The DRD2 Antagonist Haloperidol Mediates Autophagy-Induced Ferroptosis to Increase Temozolomide Sensitivity by Promoting Endoplasmic Reticulum Stress in Glioblastoma

Glioblastoma (GBM) is the most malignant central nervous system tumor and has high mortality (1). Surgery combined with radiochemotherapy is the conventional treatment approach. However, the prognosis is still relatively dismal, and the median overall survival time is 15 to 18 months (2). Because the vast majority of patients develop therapeutic resistance—the widely used chemotherapeutic agent temozolomide was found to only extend the survival time of patients with GBM by 2 months (3)—effective therapies are urgently needed. Finally, we need to study potential treatments affecting different mechanisms to inhibit GBM growth.

Ferroptosis is a recently identified form of nonapoptotic necrotic cell death that is regulated and triggered by iron-dependent lipid peroxidation and is modulated by multiple cellular metabolic pathways and a variety of disease-related signaling pathways (4, 5). Reactive oxygen species (ROS) accumulation in cells, which is regulated by the iron-dependent Fenton reaction and glutathione (GSH) loss, is the direct cause of ferroptosis (5). Impaired glutathione peroxidase 4 (GPX4) activity and the accumulation of redox-active iron are the hallmarks of ferroptosis (6). The whole ferroptotic process is often accompanied by other phenotypic interactions and often involves autophagy and endoplasmic reticulum (ER) stress (7–9). Many other processes, such as p62-Keap1-NRF2 pathway signaling, fatty acid metabolism, iron metabolism, and ROS accumulation, require both autophagy and ferroptosis to function. Therefore, treatments that increase the intensity of autophagy to a level sufficient to induce ferroptosis could be promising therapeutic strategies (10). Moreover, ferroptosis is most likely to occur in drug-resistant tumor cells, especially in mesenchymal and metastatic tumor cells (11). Numerous studies have reported that reduced ferroptosis contributes to the development of drug resistance in tumor cells (12). Moreover, the strategy of inducing ferroptosis has been used in the treatment of breast, pancreatic, and lung cancer (13–15). These results suggest that the role of ferroptosis in cancer could be complex and deserves further investigation.

Recently, dopamine D2 receptor (DRD2) was noted as a potential target for oncotherapy, as high DRD2 expression is associated with poor prognosis in many human cancers, such as pancreatic ductal adenocarcinoma (PDAC) and GBM (16, 17). Haloperidol, as a FDA-approved DRD2 antagonist, has inhibitory effects in a variety of cancers, including GBM (12, 18). However, the mechanism by which haloperidol inhibits GBM remains unclear. In this study, we identified haloperidol as a promoter of ferroptosis in GBM cells via regulation of autophagy and ER stress. Furthermore, we demonstrated that mechanistically, haloperidol reverses temozolomide resistance caused by upregulation of DRD2. Collectively, our results indicate the novel pharmacologic action of haloperidol in promoting cellular ferroptosis in the treatment of GBM.

Ferrostatin-1 (Fer-1, #S7243), haloperidol (#S1920), deferoxamine mesylate (DFO, #S5742), temozolomide (#S1237), necrostatin-1 (Nec-1, #S8037), Z-VAD(OH)-FMK (#S8102), RSL3 (#S8155), N-Nitro-L-nitroarginine methyl ester (L-NAME, #S2877), N-acetylcysteine (NAC, #S1623), Indomethacin (INDO, #S1723), 3-methyladenine (3-MA, #S2767), and bafilomycin A1 (Baf-A1, # S1413) were purchased from Selleck Chemicals (Houston, Texas). Antibodies specific for MGMT (AB_2800069), LC3 I/II (AB_2617131), GRP78 (AB_2119845), p-PERK (AB_2095853), PERK (AB_10841299), ATF-4 (AB_2616025), phosphorylated protein kinase A (p-PKA, AB_330304), and FTH1 (AB_11217441) were purchased from Cell Signaling Technology (Massachusetts). Antibodies specific for DRD2 (AB_2759776), NCOA4 (AB_2766454) and GAPDH (AB_2619673) were purchased from ABclonal Technology Co., Ltd. (Wuhan, China). Antibodies specific for 4-HNE (AB_867452) were purchased from Abcam (Cambridge, United Kingdom).

The human GBM cell lines U87MG (HTB-14), T98G (RRID:CVCL_0556), LN229 (RRID:CVCL_0393), A172 (RRID:CVCL_0131), and U118MG (HTB-15) were purchased from ATCC and the murine GBM cell line GL261 (RRID:CVCL_Y003) was purchased from the SGST (Shanghai, China). The human glial cell line HEB (RRID:CVCL_S549) was obtained from Sun Yat-Sen University (Guangdong, China). Identification of cells was confirmed by short tandem repeat analysis. Prior to the experiment, frozen early passaged cells were thawed and passed through two to three times. Cells were routinely cultured in DMEM (containing 4.5 g/L glucose, Gibco 11995065) supplemented with 10% FBS (Gibco, 16140071) and 1% penicillin/streptomycin (Gibco, 10378016). All cell lines were cultured at 37°C in a humidified atmosphere of 5% CO₂. Cell lines were tested for Mycoplasma prior to the beginning of the studies (Mycoplasma Detection Kit, Yeasen Biotechnology).

