Polysaccharides from Huaier induce autophagy-dependent ferroptosis to inhibit breast cancer stem cells in triple-negative breast cancer

Abstract

Triple-negative breast cancer (TNBC) is a highly aggressive subtype with poor prognosis and limited treatment options, largely due to the presence of breast cancer stem cells (BCSCs) that contribute to chemotherapy resistance, metastasis, and relapse. In this study, we investigated the therapeutic potential of polysaccharides from Huaier (PS-T) in targeting BCSCs via a novel mechanism involving autophagy-dependent ferroptosis. We first identified a 17-gene stemness signature from TCGA data of 167 TNBC patients, which correlated with poor prognosis. Using this signature, we developed a prognostic model that demonstrated strong predictive power for TNBC survival. In vitro, PS-T treatment significantly reduced the stemness markers POU5F1, SOX2, and NANOG, and decreased the ALDH⁺ and CD44highCD24low BCSC populations in a dose-dependent manner. PS-T also suppressed colony formation and mammosphere growth in ALDH⁺ TNBC cells. Ferroptosis induction was confirmed by increased intracellular reactive oxygen species (ROS) and lipid peroxidation, with these effects reversed by ferroptosis inhibitors. Mechanistically, PS-T downregulated GPX4, a key regulator of ferroptosis, through an autophagy-dependent pathway. This process was further enhanced by PS-T-induced upregulation of acid sphingomyelinase (ASM), which facilitated autophagic degradation of GPX4. In vivo, PS-T treatment significantly reduced tumor formation from ALDH⁺ TNBC cells. Overall, our findings suggest that PS-T inhibits BCSC stemness and promotes ferroptosis, providing a promising therapeutic strategy for improving the prognosis of TNBC patients.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-025-04102-4.

Introduction

Breast cancer is the leading cancer threat to the health of Chinese women, with triple-negative breast cancer (TNBC) being one of its most aggressive and challenging subtypes. TNBC is characterized by the absence of estrogen, progesterone, and human epidermal growth factor receptor 2 (HER-2) receptors, making it unresponsive to conventional endocrine or targeted therapies. This lack of targeted treatment options contributes to the higher rates of local recurrence and distant metastasis in TNBC patients [1].

A significant contributor to the aggressive nature of TNBC is the presence of breast cancer stem cells (BCSCs), which are responsible for tumorigenesis, self-renewal, and metastasis. These cells are notoriously resistant to chemotherapy, and although chemotherapy can eliminate the bulk of tumor cells, it often enriches the population of BCSCs, thereby increasing the likelihood of recurrence and metastasis in TNBC patients [2, 3]. Consequently, developing novel strategies to specifically target BCSCs has become a primary focus in the treatment of TNBC [4, 5].

Ferroptosis, a form of iron-dependent programmed cell death characterized by massive reactive oxygen species (ROS) accumulation and lipid peroxidation, has recently garnered attention as a potential therapeutic target for cancer treatment [6, 7]. Su et al. conducted a comprehensive multi-omics analysis identifying ferroptosis as significantly activated in breast tumors exhibiting high intratumor heterogeneity (IITH), suggesting that ferroptosis may be a predisposing factor in these cases and that its inhibition could offer promising therapeutic options for refractory breast cancer [8]. Additionally, iron is found to be present at elevated levels in BCSCs, and its metabolism plays a crucial role in maintaining their stem-like characteristics. Promoting ferroptosis in these cells has been shown to suppress stemness in various cancers, including breast, colorectal, and lung cancer [9–11]. Recent studies also highlight the regulatory role of autophagy in ferroptosis, including mechanisms such as the degradation of iron storage proteins, modulation of lipid peroxidation, and the degradation of glutathione peroxidase 4 (GPX4) [12, 13], suggesting that autophagy may serve as a key factor in regulating ferroptosis and BCSC stemness.

Recent investigations have also uncovered the anticancer properties of several natural medicinal fungi and their extracts, which offer promising benefits in preventing tumorigenesis, recurrence, and metastasis. Huaier, derived from the polysaccharide of Trametes robiniophila Murr (PS-T), has shown potential as an effective adjuvant therapy for malignancies, including breast cancer. In addition to improving the quality of life for cancer patients, Huaier is associated with minimal side effects and a high safety profile [14, 15]. A prior clinical study demonstrated that Huaier effectively combats the metastasis and recurrence of TNBC through the activation of autophagy [14, 16]. Building on these findings, we hypothesize that PS-T may suppress BCSCs by regulating autophagy-related ferroptosis in TNBC.

In this study, we provide the first evidence that PS-T promotes ferroptosis in BCSCs of TNBC, and identify the underlying mechanism by which PS-T induces GPX4 degradation via an autophagic-lysosomal pathway mediated by ASM. These findings present a novel mechanism for how PS-T may inhibit metastasis and recurrence in TNBC, while also offering a theoretical foundation and experimental support for identifying new molecular targets and therapeutic strategies for TNBC treatment through the promotion of ferroptosis.

