Background

Glioma represents the prevailing primary malignant neoplasm within the central nervous system tumors, with glioblastoma (GBM) exhibiting the most severe degree of malignancy and the worst prognosis for patients, as evidenced by a 5-year survival rate below 10%.1 2 In spite of the implementation of an extensive, conventional treatment regimen that encompasses surgical intervention, radiation therapy, chemotherapy, and TTF, patients are observed to have a median survival period of merely 12–14 months subsequent to the initial diagnosis.2 3 The imperative objective for clinical researchers is to identify efficacious treatment modalities and enhance the prognosis of patients.

Recently, the immunotherapy of glioma has received extensive attention, but there is no effective immunotherapy for GBM so far. Therefore, elucidating the underlying mechanism of glioma immune microenvironment regulation can provide new therapeutic strategies for GBM. Immunotherapy, especially targeting programmed death receptor 1 (PD-1)/PD-1 ligand (PD-L1) for immune checkpoint blockade, has brought about significant transformations in the management of various tumors.4 However, the clinical effect of targeting PD-1/PD-L1 in patients afflicted with GBM is still limited. Most patients suffering from GBM exhibit resistance to pharmaceutical interventions that specifically target PD-1/PD-L1, except for a limited subset of patients who demonstrate a positive response to immunotherapeutic approaches using this pathway.5 The response to PD-1/PD-L1 blockade was found to be related to PD-L1 expression levels in tumor cells.6 7 Moreover, PD-L1 protein levels were found to be related to glioma grades, and higher tumor cell PD-L1 levels contribute to immune evasion in glioma patients.8 9 To enhance effective strategies for glioma immunotherapy, the mechanisms behind PD-L1 regulation require further investigation.

OMA1 is a metalloprotease situated within the inner mitochondrial membrane and encoded by the OMA1 gene. Studies have shown that OMA1 can be triggered and activated by exogenous stimuli or cellular stress. OMA1 possesses the ability to cleave and degrade substrate proteins, including but not limited to OPA1, DELE1, and PINK1, owing to its protease activity. This enzymatic action allows OMA1 to modulate the morphology and functionality of mitochondria, consequently leading to alterations in various cellular biological processes, thereby resulting in a range of disorders.10 11 The upregulation of OMA1 in neoplastic cells has a correlation to a significant enhancement in both its expression and activity. This upregulation contributes to excessive cell proliferation and expedites tumor initiation and progression through its governing of mitochondrial function and cellular metabolism.12 In a study conducted by Thomas Langer and colleagues, it was observed that the integrated stress response, mediated by OMA1, plays a protective role toward ferroptosis in cases of mitochondrial cardiomyopathy.13 In GBM, further research is needed to understand how OMA1 affects immunosuppression, particularly PD-L1 expression.

Herein, our results indicated that OMA1 is related to immunosuppression and overexpressing PD-L1 in PD-1 inhibitor-resistant GBM. Further, we found that the expression of OMA1 was increased, which inhibited the formation of IP3R/HSPA9/VDAC1 complex by competitively binding with IP3R to HSPA9, resulting in the blockage of mitochondrial TCA cycle. The impaired mitochondrial function will promote mitophagy, thereby increasing mitochondrial DNA release, activating the cGAS–STING signaling pathway, and overexpressing PD-L1. PD-L1 is recognized by PD-1 of CD8+ T cells and contributes to immune escape (IE).

Materials and methods

Clinical samples

Surgical samples of GBM were obtained from Wuhan Union Hospital (Wuhan, China). Online supplemental table S1 provides details on the patient features. Each patient signed an informed consent form before specimen collection. All patients had not received chemoradiotherapy before surgery. This research followed the Declaration of Helsinki, and the necessary ethical authorizations were acquired.

Cell culture and treatment

For the detailed extraction and culture methods of GBM primary cells, please refer to the previous article.14–16 All cells underwent short tandem repeat analysis and were consistently screened for mycoplasma contamination. For more detailed steps on extraction, culture, and passage of GBM primary cells, please refer to the supplementary materials.

Plasmids and siRNAs

GeneChem (Shanghai, China) synthesized the lentiviral overexpression plasmids for OMA1 and HSPA9, as well as the lentiviral knockdown plasmids for OMA1 and c-GAS. Online supplemental table S2 lists the used sequences of shRNAs and siRNAs. PCR was used to amplify and clone the truncated structure cDNAs of human OMA1 (NM_145243.5) into the pECMV-3xFLAG-C expression vector. For transient transfections, Lipofectamine 3000 from TermoFisher Scientific in the USA was used per the protocols. Subsequently, colonies expressing cells were chosen by applying 2 mg/mL puromycin for stable expression.

