All human breast cancer cell lines (MDA-MB-231, HS578T, MCF-7, T47D, HBL-100, BT549, MDA-MB-453, and NHFB) and breast epithelial cell line HBL-100 were purchased from the American Type Culture Collection (ATCC) and stored in liquid nitrogen. MDA-MB-231 and HS578 T cells were cultured in DMEM supplemented with 10% fetal bovine serum, 4.0 mM l-Glutamine, and 100IU/ml penicillin and 100IU/ml streptomycin. MCF-7 and T47D cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum, 4.0 mM l-Glutamine, and 100IU/ml penicillin and 100IU/ml streptomycin. All cells were cultured in a humidified incubator (37°C, 5% CO₂). The cell culture media and supplements were purchased from Thermo Fisher Scientific.

The corresponding reagents erastin (#HY-15763), sulfasalazine (#HY-14655), Z-VAD-FMK (#HY-16658B), E-necrosulfonamide (#HY-100573), ferrostatin-1 (#HY-100579), deferoxamine (#HY-D0903), and ciclopirox (#HY-B0450) were purchased from MedChemExpress. Liproxstatin-1 (#S7699) was purchased from Selleck. 3-Methyladenine (#189490) was purchased from Sigma Aldrich.

Cell viability was evaluated using a Cell Counting Kit-8 purchased from Ape Bio (K1018). Cells were seeded in 96-well plates. After treatment, the medium was replaced with 100 μL fresh medium containing 10 μL CCK8 reagent and incubated in a humidified incubator (37°C, 5% CO₂) for 2 h. The OD value was measured at 450 nm using a Thermo Scientific Spectrophotometer (1510–00712).

Propidium iodide (PI, Invitrogen) was used as a fluorescent signal for cell death. Polyphyllin III with or without Fer-1 treated cells were transferred to SFM supplemented with 5 mg/ml PI for PI imaging. After a 30 min incubation, bright-field and PI images were acquired.

Cells were harvested 24 h after treatment with Polyphyllin III with or without SAS and lysed with radioimmunoprecipitation assay (RIPA) buffer supplemented with 1× PMSF. Western blotting was performed according to the standard protocol with primary antibodies including GPX4 (1:1,000, abcam, ab125066), SLC7A11 (1:500, CST, 12691s), KLF4 (1:1,000, Novus, NBP1-83940), and ACSL4 (1:1,000, Santa Cruz, sc-271800) and the secondary antibody (1:5,000, Bioss, IWR-1-endo). Equal protein sample loading was monitored using an anti-β-actin antibody (1:2,000, Santa Cruz, sc-47778). Reactive bands were visualized with Amersham Image 600.

The total RNA was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. cDNA was synthesized using SuperScript Ⅱ Reverse transcriptase. Quantitative real-time PCR was performed using SYBR GreenER qPCR SuperMix universal, and triplicate samples were run on a Stratagene MX3000P qPCR system according to the manufacturer’s protocol. The threshold cycle (Ct) values for each gene were normalized to those of β-actin, and the 2^−ΔΔCt method was used for quantitative analysis. The primers used were as follows:

Q-SLC7A11-F: TCA​TTG​GAG​CAG​GAA​TCT​TCA

Adenoviral KLF4 (Ad-KLF4) was transfected into cells to overexpress KLF4. The LacZ plasmid vector was used as a nonspecific negative control. Proteins Knockdown was performed by duplex oligos siRNA transfection with Lipofectamine 3,000 (Invitrogen, no. 2189668) in OPTI-MEM (Gibco, no. 2185849) for 24 h. The corresponding siRNA sequences used were as follows:

siACSL4#1: 5′- CCU​CUU​AUU​UGC​UGU​GAA​ATT -3′ (sense);

The relative lipid ROS level in cells was assessed using C11-BODIPY dye (Thermo Fisher Scientific, D3861). Cells were treated with 5 μM C11-BODIPY for 30 min, harvested and washed twice with PBS. The sediment was resuspended in 500 μL PBS. Oxidation of the polyunsaturated butadienyl portion of the dye results in a shift of the fluorescence emission peak from ∼590 to ∼510 nm, which was detected by flow cytometry (BD Bioscience). The relative lipid peroxidation level was calculated using the median of each peak.