To inhibit gene expression, siRNAs against the target genes were used. DRD2-, ATG5-, ATG7-, and NCOA4‐targeting shRNAs and siRNAs were designed and manufactured by GenePharma, Inc. The siRNAs were added to 125 μL of ribonuclease-free water to a concentration of 20 μmol/L, and 5 μL of the siRNA solution was mixed with transfection reagent and incubated at 25°C for 20 minutes. For transfection, 5 × 10⁶ cells were seeded in 6‐well plates and incubated overnight at 37°C in 5% CO₂ to 80% to 90% confluence and were then transfected with siRNA. The medium containing siRNAs was removed after 6 hours and replaced with medium containing 10% FBS for further incubation.

To measure cell viability, 5,000 cells per well were seeded in a 96-well plate 1 day before treatment. Upon treatment with the appropriate drugs as indicated, the medium in each well was replaced with fresh medium containing Cell Counting Kit-8 (CCK8) reagent (#B34304; Bimake). After incubation for 1 hour at 37°C, the plate was analyzed using a BMG microplate reader (BMG Labtech, CLARIOstar, RRID:SCR_019751), and the absorbance of the wells was measured at 450 nm.

An EdU incorporation assay was conducted to examine the proliferation of cells using an EdU assay kit (Beyotime Biotechnology) according to the manufacturer's protocol. In brief, cells were passaged in 24-well plates at a density of 1 × 10⁵ cells/well. After 24 hours, the cells were exposed to various concentrations of haloperidol (0, 10, 20, and 40 μmol/L) and temozolomide (400 μmol/L) for 24 hours. Then, 10 μmol/L EdU reagent was added to the medium and incubated for 2 hours. After fixation and permeabilization, the human brain microvascular endothelial cells were counterstained. The percentage of EdU-positive cells was determined using a laser scanning confocal microscope (Olympus Confocal Laser Scanning Microscope Fluoview FV3000, RRID:SCR_017015).

For the cell colony formation assay, cells (200 cells/well) were seeded into 6-well culture plates and cultured in DMEM supplemented with 10% FBS. The cells were treated with the indicated agents and incubated for 10 to 14 days at 37°C in 5% CO₂. The colonies were then stained with 0.1% crystal violet (Sigma‒Aldrich) and counted. For each set of cells, three independent assays were carried out.

Apoptosis was assessed by a FITC Annexin V Apoptosis Detection Kit I (BD Biosciences, San Diego, CA) according to the provided protocol. In brief, cells were passaged in 6-well plates at a density of 3 × 10⁵ cells/well. After 24 hours, the cells were exposed to various concentrations of haloperidol (0, 10, 20, and 40 μmol/L) and temozolomide (400 μmol/L) for 24 hours. Then, the collected cells were subjected to Annexin V-FITC/PI double staining and analyzed using a flow cytometer (FACS Canto II, BD Biosciences). The apoptosis rate was determined using FlowJo version 10.0 software (FlowJo LLC/Becton Dickinson, Ashland, OR, RRID:SCR_008520).

GBM cells were subjected to different treatments. Freshly harvested tumors from mice were fixed overnight with 2.5% glutaraldehyde at 4°C and postfixed in 1% osmic acid. The fixed samples were then dehydrated using a graded series of ethanol (70%−100%) and embedded in EPON resin. Ultrathin sections were sliced with an ultramicrotome and double stained with uranyl acetate and lead citrate. The stained sections were then examined using a Jeol JEOL JEM-1400 Flash Transmission Electron Microscope (RRID:SCR_020179).

GBM cells stably expressing GFP-LC3 were treated with haloperidol (0 and 20 μmol/L) and temozolomide (400 μmol/L). Two days after transfection, the cells were fixed with 4% paraformaldehyde (PFA). GFP-LC3 puncta formation was visualized under a confocal laser scanning microscope (Olympus Confocal Laser Scanning Microscope Fluoview FV3000). The average number of GFP-LC3 puncta/cell was determined in at least 200 cells.

The relative iron concentration in cell lysates was determined with an Iron Assay Kit (#ab83366, Abcam) according to the kit instructions.

Two million cells were seeded on 10-cm dishes. The following day, cells were treated with compounds to deplete GSH before being harvested to determine cell number. Each sample had two million live cells transferred to new tubes. A GSH Assay Kit (#S0053; Beyotime Institute of Biotechnology, Shanghai, China) was used to measure the intracellular total GSH levels according to the manufacturers' protocols. Results normalized to cell viability.

On the basis of the reaction between malondialdehyde (MDA) and thiobarbituric acid (TBA), we used a lipid peroxidation MDA assay kit (#S0131S; Beyotime, Shanghai, China) to measure changes in lipid peroxidation. In addition, we used another C11-BODIPY581/591 probe as a lipid peroxide indicator. After different treatments, cells were stained with 5 μmol/L C11-BODIPY581/591 probe (#D3861; Thermo Fisher Scientific, Waltham, MA) in accordance with the manufacturer's instructions. C11-BODIPY581/591 fluorescence was evaluated by flow cytometry (FACS Canto II, BD Biosciences). Data analysis was conducted using FlowJo Software.