Western blotting (WB) and antibodies

Our previous studies published the related protocol details.14–17 Online supplemental table S3 provides details on all antibodies. For more detailed experimental procedures, please refer to the supplementary material.

Nuclear and cytoplasmic extraction, RNA extraction, and qRT-PCR

An RNA purification kit (AM1921, Thermo Fisher Scientific) was used to separate nuclear and cytoplasmic RNA. The total RNA extraction from tissues or cell lines was conducted through Trizol (Takara, Otsu, Japan) per the protocols. With SYBR Green premix Pro Taq HS qPCR Kit (AG11728, Accurate Biotechnology, Hunan, China), the reverse transcription of 1 µg of total RNA into cDNA was performed using Evo M-MLV RT Kit. Real-time PCR reactions were carried out using the BioRad system (BioRad, USA). The internal reference GAPDH was employed to normalize the RT-qPCR results, calculating the relative expression via the 2(−ΔΔCT) method. Online supplemental table S4 lists the primer sequences.

Transmission electron microscopy

Briefly, GBM primary cells were plated in T25 culture flasks and treated for 48 hours with different treatments. After centrifugation and precipitation, cells are collected. The cells went through resuspension in IEM fixative followed by fixation at 4°C for preservation. An electron microscope (Hitachi, HT7700, Japan) was used to obtain the images.

Bioinformatics analysis

The TCGA (https://cancergenome.nih.gov/) was accessed to download transcript data of glioma samples and clinical information while accessing the Genotype-Tissue Expression (https://gtexportal.org/home/) to download the transcript data of normal brain tissues.18 For more detailed experimental procedures, please refer to the supplementary material.

Co-immunoprecipitation (Co-IP)

We conducted Co-IP assays, as reported earlier.14 15 19 We first enriched mitochondrial proteins before performing immunoprecipitation to study the interaction between OMA1 and HSPA9. In brief, Protein A+G magnetic beads were used to hatch the indicated cell lysates for a whole night at 4°C with specific primary antibodies. Employing the corresponding antibodies, the immunocomplexes were then analyzed via immunoblotting after being rinsed with lysis buffer. Online supplemental table S3 provides detailed information on the antibodies used herein.

Determination of total cellular ATP

Aspirate the culture medium, add lysis liquid at a ratio of 200 µL to each well of the 6-well plate (equivalent to 1/10 of 2 mL of cell culture medium), and lyse the cells. When lysing cells, in order to fully lyse them, you can use a pipette to repeatedly pipet or shake the culture plate to fully contact the lysis solution and lyse the cells. Normally cells lyse immediately after contact with the lysis buffer. After lysis, centrifuge at 12 000g for 5 min at 4°C, and take the supernatant for subsequent determination. Next, strictly follow the instructions of the ATP Assay Kit (YT361) to prepare the measurement of the standard curve, the preparation of the ATP detection working solution, and the measurement of the ATP concentration. For more detailed experimental procedures, please refer to the supplementary material.

Immunofluorescence (IF) and immunohistochemistry (IHC) staining

The study performed IF and IHC assays, as reported earlier.17 19 20 In brief, sectioned human or mouse tissue specimens were subjected to fixation in 4% paraformaldehyde, embedding in paraffin, and immunostaining with specific antibodies. The IF staining assays used 4% paraformaldehyde for 15 min at room temperature, 0.5% Triton X-100 for 10 min, and 5% BSA for 1 hour. Afterward, secondary antibodies with fluorine labels (1:200, Thermo Fisher Scientific) were applied. Then we stained nuclei with DAPI (C1002, Beyotime) using a fluorescent microscope (Nexcope NE930, Ningbo, China). Online supplemental table S3 lists the corresponding antibodies used herein.