Materials and methods

Chemicals and reagents

PS-T was isolated from the crude extract of Huaier, which was provided by Qidong Gaitianli Pharmaceutical Co., Ltd. (Qidong, China). The polysaccharides were prepared as previously described, including processes such as isolation, dialysis, extraction, lyophilization, elution, and purification. It is a homogeneous, neutral, water-soluble polysaccharide purified through a multi-step process involving DEAE-52 cellulose ion-exchange chromatography and gel-filtration chromatography on Sephadex G-100 and Sepharose CL-6B. The average molecular weight of PS-T is approximately 2.5 × 10⁴ Da, as determined by high-performance gel-permeation chromatography (HPGPC), which also revealed a single, sharp, and symmetric peak, confirming its high homogeneity. The total carbohydrate content is 96.5%, as measured by the phenol-sulfuric acid method. α-Tocopherol phosphate (HY-16686, 100 µM) and Erastin (HY-15763, 10 µM) were obtained from MedChemExpress (USA). Liproxstatin-1 (T2376, 200 µM), 3-methyladenine (T1879, 5 mM), and ASM-IN-1 (T74778, 2 µM) were purchased from TargetMol (USA). Z-VAD-FMK (S7023, 30 µM), Necrostatin-1 (S8037, 20 µM), and LDC7559 (S9622, 5 µM) were acquired from Selleck Chemicals (USA).

Anti-human CD24 (PE, 311106, 5 µL/test) and anti-mouse/human CD44 (FITC, 103022, 5 µL/test) antibodies were obtained from BioLegend (USA). BODIPY™ 581/591 C11 (D3861, 5 µM/test), GPX4 (PA5-102521), and ASM antibodies (PA5-77047) were sourced from Thermo Fisher (USA). The following antibodies were purchased from Cell Signaling Technology (USA): GPX4 (52455), SLC7A11 (98051), FTH1 (3998), NRF2 (12721), DMT1 (15083), NCOA4 (66849), and β-actin (4970).

Cell culture

The MDA-MB-231, HCC1806, and HCC1937 cell lines were purchased from the FuHeng Cell Centre (Shanghai, China). MDA-MB-231 cells were maintained in L-15 medium (LA9510; Solarbio, Beijing, China) at 37 °C in 100% air, while HCC1806 and HCC1937 cells were cultured in RPMI-1640 medium (C11875500B; Gibco, USA) at 37 °C in 5% CO₂, supplemented with 10% fetal bovine serum (10091148; Gibco, New Zealand) and 1% L-glutamine (C0222; Beyotime, Shanghai, China). Sorted ALDH + cells were maintained in 3D GRO Sphere medium under standard culture conditions and digested at a density of 90%.

Transfection and gene overexpression

The ASM (Acid Sphingomyelinase) gene was overexpressed in the MDA-MB-231 cell line. Primers were designed using the PrimerBank website (https://pga.mgh.harvard.edu/primerbank/): Forward primer: 5’-CTGTCTGACTCTCGGGTTCTC-3’; Reverse primer: 5’-CTATGCGATGTAACCTGGCAG-3’. The ASM gene was amplified by PCR, digested with restriction enzymes, and inserted into the expression vector using ligase. The ligation product was transformed into Escherichia coli, and positive clones were selected via antibiotic resistance. MDA-MB-231 cells were seeded in 6-well plates to achieve 70–80% confluence before transfection. Lipofectamine 3000 was used as the lipid-based transfection reagent. After 48 h, appropriate antibiotics were added to select for cells stably expressing ASM. Single-cell clones were obtained through limiting dilution and expanded for further culture.

mRFP-GFP-LC3 adenovirus infection

5 × 10⁴ MDA-MB-231 cells, either overexpressing ASM or not, were seeded in 24-well plates and treated with or without 10 µg/mL PS-T for 24 h. Once the cells reached approximately 70%−80% confluence, they were transfected with the mRFP-GFP-LC3 adenoviral vector. A negative control group was established. RFP and GFP puncta in the cells were observed and quantified using laser confocal microscopy to assess autophagic activity.

Flow cytometry analysis and cell sorting

Cells were exposed to distinct drugs for 24 h after adherence. The cells were adjusted to 1 × 10⁶ cells/mL using assay buffer and incubated with CD24 and CD44 antibodies at 37 °C, avoiding light, for 30 min. Subsequently, 300 µL of buffer (70-S1001; MultiSciences, Zhejiang, China) was added. For ALDEFLUOR Stem Cell Identification (01700; STEMCELL, Canada), activated ALDEFLUOR was mixed with the sample, and half was transferred to a DEAB control tube before antibody incubation. The samples were analyzed using a BD LSR Fortessa or sorted using a BD FACS Aria III.

Colony formation and mammosphere formation assays

For colony formation, 2,000 cells were seeded per well in 6-well plates, treated with drugs for 24 h, and cultured for 7–10 days. Medium was refreshed every two days. Before observation, plates were fixed (BL539A; Biosharp, Shanghai, China) and stained (C0121; Beyotime, Shanghai, China) following the manufacturer’s protocol. For mammosphere assays, 2,000 cells were seeded in ultra-low attachment multi-well plates (CLS3471; Corning^® Costar^®, USA) and treated with drugs 24 h post-seeding. Cells were cultured for 10 days, with half the medium refreshed every two days. Mammospheres were observed under a microscope.