Total RNA was extracted from cultured tumor cells or tumor samples using RNAiso Plus* (TaKaRa, 9109; Shiga, Japan) according to the manufacturer's instructions. For each sample, 1 μg of total RNA was used in a 20 μL reaction volume to synthesize cDNA using a PrimeScript RT Reagent Kit with gDNA Eraser (TaKaRa, RR047A). Then, 1 μL of the cDNA library was included in a 20 μL volume of PCR mixture. SYBR Premix Ex Taq II Tli RNaseH Plus (TaKaRa, RR820A) was used to determine the threshold cycle (Ct) value of each sample using the Quant Studio 5 Real-Time PCR System (Thermo Fisher Scientific). The primer sequences were as follows: human PTGS2, forward: 5′-CTGATGATTGCCCGACTCCC-3′, reverse: 5′-TCGTAGTCGAGGTCATAGTTC-3′; human DRD2, forward: 5′-TGTACAATACGCGCTACAGCTCCA-3′, reverse: 5′-ATGCACTCGTTCTGGTCTGCGTTA-3′; human GAPDH, forward: 5′-GGAGCGAGATCCCTCCAAAAT-3′, reverse: 5′-GGCTGTTGTCATACTTCTCATGG-3′. Gene expression was calculated with the 2- Δ ΔCt method and normalized to the internal control, GAPDH.

The relative cyclic adenosine monophosphate (cAMP) concentration in cell lysates was determined with a cAMP Direct immunoassay kit (#K371–100, Biovision), and the experimental method was carried out according to the instructions. In brief, 50 μL of neutralization buffer, 5 μL of acetylation mix and 845 μL of 1× cAMP assay buffer were added to 100 μL of each standard and sample and mixed well. Then, 50 μL of the mixture was transferred to a protein G–coated plate, and 10 μL of a rabbit anti-cAMP polyclonal antibody was added to each well. After incubation for 1 hour at room temperature, 10 μL of cAMP-horseradish peroxidase (HRP) was added and incubated again for 1 hour at room temperature. The plate was washed, 100 μL of HRP developer was added and incubated with shaking for 1 hour, 100 μL of 1 M HCl was added to terminate the reaction, and the OD450 nm was measured. The concentration of cAMP in each sample was calculated after plotting the standard curve based on the standard.

Cells and tumor samples were lysed in RIPA buffer (Sigma, R0278), and protein concentrations were determined using a BCA Protein Assay Kit (Solarbio Life Sciences, PC0020; Beijing, China). Total protein (40 μg) was separated by SDS‒PAGE on 10% to 15% gels and electrotransferred to PVDF membranes (Millipore, IPVH00010; Billerica, MA). The membranes were then blocked with 5% skim milk (BD Biosciences, 232100; San Jose, CA) or 5% BSA (Solarbio Life Sciences, A8020) in 0.1% Tween 20 (Sigma, P9416) in TBS, incubated with the primary antibodies overnight at 4°C, and then incubated with an HRP-conjugated secondary antibody (CST) for 1 hour at room temperature. The band intensities were quantified using the Tanon 5500 Chemiluminescence Imaging System (Tanon Science & Technology; Shanghai, China) with Immobilon ECL Ultra Western HRP Substrate (Millipore, WBULS0500) as the chemiluminescent substrate. GAPDH served as the loading control.

IHC was used to detect protein expression in xenograft tumors from C57BL/6. Target tumors were fixed with 4% PFA for 24 hours. Paraffin-embedded blocks were used to prepare 4 μmol/L thick sections. After deparaffinization, dehydration, treatment with 3% hydrogen peroxide to block endogenous peroxidase activity, antigen repair, and conventional serum blocking, the sections were incubated overnight at 4°C with specific primary antibodies, including anti-DRD2 (1:200), anti-p-PKA (1:400), anti-GRP78 (1:200), anti-4-HNE (1:200), and anti-LC3B (1:250) antibodies. Then, the sections were incubated with a secondary antibody. The peroxidase reaction was performed using DAB, and the sections were then counterstained with Mayer's hematoxylin. Finally, the sections were visualized, and images were acquired under a microscope (Olympus IX71, RRID:SCR_022185). An immunoreactive scoring system was used to conduct semiquantitative analysis of IHC staining. The percentage of positive cells was scored as follows: no stained cells: 0; 1% to 10% stained cells: 1; 10% to 50% stained cells: 2; 51% to 80% stained cells: 3; and 81% to 100% stained cells. The staining intensity was scored as follows: no color reaction: 0; weak reaction: 1; moderate reaction: 2; and intense reaction. The final immunoreactive score was calculated as follows: (staining intensity score) × (percentage of positive cells score).

Male C57BL/6 (RRID:MGI:2159769) mice were purchased from Guangdong Medical Laboratory Animal Center. To establish the orthotopic brain tumor model, GL261 (1 × 10⁶) were injected intracranially (in 5 μL of PBS per mouse) into 8-week-old male C57BL/6 mice. The cells were injected into the frontal cortex (coordinates: X = 2 mm, Y = −3 mm, Z = −2.5 mm, with the bregma serving as the origin for the X and Y coordinates). The mice were kept in a pathogen-free environment with unlimited access to food and drink. Four weeks after injection, tumors were confirmed using MRI (Bruker Medical Inc., Billerica, MA), and the animals were randomly divided into five groups of 12 mice each. These mice were subsequently treated with the following drugs: temozolomide (50 mg/kg), haloperidol (10 mg/kg), and DFO (100 mg/kg). Mice were sacrificed at defined asymptomatic time points or defined humane endpoints (significant weight loss; neurological signs; changes in drinking or eating; signs of pain). The mice were sacrificed in the terminal phase under abdominal anesthesia with pentobarbital sodium. The brains were then removed after transcardiac perfusion with heparin-saline and 4% PFA. The tumor volume was calculated using the following equation: tumor volume (V) = length × width × width × 1/2. The animal experiments were approved by the Animal Experiment Center of Southern Medical University.