The relative GSH concentration in the cell lysates was assessed using a total Glutathione Assay Kit (Beyotime, S0052) according to the manufacturer’s instructions. The total glutathione content can be calculated by measuring the OD value at 412 nm.

Cells cultured in a 6-well plate were fixed with a solution containing 2.5% glutaraldehyde in PBS for 24 h. After being washed in PBS, the cells were treated with 0.1% Millipore-filtered cacodylate-buffered tannic acid, postfixed with 1% buffered osmium, and stained with 1% Millipore-filtered uranyl acetate. After dehydration and embedding, samples were incubated in a 60°C oven for 24 h. Digital images were obtained using a transmission electron microscope [Thermo Fisher (FEI)].

Tissue sections from clinical specimens (collected from Sir Run Shaw Hospital) or the indicated mouse models were fixed in 10% buffered formalin and embedded in paraffin. For IHC staining, tissue slides were deparaffinized in xylene and rehydrated in alcohol. Endogenous peroxidase was blocked with 3% hydrogen peroxide. Antigen retrieval was performed in a microwave with 0.1 M sodium citrate buffer (pH 6.0). Sections were then incubated overnight at 4 °C with antibodies against KLF4 (1:1,000, Novus, NBP1-83940), SLC7A11 (1:500, Abcam, ab37185) and 4-HNE (1:500, Abcam, ab48506). Antibody binding was detected using an HRP-DAB kit. Sections were then counterstained with Mayer’s hematoxylin and mounted onto coverslips. Images were acquired using polarized-light microscopy (Nikon, Eclipse 80I).

The cells were washed with PBS, and ice-cold lysis buffer (300 μL per 55 cm² dish) was added. Adherent cells were scraped off the dish using cell scrapers, and the cell suspension was held at 4°C for 30 min. The suspension was centrifuged in a microcentrifuge at 4°C, and the supernatant was transferred into a clear tube. Then 5× loading buffer was added to 50 μL lysate to serve as the input. Next, 40 μL of bead slurry was added to another 250 μL lysate and incubated overnight at 4°C. The mixture was centrifuged in a microcentrifuge at 4°C, and 1.5× loading buffer was added to the sediment to serve as the IP. Western blotting was performed according to the standard protocol with primary antibodies including SLC7A11 (1:500, CST, 12691s) and KLF4 (1:1,000, Novus, NBP1-83940) and the secondary antibody (1:5,000, Bioss, IWR-1-endo). Equal protein sample loading was monitored using an anti-β-actin antibody (1:2,000, Santa Cruz, sc-47778). Reactive bands were visualized with an Amersham Image 600.

Animal studies were reviewed and approved by the Ethics Committee for animal studies of Sir Run Shaw Hospital affiliated to Zhejiang University. MDA-MB-231 xenografts were established in 5 week-old BALB/C nude mice (Shanghai SLAC Laboratory Animal Corporation) by inoculating 1 × 10⁶ cells mixed with Matrigel (BD Biosciences) at a 1:1 ratio into the abdominal mammary fat pad. When the tumor reached 50–100 mm³, the mice were assigned randomly into different treatment groups (DMSO, PPIII, SAS, and PPIII + SAS groups), and each group consisted of 5 mice. PPIII (5 mg/kg/day) and SAS (200 mg/kg/day) were dissolved in dimethyl sulfoxide (DMSO), diluted in PBS, and then intraperitoneally injected into mice at a dose of 10 ml/kg/d once a day. The tumor sizes were measured every three days, and the tumor volumes were calculated as follows: length × width² × 0.5. After 21 days of treatment, all mice were euthanized, and the tumors were surgically removed. Portions of the tumors were immediately fixed in 10% buffered formalin for immunohistochemistry.

