Erastin (cat. no. S7242), RAS-selective lethal 3 (RSL3; cat. no. S8155), ML162 (cat. no. S4452), sulfasalazine (SAS; cat. no. S1576), APY29 (cat. no. S6623), sunitinib (cat. no. S7781) and IRE1α kinase-inhibiting RNase attenuator (KIRA6; cat. no. S8658) were purchased from Selleck Chemicals. L-Glutamic acid (glutamate; cat. no. G8415), L-Cystine (cat. no. C7602), L-Methionine (cat. no. M5308) and L-Glutamine (cat. no. G8540) were purchased from Sigma-Aldrich (Merck KGaA). Lipofectamine^® 3000 Transfection Reagent (cat. no. L3000015), 0.25% Trypsin-EDTA (cat. no. 25200056), Hank's Balanced Salt Solution (HBSS; cat. no. 14175095), Opti-MEM^™ (cat. no. 11058021), cystine-deficient DMEM (cat. no. 21013024), BODIPY^™ 581/591 C11 (cat. no. D3861), TRIzol^® reagent (cat. no. 15596018) and the SuperScript^™ IV First-Strand Synthesis System Kit (cat. no. 18091050) were purchased from Thermo Fisher Scientific, Inc. Unless specified, the rest of the reagents used were purchased from Beyotime Biotechnology.

The MDA-MB-231 (cat. no. SCSP-5043), MDA-MB-468 (cat. no. SCSP-5053), 4T1 (cat. no. SCSP-5056) and HT1080 (cat. no. TCHu170) cell lines were purchased from The Cell Bank of Type Culture Collection of The Chinese Academy of Sciences. All cells were cultured under standard conditions in DMEM supplemented with 10% fetal bovine serum (catalog no. SFBE; NATOCOR,) and maintained in a humidified incubator at 37°C with 5% CO₂. The medium was replaced with fresh medium every 3 days. Upon reaching 80–90% confluency, the cells were passaged using 0.25% trypsin-EDTA solution to maintain optimal growth conditions.

The aforementioned cell lines (MDA-MB-231, MDA-MB-468, 4T1 and HT1080) were seeded in 96-well plates (3,000 cells/well). After plating, cells were treated with the indicated compounds (Erastin, RSL3, ML162, SAS, glutamate, APY29, sunitinib and KIRA6) at the concentrations detailed in the corresponding figures. After 24 h of treatment at 37°C, the culture medium was replaced with a mixture of serum-free medium and Cell Counting Kit-8 reagent (cat. no. 96992; MilliporeSigma) at a 9:1 ratio, followed by incubation at 37°C for 2 h. Cell viability was assessed by measuring the optical density at 450 nm, and the data were normalized to the control group, as previously described (12).

Intracellular lipid reactive oxygen species (ROS) were measured using flow cytometry, as previously described (13). A total of 100,000 MDA-MB-231 cells were seeded per well in a 6-well plate. After 6 h of treatment with the indicated compounds (erastin, 2 µM; SAS, 1 mM; glutamate, 20 mM; APY29, 200 nM) at 37°C, cells were harvested, washed with PBS and incubated with 1 µM BODIPY 581/591 C11 in serum-free medium for 30 min at 37°C. Cells were then resuspended in 500 µl HBSS and filtered through a 40-µm cell strainer (BD Biosciences). Flow cytometry analysis was immediately performed using an LSRFortessa instrument (BD Biosciences) with a 488 nm laser. Fluorescence signals from both unoxidized (PE channel) and oxidized (FITC channel) C11 were monitored. The ratio of the mean fluorescence intensity of FITC to PE was calculated. Data analysis was performed using FlowJo v10 software (BD Biosciences).

