Turning Glioblastoma Stem Cells Into Dendritic-Like Cells to Train the Immune System
Mouse study: a gel-delivered lipid nanoparticle reprograms glioblastoma stem cells into antigen-presenting dendritic-like cells to spark antitumor immunity.

Summary
Glioblastoma resists immunotherapy partly because its cells, especially stem-like ones, present few tumor antigens to the immune system. Researchers designed a lipid nanoparticle carrying a plasmid for three transcription factors (PU.1, IRF8, BATF3) plus the T-cell costimulatory ligand TNFSF9. Embedded in an injectable hydrogel for the post-surgical cavity, it was meant to convert glioblastoma stem cells into type 1 conventional dendritic cell (cDC1)-like cells. In cell culture, treated cells took on dendritic-cell markers, secreted inflammatory cytokines, took up and processed antigen, and activated T cells. In orthotopic mouse models, the particles preferentially transfected stem-like tumor cells and produced early signs of reprogramming. The abstract reports tumor suppression, immune memory and prevention of postoperative relapse in mice. This is early preclinical work.
Detailed Summary
Glioblastoma (GBM) is the most aggressive primary brain tumor and has largely escaped the benefits of modern immunotherapy. One reason is poor antigen presentation: tumor cells, particularly the stem-like cells (GSCs) that drive growth and recurrence, are hypoimmunogenic, and professional antigen-presenting cells such as cDC1s are scarce or dysfunctional in the tumor. This study asks whether the tumor's own stem cells can be turned into antigen-presenting cells inside the brain.
The team built a plasmid, pPIBT, encoding the transcription factors PU.1, IRF8 and BATF3 (which can push cells toward a cDC1 fate) plus TNFSF9 (4-1BBL), a costimulatory ligand meant to sustain CD8+ T-cell function and limit exhaustion. They screened 64 lipid nanoparticle (LNP) formulations built on new nucleobase-derived ionizable lipids and chose a lead (TNO12). It was modified with a GSC-targeting peptide (CP), giving particles of about 142 nm with about 91% plasmid encapsulation and stability over 21 days. The particles were loaded into an injectable thymopentin-chitosan/alginate hydrogel that gels in about 3 minutes at 37 °C. The gel passes through a 25-gauge needle, degrades over roughly 14 days, and released about 86% of its LNPs in that time in artificial cerebrospinal fluid.
In vitro, treated mouse GSCs acquired cDC1-like features. Nine days after treatment, 32.27% of transfected cells were reprogrammed, comparable to lentiviral delivery. Among the reprogrammed cells, about 68% expressed CLEC9A and about 61% expressed CD11c. RNA sequencing showed upregulation of antigen-presentation, cDC1 identity (including Xcr1) and inflammatory genes, but only a partial cDC1 signature rather than full transcriptional convergence. The cells raised CD80, CD86, CD40 and MHC-I, secreted IL-12p70, IL-6, TNF-α and CXCL10, engulfed apoptotic cells, took up and processed ovalbumin, and activated and expanded OT-I CD8+ T cells. Adding TNFSF9 was justified because PIB reprogramming alone did not raise TNFSF9.
In orthotopic GL261 glioma mice, the particles transfected GSCs (about 37% of EGFP-positive cells) far more than DCs (about 3%) or macrophages (about 9%). Nine days post-treatment, treated tumors had more CD45+MHC-II+ cells among CD133+ cells. The abstract reports that treatment inhibited tumor growth, produced long-term antitumor memory and prevented postoperative relapse in mice, with experiments also in SB28 tumors and with OVA-tracking of antigen-specific T cells in lymph nodes. The supplied text cuts off before the detailed in vivo efficacy, survival and memory data, so those figures cannot be verified here.
Caveats are substantial. All evidence is from mouse cells and mouse models. Reprogramming was incomplete, with mixed intermediate populations, and the long-term safety of converting tumor cells into immune-like cells is unknown. Plasmid delivery was transient and not fully GSC-specific, and human translation has not been tested.
Key Findings
- Nucleobase-derived LNP carrying PU.1, IRF8, BATF3 and TNFSF9 reprogrammed mouse GSCs into cDC1-like cells at about 32% efficiency among transfected cells.
- Reprogrammed GSCs raised CD80/CD86/CD40, secreted IL-12p70 and CXCL10, processed antigen, and activated and expanded naïve OT-I CD8+ T cells in vitro.
- In GL261 brain tumors, the targeted particles transfected GSCs (~37%) much more than DCs (~3%) or macrophages (~9%).
- An injectable chitosan/alginate hydrogel released about 86% of LNPs over 14 days, designed to fill the postoperative cavity.
- Reprogramming was only partial: transcriptomes gained some cDC1 features without full convergence, and PIB alone did not raise TNFSF9.
Methodology
Preclinical study combining in vitro work with mouse GSCs (flow cytometry, RNA-seq, ELISA, antigen uptake and OT-I T-cell priming assays) and in vivo testing in orthotopic GL261 and SB28 glioma mouse models. A 64-LNP library was screened, and the lead nucleobase-lipid LNP was combined with a thymopentin-chitosan/alginate hydrogel. Group sizes were small (n=3–6 mice), and the supplied text ends before the full in vivo efficacy results.
Study Limitations
All data are from mouse cells and murine models, which poorly capture the heterogeneity, immunosuppression and recurrence patterns of human GBM. Reprogramming was incomplete and heterogeneous, targeting was not exclusive to GSCs, and the safety of creating tumor-derived antigen-presenting cells is unaddressed. The in vivo efficacy and memory data could not be reviewed in detail from the supplied text.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe:
