UA (cat. no. U6753) and deferoxamine (DFO; cat. no. D9533) were procured from Sigma-Aldrich; Merck KGaA. Z-VAD-FMK (ZVAD; cat. no. S7023) and ferrostatin-1 (fer; cat. no. S7243) were obtained from Selleck Chemicals. Chloroquine (CQ; cat. no. HY-17589A) and PTX (cat. no. HY-B0015) were purchased from MedChemExpress.

MDA-MB-231 and BT-549 cell lines were sourced from the Shanghai Institute of Biotechnology, Chinese Academy of Science. Cell line expansion and cryopreservation followed the guidelines of the American Pharmacopoeia Commission, with periodic mycobacterial inspection. MDA-MB-231 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (HyClone; Cytiva), while BT-549 cells were cultured in Roswell Park Memorial Institute-1640 media (HyClone; Cytiva). Both cell lines were supplemented with 10% fetal bovine serum (Bioexplorer, Inc.) and maintained in a humid environment with 5% CO₂ at 37°C. Subsequently, cells were kept under stem cell conditions using serum-free DMEM/F12 (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10 ng/ml human recombinant epidermal growth factor (EGF) (Invitrogen; Thermo Fisher Scientific, Inc.). Adherent cells of MDA-MB-231 and BT-549 were seeded into ultra-low attachment 6-well plates (Corning, Inc.) to form non-adherent spheroids at a density of 8×10⁴ cells/well, with medium replacement every four days.

Non-adherent spheroids were stained with CD44-APC (cat. no. 397517; BioLegend, Inc.), CD24-FITC (cat. no. 311117; BioLegend, Inc.) and respective isotype controls. After a 40-min incubation in the dark, cells were detected using the NovoCyte flow cytometer (ACEA Bioscience, Inc.) and analyzed by NovoExpress 1.5.8 software (Agilent Technologies, Inc.).

The manufacturer's protocol for the CD24 Microbead Kit (cat. no. 130-095-951; Miltenyi Biotec GmbH) was adhered to for the cultivation of 3×10⁷ MDA-MB-231 and BT-549 mammospheres using a primary antibody against CD24. Subsequently, the labeled cells were incubated with IgG microbeads (Miltenyi Biotec GmbH) and underwent magnetic separation using MiniMACS columns (Miltenyi Biotec GmbH). The purified CD24⁻ cells were then incubated with CD44 microbeads (cat. no. 130-095-194; Miltenyi Biotec GmbH), washed and subjected to magnetic separation once more.

Cellular proliferation was monitored according to the manufacturer's protocol using an optical microscope and quantitatively assessed via the Cell Counting Kit-8 (CCK-8) assay (Chongqing Baoguang Biotech Co., Ltd.; http://www.bgbiotech.com/) across various experimental groups. MDA-MB-231 and BT-549 adherent cells and spheroids were seeded into 96-well microplates or ultra-low attachment 96-well plates at a density of 1×10⁴ cells per well. After incubation at 37°C for 24 h, cells were treated with varying concentrations of UA or PTX (0, 10, 20, 40 µM) for 24 h, respectively. Subsequently, 10 µl of CCK-8 reagent was added per well, and the plates were incubated for 2 h. Cell viability was determined using the formula: Cell viability (%)=[(As-Ab)/(Ac-Ab)]x100%, where Ab, As, and Ac respectively denote the absorbance values at 450 nm wavelength for the blank medium, experimental group and control group.

The ALDH assay kit (cat. no. MAK082; Sigma-Aldrich; Merck KGaA) was employed for assessing ALDH activity, using 1×10⁶ cells per group. The colorimetric product measured at 450 nm was directly proportional to the ALDH activity present. The experiments were repeated at least three times.

The ROS assay kit (cat. no. S0033S; Beyotime Institute of Biotechnology) was utilized. Cells were seeded at a density of 5×10⁴ cells per well in either a standard 24-well plate or an ultra-low attachment 24-well plate. Intracellular ROS were measured using the DCFH-DA probe. Images of randomly selected fields were captured using a fluorescence microscope. The experiments were repeated at least three times.

The GSH (cat. no. KTB1600) and MDA (cat. no. KTB1050) assay kits were obtained from Abbkine Scientific Co., Ltd. The relative contents of MDA and GSH were detected following the manufacturer's recommended protocol. Each group prepared 1×10⁶ cells. The GSH and MDA levels of each group were determined by spectrophotometer detection at 412, 532 and 600 nm. The experiments were repeated at least three times.