Xenograft tumor model

Female nude mice (6 weeks old, n = 7) were purchased from Byrness Weil Biotech Ltd. (Chongqing) and randomly assigned to either the control or PS-T-treated group. To ensure blinding, the mice were assigned unique identifiers, and researchers performing experiments were unaware of the treatment assignments. ALDH + MDA-MB-231 cells (1 × 10⁵) were injected into the mammary fat pads of each mouse. The PS-T-treated group received 3 mg PS-T by intragastric administration every two days, starting from day 2 post-injection. The control group received an equivalent volume of vehicle. After 4 weeks, mice were euthanized, and tumors were inspected and photographed. For the survival analysis, female nude mice bearing orthotopic xenografts derived from ALDH + MDA-MB-231 cells were randomly divided into control and PS-T-treated groups (n = 10 per group). The treatment regimen was identical to the initial xenograft study. Instead of a fixed endpoint, mice were monitored every two days for body weight, tumor volume, and overall health. The primary endpoint for each mouse was defined as the time to reach a humane endpoint, which included any of the following criteria: (1) tumor volume exceeding 1500 mm³, (2) ulceration or necrosis of the tumor, or (3) body weight loss exceeding 20% of the initial weight. The survival time for each mouse was recorded as the number of days from tumor cell inoculation to the humane endpoint. Survival curves were plotted using the Kaplan-Meier method, and statistical differences between groups were analyzed using the Log-rank test. A P-value of less than 0.05 was considered statistically significant.

Cell viability assay (CCK-8)

Cells (2 × 10⁴ cells/mL) were seeded into 96-well plates, treated with indicated drugs for 24 h, and then 10 µL of CCK-8 solution (CK04, DOJINDO) was added. After 1 h at 37 °C, the optical density (OD) was measured using a Varioskan Flash spectrophotometer (Thermo Scientific).

Reactive oxygen species (ROS) detection

Cells were seeded on coverslips in 6-well plates (2.5 mL) and cultured for 24 h with the indicated drugs. After incubation with DCFH-DA from the ROS Assay Kit (S0033; Beyotime, Shanghai, China) in serum-free medium at 37 °C for 30 min, cells were fixed in 4% paraformaldehyde and mounted with antifade medium containing DAPI (P0131; Beyotime, Shanghai, China). ROS production was observed under a fluorescence microscope (FV-1000; Olympus, Tokyo).

Lipid peroxidation and MDA quantification

MDA levels in cells and tissues were quantified using the Lipid Peroxidation MDA Assay Kit (S0131; Beyotime, Shanghai, China). The MDA detecting working solution was mixed with samples and incubated at 100 °C for 15 min. After cooling and centrifugation, 200 µL of supernatant was transferred into a 96-well plate, and absorbance at 532 nm was measured using Varioskan Flash. For the BODIPY™ 581/591 C11 probe, cells were incubated at 37 °C in the absence of light for 30 min. After digestion and washing, cells were resuspended in buffer and fluorescence was measured with BD LSRFortessa.

Western blotting

Cells were harvested and proteins were extracted using RIPA lysis buffer (P0013B; Beyotime, Shanghai, China). Protein concentration was quantified using an enhanced BCA protein assay kit (P0010; Beyotime, Shanghai, China). Protein samples were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane was blocked and incubated with primary and secondary antibodies, followed by visualization with the ECL detection system.

RNA isolation and quantitative PCR (qPCR)

Total RNA was extracted using the TaKaRa MiniBEST Universal RNA Extraction Kit (9767; Takara, Japan). cDNA was synthesized with the PrimeScript™ RT Master Mix (RR036A; Takara, Japan). PCR primers are listed in Table S1. qPCR was performed using TB Green^® Premix Ex Taq™ II FAST qPCR (CN830S; Takara) on an Eppendorf Mastercycler realplex. mRNA expression was calculated using the 2-∆∆CT method.

Immunohistochemistry (IHC) staining

IHC staining was performed following the SP kit instructions (SP-9000; ZSGB-BIO, Beijing, China). Staining intensity (SI) was scored as: 0 = negative; 1 = weakly positive; 2 = positive; 3 = strongly positive. The percentage of positive cells (PP) was scored as: 0 = < 5%, 1 = 6–25%, 2 = 26–50%, 3 = 51–75%, 4 = > 75%. IHC staining was quantified by multiplying SI by PP. Mean fluorescence intensity (MFI) was determined using ImageJ.

Bioinformatics analysis

RNA-sequencing data and clinical information for TNBC patients were sourced from the TCGA database. LASSO regression was applied using the R package glmnet, and nomogram models were constructed using the Cox method in the R package rms. ROC curve and ssGSEA enrichment analysis were performed using Sangerbox 3.0 [18]. Gene sets PECE_MAMMARY_STEM_CELL_UP, PECE_MAMMARY_STEM_CELL_DN, and WP_FERROPTOSIS were obtained from the Molecular Signatures Database (https://www.gsea-msigdb.org/gsea/msigdb). Differentially expressed genes (DEGs) were analyzed and visualized with GSEA 4.1 software. Survival analysis was performed with Kaplan-Meier curves, generated using Kaplan-Meier Plotter (http://kmplot.com/analysis/) and bc-GenExMiner v5.0 (http://bcgenex.ico.unicancer.fr/BC-GEM/GEM-Accueil.php? js=1).

Statistical analysis

Data are presented as mean ± standard deviation. The ‘n’ value represents the number of independent biological replicates. Statistical significance was assessed using the Student’s t-test or the Mann–Whitney U test for two-group comparisons. One-way ANOVA followed by Bonferroni’s post-hoc test was used for multiple comparisons. Univariate and multivariate logistic regression analyses were conducted. A p-value < 0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism 8.0 and SPSS 26.0 software.