Cell counting kit‐8 (CCK‐8) assay

The first step is to prepare the cell suspension. Cells were collected by trypsin digestion and centrifugation. The collected cells were resuspended in serum-containing medium, counted using a hemocytometer, and then diluted to a single cell suspension of 5×103–5×104 cells/mL. The second step is to inoculate the cell suspension in a 96-well plate, 100 µL per well, and design different concentration groups. Each concentration group can be designed with 3–6 duplicate wells and set up a blank group and a control group. The third step is to put the culture plate into the incubator and preculture it for about 24 hours (37℃, 5% CO2). The fourth step is to add different concentrations of lentivirus to each well of the culture plate and place it in the incubator for 6–96 hours. Step 5: Add 10 µL of CCK-8 solution to each well. After adding the reagent, gently shake the culture plate to help mix (to prevent errors caused by CCK-8 reagent sticking to the wall of the well). Try as much as possible during the adding process. Do not create bubbles to avoid affecting the OD value reading. Then put the culture plate into the incubator and incubate it for 1–4 hours. Finally, the absorbance (OD) at 450 nm was measured using a microplate reader.

Colony formation assay

To assess the colony formation ability of each cell line, the following experimental procedure was carried out. In short, cell plating, incubation, fixation, staining, colony counting and photography and replicate experiments. For more detailed experimental procedures, please refer to the supplementary material.

Confocal microscopy

The HBAD-EGFP-LC3 and HBAD-h-mito-dsRed adenoviral particles were obtained from HanBio (Shanghai, China). The cells were cultured for another 24 hours after infection with adenoviral particles. Following three PBS washes, the glioma cells were incubated at 37°C for 1 hour in the dark with 4% paraformaldehyde. After that, the sections were mounted with VECTASHIELD Antifade Mounting Medium containing DAPI (H-1200, Vector Laboratories, Burlingame, California, USA). As a final step, samples were imaged using a Nikon A1+/A1R+ confocal laser microscope (Nikon, Tokyo, Japan).

Xenograft model

In view of the advantages of fast reconstruction, high conversion rate, and low cost of the humanized mouse immune reconstitution model of PBMC, the authors used this model in this study to explore and verify the role of the main molecule OMA1 in tumor immunity. Briefly, following the random categorization into distinct groups, female NOG-dKO mice (aged 6–8 weeks, n=5/group) went through anesthesia. Subsequently, a suspension of 5 µL GBM cells (5×106 cells) (with indicated treatment) was injected into the mouse brain with the stereotaxic device. Tumor size estimation was conducted using the formula V=(D×d2)/2, where D and d refer to the longest and shortest diameter, respectively. For more detailed experimental procedures, please refer to the supplementary material.

Statistical analysis

The study employed SPSS V.25.0 (SPSS, Chicago, Illinois, USA) and GraphPad Prism (V.8.0; GraphPad, La Jolla, California, USA) for performing all statistical analyses. For more detailed experimental procedures, please refer to the supplementary material.

ResultsOMA1 promotes GBM progression and is closely associated with prognosis

Clinically, we found that some patients with GBM are sensitive to PD-L1 inhibitors, while others are resistant. In order to further elucidate the molecular mechanism behind this phenomenon, we collected tumor samples from these patients and carried out primary isolation and culture. The protein chip technology was used to screen the differential proteins between the two. The heatmap shows the top 15 differentially expressed proteins in PD-1 inhibitor sensitive or resistant GBM (online supplemental figure S1A). Subsequently, we used methods such as bioinformatics analysis and the novelty of candidate molecules, and finally, we locked the research molecule OMA1. Then, the study used WB for validating OMA1 expression in PD-1 inhibitor-sensitive or resistant GBM, revealing that OMA1 was overexpressed in PD-1 inhibitor-resistant GBM (online supplemental figure S1B). From TCGA GBM database results, we found OMA1 was upregulated in glioma, was proportional to tumor grade, and was inversely proportional to the patient’s prognosis (online supplemental figure S1C,D). Herein, about online supplemental figure S1D, we only use two standardized methods, FPKM and TPM, for quantitative grouping, and adopt different statistical assumptions. Currently, TPM is more recommended. These two pictures actually express the same meaning. Aiming to verify OMA1 biological function, we knocked down and overexpressed OMA1, respectively, followed by employing qRT-PCR and WB for validating the efficiency of knockdown and overexpression (online supplemental figure S1E). CCK-8, colony formation experiments, and EdU experiments all showed that overexpression of OMA1 (oeOMA1) promoted GBM proliferation while knocking down OMA1 had the opposite effect (figure 1A–C and online supplemental figure S2A–C). Similarly, animal experiments on intracranial orthotopic tumorigenesis of glioma showed that oeOMA1 promoted GBM growth while knocking down OMA1 had the opposite effect. The results of KI67 IF-staining on the tumor specimens of the above two groups of animals also confirmed the above results again (figure 1D and online supplemental figure S1F and S2D). The above results indicate that OMA1 can promote the progression of GBM.