The iron assay kit (cat. no. ab83366) was supplied by Abcam. In a physiological context, Fe²⁺ reacts with the iron probe to form a stable colored complex exhibiting absorbance at 593 nm. Additionally, Fe³⁺ can be reduced to Fe²⁺, enabling the accurate measurement of total iron (both Fe²⁺ and Fe³⁺) by adding iron reducer. The concentration of Fe³⁺ was determined by subtracting the Fe²⁺ content from the total iron level. To assess the iron content, 2×10⁶ cells were collected from each experimental group. The experiments were repeated at least three times.

Primary antibodies against CD44 (1:1,000; cat. no. ab189524), CD24 (1:1,000; cat. no. ab179821), TFR1 (1:1,000; cat. no. ab214039), KEAP1 (1:1,000; cat. no. 10503-2-ap) and LaminB1 (1:1,000; cat. no. 10503-2-ap) were supplied by Abcam, while the phosphorylated (p-) NRF2 (1:1,000; cat. no. ap1133) antibody was supplied by ABclonal Biotech Co., Ltd. and NRF2 (1:1,000; cat. no. 12721s) was bought from Cell Signaling Technology, Inc. Total protein was extracted from cells or tissues of each sample group using WB and IP cell lysates. Quantification was achieved using the BCA Protein Assay kit (cat. no. 23227; Thermo Fisher Scientific, Inc.). Subsequently, protein samples (20 micrograms per well) were separated by SDS-PAGE on a 10% gel and transferred onto PVDF membranes via electrophoresis. The membranes were blocked with blocking buffer (Vazyme Biotech Co., Ltd.) for 30 min at room temperature, followed by overnight incubation with primary antibodies at 4°C. The next day, the membranes were incubated with HRP-conjugated anti-rabbit secondary antibodies (1:10,000; cat. no. ab6721; Abcam) for 1 h at room temperature. The target protein was then detected by exposure to an ECL luminescent solution (Shanghai Yeasen Biotechnology Co., Ltd.). Quantitative protein analysis was performed using ImageJ 1.8.0 software (National Institutes of Health).

The Nuclear Extract Kit (cat. no. 40010) was purchased from Active Motif, Inc. MDA-MB-231 spheroids were plated in an ultra-low attachment dish (100 mm) at a density of 5×10⁶ cells/well and incubated for 4 h. The cells were then treated with 20 µM UA, 2 µM fer, or 10 µM DFO for 48 h. After washing the cells twice with ice-cold PBS containing phosphatase inhibitors, the cells were harvested by scraping the wells. The harvested suspension was centrifuged at 200 × g at 4°C for 5 min. To extract nuclear protein, the cell pellets were initially treated with hypotonic buffer for 15 min and then centrifuged at 14,000 × g at 4°C for 30 sec. The supernatants, representing the cytosolic fractions, were discarded. The pellets, designated as nuclear fractions, were lysed again in Complete Lysis Buffer (Active Motif, Inc.). The cell lysates were vortexed every 5 min for a total of 30 min and then centrifuged at 14,000 × g at 4°C for 10 min. The supernatants were isolated and collected as the nuclear protein extract.

Total RNA was isolated from MDA-MB-231 and BT-549 cells, as well as their respective sphere cells, using the Tissue RNA Miniprep Kit (cat. no. BW-R6311; Biomiga, Inc.). According to the manufacturer's instructions, a total of 500 ng of total RNA from each sample was used for the reverse transcription reaction using the HiScript III RT SuperMix (+gDNA wiper) (cat. no. R323-01; Vazyme Biotech Co., Ltd.). qPCR was performed on cDNA templates using ChamQ SYBR Color qPCR Master Mix (cat. no. Q431-02; Vazyme Biotech Co., Ltd.) on a realplex RT-qPCR system (Eppendorf SE). All qPCR reactions were conducted in a total volume of 10 µl. The qPCR protocol started with an initial denaturation step at 95°C for 5 min, followed by 40 amplification cycles, each consisting of denaturation at 95°C for 30 sec, annealing at 60°C for 30 sec, and extension at 72°C for 30 sec. The primer sequences used for the analysis of each gene are comprehensively listed in Table SI.

