What the study is

Mu, Lu, and colleagues at the Hospital of Stomatology, Sun Yat-sen University, published this work in Stem Cell Research & Therapy (volume 17, 2026, article 10.1186/s13287-026-05078-2, published 2026-06-04, open access). The group has a running program on extracellular vesicles from stem cells of human exfoliated deciduous teeth (SHED-EVs), including a 2024 paper in which SHED-EVs in a GelMA scaffold drove odontogenic differentiation of pulp stem cells for pulp regeneration. Here they ask a different question: can the same vesicles, given cell-free, control the inflammatory microenvironment that disables dental pulp stem cells (DPSCs) during pulpitis? Vital pulp therapy depends on DPSCs staying alive and reparative while the pulp is inflamed, so the readouts are DPSC function in an LPS-inflamed dish and a short LPS pulpitis model in rats. It sits in the pulp-dentin repair program.

What the vesicles did in inflamed pulp cells

SHED came from non-carious exfoliated deciduous teeth of 12 donors aged 6 to 8, and DPSCs from caries-free premolars and third molars of 12 adults aged 18 to 24; cells at passages 3 to 5 carried standard MSC markers (CD29, CD44, CD73, CD90, CD105 up to 99.93 percent, CD34 and CD45 below 0.14 percent). Vesicles isolated by differential ultracentrifugation were cup-shaped, 50 to 200 nm with a main NTA peak near 121.9 nm, and positive for CD63 and TSG101; DiI-labeled vesicles were taken up by both healthy and inflamed DPSCs. In dose-finding, 10 µg/mL was the minimum effective concentration and effects plateaued above it.

The core experiments used DPSCs inflamed with 1 µg/mL LPS for 24 hours, then 48 hours with or without 10 µg/mL SHED-EVs. Vesicles raised proliferation in both healthy and inflamed cells (in healthy DPSCs the EdU-positive fraction was about 15 percent versus under 10 percent in controls, p less than 0.0001; inflamed cells also gained proliferation, p less than 0.05), but the migration benefit appeared only in healthy DPSCs, not in LPS-inflamed ones. On inflammation, qRT-PCR showed IL-1β, IL-6, and TNF-α down and IL-10 and TGF-β up after vesicle treatment; ELISA put the IL-6 rise from LPS at up to 10-fold and the vesicle effect at roughly a 2.5-fold reduction relative to inflamed cells; Western blot confirmed lower IL-1β, IL-6, and TNF-α protein. Apoptosis measured by flow cytometry rose with LPS and was attenuated by vesicles (p less than 0.01), with Bcl-2 up and cleaved caspase-3 down. The mitochondrial readouts moved the same way: mitochondrial ROS accumulation and fragmentation fell, and the JC-1 red/green ratio, a membrane-potential proxy that drops on depolarization, was partially restored (p less than 0.0001 for the ROS and membrane-potential comparisons).

The pathway claim and its limit

RNA sequencing of inflamed DPSCs with and without vesicles (three replicates per group, fold change beyond 1.5-fold in either direction, p and FDR below 0.05) found 296 genes down and only 5 up, enriched in KEGG terms for apoptosis, FoxO, and NF-κB signaling, with the NLRP3 inflammasome flagged in the cellular-component analysis. Western blots matched: FoxO1, which LPS had suppressed, was restored by vesicle treatment; phosphorylated p65 and NLRP3 fell; total p65 did not change. The authors frame the story as vesicle-mediated FoxO1 activation plus NF-κB/NLRP3 suppression. That is an association built on transcriptomics and protein levels. There is no inhibitor or epistasis experiment here showing that blocking FoxO1 or NF-κB abolishes the vesicle effect, and no identification of which vesicle cargo does the work, both of which the authors list as open questions. Treat the mechanism as a pointer, not a demonstration.

Three days in a rat molar

Twelve male Sprague-Dawley rats (8 weeks, 250 to 280 g) had the pulp of the maxillary first molar exposed under anesthesia; the pulpitis group received 2 µL of LPS at 10 mg/mL into the chamber and a PBS-soaked gelatin sponge, while the treatment group received the same LPS followed by a sponge carrying 3 µL of SHED-EVs at 1 µg/µL. All teeth were capped with iRoot BP Plus and sealed; three days later the animals were euthanized, with n = 4 per group and blinded analysis. H&E showed dense inflammatory infiltration and dilated vessels in the pulpitis group and only slight infiltration with mild capillary dilation after vesicles. Immunohistochemistry on the pulp showed lower IL-1β and IL-6 (p less than 0.05) and higher FoxO1 fluorescence (p less than 0.01) in the vesicle group, mirroring the dish results. The total in vivo dose came to 3 µg per tooth, well below the per-mL minimum effective concentration in dishes, a gap the paper does not discuss. This is an inflammation-dampening readout at 72 hours: no dentin bridge, no reparative dentin, no survival or function endpoint, and the model is a single LPS insult, not caries with live bacteria.

Two caveats

Two details qualify the tidy story. First, secreted IL-1β protein by ELISA showed no significant difference between the LPS-inflamed cells and untreated controls, even though IL-1β mRNA and the EV-treated comparisons did shift; the authors invoke mesenchymal-cell negative feedback on IL-1β secretion, and the anti-inflammatory case here rests more on IL-6 and TNF-α than on secreted IL-1β. Second, inflamed DPSCs did not gain migration from the vesicles, which matters if the clinical idea is recruiting cells into an injured site; the authors attribute the flat result to an already-activated baseline state, but it is a measured null, not a success.

Boundary and what it changes

Everything is LPS inflammation: human cells in dishes and one 72-hour rat model, n = 4 per group, with no long-term repair endpoint and no natural infectious model. The mechanism is associative, the cargo is unidentified, and vesicle donor variability is not addressed. For the pulp-dentin repair program, the piece of evidence is that a cell-free biologic can, at least acutely, damp an LPS-inflamed pulp environment in rats, and in dishes keep inflamed pulp stem cells alive with intact mitochondria. It sits alongside this site’s recent coverage of the mitochondrial axis in pulp repair, where glutathione primed PINK1/Parkin mitophagy to steer dentin repair in a mouse injury model (2026-09-05), and with earlier work linking pulp stiffness in deep caries to DPSC exosome release (2026-08-19). Together these lines suggest mitochondrial health is becoming a shared therapeutic target in pulpitis, with vesicles as one delivery idea among several. That is a research-stage observation, not a clinical claim, and the field assessment at /field/ stands.

Provenance: primary source read in full at the open-access journal page (Stem Cell Research & Therapy 17, 2026, published 2026-06-04); every number above was checked against the article text. See /method/ for the site’s evidence standards.