Flow cytometry data were analyzed using Flow Jo (X.10.0.7r2). Statistical analysis was performed using GraphPad Prism 6. The values are presented as the mean ± S.D. Statistical significance was determined using Student’s t-test. The significance of the correlation was determined using Pearson correlation analysis. p ≤ 0.05 was denoted as statistically significant.

To detect the cytotoxicity of Polyphyllin III toward triple-negative breast cancer (TNBC), we treated MDA-MB-231 and HS-578 T human TNBC cells with various concentrations of Polyphyllin III (PPIII, from 2.5 to 15.0 μM) for 24, 48 and 72 h and determined the cell viability using CCK8 assays. Polyphyllin III exerted dose- and time-dependent toxicity and proliferation inhibitory effects on both cancer cell lines (Figure 1A). Polyphyllin III exerted similar dose and time-dependent toxicity on MCF-7 and T47D breast cancer cells (Figure 1A) but not HBL-100 normal breast epithelial cells (Figure 1B), indicating Polyphyllin III had exclusively an anticancer activity. Next, we explored which form of cell death dominated in Polyphyllin III-induced MDA-MB-231 cell death. During the expanding experiment, we observed that there were not enough cancer cells left for subsequent studies with 7.5 μM Polyphyllin III. Referring to relevant research (Wang et al., 2016; Ding et al., 2020), we combined 5 μM Polyphyllin III with various cell death inhibitors, including Z-VAD-FMK (an apoptosis inhibitor), (E)-necrosulfonamide (a necroptosis inhibitor), 3-methyladenine (3-MA, an autophagy inhibitor), ferrostatin-1 (Fer-1, a ferroptosis inhibitor), liproxstatin-1 (Lipo-1, a ferroptosis inhibitor), deferoxamine (DFO, an iron chelating agent), and CPX (ciclopirox, an iron chelating agent). The results showed that only the ferroptosis inhibitors Fer-1, Lipo-1, and CPX could significantly improve the cellular viability, while the others could not (Figure 1C). Moreover, it was observed via propidium iodide (PI) staining that Polyphyllin III-induced cell death could be inhibited by Fer-1 (Figure 1D), suggesting that Polyphyllin III might exert its cytotoxicity through ferroptosis in MDA-MB-231 breast cancer cells.

To clarify that Polyphyllin III induced ferroptosis, we first investigated the morphological changes in MDA-MB-231 cells. Transmission electron microscopy revealed that cancer cells after 24 h of Polyphyllin III treatment exhibited an increased density of the mitochondrial membrane and a decreased density of the mitochondrial cristae (red arrows), which are typical morphologic features of ferroptosis, compared to those after DMSO treatment (control). We did not observe cell shrinking or swelling after Polyphyllin III treatment, and no significant shrinkage of nuclei (blue arrows) or aggregation of chromatin was observed (Figure 1E). Other hallmarks of ferroptosis are the accumulation of intracellular lipid ROS and the decrease of GSH. Thus, we investigated the level of total glutathione, and the results showed a decrease in GSH levels after Polyphyllin III treatment (Figure 1G), followed by a decrease in GPX4 protein expression (Figure 1H). Lipid peroxidation assayed by C11-BODIPY fluorescence also showed that lipid peroxidation was increased after Polyphyllin III treatment, while the combination of Fer-1 could impede this phenomenon (Figure 1I).