Results

Elevated stem-related signature risk score predicts unfavorable prognosis in TNBC

We employed a LASSO regression model to analyze data from 167 TNBC patients in the TCGA database to identify genes most closely associated with stemness. This analysis revealed 17 genes that were predominantly linked to prognosis: CDC5L, IL4R, NDP, REXO2, TBCA, RGS17, RAB2B, ABHD6, PUM2, UQCR11, TCF7L2, SFPQ, TXN, EEF1D, SHQ1, FHL1, and PLBD1 (Fig. 1A-B). Risk factor analysis indicated a significant correlation between the risk score and the prognosis of TNBC patients (Fig. 1C). The prognostic model showed excellent performance, with areas under the time-dependent ROC curve for 1, 3, and 5 years of 0.80, 0.86, and 0.88, respectively (Fig. 1D). Subsequently, the 167 samples were divided into high-risk (n = 91) and low-risk (n = 76) groups. As shown in Fig. 1E, the mortality rate in the high-risk group was significantly higher than in the low-risk group (HR = 22.89, 95% CI: 5.44–96.3, P = 5.8e-10). Furthermore, the prognostic impact of these 17 genes was assessed through overall survival (OS), relapse-free survival (RFS), and distant metastasis-free survival (DMFS) data, with results presented in forest plots (Fig. 1F-G, and Fig. S1).

Fig. 1.

High stem-related signature risk score predicts worse prognosis in TNBC. (A-B) LASSO regression prognostic model for 167 TNBC patients based on TCGA database. (C) Risk factor analysis of TNBC patients in TCGA database (n = 167). (D) Validation of the prognostic model by the time-dependent ROC curves. (E) Kaplan-Meier survival analysis of TNBC patients in the high (blue) and low-risk group (red). (F-G) TNBC prognostic forest plot of 17 stem-related genes from the TCGA database

Univariate analysis revealed that primary tumor, nodal status, TNM stage, and risk factor score were independent predictors of TNBC prognosis (Fig. 2A). Multivariate analysis confirmed that the risk factor scores outperformed conventional clinical features like TNM stage in predicting prognosis (Fig. 2B). Additionally, we constructed a nomogram model using survival time, survival status, and other clinical characteristics, demonstrating that the risk score is a robust predictor of mortality risk (C-index = 0.891, 95% CI: 0.85–0.94, P = 1.33e-62) (Fig. 2C). The calibration curves further validated the model’s accuracy in predicting outcomes at 3 and 5 years (Fig. 2D). External validation using the GSE20711, GSE58812, and GSE21653 datasets confirmed the prognostic capability of the model, with areas under the ROC curve at 9 years of 0.78, 0.63, and 0.70, respectively (Fig. 2E).

Fig. 2.

Validation of the stem-related prognostic risk model. (A-B) Univariate and multivariate analysis of the stem-related prognostic risk model. (C) The nomogram for the prognostic risk model based on TCGA database. (D) Bootstrap analysis for indicating the probability of 3 and 5 years based on the prognostic model. (E) Validation of the prognostic model in external datasets (GSE20711, GSE58812, and GSE21653) by using the ROC curves at 9 years. (mean ± standard deviation; ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001)

PS-T diminishes the stemness phenotype of BCSCs in TNBC

To explore the effect of PS-T on cell stemness in TNBC, we conducted ssGSEA enrichment analysis and observed a reduction in ssGSEA scores in the PS-T-treated group (Fig. 3A). A dose-dependent decrease in the expression of stem cell markers, including POU5F1, SOX2, and NANOG, was noted following PS-T treatment (Fig. 3B). Notably, a similar suppressive effect of PS-T on the mRNA expression of POU5F1, SOX2, and NANOG was also observed in the HCC1806 TNBC cell line (Fig. S2), indicating that the anti-stemness effect of PS-T is not cell line-specific. Additionally, the proportions of ALDH⁺ and CD44^{^}highCD24^{^}low MDA-MB-231 cells significantly decreased in a dose-dependent manner after PS-T treatment (Fig. 3C). ALDH⁺ cells were sorted for further analysis in colony formation and mammosphere assays. As the PS-T concentration increased, the number and size of the colonies and mammospheres gradually diminished (Fig. 3D-E). Furthermore, when ALDH⁺ MDA-MB-231 cells were orthotopically injected into female nude mice, PS-T treatment significantly reduced the probability of tumor formation (Fig. 3F). We next investigated its effect on overall survival in mice bearing TNBC xenografts. ALDH-positive MDA-MB-231 cells were orthotopically injected into nude mice and then treated with PS-T or a control group (n = 10 per group). Kaplan-Meier survival analysis showed that PS-T treatment significantly prolonged the median survival of mice compared with the control group (Fig. S3). These data suggest that PS-T attenuates the stemness of BCSCs in TNBC.

Fig. 3.

PS-T reduces the stemness characteristics of BCSCs in TNBC. (A) The ssGSEA scores of MAMMARY-STEM-CELL gene set in MDA-MB-231 cells treated with or without PS-T (n = 3). (B) The mRNA levels of POU5F1, SOX2, and NANOG in 0, 10, and 50 µg/mL PS-T–treated MDA-MB-231 cells were detected by quantitative PCR (n = 9). (C) The proportion of ALDH⁺ (up) and CD44^highCD24^low MDA-MB-231 cells (down) treated with 0, 10, 20, and 50 µg/mL PS-T was determined by flow cytometry (n = 9). (D-E) Colony formation (D) (n = 3) and mammosphere assays (E) (n = 9) of ALDH⁺ MDA-MB-231 cells treated with indicated concentration of PS-T. Scale bars = 50 μm. (F) Xenograft assay using ALDH⁺ fraction upon PS-T treatment in MDA-MB-231 cells (n = 7/group). (B-E) One-way analysis of variance and post-hoc Bonferroni tests were utilized for the statistical analysis. Each experiment was repeated three times, and data was pooled. (mean ± standard deviation; For (B-E), one-way ANOVA with Bonferroni’s post-hoc test was used for statistical analysis. ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001)