Single cell suspensions of MDA-MB-231 and BT-549 spheroids were prepared by pipetting. In DMEM/F-12 medium supplemented with 20 ng/ml EGF and 10% B27, the single cell suspensions were cultured in ultra-low adhesion 6-well plates at a density of 2×10⁴ cells per well to generate mammospheres for 7 days. A total of 3–5 randomly selected fields of view per well were counted, and floating aggregates >50 µm in diameter were chosen as mammospheres for further analysis. The total mammosphere area was used for statistical analysis.

Female Balb/C mice, aged 5 weeks and weighing between 16–18 grams, were obtained from Shanghai Jihui Laboratory Animal Care Co., Ltd. The mice were housed in specific pathogen-free conditions with 24–26°C temperature, 50–60% humidity and 12/12-h light/dark cycle at the Animal Resource Center of the Shanghai University of Medicine and Health Sciences, where they were provided with autoclaved water and food ad libitum. The efficacy of UA in vivo was assessed using an MDA-MB-231 spheroid xenograft model. Animals underwent a one-week acclimatization period before being used in experiments. Dissociated MDA-MB-231 spheroids (5×10⁶) were resuspended in 100 µl PBS and injected subcutaneously into the right flank of mice. Engrafted mice were monitored for tumor development through visual inspection until tumor formation occurred. On the 14th day post-injection, the mice were randomly assigned to 7 treatment groups, with 5 mice per group. Mice received intraperitoneal injections of 30 µl 10% dimethyl sulfoxide (DMSO) (22–24), which was freshly diluted in a saline containing 0.9% NaCl, containing UA (20 mg/kg body weight, daily) (25), or PTX (20 mg/kg body weight, daily) (26), or together with fer (5 mg/kg body weight, daily) (27) or DFO (100 mg/kg body weight, daily) (28) for 30 days. Body weight, tumor mass and the health indexes of mice (strong appetite, bright eyes, quick reaction, power of muscles) were assessed every 5 days. Tumor volume was measured using Vernier calipers and calculated using the following formula: Width² × length × π/6. According to the Laboratory animal Guidelines for euthanasia in China, the criteria used to determine when animals should be euthanized included: i) The maximum tumor diameter approaching 15 mm; ii) loss of weight >20%; iii) difficulty to move or breath. When the maximum tumor diameter approached 15 mm (the 30th day of treatment), mice were administered 2.5% isoflurane for induction of anesthesia by inhalation for 3 min and then 1.5% isoflurane for maintenance of anesthesia before being euthanized. Mice were euthanized by cervical dislocation under anesthesia to ensure humane conditions for the study. The mice were observed to have no response, including limb paralysis and no rise and fall of the chest to confirm death. The duration of the in vivo experiment was 44 days totally, including 14 days for tumor cells to proliferate in vivo and 30 days for the treatment with individual drug. Totally, 35 mice were used in the present study with no mice succumbing before the end of study.

The expression of KEAP1, NRF2 and TFR1 in cancer tissues was assessed using IHC staining. The tumors were fixed with 4% paraformaldehyde overnight at 4°C and then embedded in paraffin. Paraffin sections (4–6 µm) were heated in an oven at 60°C for 60 min and then dehydrated in xylene, followed by a series of ethanol washes (anhydrous ethanol, 95, 85 and 75% ethanol). Subsequently, the slices underwent three consecutive washes with PBS. Antigen retrieval was performed in sodium citrate buffer solution at 120°C for 15 min. After blocking with goat serum (cat. no. 16210064; Gibco; Thermo Fisher Scientific, Inc.) for 10 min at room temperature, the sections were incubated with primary antibodies against KEAP1 (1:500), NRF2 (1:500) and TFR1 (1:350) overnight at 4°C. The following day, the slices were washed with PBS and then incubated with biotin-labeled goat anti-rabbit antibody (1:1,000; cat. no. R-21234; Invitrogen; Thermo Fisher Scientific, Inc.) at 37°C for 1 h. DAB solution was applied to stain the samples, followed by counterstaining with hematoxylin at room temperature for 3 min. Finally, images were captured using a light microscope and quantitatively analyzed by IHC mean optical density (MOD) values. MOD value=(staining intensity × staining area)/total tissue area.

