Clinical tissues (Supplementary Table S1) were collected from patients with NF (18 cases) and MPNST (12 cases) at the Department of Neurosurgery at Beijing Tiantan Hospital Affiliated with Capital Medical University (Beijing, China), between 2019 and 2023. For four of the MPNST patients, adjacent non-tumor tissues (referred to as peritumor) were also obtained. The inclusion criteria were tumor specimens resected from patients initially diagnosed with NF or MPNST. The exclusion criteria were patients initially diagnosed with non-NF or non-MPNST conditions, or those with recurrent tumors. Human tissue samples were gathered from every patient who had offered written informed consent. The standardized procedures for human samples collection and utilization were approved by the Internal Review Board of Beijing Tiantan Hospital Affiliated with Capital Medical University (KYSQ 2025-334-01). Clinical trial number: not applicable.

The human plexiform neurofibroma cell line ipNF05.5 (RRID: CVCL_UI71) and MPNST cell line sNF96.2 (RRID: CVCL_K281), purchased from the American Type Culture Collection (Manassas, VA, United States), were cultured in DMEM (Gibco) and RPMI 1640 (Gibco) medium respectively, supplemented with 1% penicillin and streptomycin and 10% foetal bovine serum (Gibco). The cells were cultured at 37°C in a humidified, 5% CO₂ environment. Both human cell lines have been authenticated using STR (or SNP) profiling, and all experiments were performed with mycoplasma-free cells.

Tissue samples embedded in paraffin were precisely sectioned into 4 mm sections. These sections were then meticulously dewaxed and hydrated. Afterwards, they were incubated with following primary antibodies at 4°C overnight. TSPO (ABclonal, A4881, 1:200), FTH1 (Abcam, ab287968, 1:100), and AKR1C1 (Abcam, ab192785, 7.5 μg/mL) antibodies. The signals were visualized and images were analyzed as described previously (Zhang et al., 2021).

The total RNA was isolated from tissues cryopreserved in liquid nitrogen or cells and reverse transcribed into cDNA as described previously (Zhang et al., 2022). The results underwent normalization against the expression of glyceraldehyde-3-phosphate dehydrogenase. The sequences of the GAPDH, TSPO, AKR1C1, FTH1, GPX4, TP53, P21 and NRF2 primers used in this study are presented in Supplementary Table S2.

Proteins were isolated from tumor tissues cryopreserved in liquid nitrogen or cells and protein concentrations were quantified as described previously (Zhang et al., 2021). 100 μg lysate was subjected to separation on a 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel. Following electrophoretic separation, the proteins were transferred onto polyvinylidene difluoride membranes. Subsequently, the membranes were incubated with following primary antibodies during an overnight incubation at 4°C. TSPO (ABclonal, A4881, 1:1000), GPX4 (Proteintech, 67763-1-Ig, 1:1000) and β-actin (Invitrogen, A1978, 1:10000) antibodies. Subsequently, the membranes were incubated with a peroxidase-conjugated secondary antibody (Cell Signaling Technology, 1:10000) for 1.5 h. The signal was visualized through the application of enhanced chemiluminescence reagent (Millipore).

We used an Fe²⁺ indicator (FeRhoNox-1, MKbio, MX4558) to detect ferrous iron in tumor cells as described previously (Zhang et al., 2022). Cells were incubated for 1 h at 37°C with FeRhoNox-1. The cells were then harvested by trypsinization, and the level of ferrous iron was determined by imaging using a confocal microscope or by flow cytometry analysis.

The extraction of lipids from approximately one million cells was conducted utilizing an adapted Bligh and Dyer’s method as described previously (Song et al., 2020). The lipidomic analysis was carried out at LipidALL Technologies by utilizing an ExionLC-AD-Sciex QTRAP 6500 PLUS mass spectrometer as previously reported (Lam et al., 2021). The separation of distinct lipid classes within oxidized phospholipids was accomplished through normal phase-high performance liquid chromatography. A TUP-HB silica column (i.d. 150 × 2.1 mm, 3 μm) was employed for this separation process. The chromatographic conditions were as follows: mobile phase A consisted of a mixture of chloroform, methanol, and ammonium hydroxide in a ratio of 89.5:10:0.5, and mobile phase B comprised chloroform, methanol, ammonium hydroxide, and water in a ratio of 55:39:0.5:5.5. Multiple reaction monitoring transitions were configured to facilitate the comparative analysis of diverse oxidized phospholipids. Quantification of individual lipid species was achieved by referencing the responses against spiked internal standards.

We used a reduced glutathione assay kit to measure the reduced GSH as described previously (Zhang et al., 2021). The cells were collected, washed twice with ice-cold PBS, lysed by freezing and thawing twice in liquid nitrogen at 37°C, and centrifuged. The supernatant was collected and used for the assays. GSH standard solutions and samples were loaded into 96-well plates. Next, a buffer solution was added to each well, and the plate was incubated at 25°C for 5 min. NADPH solution was then added to each well and the plate was incubated at 25°C for 30 min, and the absorbance was read at 412 nm using microplate reader. The GSH concentrations were normalized to the total protein levels in each sample.

Unless explicitly indicated otherwise, all experimental procedures were independently replicated three times. Statistical analyses were conducted utilizing the GraphPad Prism software (version 8.0), a widely recognized and validated tool in the field of data analysis. The experimental data are presented as the mean ± standard deviation (SD). For the comparison of two groups, the Student’s t-test for unpaired data was applied. Comparisons among more than two groups were performed using ANOVA and Tukey’s test. For all statistical tests conducted in this study, a significance level of p ≤ 0.05 was adopted, and results with p values below this threshold were denoted with an asterisk (*p < 0.05) to indicate statistical significance.