Figure 1

OMA1 promotes the growth of GBM. (A–C) The results of CCK-8, clone formation assays, and EdU assays all showed that oeOMA1 promoted the proliferation of GBM primary cells. The right side represents a typical histogram. (D) The results of animal experiments on orthotopic tumorigenesis of glioma also showed again that OMA1 promotes the intracranial growth of GBM. Representative Ki-67 immunofluorescent staining and corresponding histograms of intracranial tumor specimens of mice in various animal experiments (n=5). Scale bar, 50 µm. The means±SDs are provided (n=5). *p<0.05, **p<0.01, and ***p<0.001 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; CCK-8, cell counting kit‐8; GBM, glioblastoma; oeOMA1, overexpression of OMA1.

OMA1 mediates IE in GBM

In order to further explore whether OMA1 can mediate the IE of GBM, we cocultured primary GBM cells overexpressing or knocking down OMA1 with CD8+ T cells in vitro. The number of surviving primary cells with OMA1 overexpressing increased, whereas the surviving number of GBM primary cells with OMA1 knockdown decreased (figure 2A and online supplemental figure S3A). Consequently, CD8+ T cell growth and IFN-γ, TNF-α, and Gzmb expressions were determined by employing flow cytometry and qPCR. The outcomes exhibited that CD8+ T cells exhibited lower TNF-α, IFN-γ, and Gzmb expression levels and proliferation in the oeOMA1 groups (figure 2B–E). However, the OMA1 knockdown groups showed that CD8+ T cells exhibited higher TNF-α, IFN-γ, and Gzmb expression levels and proliferation (online supplemental figure S3B–E). The following procedures were followed to establish an intracranial orthotopic tumor model in a nude mouse: The primary GBM cells were implanted orthotopically into the mouse brain while being subjected to various treatment conditions. Following a 15-day period, the activated CD8+ T cells that were isolated from healthy human peripheral blood were administered through the tail vein at intervals of 3 days, revealing that mice subjected to oeOMA1 injection showed a significantly increased tumor size more than the control. Yet, OMA1 knockdown groups showed that tumor size significantly decreased. As anticipated, the IF analysis of Ki-67 in the transplanted animal tumor samples revealed that the oeOMA1 groups exhibited a higher Ki-67 expression than the mice in the Vector groups. However, OMA1 knockdown groups showed weakened expression of Ki-67 (figure 2F,G and online supplemental figure S3F,G).

Figure 2

OMA1 promotes immune escape in GBM. (A) oeOMA1 significantly inhibited T cell-mediated tumor cell killing compared with Vector. (B–E) Flow cytometry and qPCR results indicated that CD8+ T cells cocultured with GBM primary cells overexpressing OMA1 showed lower proliferation and expression of IFN-γ, TNF-α, and Gzmb. (F) Schematic diagram of animal experiments. (G) Typical animal experiment tumor slices, Ki-67 IF pictures, and typical column chart in different groups. Scale bar, 50 µm. The means±SDs are provided (n=5). *p<0.05 and **p<0.01 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; GBM, glioblastoma; oeOMA1, overexpression of OMA1.

Furthermore, ELISA outcomes revealed that CD8+ T cells supernatants that were cocultured with GBM primary cells overexpressing OMA1 exhibited reduced IFN-γ, TNF-α, and Gzmb secretion levels, while the OMA1 knockdown groups exhibited the opposite effect (online supplemental figure S4A). Moreover, the IF of CD8 on the transplanted animal tumor samples in each group revealed that the proportion of CD8+ cells in the oeOMA1 groups was significantly lower than in the control. However, OMA1 knockdown groups showed higher expression of CD8 (online supplemental figure S4C).

OMA1 promotes tumor IE via mitophagy

Mitochondria are organelles that contribute to multiple functions, including energy metabolism, cell signal regulation, and apoptosis in eukaryotes.21 Mitophagy controls mitochondrial mass and maintains cellular homeostasis by selectively degrading excess or damaged mitochondria.22 23 Mitophagy is involved to a great extent in regulating immune-related diseases, including tumors,24 25 neurodegenerative diseases,26 27 and cardiovascular diseases.28 29 Interestingly, it has been reported that BNIP3L, a key protein of mitophagy, is highly expressed in glioma, so it is speculated that the mitophagy pathway is activated in glioma.30 31 Nevertheless, the mitophagy and tumor IE association remains unelucidated. Therefore, we wanted to explore whether mitophagy possesses a regulatory role in the process of OMA1 promoting IE in glioma.

