What was tested
Kai Zeng, Lijing Wu and colleagues at the Hospital of Stomatology, Sun Yat-sen University, Guangzhou, followed up their own earlier finding that the X-linked long noncoding RNA FTX suppresses proliferation and multilineage differentiation of human dental pulp stem cells (hDPSCs) by inhibiting OCT4 (DOI 10.1186/s13287-025-04685-9, Stem Cell Research & Therapy, volume 16, article 567, published 15 October 2025, open access). The question in this paper is what sits between FTX and the phenotype. The cells were hDPSCs isolated from caries-free premolars extracted for orthodontic treatment or third molars from donors aged 18 to 25, used at passages 3 to 5. The screen was small RNA sequencing on RNA pooled from 3 patients per group into a single sample per group, which returned 115 differentially expressed miRNAs between FTX-overexpressing and control cells (26 up, 89 down; fold change at least 1.5, p less than 0.05). miR-122-5p was the candidate taken forward. The chain was built with luciferase reporter and mutation assays and chromatin immunoprecipitation: OCT4 transcriptionally activates miR-122-5p, and miR-122-5p directly targets the FOXO3 3-prime untranslated region. Functional readouts were CCK-8 proliferation assays, qRT-PCR, western blots, alizarin red and oil red O staining, all at n = 3. The in vivo arm was a heterotopic dentin-formation assay: 5 times 10 to the 6 hDPSCs per mL on hydroxyapatite/tricalcium phosphate scaffolds, one week in odontoblastic differentiation medium, then subcutaneous implantation into the backs of 6-week-old BALB/c nude mice in four groups (FTX plus mimic, FTX plus mimic negative control, control plus mimic, control plus mimic negative control), harvested at 8 weeks and read out with Masson’s trichrome, DSPP immunofluorescence and FOXO3 immunofluorescence. The work was funded by the National Natural Science Foundation of China (grants 82370943, 81800956, 82100972 and 82071162) and the Guangdong Basic and Applied Basic Research Foundation (2021A1515010562).
What they found
The proposed axis runs FTX to OCT4 to miR-122-5p to FOXO3. FTX overexpression reduced OCT4; reduced OCT4 meant less transcriptional activation of miR-122-5p; and the resulting derepression of FOXO3 (a direct miR-122-5p target) dragged down proliferation and differentiation. The upstream link from FTX to OCT4 rests on this group’s prior paper plus supplementary data here, with lentiviral OCT4 silencing as corroboration rather than a rescue; the two true rescue experiments used a miR-122-5p mimic and FOXO3 siRNA. The paper’s own verbs are hedged: the abstract says the mimic “partially alleviated” FTX’s inhibition of proliferation and of odontoblastic (mineralization) and adipogenic differentiation markers, the FOXO3 siRNA rescue is likewise described as partial, and the results text elsewhere says the mimic restored or attenuated the affected markers. In the mouse assay, the FTX-plus-control-mimic group showed less collagen fiber deposition on Masson’s trichrome than the empty-vector control group, miR-122-5p restoration reversed that deficit, FTX suppressed the odontoblast marker DSPP in vivo with the mimic attenuating the suppression, and FOXO3 protein tracked in the expected direction (up with FTX, down with the mimic). The authors frame the work as two novel regulatory mechanisms, FTX/OCT4/miR-122-5p and FTX/miR-122-5p/FOXO3, that could inform RNA-based therapy and gene editing for dentin regeneration.
What it does not show
Three bounds matter. First, FTX itself is manipulated in only one direction: every experiment overexpresses FTX from a lentiviral construct, and there is no FTX knockdown, so the paper cannot say whether lowering FTX in normal cells would raise potency, which is the direction any therapy would need. Second, the screen that nominated miR-122-5p ran on one pooled RNA sample per group (3 patients pooled into each), so the 115-miRNA list has no biological replication behind it; the downstream validation is solid standard molecular biology but all in vitro readouts are n = 3, and the methods do not state how many mice were used, with the figure caption for the in vivo immunofluorescence reporting n = 3 without saying whether that means mice or image fields. Third, the in vivo model is heterotopic: subcutaneous space in immunodeficient mice, not a pulp chamber, a root canal, or an injured tooth. It shows that FTX-overexpressing hDPSCs deposit less collagen and DSPP in a scaffold on a mouse back, not that tuning this axis regenerates pulp or dentin in a tooth. There is no delivery vehicle, no dosing, no inflamed or necrotic pulp environment, and no functional test of the deposited matrix.
Where it sits in the field
For the pulp and dentin repair program, this is the intracellular regulation layer that sits underneath the cell-therapy, secretome and scaffold papers the pass usually tracks. It names concrete candidate drug targets (a lncRNA, a miRNA, a transcription factor) in the potency circuitry of the exact cell type most regenerative endodontics approaches rely on, and it is one of the few papers in the backlog that carries its own in vivo confirmation, even if that confirmation is heterotopic. It does not move any translational tier: it is one group’s mechanism chain, built on overexpression, with partial rescue effects, in cells from young healthy donors. The current field assessment stands at /field/.
Where we differ from the coverage
We found no press or popular coverage of this paper to differ from. Against the paper’s own framing, we hold back the abstract’s suggestion of “promising RNA-based therapeutic strategies and gene-editing targets.” The data support a contributory regulatory axis in cultured pulp stem cells and a scaffold assay on a mouse back; they do not show that modulating this axis works in a tooth, and the absence of any FTX loss-of-function experiment means the therapeutic direction (lowering FTX) is untested even in a dish. We also note the sequencing screen lacks biological replication, a caveat the paper does not raise.
Provenance: grounded in the open-access full text of record (Stem Cell Research & Therapy 16:567, PMC12522884), retrieved 19 October 2026; every number above is taken from the article’s abstract, methods, results text and figure captions. Method and sourcing standard at /method/.