Five patients with primary GBM and 5 patients with recurrent GBM who underwent tumor resection at the Nanfang Hospital of Southern Medical University (Guangzhou, China) were included in this study. Written informed consent was obtained from each patient. All the research was carried out in accordance with the provisions of the Declaration of Helsinki of 1975. Fresh samples were immediately preserved in liquid nitrogen. All specimens had confirmed pathologic diagnosis and were classified according to the 2016 World Health Organization Classification of Tumors of the Central Nervous System. Informed consent was obtained from each patient. The use of human brain tumor specimens and the database was by written consent and ethically approved by the Institutional Review Board at Nanfang Hospital of Southern Medical University (Guangzhou, China).

All experiments were performed in triplicate, and the mean and standard error of the mean were reported where appropriate. Analysis of variance (ANOVA) followed by a post hoc Dunnett test (to compare multiple groups to one control group) or post hoc Tukey test (to identify differences among subgroups) was conducted for multigroup comparisons. Where appropriate, direct comparisons were conducted using an unpaired two-tailed Student t test. A Spearman rank test was applied to verify the correlations of the grading information. Survival curves were estimated with the Kaplan–Meier product-limit method, and survival distributions were compared across groups with the log-rank test. Significance is denoted as follows: ns, P ≥ 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Any data generated, acquired, or used in this study will be made available upon request, unless the 3 databases are already accessible via public repositories as indicated earlier. The 3 datasets used for data comparison with the current study were CGGA (http://www.cgga.org.cn/), lvy GAP (http://glioblastoma.alleninstitute.org), and Freije (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE4412). The MS proteomics data (.RAW files) have been deposited to the Proteome Xchange Consortium via the PRIDE partner repository with the dataset identifier PXD042080 (19). Raw data generated in this study are available upon reasonable request from the corresponding authors.

To determine whether DRD2 expression is associated with prognosis in GBM, we performed bioinformatics analysis using the GlioVis website (http://gliovis.bioinfo.cnio.es/) and found a correlation of DRD2 mRNA expression with prognosis in GSE4412 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE4412) released by Freije and colleagues (Fig. 1A; Supplementary Table S1). Next, expression analysis based on the lvy_GAP dataset (http://glioblastoma.alleninstitute.org/) indicated that DRD2 was expressed at lower levels in the mesenchymal type of GBM than in the classical and proneural types (Fig. 1B; Supplementary Table S1). We further confirmed that DRD2 was more highly expressed in patients with MGMT promoter methylation than in those without MGMT promoter unmethylation (Fig. 1C; Supplementary Table S1). We then investigated the relationship between DRD2 expression and MGMT-independent temozolomide resistance in human GBM cell lines and revealed that DRD2 was highly expressed in MGMT-negative GBM cell lines with a higher IC₅₀ of temozolomide (Fig. 1D and E). Moreover, the DRD2 level was found to be significantly increased in the established temozolomide-resistant (TMZ-R) cells compared with the parental cells (Fig. 1F). We further confirmed that DRD2 was also expressed at varying levels in patient GBM tissue samples by Western blotting (Fig. 1G) and IHC (Fig. 1H). The results indicated high DRD2 expression in tumor compared with normal brain tissue and in recurrent tumor compared with primary tumor tissue. These results suggested that DRD2 is positively correlated with GBM malignancy and would likely be useful as a therapeutic target in GBM.

Tumor cells exhibit adaptive resistance to chemotherapy, and previous research has shown that dopamine receptors are involved in chemotherapeutic resistance (20); therefore, we investigated whether temozolomide induces DRD2 upregulation. As expected, temozolomide treatment resulted in significantly high DRD2 protein expression in a time-dependent manner in both U87MG cells (with low DRD2 expression) and LN229 cells (with high DRD2 expression), as determined by Western blotting and qPCR (Fig. 2A and B). To exclude the influence of culture conditions in vitro, we treated C57BL/6 tumor-bearing mice with temozolomide. IHC experiments indicated the upregulation of intracellular DRD2 (Fig. 2C). In addition, proteome analysis of LN229 cells treated with temozolomide confirmed the upregulation of DRD2 (Fig. 2D). The above results revealed that DRD2 expression was upregulated in GBM cells after temozolomide chemotherapy. Furthermore, DRD2 overexpression was shown to significantly suppress temozolomide-induced growth inhibition in U87MG cells, but DRD2 knockdown increased temozolomide sensitivity. This sensitization effect was attenuated by the ferroptosis inhibitor Fer-1(Fig. 2E). Ferroptosis plays an important role in drug resistance in various types of cancer, and we also found that DRD2 expression was inversely correlated with the ferroptosis-related genes ACSL4, GLS2, ALOX12, ALOX15, EMC2 and SAT1 expression in the CGGA (http://www.cgga.org.cn/) database (Supplementary Fig. S1A and Supplementary Table S1); it is further suggested that DRD2 is associated with ferroptosis sensitivity.

To confirm whether DRD2 affects the ferroptosis sensitivity of GBM cells, we first determined the mRNA level of PTGS2, an important biomarker for ferroptosis, in DRD2 knockdown cells. Cells treated with both shDRD2 and temozolomide had higher PTGS2 mRNA levels than those treated with either shDRD2 or temozolomide alone (Fig. 2F). As ferroptosis is characterized by iron accumulation, lipid peroxidation and antioxidant system damage (21), we examined the levels of MDA, GSH and lipid peroxidation. As expected, the combined treatment increased the levels of MDA and lipid peroxidation (Fig. 2G and I) but reduced the relative GSH expression level (Fig. 2H). Taken together, these results indicate that temozolomide reduces the ferroptosis sensitivity of GBM cells by upregulating DRD2, thereby promoting acquired drug resistance.