Cellular MDA levels were assessed using a Lipid Peroxidation Assay Kit (cat. no. ab233471; Abcam), as previously described (14). Briefly, cells were treated with compounds (Erastin, 2 µM; SAS, 1 mM; glutamate, 20 mM; APY29, 200 nM) for 12 h at 37°C. Samples and the MDA color reagent were added to the wells and incubated for 20 min at room temperature. Following this, the reaction solution was introduced, and the mix was incubated for 60 min at room temperature. The resulting product was subsequently assessed at 695 nm using an absorbance microplate reader.

MDA-MB-231, MDA-MB-468, and 4T1 cells were treated with compounds (Erastin, 2 µM; SAS, 1 mM; glutamate, 20 mM; APY29, 200 nM) for 24 h at 37°C and then harvested. GSH was assayed using the GSH and GSSG Assay Kit (cat. no. S0053; Beyotime Biotechnology), as previously described (15). Briefly, cells were washed with ice-cold PBS, harvested and resuspended in three volumes of Protein Removal Reagent M solution. Cell samples were subjected to two rapid freeze-thaw cycles by alternating between liquid nitrogen and a 37°C water bath. The corresponding assay reagents were added to the appropriate amount of cell sample. After 25 min, absorbance was measured at 412 nm using a microplate reader. GSH content was then calculated based on a standard curve.

The restricted medium was prepared by supplementing glutamine-, methionine- and cystine-deficient DMEM with 4 mM glutamine, 200 µM methionine and 10% fetal bovine serum, as previously described (16). For media with specific cystine concentrations, L-cystine was added to this base restricted medium. Prior to the experiments, cells were washed three times with ice-cold PBS to remove residual cystine and then cultured in the cystine-restricted medium. All subsequent assays, including lipid peroxidation and intracellular GSH measurement, were performed under a cystine-restricted condition of 50 µM.

MDA-MB-231, MDA-MB-468, and 4T1 cells were harvested and washed twice with ice cold PBS. Cells were then lysed on ice using RIPA lysis buffer (cat. no. P0013; Beyotime Biotechnology) containing the protease inhibitor PMSF (cat. no. ST507; Beyotime Biotechnology). Protein samples were quantified using a BCA protein concentration assay kit (cat. no. P0011; Beyotime Biotechnology). Equal amounts of protein (20 µg per lane) were separated using 4–20% SDS-PAGE and transferred to PVDF membranes using wet blotting. To reduce nonspecific binding, the membranes were blocked with 5% skim milk for 1 h at room temperature, followed by incubation with the corresponding primary antibodies overnight at 4°C with gentle shaking. The following day, the membranes were washed five times with TBST buffer (containing 0.1% Tween-20) for 5 min each and then incubated with horseradish peroxidase-conjugated secondary antibodies: Goat Anti-Mouse IgG (1:5,000; cat. no. ab205719; Abcam) and Goat Anti-Rabbit IgG (1:5,000; cat. no. ab205718; Abcam) at 30°C for 2 h. After washing three times with TBST (containing 0.1% Tween-20), the membranes were developed using an enhanced chemiluminescence detection reagent (cat. no. P10300; NCM Biotech, China) and images were captured using a ChemiScope 6100 Chemiluminescence Imaging System (CLiNX Science Instruments Co., Ltd.). Finally, grayscale analysis of target bands was performed using ImageJ software (v1.53a; National Institutes of Health). The primary antibodies used in this experiment were as follows: glutathione peroxidase 4 (GPX4; 1:5,000; cat. no. ab125066; Abcam), acyl-CoA synthetase long chain family member 4 (ACSL4; 1:5,000; cat. no. ab155282; Abcam), IRE1α (1:1,000; cat. no. 3294T; Cell Signaling Technology, Inc.), solute carrier family 7 member 11 (SLC7A11; also known as xCT; 1:1,000; cat. no. 12691T; Cell Signaling Technology, Inc.), and the internal controls β-actin (1:1,000; cat. no. sc-47778; Santa Cruz Biotechnology, Inc.) and GAPDH (1:1,000; cat. no. 2118T; Cell Signaling Technology, Inc.).

