Plant-Derived Nanoparticles From a Medicinal Herb Trigger Ferroptosis in Liver Cancer Models

Scientists have isolated nanoscale vesicle-like particles from Selaginella doederleinii, a medicinal herb long used in traditional Asian medicine, and shown that they suppress hepatocellular carcinoma cells in laboratory dishes and shrink tumors in mice. The particles, dubbed SD-EVLP, appear to kill cancer cells by inducing ferroptosis, an iron-dependent form of regulated cell death, through suppression of the xCT/GPX4 antioxidant defense axis. The study, published in Cancer Nanotechnology, also reports that treatment was associated with widespread metabolic reprogramming in tumor tissue and shifts in the gut microbiota of tumor-bearing mice, although the authors caution that these microbial and metabolic findings remain associative.

The research team, led by Jingting Zhang, Tianhao Tong and colleagues with senior author Puhua Zeng, isolated the particles from fresh plant material using ultracentrifugation followed by sucrose density gradient purification. Nanoparticle tracking analysis revealed a mean hydrodynamic diameter of 147.4 nanometers with a single, uniform size distribution, while transmission electron microscopy showed spherical structures with the typical bilayered membrane characteristic of extracellular vesicles. Coomassie staining demonstrated reproducible protein band patterns, and a BCA assay measured a total protein concentration of 400 micrograms per milliliter. When the particles were labeled with the fluorescent dye DiI and added to cultures, confocal microscopy confirmed that they were efficiently internalized by both human HepG2 and murine Hepa1-6 liver cancer cells, with red fluorescent signals accumulating in the cytoplasm. A control experiment in which cells were exposed to free DiI dye alone produced negligible intracellular fluorescence, ruling out the possibility that the signal came from unbound dye.

Functional testing showed that the particles suppressed multiple malignant behaviors in both cell lines. In CCK-8 viability assays, concentrations of 25 micrograms per milliliter or higher significantly reduced cell viability in a dose-dependent manner after 48 hours, with the inhibitory effect plateauing above 100 micrograms per milliliter. Wound healing assays revealed significantly delayed wound closure at medium and high doses, though the authors note that this assay reflects combined effects of migration and proliferation rather than motility alone. EdU incorporation assays showed marked reductions in DNA synthesis at medium and high doses, and colony formation assays demonstrated significant suppression of long-term clonogenic capacity. Transwell Matrigel invasion assays similarly showed fewer invading cells after treatment. In several assays, the high-dose particle treatment performed comparably to sorafenib, the approved multikinase inhibitor used as a positive control, although the magnitude of inhibition varied between the two cell lines, suggesting that responses may differ among hepatocellular carcinoma models.

To test whether these effects translated to living animals, the researchers established subcutaneous xenografts by injecting HepG2 cells into four-week-old male BALB/c nude mice. All 30 inoculated mice developed tumors within ten days, and the animals were randomly assigned to five groups of six: untreated control, three doses of SD-EVLP, and sorafenib. The particle doses were derived from the clinically recommended adult dose of the crude herb using body surface area conversion, yielding 30.75, 61.5 and 123 micrograms per gram for the low, medium and high groups, all administered by oral gavage once daily for 14 days. Tumor growth was significantly inhibited in the high-dose and sorafenib groups compared with controls, while the low and medium doses showed only non-significant decreasing trends in tumor volume. Tumor weight, however, was significantly reduced at all three particle doses in a dose-dependent pattern, with no statistically significant difference between the high-dose group and sorafenib. No significant differences in body weight were observed among groups, and histological examination of the liver, heart, lungs and kidneys revealed no apparent abnormalities, suggesting acceptable tolerability within the tested dose range. One mouse in the control group and one in the medium-dose group died during the intervention period.

Histopathology of the tumors themselves told a consistent story. Tumors from the high-dose group showed reduced cell density, loosened stromal architecture and a marked decrease in mitotic figures, while medium-dose tumors displayed focal necrotic areas and chromatin condensation. Immunohistochemical staining for Ki-67, a marker of proliferative activity, was significantly reduced in the medium-dose, high-dose and sorafenib groups, although the high-dose particle group still showed higher Ki-67 positivity than the sorafenib group, indicating that the antiproliferative magnitude of the two interventions differed.