Above, we revealed that Polyphyllin III induced ferroptosis in MDA-MB-231 cells. To demonstrate the associated mechanism, we first detected the expression of proteins related to the ferroptosis pathway and found that Acyl-CoA synthetase long-chain family member 4 (ACSL4) was upregulated after Polyphyllin III treatment, both dose- and time-dependently (Figures 2A,B). ACSL4 functions as an acylase of arachidonic acid (AA) to promote lipid peroxidation and plays an important role in ferroptosis. qRT-PCR also showed that ACSL4 was upregulated at the RNA level (Figure 2C). Next, we transfected MDA-MB-231 cells with small interfering RNA (siACSL4) to silence ACSL4 expression, as shown in Figure 2E. We showed that downregulation of ACSL4 attenuated Polyphyllin III-induced ferroptosis, as confirmed by cell viability assays and lipid ROS assays. ACSL4 silencing decreased the proliferation-inhibitory effect of Polyphyllin III on MDA-MB-231 cells (Figure 2D). Moreover, it partially reversed the Polyphyllin III-induced lipid peroxidation increase (Figure 2F). Together, our results indicated that Polyphyllin III induced ferroptosis partly through upregulating ACSL4. Online UALCAN analysis (http://Ualcan.path.uab.edu/analysis) of ACSL4 expression in BRCA based on subclasses showed that ACSL4 expression in BRCA cells was significantly lower than that in normal breast cells, while TNBC exhibited the highest ACSL4 expression level among all BRCA subclasses (Figure 2G), which matches our finding in Figure 2F. Online Kaplan-Meier Plotter analysis (http://kmplot.com/analysis) showed a correlation between the low expression of ACSL4 and a poor prognosis of patients with TNBC (Figure 2H), possibly because cancer cells were less sensitive to ferroptosis. These online analyses highlighted the future perspective of using Polyphyllin III in the treatment of breast cancer.

When we detected the expression of ferroptosis-related proteins, we fortuitously observed the increasing expression of xCT after Polyphyllin III treatment (Figure 3A), which exerts an inhibitory effect on ferroptosis according to existing knowledge (Liu et al., 2020). We assumed that this protective upregulation of xCT could be an adaptive response of cancer cells to external stimulation, which might contribute to anticancer drug resistance.

We then combined Polyphyllin III with the xCT inhibitor sulfasalazine (SAS), an FDA-approved drug for ulcerative colitis, for the treatment of MDA-MB-231 cells. The results showed that after combination with SAS, the increase in xCT caused by Polyphyllin III treatment decreased (Figure 3B). More importantly, the sensitivity of cancer cells to Polyphyllin III was significantly increased, as confirmed by cell viability assays (Figure 3C) and PI staining (Figure 3F). In addition, SAS further increased the intensity of ferroptosis induced by Polyphyllin III with lower GSH levels and higher lipid peroxidation levels (Figures 3D,E). Morphologically, cancer cells treated with SAS and Polyphyllin III present more atrophic mitochondria, denser mitochondrial membranes and fewer mitochondrial cristae (Figure 3G) compared to those treated with a single application. We also used xCT siRNA (sixCT) to downregulate the expression of xCT (Figure 3I). Cell viability assay showed that sixCT strengthened the toxicity of Polyphyllin III on MDA-MB-231 cells (Figure 3H), further confirming the beneficial effect of xCT in Polyphyllin III activity, Together, our data strongly suggested that SAS sensitized MDA-MB-231 cells to Polyphyllin III, with a stronger intensity of ferroptosis.

Krüppel-like factor 4 (KLF4) is one of the most important transcription factors in eukaryotes and plays an important regulatory role in the occurrence and development of breast cancer. Studies have shown that changes in intracellular ROS could increase the expression of KLF4. We speculated that KLF4 might play a certain role in the induction of intracellular accumulation of lipid ROS and ferroptosis by Polyphyllin III.

First, we used online Kaplan-Meier Plotter analysis (http://kmplot.com/analysis) to analyse the relationship between breast cancer prognoses and mRNA expression levels of KLF4 or SLC7A11. In 414 breast cancer cases, both KLF4 and SLC7A11 mRNA levels were negatively related to patients’ post progression survival (PPS) (Figure 4A); that is, the higher the KLF4 or SLC7A11 mRNA expression level, the lower the survival rate after progression. Next, we performed data analysis on the GEO database. The results showed that KLF4 and SLC7A11 were positively correlated in basal-like breast cancer cells (Figure 4B). Then, KLF4 and xCT protein levels were measured in different breast cancer cell lines using western blotting analysis. We observed that the expression levels of both KLF4 and xCT in MCF-7, T47D and HS578T breast cancer cells were relatively high, while in BT549, MDA-MB-453 and NHFB breast cancer cells, they were relatively low (Figure 4C). Pearson correlation analysis showed that there was a significant positive correlation between KLF4 and xCT with a Pearson r value of 0.6774 and p < 0.05. Finally, 10 clinical specimens of breast cancer cases underwent immunohistochemical analysis of KLF4 and xCT. As shown in Figure 4D, the expression levels of KLF4 and xCT were also positively correlated, with the Pearson χ² value of 4.286 and p < 0.05.