All data were presented as the mean ± SD from three independent experimental replicates. Comparison between two groups was performed using unpaired t-test. Statistical analysis of multiple groups was performed using one-way analysis of variance with Tukey's post hoc test using GraphPad Prism 9.0 (GraphPad Software; Dotmatics). P<0.05 was considered to indicate a statistically significant difference.

BCSCs, enriched through the formation of non-adherent spheroids, exhibit characteristics of CSCs (29). CD44⁺/CD24⁻ cells and ALDH^high cells are widely recognized as BCSCs (30). In the present study, BCSCs were collected from MDA-MB-231 and BT-549 cells within non-adherent spheroids, showing higher CD44 expression and lower CD24 expression compared with adherent cells, as detected by flow cytometric assay (Fig. 1A) and WB (Fig. 1B). ALDH activity was significantly elevated in spheroids compared with adherent cells (Fig. 1C). Moreover, the expression levels of putative stemness-associated markers CD133, EpCAM and transcription factors OCT4, Sox2 and c-myc were increased in spheroids (Fig. 1D). To further purify the CSCs, CD44⁺/CD24⁻ cells were isolated from MDA-MB-231 and BT-549 mammospheres using magnetic-activated cell sorting (Fig. S1). The intracellular ROS levels were slightly increased in spheroids and CD44⁺/CD24⁻ CSCs with no significant difference compared with adherent cells (Fig. 1E). Additionally, the amount of GSH, an intracellular antioxidant, was significantly higher in spheroids and CD44⁺/CD24⁻ CSCs than in adherent cells (Fig. 1F). However, lipid peroxidation levels were increased in CD44⁺/CD24⁻ CSCs compared with adherent cells, as evidenced by the amount of MDA in these cells (Fig. 1G). These findings suggested that BCSCs may possess a protective mechanism against lipid peroxidation.

Given that lipid peroxidation leads to ferroptosis, an iron-dependent form of cell death, several key regulators of ferroptosis were examined. Inhibitors of ferroptosis such as SLC7A11, GPX4 and NRF2 were significantly upregulated in spheroids and CSCs compared with adherent cells (Fig. 2A), indicating that BCSCs undergo less ferroptosis. Importantly, spheroids and CSCs exhibited enhanced iron uptake with higher TFR1 expression, increased iron storage with higher FTH1 expression, and reduced iron export with lower FPN expression (Fig. 2A). The expression of TFR1 was confirmed by WB (Fig. 2B). Indeed, spheroids and CSCs showed evident iron accumulation in ferric form (Fig. 2C). Conversely, ferrous iron, an activator of ferroptosis, was significantly lower in spheroids and CSCs compared with adherent cells (Fig. 2C). Thus, it was indicated that BCSCs might employ a protective mechanism against ferroptosis to maintain their stemness characteristics in an iron homeostasis-related manner.

Since BCSCs exhibit resistance to traditional chemotherapy drugs such as PTX, there is an urgent need to develop novel drugs targeting BCSCs. UA has gained attention as a potent anticancer agent in various cancers through inducing apoptosis or ferroptosis (27,31). However, whether UA could target CSCs by overcoming ferroptosis resistance remains unclear. In the present study, it was found that PTX and UA had no significant difference in inhibiting cell proliferation in adherent cells (Fig. 3A). However, breast cancer cell spheroids were more sensitive to UA-induced cell death than to PTX-induced cell death (Fig. 3A), indicating that UA is more effective than PTX in inducing cell death in BCSCs. Since the IC₅₀ of UA on MDA-MB-231 and BT-549 spheroids was 25.8 and 18.6 µM, respectively (Fig. S2A and B), 20 µM of UA were then used for the treatment of breast cancer cells. To elucidate the mechanisms by which UA induces cell death in BCSCs, several cell death inhibitors were added. As demonstrated in Fig. 3B, the caspase inhibitor ZVAD failed to reverse UA-induced cell death in BCSCs. The autophagy inhibitor CQ efficiently reversed UA-induced cell death in BCSCs. Importantly, ferroptosis inhibitors fer (which blocks lipid peroxidation) and DFO (an iron chelator) significantly reversed UA-induced cell death in BCSCs. Thus, UA might induce cell death in BCSCs mainly through ferroptosis. Furthermore, mammosphere formation, a typical property of CSCs in vitro, was investigated upon UA treatment. UA significantly reduced the sphere formation ability of BCSCs (Fig. 3C). Fer and DFO significantly reversed the inhibitory effect of UA. The expression of stemness markers, CD44 and CD24, was also analyzed in BCSCs upon UA treatment. UA treatment led to a decrease in CD44 and an increase in CD24 in BCSCs, which could be reversed by fer and DFO (Fig. 3D). The stemness regulators OCT4, SOX2 and c-myc were downregulated by UA, which could be partly reversed by fer (Fig. 3E). These data indicated that UA efficiently inhibits stemness characteristics and proliferation of BCSCs through ferroptosis-related cell death.