Human IRE1α WT (cat. no. 20744; Addgene, Inc.) and human IRE1α kinase-dead mutants (K599A; cat. no. 20745; Addgene, Inc.) were transfected into MDA-MB-231 cells as previously described (17,18). Lentiviral particles were produced using the second-generation packaging plasmids psPAX2 (cat. no. 12260; Addgene, Inc.) and pMD2.G (cat. no. 12259; Addgene, Inc.). For each transfection, 4 µg of the IRE1α transfer plasmid was combined with 3 µg of the packaging plasmid (psPAX2) and 1 µg of the envelope plasmid (pMD2.G). Specifically, the plasmid mix was diluted in 250 µl Opti-MEM containing 5 µl P3000 reagent, and 10 µl Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) was diluted in 250 µl Opti-MEM. The two solutions were mixed, incubated at room temperature for 15 min and then added to the 293T cell (ATCC) culture. Transfected cells were maintained at 37°C with 5% CO₂ for 48–72 h. The lentiviral supernatant was collected at 24 and 48 h post-transfection, pooled, filtered (0.45 µm), and concentrated. For transduction, MDA-MB-231 cells were infected at a multiplicity of infection (MOI) of 5 in the presence of 8 µg/ml polybrene for 24 h. Stable cell lines were subsequently selected using 2 µg/ml puromycin for 7 days, starting 48 h post-transduction. Following a 5–7 day recovery period, overexpression was verified by western blotting.

Bioinformatics analysis was performed using web-based bioinformatics tools. ERN1 levels in normal and breast cancer samples, as well as the profiles of ERN1 levels in different molecular subtypes of breast cancer, were analyzed using the online University of ALabama at Birmingham CANcer data analysis Portal (UALCAN) (19) (https://ualcan.path.uab.edu/). The levels of ERN1 in several malignant tumors were analyzed based on the web tool Gene Expression Profiling Interactive Analysis 2 (20) (http://gepia2.cancer-pku.cn). The online Kaplan-Meier Plotter (21) (http://kmplot.com/analysis) was employed to analyze recurrence-free survival differences between ERN1-high and -low groups in both overall breast cancer and HER2-positive subtypes, using the default parameters (probe: 227755_at; cut-off: median) based on its integrated breast cancer gene expression dataset. For TNBC-specific analysis, the GSE21653 dataset (available within the Kaplan-Meier Plotter tool) (22) was selected, applying the ‘auto-select best cut-off’ function.

MDA-MB-231, MDA-MB-468 and 4T1 cells were treated with APY29 (200 nM) for 24 h at 37°C and subsequently harvested. Total RNA was isolated using TRIzol reagent. cDNA was synthesized using the SuperScript IV First-Strand Synthesis System with the following protocol: 25°C for 10 min, 50°C for 10 min, and 80°C for 10 min. qPCR was performed using the SsoAdvanced Universal SYBR^® Green SuperMix (cat. no. 1725270; Bio-Rad Laboratories, Inc.). The thermocycling conditions were as follows: initial denaturation at 95°C for 30 sec, followed by 40 cycles of 95°C for 15 sec and 60°C for 30 sec. Gene expression was normalized to the endogenous reference gene GAPDH using the 2^−∆∆Cq method (23). The results are expressed as fold changes relative to the control group. Primer sequences used for qPCR were as follows (24): xCT forward, 5′-TGTGTGGGGTCCTGTCACTA-3′; xCT reverse, 5′-CAGTAGCTGCAGGGCGTATT-3′; GAPDH forward, 5′-ATGGGGAAGGTGAAGGTCG-3′; and GAPDH reverse, 5′-GGGGTCATTGATGGCAACAATA-3′.

All experimental data are presented as mean ± standard error of the mean. Unpaired t-test and one- or two-way ANOVA with Tukey's post hoc test were performed using GraphPad Prism 8.0 software (Dotmatics). P<0.05 was considered to indicate a statistically significant difference. All cell culture experiments were independently repeated at least three times to ensure data reproducibility and reliability.