To improve the chemosensitivity of GBM cells, we used haloperidol, a specific antagonist of DRD2, in combination with temozolomide. Compared with treatment with haloperidol or temozolomide alone, treatment with the haloperidol and temozolomide combination significantly reduced the survival of U87MG cells (Fig. 3A), without cytotoxicity to normal glial cells (Supplementary Fig. S1B). After measurement of the IC₅₀ value (Supplementary Fig. S1C), cell viability was measured in the U87MG cell line following treatment with 400 μmol/L temozolomide in the absence or presence of 20 μmol/L haloperidol. As shown in Fig. 3B, haloperidol promoted temozolomide-induced cell death in a time-dependent manner in U87MG cells. The EdU incorporation assay (Fig. 3C) and colony formation assay (Fig. 3D) showed a similar result: in combination with temozolomide treatment, haloperidol treatment significantly inhibited the growth of GBM cells compared with treatment with either temozolomide or haloperidol alone. Having established that the drug combination of haloperidol and temozolomide reduced cell survival, we next sought to characterize the mechanism of cell death.

Since we showed that DRD2 knockdown increased temozolomide's inhibitory effect on GBM cell growth via ferroptosis, to confirm whether haloperidol also induces ferroptosis, we examined morphologic changes in mitochondria by transmission electron microscopy (TEM) and found that the combined treatment increased the mitochondrial matrix electron density and damaged mitochondrial membranes and ridges (Fig. 3E). Moreover, treatment with combinations of haloperidol and temozolomide significantly increased events associated with ferroptosis, including an increase in PTGS2 mRNA expression (Fig. 3F), MDA generation (Fig. 3G), a reduction in GSH (Fig. 3H), and lipid peroxidation (Fig. 3I). Furthermore, treatment with the ferroptosis inhibitor Fer-1 and the ROS scavenger NAC markedly reversed the temozolomide+haloperidol–induced growth inhibition (Supplementary Fig. S2A–S2C), whereas treatment with either the pancaspase inhibitor Z-VAD-(OH)-FMK or the necroptosis inhibitor necrostatin-1 failed to reverse this process (Fig. 3J). In addition, Western blot analysis (Supplementary Fig. S1D) indicated that haloperidol treatment did not activate caspase-3–mediated apoptosis and necrosis, and annexin-V FITC/PI staining (Supplementary Fig. S1E) revealed that the reduced survival of U87MG cells by haloperidol treatment was not associated with apoptosis. Because ferroptosis involves disruption of the reduction homeostasis of GPX4 and overproduction of ROS, we confirmed that temozolomide+haloperidol caused an imbalance in intracellular redox levels, resulting in ferroptosis by nitric oxide synthase inhibitor (N-Nitro-L-nitroarginine methyl ester, L-NAME) and overexpression of GPX4 and CAT (Supplementary Fig. S2D). These results indicate that haloperidol increases the temozolomide sensitivity of GBM cells by inducing ferroptosis. Ferroptosis can also be used to treat GBM cells that are resistant to radiotherapy, and Lei and colleagues indicate that inactivating SLC7A11 or GPX4 with ferroptosis inducers (FIN) sensitizes radioresistant cancer cells and xenograft tumors to ionizing radiation (22). We synoptically verified that haloperidol also enhances the radiosensitivity of GBM cells (Supplementary Fig. S3A and S3B).

Autophagy is extensively involved in temozolomide resistance in GBM (23) and has been shown to act as a double-edged sword: low levels of autophagy may serve a protective function in preserving cell survival under genetic stress, but excessive levels of autophagy ultimately mediate cell death. Increasing evidence suggests that ferroptosis is an autophagic cell death process (24). As shown in Fig. 4A and B, autophagic activity was upregulated in GBM cells exposed to haloperidol and was even more significantly enhanced in cells treated with temozolomide (Fig. 4C) as increased LC3-II levels. In addition, TEM was used to quantify the number of autophagic vacuoles per cell. Cells treated with temozolomide+haloperidol had significantly more autophagic vacuoles per cell (Fig. 4D). Consistent with this finding, temozolomide+haloperidol treatment strongly triggered GFP-LC3 puncta formation in T98G cells (Fig. 4E). As autophagy is a dynamic process, LC3-II accumulation can result from either autophagy induction or autophagic flux inhibition at a later stage in the pathway. We further monitored autophagic flux by detecting LC3-II in the presence of 3-MA or Baf-A1. In addition, we used ATG5/7 siRNA to inhibit autophagy. These inhibitors significantly inhibited the various stages of temozolomide+haloperidol–induced autophagy, indicating that temozolomide+haloperidol induced complete autophagic flux (Fig. 4F and G).

Substantial evidence indicates that autophagy is inextricably linked to ferroptosis (24). Thus, we examined whether temozolomide+haloperidol–induced ferroptosis is caused by a high level of autophagy. As shown in Fig. 4H and M, pretreatment with autophagy inhibitors dramatically attenuated or suppressed cell death induced by the combination of temozolomide and haloperidol in U87MG cells. Moreover, pretreatment with these inhibitors interfered with ferroptosis induced by temozolomide and haloperidol (Fig. 4H–Q and Supplementary Fig. S4A and S4B). Altogether, these results suggest that the ferroptosis induced by the combination of haloperidol and temozolomide depends on autophagy in GBM cells. Furthermore, the role of haloperidol in sensitivity enhancement is reflected by the increase in temozolomide-induced autophagy and consequent promotion of ferroptosis.