We then used small interfering RNA siKLF4 to silence or plasmid transfection to overexpress KLF4. Silencing KLF4 induced a decrease in xCT protein levels but not mRNA levels. Similarly, overexpressing KLF4 induced an increase in xCT protein levels but not mRNA levels (Figures 4E,F). Moreover, the protein interaction between KLF4 and xCT was detected by immunoprecipitation (IP), confirming that KLF4 had a direct interaction with xCT (Figure 4G).

Western blotting analysis confirmed that KLF4 was upregulated at both the protein (Figure 5A) and mRNA levels (Figure 5B). To demonstrate the hypothesis that KLF4 mediated Polyphyllin III-induced protective upregulation of xCT, transfection was conducted to silence or overexpress KLF4.

On one hand, silencing KLF4 decreased the xCT protein upregulation effect induced by Polyphyllin III (Figure 5C) and sensitized MDA-MB-231 cells to Polyphyllin III (Figure 5E). Silencing KLF4 further decreased the GSH levels (Figure 5F) and increased the lipid peroxidation levels (Figure 5G) compared to Polyphyllin III treatment only.

On the other hand, overexpressing KLF4 increased xCT protein expression (Figure 5D) and blocked the proliferation inhibitory effect caused by Polyphyllin III with (Figure 5H) or without SAS (Figure 5I). Overexpression of KLF4 compensated for the GSH reduction induced by Polyphyllin III (Figure 5J). Overexpression of KLF4 did decrease the lipid ROS elevation induced by Polyphyllin III (Figure 5K), but the difference was almost significance.

Together, our data showed that KLF4 was involved in Polyphyllin III-induced protective upregulation of xCT.

Next, we explored the effect of Polyphyllin III with or without SAS in vivo by establishing a xenograft model of breast cancer in immunodeficient nude mice. MDA-MB-231 human breast cancer cells were implanted subcutaneously into the left axilla. Two weeks after implantation, the tumor-bearing mice were randomly divided into four groups: a negative control group (equal amount of DMSO dissolved in PBS), a PPIII group (5 mg/kg/day), an SAS group (200 mg/kg/day) and a PPIII + SAS group. Each group was treated daily with the indicated drugs intraperitoneally at a dosage of 200 ml/kg for 21 days. The tumor volume (Figures 6A,E) and tumor weight (Figure 6D) of the mice were recorded for assessment of the drug efficacy. As expected, Polyphyllin III potently suppressed tumor growth, and its combination with SAS further enhanced this effect on breast cancer (Figures 6B–E). Notably, SAS alone did not suppress the tumor growth of the xenografts. Immunohistochemical analysis of the tumor tissue confirmed the increase in ACSL4 expression after Polyphyllin III treatment (Figure 6F). Moreover, Polyphyllin III treatment significantly increased the expression of KLF4 and xCT, while the xCT inhibitor SAS combination reversed this phenomenon and enhanced the upregulation effect of Polyphyllin III on 4-HNE, which is an indicator of intracellular lipid peroxidation and oxidative stress (Figure 6G). Together, our data showed that Polyphyllin III exerted a tumor suppression effect in vivo. Moreover, the xCT inhibitor SAS significantly sensitized MDA-MB-231 breast cancer cells to Polyphyllin III, likely by enhancing intracellular lipid peroxidation and ferroptosis.