Previous research reported that the activation of IRE1α can promote breast cancer development (25). The present study thus hypothesized that IRE1α may be pathologically associated with breast cancer and validated this using existing clinical datasets. Using the online tool UALCAN, ERN1 (which encodes IRE1α) levels were analyzed. The results revealed that ERN1 expression was significantly downregulated in breast cancer tissues compared with that in normal tissues (Fig. 1A). Notably, the ERN1 levels were significantly lower in the more aggressive HER2-positive and TNBC subtypes (26), than in the luminal subtype (Fig. 1B). This suggests that ERN1 may serve a critical role in the pathological process of HER2-positive and TNBC. Furthermore, pan-cancer analysis further demonstrated that ERN1 is generally downregulated in multiple cancer types compared with that in normal tissues (Fig. S1). Kaplan-Meier survival analysis also revealed that a low expression of ERN1 was significantly associated with a worse prognosis of patients with breast cancer (Fig. 1C). In patients who were HER2-positive and in those with TNBC with worse clinical outcomes (27), a lower expression of ERN1 was also significantly associated a worse prognosis (Fig. 1D and E). Collectively, these findings highlight the potential of targeting IRE1α for breast cancer therapy.

Ferroptosis has emerged as a promising therapeutic target in breast cancer (3), and TNBC exhibits a heightened susceptibility to ferroptosis compared with other molecular subtypes (7). Furthermore, IRE1α possesses both a kinase domain and an RNase domain (8), and whilst previous research has established the involvement of its RNase in modulating ferroptosis sensitivity (11), the potential role of its kinase in ferroptosis regulation remains unexplored. In the present study, to investigate the association between IRE1α kinase and ferroptosis in TNBC, MDA-MB-231 cells were treated with the type I kinase inhibitor APY29 (28) and a dose-effect curve was determined (Fig. S2A). Following subsequent experiments at a non-toxic concentration of APY29 (0.2 µM), it was observed that treatment with APY29 significantly enhanced the resistance of MDA-MB-231 cells to multiple ferroptosis-inducing agents, including erastin, SAS, glutamate and cystine restriction (Fig. 2A-D; P<0.0001 vs. Control at key concentrations). In contrast to the dose-dependent decrease in viability observed with erastin, SAS, and glutamate (which inhibit cystine uptake), cell viability increased with higher cystine concentration in the deprivation assay (Fig. 2D), as cystine itself is a critical substrate for the cellular antioxidant defense against ferroptosis. Notably, these inducers all trigger ferroptosis by directly or indirectly inhibiting system Xc⁻, the cystine/glutamate antiporter critical for antioxidant defense (29). This protective effect was further confirmed in MDA-MB-468 and 4T1 TNBC cells, as well as in HT1080 fibrosarcoma cells (Figs. 2E-H, S2B-D and S3A-D; P<0.0001 vs. Control at key concentrations). Notably, APY29 did not exert a significant effect on ferroptosis induced by the GPX4 inhibitors RSL3 or ML162 (Fig. S3E-H; P>0.05 vs. Control). These results suggest that IRE1α kinase modulates the susceptibility to ferroptosis through the regulation of system Xc⁻ rather than through GPX4.

The present study further selected two known IRE1α modulators, sunitinib [a type I inhibitor that suppresses IRE1α kinase activity without affecting its RNase function (30)] and KIRA6 [an ATP-competitive IRE1α kinase inhibiting RNase attenuator (31)], for subsequent investigations. First, the cytotoxicity of these compounds was evaluated in MDA-MB-231 cells (Fig. S2E and F). Subsequent experiments were then performed at non-toxic concentrations (sunitinib at 0.2 µM; KIRA6 at 0.5 µM). The results revealed that sunitinib significantly suppressed the ferroptosis induced by erastin, SAS, glutamate or cystine restriction (Fig. 3A-D; P<0.0001 vs. Control at key concentrations). Similarly, KIRA6 exerted marked cytoprotective effects (Fig. 3E-H; P<0.0001 vs. Control at key concentrations). These findings suggest that targeting the IRE1α kinase signaling pathway may represent an effective strategy for modulating ferroptosis.