To explore TSPO expression in tumors, we analyzed data from the TIMER database and found that TSPO mRNA expression varied depending on tumor type, with notably high expression in liver cancer and comparatively low expression in lung cancer (Supplementary Figure S1A). We performed RT-qPCR and western blotting on four pairs of MPNST tissue samples, with each pair originating from a single patient. The results showed that TSPO protein expression in MPNST tissues was lower than that in the adjacent non-tumor tissues (Figures 1A,B and Supplementary Figure S1B). We further validated the decreased expression of TSPO in MPNST clinical specimens using IHC staining (Figure 1C). Moreover, we extracted RNA from 18 benign and 12 malignant patient tissues and detected low TSPO expression in MPNST tissues using RT–qPCR (Figure 1D). IHC staining also confirmed that TSPO was expressed at low levels in malignant tumor samples (Figure 1E). These findings indicate that TSPO is downregulated in malignant tumor specimens.

To investigate the relationship between TSPO and ferroptosis in MPNST, we used RT-qPCR to detect the differential mRNA expression levels of two classic ferroptosis biomarkers, AKR1C1 and FTH1, in tissues from patients with benign and malignant tumors. Compared with benign specimens, ferroptosis biomarkers were downregulated in malignant tumors (Figures 2A,B), which was consistent with the anti-death characteristics of tumor cells. Notably, TSPO expression positively correlated with ferroptosis markers (Figures 2C,D). IHC staining showed that malignant tumors with high TSPO expression had significantly higher AKR1C1 and FTH1 expression levels. Conversely, in tumor tissues with low TSPO expression, AKR1C1 and FTH1 are similarly found to have low expression (Figure 2E). These findings indicate that TSPO may play a vital role in regulating ferroptosis in MPNST.

To address the putative role of TSPO in the ferroptosis, we first established ferroptosis models using a ferroptosis inducer (1S,3R-RSL3) in the MPNST cell line sNF96.2. In alignment with prior reports (Zhang et al., 2021), the cell death triggered by RSL3 was fully reversed by Fer-1, a specific inhibitor of ferroptosis, but not by Z-VAD-FMK (an inhibitor of apoptosis) (Supplementary Figure S2A). The MPNST cells sNF96.2 were more resistant to ferroptosis induced by various concentrations of RSL3 than neurofibroma cells ipNF05.5 (Supplementary Figure S2B). The effect of TSPO knockdown on ferroptosis was assessed by trypan blue staining. As shown in Figure 3A, TSPO deficient tumor cells are insensitive to ferroptosis. The inhibition of ferroptosis by TSPO deficiency was supported by the measurement of ferroptosis markers using RT-qPCR (Figure 3B, and Supplementary Figure S2C). In addition, TSPO knockdown decreased the levels of intracellular Fe²⁺ and ROS, as measured using FeRhoNox-1 and DCFH-DA, respectively (Figures 3C–F). These results suggest that TSPO deficiency attenuates ferroptosis in human MPNST cells.

Ferroptosis is fundamentally orchestrated by the peroxidation of phospholipids bearing polyunsaturated fatty acids, leading to the deleterious accumulation of lipid peroxides on cellular membranes and eventually resulting in membrane disruption (Lei et al., 2022). To understand how loss of TSPO provides protective effects, we analyzed the difference in oxidized phospholipids in wild-type (WT) and TSPO knockdown sNF96.2 cells by liquid chromatography-mass spectrometry. TSPO-knockdown cells exhibited distinctive oxidized phospholipid (PL) profiles that differed from those of WT cells (Supplementary Figure S3A). Phosphatidylethanolamine (PE) and oxidized PE levels significantly decreased in TSPO-knockdown cells (Figures 4A,B, and Supplementary Figure S3B). Although TSPO deficiency increased phosphatidylcholine (PC) levels, it reduced oxidized PC levels, which play a crucial role in ferroptosis (Figures 4C,D, and Supplementary Figure S3B). AA (20:4) and AdA (22:4), which are present in phospholipids, are the preferred substrates for oxidation (Doll et al., 2017). A detailed analysis of the potentially oxidized phospholipids was performed. We found that the respective oxidized species were more abundant in the WT cells, especially the hydroperoxy-PE (18:1/22:4) and hydroperoxy-PC (18:1/22:4) molecular species (Figures 4E,F). Overall, these findings indicate that TSPO deficiency is accompanied by a substantial decrease in oxidized phospholipids, which are used as substrates to trigger ferroptosis.

To further elucidate the mechanisms by which TSPO deficiency inhibits ferroptosis, transcriptome analysis was performed in WT and TSPO-knockdown ipNF05.5 and sNF96.2 cells, with three replicates for each group. We identified 13 genes by intersecting the differentially expressed genes in each group (shTSPO vs. shCtrl) with 485 ferroptosis-related genes from the FerrDB database (Figure 5A). Among these, the GPX4 mRNA levels were significantly increased in the TSPO-knockdown groups (Figure 5B). GPX4 is the key antioxidant and ferroptosis defence system that convertes PL hydroperoxides to PL alcohols to protect cells against ferroptotic death. We further validated the marked upregulated in TSPO-knockdown cells by using RT-qPCR and western blotting (Figures 5C,D). Furthermore, GPX4 is elevated in malignant tumors compared to benign specimens and is negatively correlated with TSPO expression in tumor specimens (Figure 5E and Supplementary Figure S4). TSPO deficiency increased the levels of reduced glutathione (GSH), a cofactor used by GPX4 (Figure 5F). These observations suggest that TSPO deficiency blocks the ferroptotic death of MPNST cells by upregulating the GPX4 antioxidant system and subsequently decreasing lipid peroxidation.