Next, in the GBM primary cells overexpressing OMA1, the mitophagy markers PINK1, p-Parkin, BNIP3, and BNIP3L were detected by WB, and the expressions were all increased. In addition, WB was employed for detecting the expression levels of marker proteins LC3-II/I of autophagy flow, and it was observed that the LC3-II/I ratio increased. Additionally, to further confirm the alterations in autophagy flux, the inhibition of lysosomal degradation by bafilomycin A1 should be employed (figure 3A). By combining TOM20 and LC3 fluorescent signals, OMA1 enhanced autophagosome-mitochondrion colocalization (figure 3B,C). PINK1/Parkin represents a key signaling pathway mediating mitophagy in mammals, and it is involved in autophagosome formation.32 After overexpressing OMA1, PINK1 and Parkin expression were detected. As shown, oeOMA1 enhanced PINK1, p-Parkinser65, and LC3 expressions in GBM primary cells (figure 3D). The Parkin translocation to mitochondria represents a well-known mitophagy hallmark.33 Therefore, we proceeded to investigate this translocation in flubendazole-treated cells through analysis of cellular fractionations. Just like expectation, Parkin exhibited enrichment in the mitochondria fraction in oeOMA1 groups (figure 3E). Consistently, the aforementioned findings were additionally corroborated by the heightened level of PINK1 and Parkin colocalization in oeOMA1 groups (figure 3F). Based on these findings, OMA1 triggers mitophagy through PINK1/Parkin signaling in GBM.

Figure 3

OMA1 promotes mitophagy via PINK1/Parkin signaling. (A) The results of WB experiments showed that oeOMA1 promoted the expression of PINK1, p-Parkinser65, LC-3, BNIP3, and BNIP3L. (B, C) The autophagosomes are labeled by LC3 (green fluorescence) protein, and the mitochondria are labeled by TOM20 (red fluorescence) protein. The number of colocalized LC3 and TOM20 was quantified. Scale bar, 50 µm. (D) WB of PINK1, Parkin, p-Parkinser65, and LC3 in GBM primary cells treated with the indicated concentrations of OMA1 for 24 hours. (E) WB of Parkin in the cytosolic and mitochondrial fractions of GBM primary cells treated with or without oeOMA1 for 24 hours. β-Actin (cytoplasmic fraction) and VDAC1 (mitochondrial fraction) were used as the loading controls. (F) Colocalization of PINK1 (green fluorescence) protein and Parkin (red fluorescence) protein in GBM primary cells following oeOMA treatment or not. The number of colocalized PINK1 and Parkin was quantified. Scale bar, 50 µm. The means±SDs are provided (n=3). **p<0.01 and ***p<0.001 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; GBM, glioblastoma; NS, not statistically significant; oeOMA1, overexpression of OMA1; WB, Western blotting.

Furthermore, we assessed the influences of OMA1 on the mitochondrial permeability transition pore (mPTP) opening in GBM primary cells. In comparison with the Vector groups, the oeOMA1 groups significantly decreased Calcein AM fluorescence intensity, indicating an accelerated rate of opening of mPTP (online supplemental figure S5A). Permeability in the mitochondria’s outer membrane is often accompanied by morphological changes and dysfunction.34 To assess OMA1-induced mitochondrial changes in GBM primary cells, we analyzed mitochondrial morphology, quantity, and function. MitoTracker Deep Red FM probe was used to stain GBM primary cells. It has been observed that the oeOMA1 groups significantly enhanced the number of mitochondria displaying ring-shaped structures in GBM primary cells more than in the control. This observation suggests the presence of mitochondrial fission or even fragmentation (online supplemental figure S5B). Additionally, mitochondrial DNA copy number was used to determine the relative mitochondrial number. Consequently, oeOMA1 decreased mtDNA copy number, suggesting mitochondrial loss (online supplemental figure S5C). Additionally, oeOMA1 enhanced mitochondrial fission (online supplemental figure S5D). In addition, oeOMA1 induced an increased ATP level and an increase in superoxide in GBM primary cells (online supplemental figure S5E,F). Of course, we also examined mitochondrial ATP energy production and total cellular ATP production. The results showed that overexpression of OMA1 reduced mitochondrial ATP energy production but increased total cellular ATP energy production. This is also consistent with the metabolic characteristics of tumor cells. Normally differentiated cells mainly rely on oxidative phosphorylation of mitochondria to supply energy to cells, while most tumor cells rely on aerobic glycolysis. This phenomenon is called the “Warburg effect.” That is to say, tumor cells rely on aerobic glycolysis that occurs in the cytoplasm for energy. To further clarify OMA1 promotes tumor IE through mitophagy. We added the mitophagy inhibitor Brefeldin A to the coculture system of OMA1-overexpressing GBM primary cells and CD8+ T cells and found that Brefeldin A could reduce OMA1-mediated IE (figure 4A–G).