To further confirm ferroptosis induced by UA in BCSCs, intracellular ROS levels, lipid peroxidation levels and the amount of iron were detected. UA treatment led to a prominent increase in intracellular ROS in BCSCs (Fig. 4A). The UA-induced ROS eruption in BCSCs was significantly attenuated by fer and DFO (Fig. 4A). Lipid peroxidation levels, as indicated by the amount of MDA, were significantly enhanced in UA-treated BCSCs, which were alleviated by fer and DFO (Fig. 4B). Intracellular total iron and ferrous iron were significantly increased by UA treatment, and this increase was efficiently blocked by fer and DFO (Fig. 4C). Several key regulators of iron metabolism were therefore examined, including TFR1 for ferric iron uptake, FTH1 for ferrous iron storage, nuclear receptor coactivator 4 (NCOA4) for releasing ferrous iron in the lysosome, and FPN for iron export. The expression of TFR1, which was upregulated in BCSCs, was significantly decreased upon UA treatment as detected by RT-qPCR and WB (Fig. 4D and E). Fer and DFO efficiently reversed UA-induced TFR1 downregulation, indicating that TFR1 expression is mainly regulated by intracellular iron concentration (Fig. 4D). Consistent with the increase in ferrous iron in UA-treated BCSCs (Fig. 4C), FTH1 and NCOA4 were significantly enhanced upon UA treatment, and this was reversed by fer and DFO (Fig. 4D). FPN also increased prominently upon UA treatment and this was reversed by fer and DFO (Fig. 4D). Thus, UA-induced ferrous iron accumulation might be related to ferritin formation and degradation. These data indicated that UA might induce ferroptosis through stimulating ferrous iron and ROS accumulation in BCSCs.

NRF2, the master regulator of the antioxidant system, is known to play a crucial role in the transcriptional regulation of ferroptosis by targeting FTH1, TFR1, FSP1, GPX4 and SLC7A11 (32). KEAP1 acts as a negative regulator of NRF2 through the ubiquitin-proteasome pathway. In the present study, it was observed that the expression of KEAP1 was enhanced by UA treatment and this was reversed by fer and DFO in MDA-MB-231 spheroids (Fig. 5A). Similarly, the total level of p-NRF2/NRF2 and p-NRF2/NRF2 in the nucleus was decreased by UA treatment as well, and this was reversed by fer and DFO (Fig. 5B and C). Thus, UA may induce ferroptosis through stabilizing KEAP1 and inhibiting NRF2 activation. The main ferroptosis-suppressing pathways were further investigated; namely the GSH/GPX4 pathway and the FSP1 pathway, which is GPX4 independent. As demonstrated in Fig. 5D and E, SLC7A11, GPX4 and FSP1 were significantly reduced upon UA treatment in MDA-MB-231 and BT-549 spheroids. Fer and DFO efficiently reversed the UA-induced decline of SLC7A11, GPX4 and FSP1. These results indicated that UA may induce ferroptosis through the KEAP-NRF2-mediated pathway.

A subcutaneous tumor model was established to study the therapeutic potential of UA in vivo, and tumor growth was observed periodically (Fig. 6A-C). It was found that compared with DMSO treatment, UA treatment significantly slowed tumor growth, which was significantly more efficient than PTX treatment. The inhibitory effects of UA were attenuated by fer and DFO. Furthermore, TFR1, KEAP1 and NRF2 expression levels were detected in tumor tissues. Consistent with WB analyses, the expression levels of TFR1 and NRF2 were declined, and KEAP1 was enhanced by UA treatment, which could be reversed by fer and DFO administration (Fig. 6D). The schematic model depicting how UA regulates the KEAP1/NRF2 pathway and ferroptosis is shown in Fig. 6E.