To further investigate the role of IRE1α kinase in ferroptosis in MDA-MB-231, the present study established stable cell lines overexpressing either IRE1α-WT or IRE1α-K599A, which abolishes the kinase function of IRE1α (Fig. S4). Subsequent drug resistance assays demonstrated that, compared with the control cells, the overexpression of IRE1α-WT significantly enhanced cellular sensitivity to ferroptosis induced by several stimuli, including erastin, SAS, glutamate and cystine restriction. By contrast, the overexpression of the kinase-dead mutant (IRE1α-K599A) did not exert a significant effect on ferroptosis sensitivity (Fig. 4; P<0.0001 for IRE1α-WT vs. NC, P>0.05 for IRE1α-K599A vs. NC at key concentrations). Taken together, these findings demonstrate that the overexpression of IRE1α-WT, but not the kinase-dead mutant increased the sensitivity of TNBC cells to ferroptosis.

As lipid peroxidation is a critical process in the execution of ferroptosis (3), the present study examined the effects of IRE1α kinase inhibition on MDA [an end product of lipid peroxidation (32)] and lipid ROS levels. The findings demonstrated that treatment with APY29 significantly inhibited the accumulation of MDA (Fig. 5A-D) and lipid ROS (Fig. 5E-H) induced by erastin, SAS, glutamate or cystine restriction (compared with the respective inducer-only groups), whereas IRE1α kinase inhibition with APY29 did not affect the baseline levels of MDA or lipid ROS (compared with the control group) (Fig. 5).

Further investigations revealed that under ferroptosis-inducing conditions, the overexpression of IRE1α-WT significantly increased the accumulation of MDA (Fig. 6A-D) and lipid ROS (Fig. 6E-H) compared with that in the control cells and in cells expressing the kinase-dead mutant. Taken together, these results indicate that the inhibition of IRE1α kinase alleviates lipid peroxidation, thereby suppressing ferroptosis.

xCT is a key component of system Xc⁻ and serves a crucial role in regulating intracellular redox homeostasis and determining cellular sensitivity to ferroptosis (1). In the present study, the results revealed that the inhibition of IRE1α kinase activity significantly alleviated ferroptosis induced by system Xc⁻ suppression (Fig. 2). Specifically, Fig. 2 demonstrates that co-treatment with APY29 markedly improved cell viability compared with cells treated with erastin, SAS, glutamate, or under cystine restriction alone. Thus, it was hypothesized that IRE1α kinase may regulate ferroptosis through xCT. In comparison with controls, treatment of MDA-MB-231 cells with the IRE1α kinase inhibitor, APY29, significantly increased xCT mRNA levels and protein expression (Figs. 7A and S5A). This effect was also observed in MDA-MB-468 (Figs. S5B and S6A) and 4T1 cells (Figs. S5C and S6B). However, the expression of GPX4 and ACSL4, two key ferroptosis regulatory genes (1), remained unaltered (Fig. 7B and C).

The system Xc⁻ inhibits ferroptosis by maintaining GSH synthesis (33). Further experiments revealed that treatment with APY29 increased baseline GSH levels in MDA-MB-231 cells compared with the untreated control and effectively reversed GSH depletion induced by erastin, SAS, glutamate and cystine restriction compared with the respective inducer-only groups (Fig. 7D-G). This effect was also observed in MDA-MB-468 and 4T1 cells (Fig. S6C-F). Moreover, under these ferroptosis-inducing conditions, the overexpression of wild-type IRE1α significantly reduced GSH levels compared with that in both the control and kinase-dead mutant overexpression groups (Fig. 7H-K). Full original western blot images from Figs. 7 and S6 are presented in Fig. S7. Collectively, the aforementioned results indicate that the inhibition of IRE1α kinase activity attenuates GSH depletion, thereby enhancing cellular resistance to ferroptosis.