Figure 4

OMA1 promotes GBM immune escape through mitophagy. (A) oeOMA1 significantly inhibited T cell-mediated tumor cell killing compared with Vector. However, the mitophagy inhibitor Brefeldin A could rescue the above phenomenon. (B–E) Flow cytometry and qPCR results indicated that CD8+ T cells cocultured with GBM primary cells overexpressing OMA1 showed lower proliferation and expression of IFN-γ, TNF-α, and Gzmb. However, the mitophagy inhibitor Brefeldin A could rescue the above phenomenon. (F, G) Schematic diagram of animal experiments. Typical animal experiment tumor slices, Ki-67 IF pictures, and typical column charts in different groups. Scale bar, 50 µm. The means±SDs are provided (n=5). *p<0.05, **p<0.01, and ***p<0.001 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; GBM, glioblastoma; IF, immunofluorescence; oeOMA1, overexpression of OMA1.

Although the reduction in Calcein AM fluorescence can indicate changes in mitochondrial membrane permeability, it also reflects decreased cell viability and apoptotic features. However, we assert that cells under this condition exhibit higher activity and increased proliferative capacity. According to the literature, OMA1 is a regulator of cell apoptosis, with its overexpression promoting the release of cytochrome c. Herein, we used WB to detect the expression of OMA1 in normal astrocytes (NHA) and glioma cell lines (T98G, LN-18, LN-229, A-172 and U-87). The results showed that compared with astrocytes, OMA1 was highly expressed in glioma cell lines (online supplemental figure S6A). In addition, as the tumor grade increases, the expression level of OMA1 also increases (online supplemental figure S6B). We selected T98G, which has a low expression of OMA1, as the cell line that overexpresses OMA1, and LN-229, which has a high expression of OMA1, as the cell line that knocks down OMA1. The results showed that knocking down OMA1 inhibited the expression of cytochrome c while overexpressing OMA1 promoted the expression of cytochrome c (online supplemental figure S6C). Because cGAS/STING is required, we also observed type I IFN in culture (online supplemental figure S6D). Besides, we have conducted experiments to investigate whether STING impacts CD8 recruitment, particularly through the CXCL10–CXCR3 pathway. Our findings indicate that compared with the control group, adding CXCL10 to the lower chamber significantly promoted the migration of CD8+ T cells into the lower chamber through the Transwell membrane. The data suggest a notable influence of STING on CD8 recruitment, with a specific involvement of the CXCL10–CXCR3 pathway (online supplemental figure S6E). Of course, the main effect of OMA1 targeting is an effect on proliferation so it is very difficult to claim that immune system have are role in the delay in tumor growth. We will perform and supplement these data on syngeneic models like GL261. Specifically, we will perform targeting experiments of OMA1 in WT and RAG mice to confirm the current hypothesis. Such experiments will provide us with deeper insights into a more comprehensive understanding of the impact of OMA1 targeting on tumor growth, thereby strengthening our study conclusions. Animal experimental results show that overexpression of OMA1 promotes tumor growth while knocking down OMA1 inhibits tumor growth (online supplemental figure S6F,G). This study shows that OMA1 can effectively promote the intracranial growth of mouse GL261 glioma cells by inhibiting the infiltration of CD8+ T cells into the tumor microenvironment and weakening the toxicity of CD8+ T cells. When anti-CD8 monoclonal antibodies (A2102) were used to eliminate CD8+ T cells in mice, the tumor-promoting effect of OMA1 disappeared, proving that OMA1 relies on CD8+ T cells to exert its tumor-promoting effect in tumor immunotherapy (online supplemental figure S6H).