The mechanistic core of the study centers on ferroptosis, a form of regulated cell death distinguished by iron-dependent lipid peroxidation and collapse of antioxidant defenses. Transmission electron microscopy of tumor tissue revealed that mitochondria in treated groups displayed progressive, dose-dependent structural abnormalities, including membrane separation, cristae disruption, swelling, vacuolization and matrix dissolution, features consistent with ferroptosis-associated ultrastructural damage. Biochemical assays of tumor tissue showed that glutathione levels were significantly depleted at medium and high doses, malondialdehyde, a lipid peroxidation product, was significantly elevated at those doses, and ferrous iron accumulated dose-dependently across all treatment groups. Immunohistochemistry and Western blotting both confirmed that expression of xCT, the key subunit of the cystine/glutamate antiporter system Xc-, and its downstream effector GPX4, glutathione peroxidase 4, were significantly downregulated in the high-dose group, with no significant difference between the high-dose particles and sorafenib.

Parallel experiments in cultured cells reinforced these findings. The FerroOrange probe revealed significant dose-dependent increases in intracellular ferrous iron in both cell lines, the C11-BODIPY 581/591 probe showed significant lipid peroxidation at medium and high doses, and JC-1 staining indicated mitochondrial membrane depolarization after treatment. Western blotting confirmed dose-dependent downregulation of xCT and GPX4 in both cell lines. Critically, rescue experiments using ferrostatin-1, a ferroptosis inhibitor, partially reversed these changes: co-treatment significantly attenuated the iron accumulation, reduced the lipid peroxidation ratio in HepG2 cells, and partially restored GPX4 and xCT expression. The authors acknowledge, however, that ferrostatin-1 rescue of cell viability or cell death was not directly assessed, that genetic perturbation of ferroptosis regulators was not performed, and that apoptosis and necroptosis were not systematically excluded, so ferroptosis is supported as a contributor but not established as the sole mode of cell death.

Two additional layers of analysis broadened the picture. Untargeted metabolomics of tumor tissue from the control and high-dose groups identified 684 differential metabolites, 416 downregulated and 268 upregulated, enriched in arginine and proline metabolism, branched-chain amino acid biosynthesis, pyrimidine metabolism and the tricarboxylic acid cycle. Several metabolites feeding the glutamate cycle and glutathione synthesis, including L-proline, trans-4-hydroxyproline and L-glutamic acid itself, were significantly reduced, a pattern the authors interpret as depleting antioxidant precursors and creating a metabolic state permissive to ferroptosis. Meanwhile, 16S rRNA gene sequencing of intestinal contents showed no significant differences in alpha diversity indices, but compositional differences at the phylum and genus levels, with higher Firmicutes and lower Bacteroidota in the treated group, and LEfSe analysis identified several control-enriched taxa. The authors emphasize that because the supervised PLS-DA separation requires corroboration and the metabolomic and microbiota findings are associative, targeted flux analysis, microbiota-transfer experiments and independent validation cohorts will be needed to establish functional links.

Network pharmacology, integrating mass spectrometry identification of 25 bioactive components with target prediction and differential expression analysis of the public GSE84402 dataset, yielded 118 candidate genes, with hub genes including PPARG, ESR1, HSP90AB1, CYP3A4 and RHOA, and enrichment in oxidative stress, lipid metabolism and iron-binding functions. The study has important limitations beyond those already noted: the selectivity and immunogenicity of the particles were not tested in non-malignant hepatic or immune cells, and efficacy was evaluated only in a subcutaneous xenograft model in immunodeficient mice, which does not recapitulate the hepatic or immune microenvironment of human liver cancer. The authors position SD-EVLP not as a ready therapy but as a candidate plant-derived nanovesicle platform warranting further preclinical investigation, with orthotopic and immunocompetent models as the logical next step.

Subject of Research: Plant-derived extracellular vesicle-like nanoparticles that induce ferroptosis in hepatocellular carcinoma models

Article Title: Selaginella doederleinii–derived extracellular vesicle-like particles induce ferroptosis via the xCT/GPX4 axis and are associated with metabolic and gut microbiota alterations in hepatocellular carcinoma

Article References: Zhang, J., Tong, T., Chen, H., Wang, Z., Wu, L., Tang, J., Peng, W., Yang, R., & Zeng, P. (2026). Selaginella doederleinii–derived extracellular vesicle-like particles induce ferroptosis via the xCT/GPX4 axis and are associated with metabolic and gut microbiota alterations in hepatocellular carcinoma. Cancer Nanotechnology, 17(1), Article 56. https://doi.org/10.1186/s12645-026-00424-8

Image Credits: AI Generated

DOI: 10.1186/s12645-026-00424-8

Keywords: Selaginella doederleinii, extracellular vesicles, ferroptosis, hepatocellular carcinoma, xCT/GPX4 axis, nanomedicine, metabolomics, gut microbiota, xenograft model, lipid peroxidation, preclinical study, Cancer Nanotechnology