PS-T restrains BCSCs in TNBC by activating ferroptosis

To investigate the mechanism by which PS-T regulates BCSCs in TNBC, we assessed cell viability in PS-T-treated ALDH⁺ cells following treatment with various cell death inhibitors. PS-T significantly suppressed cell proliferation, with only a slight recovery in proliferation upon inhibition of apoptosis, necrosis, or pyroptosis. However, this effect was substantially reversed when ferroptosis inhibitors, α-tocopherol phosphate (α-toc) and Liproxstatin-1 (Lip-1), were applied (Fig. 4A-B, Fig. S4A). GSEA enrichment analysis revealed that the ferroptosis pathway was significantly more active in PS-T–treated TNBC cells compared to the control, with an enrichment score (ES) of 0.61 and a P-value of 0.024, indicating a statistically significant difference (Fig. 4C). These results were further corroborated in HCC1806 (ES = 0.69, P < 0.001) and HCC1937 (ES = 0.63, P = 0.003) cell lines, which demonstrated that PS-T activates ferroptosis in TNBC (Fig. S4B). We further validated the induction of ferroptosis by PS-T in HCC1806 cells. The PS-T-induced suppression of cell viability was significantly reversed by the ferroptosis inhibitor Ferrostatin-1 (Fer-1) (Fig. S5A). Consistently, PS-T triggered pronounced lipid peroxidation in HCC1806 cells, which was effectively suppressed by Fer-1 (Fig. S5B). Moreover, western blot analysis confirmed that PS-T treatment downregulated GPX4 protein expression and induced LC3-II accumulation in HCC1806 cells (Fig. S5C), reinforcing the role of autophagy-dependent ferroptosis across TNBC models. Ferroptosis is characterized by the accumulation of ROS and excessive lipid oxidation. Consequently, we measured ROS levels, lipid oxidation, and malondialdehyde (MDA) concentration in ALDH⁺ MDA-MB-231 cells after PS-T treatment. The mean fluorescence intensity (MFI) of DCF and BODIPY™ 581/591 C11 and MDA concentration significantly decreased following PS-T treatment (Fig. 4D-F). In vivo, PS-T also led to a reduction in lipid peroxidation (Fig. 4G-H).

Fig. 4.

PS-T restrains BCSCs by inducing ferroptosis in TNBC. (A-B) Cell viability heatmap of the effects of different cell death inhibitors on PS-T–induced cell death. ALDH⁺ MDA-MB-231 cells were treated with indicated concentration of PS-T with or without Z-VAD-FMK (30 µM), Necrostatin-1 (Necro-1, 20 µM), LDC7559 (5 µM), α-tocopherol (α-toc, 100 µM) or liproxstatin-1 (Lip-1, 200 nM) for 24 h (n = 9). (C) GSEA result of MDA-MB-231 cells treated with or without PS-T using WP_FERROPTOSIS gene set. (D-F) The ROS fluorescence (D), C11 BODIPY fluorescence (E), and relative MDA concentration (F) of 0, 10, 20, and 50 µg/mL PS-T–treated ALDH⁺ MDA-MB-231 cells (n = 9). Scale bars = 20 μm. (G-H) The relative MDA concentration (G) and C11 BODIPY fluorescence (H) of breast cancer tissue of xenograft models treated with or without PS-T (n = 9). (D, G, H) The Student’s t-test was employed. (E, F) One-way analysis of variance and post-hoc Bonferroni tests were utilized. Each experiment was repeated three times, and data was pooled. (mean ± standard deviation; ***P < 0.001)

Further experiments using α-tocopherol and Liproxstatin-1 (ferroptosis inhibitors) revealed that the lipid oxidation level and intracellular MDA concentration in PS-T-treated cells were significantly reduced after co-treatment with these inhibitors (Fig. 5A-C). Co-treatment with PS-T and ferroptosis inhibitors also prevented PS-T-induced reductions in stem cell markers (POU5F1, SOX2, NANOG), ALDH⁺ and CD44^{^}highCD24^{^}low cell populations, and colony and mammosphere formation (Fig. 5D-I). Conversely, when PS-T was co-administered with the ferroptosis inducer Erastin, a significant reduction in ALDH⁺ and CD44^{^}highCD24^{^}low cells was observed (Fig. S6). Taken together, these findings suggest that PS-T inhibits TNBC stem cells by activating ferroptosis.

Fig. 5.

Ferroptosis inhibitors reverse the effect of PS-T on BCSCs. (A-C) The ROS fluorescence (A), C11 BODIPY fluorescence (B), and relative MDA concentration (C) in ALDH⁺ MDA-MB-231 cells that were treated alone or in combination with PS-T (10 µg/mL) and ferroptosis inhibitors α-tocopherol (α-toc, 100 µM) or liproxstatin-1 (Lip-1, 200 nM) for 24 h (n = 9). Scale bars = 20 μm. (D-E) The mRNA levels of POU5F1, SOX2, and NAONG in MDA-MB-231 cells that treated with PS-T combined with α-toc (D) or Lip-1 (E) (n = 9). (F-G) The proportion of ALDH⁺ (up) and CD44^highCD24^low cells (down) in PS-T–treated MDA-MB-231 cells that co-treated with or without α-toc (F) or Lip-1 (G) (n = 9). (H-I) Colony formation (H) (n = 3) and mammosphere assays (I) (n = 9) of ALDH⁺ MDA-MB-231 cells treated with PS-T, ferroptosis inhibitors, and a combination of both. Scale bars = 50 μm. (A-I) One-way analysis of variance and post-hoc Bonferroni tests were utilized. Each experiment was repeated three times, and data was pooled. (mean ± standard deviation; (A-I) One-way ANOVA with Bonferroni’s post-hoc test was used. **P < 0.01; ***P < 0.001)