Because DRD2 deficiency leads to ER stress (16), whose severe form has antiproliferative and proapoptotic effects on cancer cells (12), we examined the possible effect of temozolomide and haloperidol on the induction of ER stress by evaluating the levels of GRP78, p-PERK (Thr 981) and ATF4. As shown in Fig. 5A and B, treatment with either temozolomide or haloperidol moderately activated ER stress, while temozolomide+haloperidol treatment further enhanced ER stress. Notably, the ferroptosis-inducing role of temozolomide+haloperidol was proven by the above results, and ER stress is a primary regulator of autophagy. Therefore, we hypothesized that ER stress triggers ferroptosis by inducing autophagy in GBM cells. We next treated the cells with the ER stress inhibitor 4-PBA prior to temozolomide+haloperidol treatment. As shown in Fig. 5C–F, pretreatment with 4-PBA dramatically suppressed ER stress and attenuated autophagy induced by the temozolomide and haloperidol combination, and the inhibitory effect of temozolomide+haloperidol on cell viability was also reversed (Fig. 5G; Supplementary Fig. S4A and S4B). In addition, ferroptosis was reduced as a result of the lower PTGS2 mRNA level (Fig. 5H) and higher GSH level (Fig. 5I). Moreover, 4-PBA treatment markedly inhibited lipid peroxidation (Fig. 5J and K). Collectively, our findings suggest that haloperidol amplifies temozolomide-induced ER stress and activates autophagy, thereby mediating the onset of cellular ferroptosis.

It has been demonstrated that cAMP and PKA play a role in ER stress (16). In addition, DRD2 has also been shown to influence intracellular cAMP formation (25). Therefore, we investigated whether the induction of ER stress by haloperidol is mediated by cAMP and PKA. The activity of PKA, which is dependent on the presence of cAMP, was measured in U87MG cells after treatment with haloperidol using Western blot analysis with an anti-phospho-PKA substrate antibody. Both haloperidol alone and temozolomide+haloperidol significantly promoted the activation of PKA (Fig. 6A and B). Moreover, SQ22536, a specific cAMP inhibitor, was used to downregulate cAMP and PKA activation in U87MG cells; alternatively, the drug H-89 was used to inhibit PKA (Fig. 6C and D). Treatment with cAMP/PKA inhibitors significantly reduced temozolomide+haloperidol–induced ER stress and autophagy in U87MG cells (Fig. 6E). Furthermore, the temozolomide+haloperidol–induced ferroptosis-related events were reversed (Supplementary Fig. S4A and S4B), as indicated by the lower PTGS2 mRNA and MDA levels and decreased lipid peroxidation (Fig. 6F–J). Moreover, SQ22536 and H89 markedly increased the GSH level (Fig. 6I). Taken together, these results indicate that haloperidol can reverse the inhibition of cAMP by temozolomide and that the combination of the two drugs can promote ER stress and autophagy by upregulating cAMP/PKA, resulting in ferroptosis.

Excessive autophagy can trigger ferroptosis by affecting iron storage through ferritinophagy (26). When compared with treatment with either temozolomide or haloperidol, temozolomide+haloperidol treatment dramatically increased the level of ferrous iron (Fe²⁺; Fig. 7A). Therefore, we analyzed the changes in the expression levels of ferritinophagy regulatory proteins, including ferritin heavy chain 1 (FTH1) and nuclear receptor coactivator 4 (NCOA4), after treatment with 400 μmol/L temozolomide with or without haloperidol. As shown in Fig. 7B, FTH1 and NCOA4 were downregulated in drug combination–treated GBM cells, indicating that the intracellular Fe²⁺ overload may be the result of increased ferritin degradation and release of free iron. To confirm the involvement of ferritinophagy, we performed an NCOA4 pulldown assay in U87MG cells and found that the amount of NCOA4-bound FTH1 increased following temozolomide+haloperidol treatment (Fig. 7C). In addition, we knocked down ATG5/7 in GBM cells before temozolomide+haloperidol treatment and measured the Fe²⁺ and ferritin phagocytosis (FTH1 and NCOA4) levels. As shown in Fig. 7D, ATG5/7 siRNA pretreatment inhibited the temozolomide+haloperidol–induced increase in the intracellular Fe²⁺ level. Moreover, both FTH1 and NCOA4 were downregulated in temozolomide+haloperidol combination–treated cells but upregulated in siRNA-treated cells (Fig. 7E and F). We then examined the role of iron in temozolomide+haloperidol–induced ferroptosis in GBM cells by pretreating cells with desferrioxamine (DFO), an iron chelator (27). Pretreatment with DFO and ATG5/7 siRNA markedly inhibited lipid peroxidation as well as PTGS2 expression, elevated the GSH level (Fig. 7H-K), and improved cell viability and colony formation in temozolomide+haloperidol–exposed GBM cells (Fig. 7G; Supplementary Fig. S2C). In addition, the downregulation of NCOA4 and FTH1 was suppressed after treatment with SQ22536 and H89 (Fig. 7L). Taken together, these results suggest that autophagic degradation of ferritin promotes ferroptosis induced by temozolomide+haloperidol in GBM cells.

