What was tested
Zhang and Yelick at Tufts University School of Dental Medicine asked whether decellularized porcine tooth bud extracellular matrix (dTB-ECM) scaffolds, reseeded with dental cells plus endothelial cells, could form tooth-like tissues after implantation in adult Yucatan minipigs (Zhang and Yelick 2025, Abstract). The scaffolds came from porcine second-molar tooth buds. Each construct paired a decellularized enamel organ half, injected with 0.5 × 10^6 porcine tooth-bud-derived dental epithelial cells and 0.5 × 10^6 human umbilical vein endothelial cells (HUVECs), with a decellularized pulp organ half, injected with 1 × 10^6 human dental pulp cells and 1 × 10^6 HUVECs. The halves were sutured together and cultured for one week in a perfusion bioreactor before implantation into fresh extraction sockets in the mandibles of 2-year-old minipigs (Zhang and Yelick 2025, Materials and Methods). The study compared three groups: cell-seeded Recell-dTB, acellular dTB, and freshly isolated natural tooth bud (nTB) controls, with eight constructs per group at each of two harvest times, 2 and 4 months.
What the imaging and histology showed
Micro-CT showed mineralized tooth-like structures in 5/8 (62.5%) Recell-dTB constructs at 2 months and 4/8 (50%) at 4 months. Across the same two harvest points, only 2/8 (25%) acellular dTB constructs and 1/8 (12.5%) natural tooth bud controls formed recognizable tooth-like structures (Zhang and Yelick 2025, Results, Supplementary Figure S1). Histology of demineralized sections revealed dentin-like and cementum-like mineralized tissues, and periodontal ligament-like tissue with Sharpey’s fibers oriented perpendicular to the cementum surface in Recell-dTB implants. DSPP immunostaining, a marker for odontoblasts and dentin, was robust in Recell-dTB-derived tissues at 2 months but not detectable at 4 months (Zhang and Yelick 2025, Results, Figure 7). Human dental pulp cells were still detectable in Recell-dTB implants at both harvest times using a human mitochondria antibody.
What the cell-seeding comparison means
The higher success rate in the Recell-dTB group than in the acellular scaffold or natural tooth bud groups suggests that the added cells, and possibly the one-week bioreactor preconditioning, matter more than the scaffold alone. In 1-week bioreactor cultures before implantation, the constructs contained roughly 40% E-cadherin-positive porcine dental epithelial cells, 45% vimentin-positive human dental pulp cells, and 15-20% Factor VIII-positive HUVECs, with HUVECs significantly less abundant than the other two cell types (P ≤ .05) (Zhang and Yelick 2025, Results, Figure 3). The authors also note that the constructs expressed fibrillin 1 and 2 and showed higher collagen IV and laminin than acellular scaffolds.
Why it matters for the field
For the bioengineered tooth germ program, this is a preclinical scaffold study rather than a clinical protocol. It is the most advanced large-animal report from the Yelick lab’s decellularized-tooth-bud line: the group switched from young, growing minipigs, where successive natural tooth eruptions dislocated implants, to mature 2-year-old hosts with fully formed dentition, and the rate of observable tooth-like structures rose from less than one-third in the earlier pilot to roughly half to two-thirds here (Zhang and Yelick 2025, Discussion). The demonstration of periodontal ligament-like tissue with Sharpey’s fibers is notable because a functional periodontal ligament is missing from titanium implants.
What it does not show
The study is in minipigs, not humans, and the bioengineered structures were smaller than natural porcine teeth. It used a xenogeneic mix of human dental pulp cells and HUVECs with porcine dental epithelial cells and a porcine scaffold, so human translation would need a different cell source and immunogenicity assessment. The authors report no eruption, occlusion, long-term function, innervation, or vascular physiology. The 4-month loss of DSPP signal is unexplained; it may reflect antibody behavior in more mature mineralized tissue, but it also means the study does not show stable, progressive dentin maturation. In addition, the natural tooth bud positive control formed recognizable structures in only 1 of 8 implants, fewer than the engineered constructs, and the authors do not explain that discrepancy; the comparison should be treated as descriptive, not as controlled evidence that the construct outperforms natural development.
Where we differ from the coverage
A February 2025 Decisions in Dentistry summary described the work as a “significant step toward developing biological tooth replacements” and said the bioengineered teeth “may better integrate with the surrounding oral environment, promoting more natural function and longevity” (Pratt Machado 2025). The peer-reviewed paper does not report functional integration, longevity, or human applicability. It reports tooth-like mineralized tissues in roughly half to two-thirds of implants in a minipig model, with constructs smaller than natural teeth and no functional periodontal or occlusal testing. The distance from this result to a biological implant alternative remains large and unspecified.
Provenance: grounded in the open-access full text of Zhang and Yelick 2025, Stem Cells Translational Medicine 14:szae076 (PMID 39729491, PMCID PMC11878782), DOI 10.1093/stcltm/szae076. Method and sourcing standard at /method/.