lesions, stem-cell pulp regeneration in a cleaned canal.
Kill tests: no human dentin gain in a controlled trial; pulp histology keeps coming back bone-like instead of pulp.
Full graded research files (Tranches 1–3) and the lab register are held in the project workspace and will be linked here as the site is stood up.