To further assess the therapeutic potential of temozolomide+haloperidol in GBMs, we evaluated the efficacy of the combination in mice bearing de novo tumors. GL261 murine GBM cells were implanted into the brains of C57BL/6 mice. Consistent with previous observations, the median survival time of GL261 tumor–bearing animals treated with temozolomide was 35 d, compared with 26 d in the control group (P = 0.0229). Treatment with haloperidol alone increased the median survival time from 26 d to 36 d (P = 0.0046). Notably, treatment with the combination of haloperidol and temozolomide significantly extended the median survival time to 53 d, but the median survival time in the combination treatment groups was decreased to only 29 d after the use of DFO. The tumor volume showed a similar pattern after treatment with temozolomide+haloperidol with or without DFO (Fig. 8A–C). To confirm the effects of haloperidol on ER stress and autophagy in vivo, tumor tissue sections were analyzed and showed increases in GRP78+ and LC3-II+ cancer cells (Fig. 8D), consistent with our in vitro observations (Figs. 4C and 5A). Moreover, DFO inhibited intracellular lipid peroxidation by chelating intracellular iron ions without affecting ER stress (Fig. 8E).

To further characterize the status of ferroptosis in temozolomide+haloperidol–treated GBM tumor–bearing mice, we performed TEM analysis of mice treated with temozolomide and haloperidol and observed altered ferroptosis signatures and increased numbers of lysosomes in the temozolomide+haloperidol–treated group (Fig. 8F and G). Taken together, these findings indicate that temozolomide+haloperidol prolonged the median survival time of tumor-bearing mice and exerted a tumor-suppressive effect in vivo. These results also confirmed the molecular mechanism revealed in the in vitro experiments: haloperidol enhances ER stress and temozolomide-induced autophagy to promote ferroptosis by inhibiting the increase in DRD2 expression mediated by temozolomide treatment in GBM cells.

Chemoresistance in GBM is an increasing concern. Importantly, Caragher and colleagues pointed out that dopamine and serotonin signaling are key players in the microenvironmental milieu in which GBM cells originate and proliferate (28). DRD2 is a D2-like dopamine receptor that belongs to the G-protein coupled receptor family and participates in the initiation and progression of GBM (29). An epidemiologic study suggested that GBM cells respond to DRD2 activation at both the gene expression and functional phenotype levels and that specific inactivation of DRD2 itself is required to provide therapeutic benefit (29). Moreover, Caragher and colleagues revealed that therapeutic stress induced by anti-glioma chemotherapy alters the epigenetic status of the DRD2 promoter and subsequently increases DRD2 protein expression (28); in addition, DRD2 activation was found to increase the sphere-forming capacity (30). Importantly, DRD2 expression was significantly elevated in temozolomide-treated samples compared with normal samples, and the DRD2 antagonist haloperidol suppressed GBM proliferation, implying that DRD2 overexpression is the primary cause of acquired resistance in GBM. In addition, He and colleagues revealed that combined treatment with the DRD2/3 antagonist ONC201 and radiation improves the efficacy of radiation against GBM in vitro and in vivo through suppression of glioma-initiating cells (GIC) without increasing toxicity (31). In this study, we demonstrated that combined treatment with the DRD2 antagonist haloperidol and temozolomide could be an effective therapeutic strategy for GBM, as this combination not only targets GBM cells but also prevents the acquisition of a more chemoresistant phenotype induced by temozolomide in GBM cells.

Ferroptosis is a recently identified form of oxidative cell death characterized by iron-dependent lipid peroxidation and is implicated in many pathological processes, including cancer, ischemia‒reperfusion injury, infection, and neurodegenerative diseases (32). Previous studies indicated that inhibition of ferroptosis can sensitize the effects of chemotherapeutic treatment with temozolomide (33). Our results demonstrated that haloperidol exposure induced ferroptosis in a dose-dependent manner in vivo and in vitro, consistent with a previous study showing that inhibition of lipid peroxidation by Fer-1 alleviated haloperidol-induced cell death in Huh-7 cells (34). Chen and colleagues revealed that dopamine and D1 receptor agonist SKF R-38393 increased nitric oxide (NO) production and induced nitrite generation in SK-N-MC cells (35). This suggests that haloperidol also promotes the production of lipid peroxidation by affecting NO production. In addition, in combination with temozolomide, haloperidol significantly promoted the ferroptosis of GBM cells compared with treatment with either temozolomide or haloperidol alone. This observation supports the important role of iron-dependent lipid peroxidation or ferroptosis in haloperidol- and temozolomide-induced toxicity in glioma cells.

Notably, numerous recent studies have emphasized the importance of autophagy as an emerging mechanism of ferroptosis and have provided new insights into regulated cell death. For example, Gao and colleagues revealed that knockout of ATG13 and ATG3, two key players in autophagy, greatly reduced the sensitivity of MEFs to ferroptosis induced by erastin or cystine starvation, and reconstituting ATG13 and ATG3 in these cells restored their ferroptosis sensitivity (7). Moreover, Park and colleagues noted that inhibition of autophagy disrupted erastin-triggered cell death (36). These results suggest that autophagy is an important player in ferroptosis. As expected, our results show that autophagic activity is upregulated in GBM cells exposed to haloperidol and is enhanced even further when haloperidol is combined with temozolomide. Then, we examined the role of autophagy in haloperidol-induced ferroptosis and found that inhibition of autophagy by 3-MA and Baf-A1 limited haloperidol + temozolomide–induced ferroptosis in glioma cells. The same results were obtained after using the autophagy-related gene ATG5/7 siRNA. These results indicate that temozolomide+haloperidol–induced ferroptosis is an autophagy-dependent process and that autophagy is a key activator of temozolomide+haloperidol–induced cell death.