PS-T promotes ferroptosis in BCSCs through autophagic degradation of GPX4

To further elucidate the molecular mechanism underlying PS-T-induced ferroptosis in BCSCs, we examined the expression levels of key ferroptosis markers through western blotting. We found that GPX4 (Glutathione Peroxidase 4) was notably downregulated after PS-T treatment, while there were no significant changes in other markers, including FTH1, SLC7A11, DMT1, NRF2, and NCOA4 (Fig. 6A). Immunohistochemical staining confirmed the PS-T-induced reduction of GPX4 protein levels in vivo (Fig. 6B-C). Notably, patients with high GPX4 expression had significantly poorer overall survival (OS) in TNBC (HR = 2.78, 95% CI: 1.45–5.31, P = 0.0012; HR = 3.29, 95% CI: 1.03–10.53, P = 0.0444) (Fig. 6D-E, and Fig. S7A). These results suggest that PS-T promotes ferroptosis by downregulating GPX4, thereby inhibiting BCSCs in TNBC.

Fig. 6.

PS-T promotes ferroptosis of BCSCs by inducing autophagic degradation of GPX4. (A) Expression levels of ferroptosis-related markers in ALDH⁺ MDA-MB-231 cells treated with 10 µg/mL PS-T for 24 h (n = 3). The grayscales were quantified using ImageJ software and calculated relative to β-actin levels. (B-C) Representative images of GPX4 protein expression in the control (up) and PS-T–treated mice (down) (n = 9). Scale bars = 100 and 50 μm. (D-E) Overall survival (OS) of low- and high-GPX4 group TNBC patients from GEO (D) and TCGA database (E). (F) Expression of GPX4 and LC3 protein in ALDH⁺ MDA-MB-231 that were divided into four groups, including control, PS-T (10 µg/mL), 3-MA (5 mM), 3-MA + PS-T (n = 3). (G-I) The ROS fluorescence (G), C11 BODIPY fluorescence (H), and relative MDA concentration (I) in ALDH⁺ MDA-MB-231 cells that were treated alone or in combination with PS-T and 3-MA for 24 h (n = 9). Scale bars = 20 μm. (J) The mRNA levels of POU5F1, SOX2, and NAONG in MDA-MB-231 cells treated with PS-T and 3-MA alone or in combination (n = 9). (K) The proportion of ALDH⁺ (up) and CD44^highCD24^low cells (down) in MDA-MB-231 cells treated with DMSO, PS-T, 3MA, and 3-MA + P-ST (n = 9). (L-M) Colony formation (L) (n = 3) and mammosphere assays (M) (n = 9) of PS-T–treated ALDH⁺ MDA-MB-231 cells that co-treated with or without treated with PS-T, 3-MA, and a combination of both. Scale bars = 50 μm. (A-C) The Student’s t-test was employed. (F-M) One-way analysis of variance and post-hoc Bonferroni tests were utilized. Each experiment was repeated three times, and data was pooled. (mean ± standard deviation; (F-M) One-way ANOVA with Bonferroni’s post-hoc test was used. ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001)

To investigate the mechanism behind PS-T-induced GPX4 downregulation, we assessed GPX4 mRNA levels in TNBC cells treated with PS-T and found no significant changes in transcription (Fig. S7B-C). Previous studies indicated that PS-T enhances autophagic flux in TNBC cells and inhibits tumor migration [14]. We thus examined whether autophagy contributes to GPX4 degradation. Using the autophagy inhibitor 3-methyladenine (3-MA), we observed that the reduction in GPX4 protein levels and lipid oxidation in PS-T-treated ALDH⁺ MDA-MB-231 cells was reversed (Fig. 6F-I). To more specifically inhibit the autophagic-lysosomal pathway and exclude the involvement of the proteasome, we employed chloroquine (CQ) and the proteasomal inhibitor MG132. Consistently, CQ, but not MG132, reversed PS-T-induced GPX4 degradation (Fig. S8), confirming the primary role of autophagy in this process. Furthermore, autophagy inhibition prevented the PS-T-induced reduction of stem cell markers, ALDH⁺ and CD44^{^}highCD24^{^}low cell populations, and colony and mammosphere formation (Fig. 6J-M and Fig.S9-S12). These findings suggest that PS-T induces autophagy-dependent ferroptosis in TNBC stem cells via GPX4 degradation.