OMA1 competitively binds HSPA9 to inhibit the IP3R/HSPA9/VDAC1 complex and mediate mitophagy

For further exploration of the specific molecular mechanism behind OMA1 activating mitophagy, we performed Co-IP combined mass spectrometry (MS) experiments in GBM primary cells overexpressing OMA1 and found a series of proteins that bind to OMA1, among which HSPA9 caught our attention (figure 5A). Studies have reported that HSPA9 can form a complex with IP3R and VDAC1 to participate in the Ca2+ transport between the endoplasmic reticulum and mitochondria, which is crucial for the maintenance of mitochondria’s normal function.35 Mitophagy removes aging and dysfunctional mitochondria.22 Therefore, we speculate that the combination of OMA1 and HSPA9 interferes with the formation of IP3R/HSPA9/VDAC1 complex, impairs the normal function of mitochondria, and promotes mitophagy to clear damaged mitochondria. To verify this conjecture, we first proved that OMA1 could bind HSPA9 through Co-IP experiments (figure 5B). Laser confocal results showed that OMA1 and HSPA9 colocalize in the cytoplasm (figure 5C). Additionally, the Co-IP results indicated that oeOMA1 weakened the binding ability of HSPA9 to IP3R and VDAC1 (figure 5D). Knockdown of HSPA9 in control GBM primary cells simulated OMA1 competitively inhibiting the IP3R/HSPA9/VDAC1 complex and found that the expression of mitophagy markers PINK1, p-Parkinser65, LC-3, BNIP3, and BNIP3L increased, suggesting that OMA1 competitively combined with HSPA9 to inhibit IP3R/HSPA9 The HSPA9/VDAC1 complex promotes mitophagy (figure 5E).

Figure 5

OMA1 competitively binds HSPA9 to inhibit the IP3R/HSPA9/VDAC1 complex and mediate mitophagy. (A) Identification of OMA1-binding proteins by WB and MS. (B) Using Co-IP to clarify that OMA1 could directly bind HSPA9. (C) The results of laser confocal experiments showed that OMA1 and HSPA9 colocalized in the cytoplasm of GBM primary cells. (D) The results of Co-IP experiments showed that oeOMA1 weakened the binding ability of HSPA9 to IP3R and VDAC1. (E) WB of PINK1, p-Parkinser65, LC-3, BNIP3, and BNIP3L in GBM primary cells treated with the indicated treatment. The means±SDs are provided (n=3). *p<0.05, **p<0.01, and ***p<0.001 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; Co-IP, co-immunoprecipitation; GBM, glioblastoma; MS, mass spectrometry; NS, not statistically significant; oeOMA1, overexpression of OMA1; WB, Western blotting.

Based on the secondary structure predicted with the InterPro (http://www.ebi.ac.uk/interpro/), PDB (https://www.rcsb.org/), and Pfam (www.pfam.org) databases, three OMA1 truncations were created for validating its binding to HSPA9. Co-IP assays provided evidence that the binding sequences specific to HSPA9 were found within the 168–354 aa region (A1) of OMA1 protein (online supplemental figure S7A). Once the binding series has undergone mutation (ΔOMA1), we employed WB to assess the autophagy-related molecule expressions in the context of OMA1 overexpression (oeΔOMA1), revealing that mitophagy-related molecule (PINK1, p-Parkin, BNIP3, and BNIP3L) expression levels did not change significantly (online supplemental figure S7B). In addition, we also used PCR and WB to verify the overexpression efficiency of HSPA9 (online supplemental figure S7C). Of course, we also confirm the precise region where HSPA9 binds to OMA1 (online supplemental figure S8A), mutate their binding sites and use WB to repeatedly verify the changes in mitophagy-related proteins after knocking down HSPA9 with mutated binding sites. As expected, there was no significant change in mitophagy-related proteins after knocking down HSPA9 with a mutated binding site (online supplemental figure S8B). Of course, we also performed IHC testing to detect the expression of PDL1 and OMA1 at different levels of Glioma (online supplemental figure S8C). Besides, the functional experiment results also indicated that oeΔOMA1 failed to promote the IE of GBM (online supplemental figure S9A–G).

To further demonstrate that OMA1 competes with HSPA9 to mediate mitophagy. First, we examined mitochondrial components (including Mitofusin-2, SOD2, VDAC1, COX IV, and TOM20) by WB in primary GBM cells overexpressing OMA1. The results showed a dose-dependent decrease in these proteins (online supplemental figure S10A). The maintenance of mitochondrial integrity is achieved through the regulation of a delicate equilibrium between the mechanisms of fission and fusion.36 Our previous experimental results also proved that OMA1 can promote mitochondrial fission (online supplemental figure S5). We found that oeOMA1 increased p-DRP1Ser616 in GBM primary cells (online supplemental figure S10A). This finding supports the hypothesis that Drp1 phosphorylation at the Ser616 site enhances its mitochondrial translocation, thereby inducing mitochondrial fission. Subsequently, an investigation was conducted to determine the significance of HSPA9 in the process of OMA1-induced mitophagy in GBM. Overexpression of OMA1 and simultaneous overexpression of HSPA9 (oeHSPA9) in GBM primary cells. HSPA9 overexpression significantly reduced OMA1-induced colocalization of the autophagosome within the mitochondria (online supplemental figure S10B). Similarly, the utilization of WB and IF analysis demonstrated that oeHSPA9 resulted in a reduction in PINK1 and Parkin expressions when subjected to OMA1 treatment (online supplemental figure S10C,D).