Oxidative stress and lipid peroxidation are associated with perturbed proteostasis, termed ER stress, which is a cascade of adaptive pathways that aim to maintain cellular homeostasis and normal ER function (37–39). D2-like dopamine receptors can inhibit cAMP and PKA by inhibiting adenylate cyclase activity (40). Several studies have shown that the cAMP/PKA signaling pathway can activate ER stress (41–43). In PDAC, inhibiting DRD2 results in anticancer activity via activation of the cAMP/PKA pathway, regulating Ca²⁺ levels and subsequently increasing ER stress, thus inducing apoptosis (16). Here, we indicate that haloperidol activates PERK-mediated ER stress and that temozolomide+haloperidol treatment further enhances ER stress. Moreover, we demonstrated that temozolomide+haloperidol induces ER stress by modulating the cAMP/PKA pathway. Furthermore, ER stress is related to autophagy in various ways, including ER stress–mediated autophagy activation and the formation of autophagosomes at the ER (44–46). Interestingly, increased ER stress was associated with autophagy and haloperidol-induced toxicity, consistent with previous studies reporting that DRD2 blockade has an antiproliferative effect in pancreatic cancers while activating ER stress (16).

Certain forms of autophagy, especially NCOA4-mediated ferritinophagy, RAB7A-mediated lipophagy, SQSTM1-mediated clockophagy and HSP90-mediated chaperone-mediated autophagy, can facilitate ferroptosis through lipid peroxidation by increasing the iron load and impairing the antioxidant system (24, 47, 48). Moreover, Zhang and colleagues revealed that ZFP36 overexpression can result in ATG16L1 mRNA decay, thus triggering autophagy inactivation, blocking autophagic ferritin degradation, and eventually conferring resistance to ferroptosis (49). NCOA4 is a selective cargo receptor for autophagic degradation of ferritin (50). The NCOA4-ferritin complex binds to autophagy-associated genes, which ultimately results in degradation of ferritin and the release of free iron through the autophagic pathway (26). The expression of both NCOA4 and FTH1 was reduced and the iron ion content was increased in U87MG and LN229 cells after treatment with temozolomide and haloperidol. In addition, inhibition of ferritinophagy and iron accumulation alleviated temozolomide+haloperidol–induced ferroptosis in U87MG cells. These findings indicate that haloperidol-induced ferritinophagy triggers FTH1 degradation and subsequent ferroptosis mediated by iron accumulation. These results may explain the excessive activation of autophagy during haloperidol- and temozolomide-induced toxicity. However, we cannot exclude the possibility that other forms of autophagy may contribute to haloperidol- and temozolomide-induced ferroptosis in glioma cells.

Originally used to treat schizophrenia, haloperidol is a butyrophenone psychostimulant known as a high-affinity dopamine antagonist. Clinical data have shown that long-term use of tricyclic antidepressants (TCA) is associated with a decreased incidence of glioma (51). The awareness of haloperidol as a TCA has led to an increasing number of studies on its molecular mechanisms in glioma. Furthermore, the fact that oral haloperidol undergoes first-pass metabolism, resulting in systemic distribution, and that it remains at high levels in human brain tissue after crossing the blood-brain barrier without causing toxic effects on normal glial cells has prompted us to investigate haloperidol's role in tumor therapy. In this study, we conclude that the dopamine receptor antagonist haloperidol enhances the radiosensitivity of GBM cells and amplifies temozolomide-induced ER stress by activating the cAMP signaling pathway, which in turn activates autophagy and thus mediates ferroptosis, resulting in effective inhibition of glioma cell survival and malignant progression (Fig. 9). Our research not only provides new insights into the mechanism of action of haloperidol but also merits consideration for future clinical trials with good translational prospects.

No disclosures were reported.

L. Shi: Data curation, software, formal analysis, supervision, writing–original draft, project administration, writing–review and editing. H. Chen: Data curation, software, formal analysis, supervision. K. Chen: Data curation, software, visualization. C. Zhong: Resources, data curation, software, formal analysis. C. Song: Supervision, funding acquisition, validation. Y. Huang: Conceptualization, resources, data curation, software. T. Wang: Formal analysis, supervision, validation. L. Chen: Investigation, visualization, methodology. C. Li: Investigation, visualization, methodology. A. Huang: Methodology, writing–original draft, project administration, writing–review and editing. S. Qi: Writing–original draft, project administration. H. Li: Project administration, writing–review and editing. Y. Lu: Writing–original draft, project administration, writing–review and editing.

This work was supported by the General Programs from the National Natural Science Foundation of China (81902544 and 82073305 to H. Li; 81972355 to Y.T.L.), the Natural Science Foundation of Guangdong Province (2020A1515010063 and 2021A1515010711 to H. Li), the Grant for Recruited Talents to Start Scientific Research from Nanfang Hospital, the Outstanding Youth Development Scheme of Nanfang Hospital, Southern Medical University (2020J006 to H. Li), the Guangdong Basic and Applied Basic Research Foundation (2019B151502048 to Y.T.L.), the National Key Clinical Specialty Project, the Dalian Medical Science Research Program Project (2111001 to C. Song), and the Dalian Key Medical Specialty Dengfeng Project(2022ZZ211 to C. Song).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.