PS-T promotes autophagic degradation of GPX4 by upregulating ASM in BCSCs

Acid sphingomyelinase (ASM; gene name: SMPD1) is activated under stress conditions (e.g., antineoplastic treatments, oxidative stress) and has been implicated in autophagy regulation [17, 18]. Previous studies suggested that ASM can facilitate autophagic degradation of GPX4 [23]. In the current study, we found that PS-T treatment significantly upregulated ASM expression in ALDH⁺ MDA-MB-231 cells (Fig. 7A-B). To investigate the regulatory mechanism underlying ASM upregulation, we pre-treated cells with the transcriptional inhibitor actinomycin D (Act D). This treatment completely abolished PS-T-induced increases in both SMPD1 mRNA and ASM protein (Fig. S13A-B), indicating that PS-T upregulates ASM primarily through a transcriptional mechanism. To definitively establish the functional necessity of ASM in the PS-T-induced signaling axis, we performed loss-of-function experiments using siRNA targeting SMPD1 (si-ASM) in MDA-MB-231 cells. Transfection with three independent si-ASM oligonucleotides, efficiently knocked down ASM expression (Fig. S14). In control (si-NC) cells, PS-T treatment induced the expected degradation of GPX4 and the accumulation of the autophagic marker LC3-II. However, in ASM-knockdown cells, PS-T failed to degrade GPX4 and could not robustly increase LC3-II levels (Fig. S15). These results demonstrate that ASM is an essential mediator for PS-T-induced autophagic degradation of GPX4 and subsequent ferroptosis in TNBC stem cells.

Fig. 7.

PS-T promotes autophagic GPX4 degradation by upregulating ASM in BCSCs. (A-B) The expression level of SMPD1 in 0, 10, and 50 µg/mL PS-T–treated ALDH⁺ MDA-MB-231 cells was detected by western blotting (A) (n = 3) and quantitative PCR (B) (n = 9). (C-D) Immunofluorescence (C) and Immunohistochemistry (D) staining of ASM, LC3, and GPX4 protein in the control (up) and PS-T–treated mice (down) (n = 9). Scale bars = 50 μm. (E) The levels of ASM, LC3, and GPX4 in TNBC patients were assessed by immunohistochemistry staining (n = 6). Scale bars = 50 μm. (F) The fluorescence of RFP and GFP puncta in ASM-overexpressing cells that were treated alone or in combination with PS-T for 24 h (n = 3). Scale bars = 20 μm. (G) Expression level of ASM, LC3, and GPX4 were photographed after treatment with DMSO (control), PS-T (10 µg/mL), or ASM-IN-1 (2 µM) alone or in combination with both (n = 3). Relative gray density was measured and analyzed statistically and are presented as a histogram (down). (A-B, F-G) One-way analysis of variance and post-hoc Bonferroni tests were utilized. (C-E) The Student’s t-test was employed. Each experiment was repeated three times, and data was pooled. (mean ± standard deviation; *P < 0.05; **P < 0.01; ***P < 0.001)

The critical role of ASM was further corroborated by in vivo evidence. In 4T-1 xenograft models,, PS-T induced ASM expression (Fig. 7C-D). Immunohistochemical analysis also revealed a positive correlation between PS-T-induced ASM upregulation and autophagic GPX4 degradation in vivo (Fig. 7D). In a subsequent immunohistochemical study of 12 TNBC samples, we observed that higher ASM expression was associated with increased LC3 levels and decreased GPX4 levels (Fig. 7E). Additionally, the combination of PS-T and ASM overexpression synergistically enhanced autophagy, as demonstrated by an increased RFP/GFP puncta ratio in the mLC3-RFP-GFP reporter assay (Fig. 7F). Notably, treatment with the ASM inhibitor ASM-IN-1 reversed the PS-T-induced autophagic flux and GPX4 degradation (Fig. 7G). These results highlight ASM as a critical factor in PS-T-mediated degradation of GPX4 in TNBC stem cells.

Discussion

Triple-negative breast cancer (TNBC) remains one of the most aggressive and difficult-to-treat subtypes of breast cancer due to its resistance to endocrine and anti-HER2 therapies [19]. Chemotherapy, the mainstay of treatment for TNBC, is often hindered by rapid recurrence and metastasis, leading to poor patient outcomes [20–22]. A key factor contributing to this challenge is the presence of breast cancer stem cells (BCSCs), which are linked to chemotherapy resistance, metastasis, and relapse [23–25]. Thus, targeting BCSCs presents a promising strategy to improve therapeutic efficacy in TNBC. In this study, we explored the potential of PS-T (polysaccharides of Huaier) as an inhibitor of BCSCs in TNBC and uncovered a novel mechanism by which PS-T induces autophagy-dependent ferroptosis to diminish the stemness of TNBC cells.

Our data highlight the predictive value of stemness-related genes in the prognosis of TNBC. Using LASSO regression analysis, we identified a set of 17 stemness-associated genes whose expression strongly correlated with the survival outcomes of TNBC patients. This supports the growing recognition that the stem cell-like properties of TNBC cells are integral to their aggressive nature and poor prognosis. It should be noted, however, that the functional connection between this prognostic signature and the specific mechanism of PS-T action remains to be fully elucidated, as it is unclear whether PS-T directly modulates these particular genes. Previous studies from our group have also demonstrated the positive impact of Huaier granules in improving overall survival and delaying disease progression in TNBC patients [16]. Expanding on these findings, we investigated PS-T, the main active component of Huaier, and its potential to target BCSCs.

We found that PS-T significantly suppressed the expression of stemness markers and reduced the proportion of ALDH⁺ and CD44^{^}highCD24^{^}low cells, both in vitro and in vivo. A limitation of our stemness validation is the absence of gold-standard in vivo limiting dilution assays, though our data showing PS-T’s abrogation of tumor formation by ALDH⁺ cells strongly supports its effect on tumor-initiating cells. These findings suggest that PS-T can effectively inhibit BCSC proliferation and stemness. The ability of PS-T to impair tumor formation by ALDH⁺ cells further underscores its potential as a therapeutic agent targeting the core of TNBC aggressiveness. Taken together, these results provide strong evidence that PS-T diminishes the stemness of TNBC cells, offering a potential avenue for improving TNBC treatment outcomes.