OMA1 upregulates PD-L1 on the GBM surface by promoting mitophagy and activating the mtDNA–cGAS–STING pathway

Studies have reported that mitophagy can promote mitochondrial DNA release into the cytoplasm and activate the cGAS–STING pathway.37 38 Accumulating evidence has indicated that the cGAS–STING pathway contributes significantly to tumor immunity.39 In order to verify whether OMA1 activates the mtDNA–cGAS–STING pathway through mitophagy, we performed cytoplasmic isolation in GBM primary cells overexpressing OMA1 and detected the mtDNA marker genes D-LOOP, CytB, and ND4 in the cytoplasm by qPCR. The expression was found to be significantly increased. However, the expression of D-LOOP, CytB, and ND4 could be downregulated after adding the mitophagy inhibitor Mdivi-1 (figure 6A). Further, the key molecules of the c-GAS–STING pathway (cGAS, STING, p-STING, TBK1, p-TBK1, IRF3, p-IRF3) were detected by WB, and the pathway was found to be activated. Yet Mdivi-1 can downregulate the phosphorylation level of key molecules in this pathway (figure 6B). The results of Edu and CCK-8 cell function experiments also showed again that oeOMA1 significantly enhanced primary GBM cell proliferation. However, Mdivi-1 could weaken OMA1 capability to promote primary GBM cell proliferation (figure 6C,D), suggesting that OMA1 activates the cGAS–STING pathway dependent on the mitophagy pathway. To further verify that OMA1-activated cGAS–STING pathway is related to its activated IE, we simultaneously knocked down c-GAS in GBM primary cells overexpressing OMA1. The key molecule PD-L1, which mediates tumor IE, was detected in GBM primary cells overexpressing/knocking down cGAS by WB and IF, and PD-L1 expression was also increased/decreased (figure 6E,F). We verified c-GAS knockdown efficiency by WB and qPCR (online supplemental figure S11A). In addition, we simultaneously knocked down c-GAS in primary GBM cells overexpressing OMA1 and cocultured them with CD8+ T cells in vitro, revealing again that OMA1 promoted IE of GBM. However, the knockdown of c-GAS could weaken the ability of OMA1 to promote the IE of GBM (figure 7A–G). Interestingly, the TCGA database analysis showed that OMA1 expression is directly proportional to PD-L1 (online supplemental figure S11B) and inversely proportional to CD8+ T cell contents (online supplemental figure S11C).

Figure 6

OMA1 promotes mitophagy and activates the mtDNA–cGAS–STING pathway to upregulate PD-L1 on the GBM surface. (A) oeOMA1 significantly promoted the expression of mtDNA marker genes D-LOOP, CytB, and ND4 in the cytoplasm by qPCR. However, the mitophagy inhibitor Mdivi-1 could rescue the above phenomenon. (B) oeOMA1 significantly enhanced the expression of cGAS, p-STING, p-TBK1, and p-IRF3 by WB. However, the mitophagy inhibitor Mdivi-1 could rescue the above phenomenon. (C, D) The results of CCK-8 and EdU assays indicated that oeOMA1 promoted the proliferation of GBM primary cells. However, the mitophagy inhibitor Mdivi-1 could rescue the above phenomenon. The bottom represents a typical histogram. (E, F) The results of WB and IF showed that overexpression of OMA1 promoted the expression of PD-L1, while knockdown of OMA1 inhibited the expression of PD-L1. The means±SDs are provided (n=5). **p<0.01 and ***p<0.001 according to two-tailed Student’s t-tests or one-way ANOVA followed by Dunnett tests for multiple comparisons. ANOVA, analysis of variance; GBM, glioblastoma; IF, immunofluorescence; NS, not statistically significant; oeOMA1, overexpression of OMA1; PD-L1, programmed death receptor 1 ligand; WB, Western blotting.