Our subsequent experiments explored the molecular mechanisms underlying the anti-BCSC effects of PS-T. Based on cell viability assays and GSEA enrichment analysis, we hypothesized that PS-T induces ferroptosis, a form of regulated cell death driven by oxidative stress and lipid peroxidation, as a mechanism to target BCSCs. Ferroptosis has gained attention in cancer research due to its ability to selectively eliminate cancer stem cells by disrupting their iron metabolism and inducing lipid peroxidations [9, 26]. Our data demonstrated that PS-T treatment led to increased levels of ROS and lipid oxidation, both in vitro and in vivo. The reversal of PS-T’s effects by ferroptosis inhibitors further confirmed the role of ferroptosis in PS-T’s anti-BCSC activity. These findings align with recent studies that have shown ferroptosis as a potent strategy to target cancer stem cells in various malignancies, including TNBC [9–11].

The central role of GPX4 (Glutathione Peroxidase 4) in ferroptosis was also confirmed in our study. GPX4 is a key regulator of ferroptosis, preventing lipid peroxidation by reducing lipid hydroperoxides [7, 12]. We observed that PS-T treatment significantly downregulated GPX4 expression, both at the protein level in TNBC cells and in vivo. This reduction in GPX4 protein provided further evidence that PS-T promotes ferroptosis in TNBC stem cells. Notably, we found that PS-T induces the degradation of GPX4 via autophagy, as pharmacological inhibition of autophagy with 3-MA reversed both the reduction in GPX4 levels and the suppression of BCSC stemness. This suggests that PS-T leverages the autophagy-lysosomal pathway to target and degrade GPX4, thereby facilitating ferroptosis.

Moreover, our study revealed that PS-T upregulates acid sphingomyelinase (ASM), a key enzyme involved in the sphingomyelin cycle, which regulates autophagy and ferroptosis. Previous research has highlighted the role of ASM in mediating the autophagic degradation of GPX4 in response to ferroptosis inducers [27].We demonstrated that PS-T enhances ASM expression in TNBC cells and tumors, which in turn promotes autophagic flux and GPX4 degradation. While we have established that PS-T induces the transcriptional upregulation of ASM, the precise upstream signaling cascade that initiates this process remains a key unresolved question. Our study does not identify the specific transcription factors, kinase pathways, or cell surface receptors that mediate the signal from PS-T to the SMPD1 gene promoter. Elucidating whether this involves known stress-responsive pathways or novel signaling mechanisms is critical for a complete understanding of the initial steps in PS-T’s mechanism of action. Inhibition of ASM reversed the effects of PS-T on GPX4 degradation and ferroptosis, confirming the critical role of ASM in the PS-T-induced suppression of BCSC stemness. These findings highlight ASM as a novel therapeutic target in the treatment of TNBC.

In conclusion, our study provides compelling evidence that PS-T inhibits TNBC stem cells by inducing ASM-mediated autophagic degradation of GPX4, thereby promoting ferroptosis. This mechanism offers a new perspective on how PS-T can be used to overcome the challenges of TNBC therapy, particularly in combating BCSC-mediated drug resistance and metastasis. While the potential of PS-T as a treatment for TNBC is promising, further research is needed to explore the off-target effects of PS-T and its broader impact on other molecular pathways. Furthermore, this study remains unexplored regarding the identity of the direct molecular target of PS-T. As a naturally occurring polysaccharide, PS-T likely interacts with specific receptors or protein complexes on or within cells to initiate the observed signaling cascade. Our current studies have defined downstream pathways but have yet to identify the primary binding partner of PS-T. Identifying this initial trigger—whether it is a pattern recognition receptor, transporter, or other cellular component—is crucial for understanding the specificity of PS-T’s actions and represents a major goal for future research. Nonetheless, this study paves the way for future therapeutic strategies targeting ferroptosis and autophagy in TNBC.

Conclusion

Our study reveals a novel therapeutic mechanism by which PS-T targets BCSCs in TNBC. By upregulating acid sphingomyelinase (ASM), PS-T activates autophagy, leading to the degradation of GPX4 and the subsequent induction of ferroptosis. This mechanism effectively diminishes the stemness characteristics of BCSCs, inhibiting their tumorigenic potential and providing a promising strategy for improving the prognosis of TNBC patients. These findings open new avenues for therapeutic approaches aimed at targeting cancer stem cells through the modulation of autophagy and ferroptosis, offering a potential solution to the challenges of TNBC treatment, including chemotherapy resistance and metastasis.

Author contributions

Conceptualization: JJ, HLM and MHW; Methodology: LXZ, ZWW, KFL, ZH and TY; Data analysis: LXZ, ZWW, ZH and PQW; Funding acquisition: MHW and WTY; Writing—original draft preparation: LXZ and ZWW; Writing—reviewing and editing: JJ, HLM, and MHW. All authors have read and agreed to the published version of the manuscript.

Funding

This study was funded by the National Natural Science Foundation of China [Grant No. 82474127 and 82203057], the Natural Science Foundation of Chongqing [Grant No. CSTB2023NSCQ-MSX0520], and Military Key Clinical Specialty [Grant No. 41561Z23612].

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All experiments were approved by the Laboratory Animal Welfare and Ethics Committee of the Army Medical University (SYXK-20170002) and the Ethics Committee of the Southwest Hospital of the Army Medical University (KY2021118). No patient was involved in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.