Mitochondrial complex I subunit NDUFS4 overexpression drives glioma progression by regulating mitochondrial function and COX5B

Jiang Wu; Juan Li; Li Xu; Yuanyuan Liu; Li Jiang

doi:10.1038/s41698-026-01281-9

. 2026 Jan 30;10:99. doi: 10.1038/s41698-026-01281-9

Mitochondrial complex I subunit NDUFS4 overexpression drives glioma progression by regulating mitochondrial function and COX5B

Jiang Wu ^1,^#, Juan Li ^2,^#, Li Xu ^3,^#, Yuanyuan Liu ^4,^✉, Li Jiang ^5,^✉

PMCID: PMC12960656 PMID: 41617910

Abstract

The current study explores the expression, functional significance, and underlying mechanisms of the mitochondrial protein NDUFS4 (NADH:ubiquinone oxidoreductase subunit S4) in glioma cells. TCGA shows that elevated NDUFS4 expression is consistently observed in glioma tissues, correlating with advanced tumor grade and diminished patient survival. Single-cell RNA sequencing further localizes this elevated expression primarily to glioma cells, where NDUFS4 co-expressed genes are integral to cellular respiration and mitochondrial ATP synthesis. These findings were corroborated in patient tissues and various primary and established glioma cell types, confirming consistent NDUFS4 overexpression. Genetic silencing (via shRNA) or CRISPR/Cas9-mediated knockout of NDUFS4 impaired mitochondrial function, evidenced by reduced oxygen consumption rate, inhibited mitochondrial complex I activity and ATP production and increased oxidative stress. NDUFS4 depletion also suppressed glioma cell proliferation, migration, and invasion, while promoting apoptosis. This inhibitory effect is specific to malignant cells, sparing non-cancerous astrocytes. Conversely, NDUFS4 overexpression enhanced mitochondrial activity and promoted aggressive malignant phenotypes in primary and immortalized glioma cells. Further multi-omics integration and experimental investigation established COX5B (cytochrome c oxidase subunit 5B) as an important downstream effector of NDUFS4. shRNA-induced silencing of COX5B replicated the outcomes of NDUFS4 depletion in primary glioma cells, and crucially, restoring COX5B in NDUFS4-silenced glioma cells abrogated the anti-glioma effects. In vivo studies demonstrated that NDUFS4 silencing effectively impeded intracranial growth of patient-derived glioma xenografts by compromising mitochondrial function, downregulating COX5B, inhibiting proliferation and inducing apoptosis. Collectively, these comprehensive data underscore NDUFS4’s essential role in glioma progression and position it as a promising therapeutic target for this aggressive malignancy.

Subject terms: Cancer, Cell biology, Molecular biology, Neuroscience, Oncology

Introduction

Glioma constitutes the most prevalent and aggressive primary malignancies of the central nervous system (CNS)^1,2, characterized by their highly infiltrative nature and inherent resistance to conventional therapeutic modalities³. Despite advancements in surgical resection techniques, radiation therapy, and systemic chemotherapy, the prognosis for patients afflicted with high-grade glioma remains exceptionally poor, with median survival rates typically not exceeding 15–20 months^1,2,4. Fundamental limitations inherent in current treatment paradigms include the pervasive invasiveness of these tumors, the formidable impediment posed by the blood-brain barrier to effective drug delivery, and the pronounced intratumoral heterogeneity that fosters the development of therapeutic resistance^1,2,4–6. The therapeutic landscape for glioma has witnessed the emergence of targeted therapies designed to selectively inhibit specific molecular pathways integral to tumor progression^1,6,7. While these innovative approaches hold considerable promise for more personalized and efficacious treatments, potentially offering reduced systemic toxicity compared to traditional cytotoxic agents, they frequently encounter formidable challenges^5,8,9. Consequently, there is an unequivocal and pressing imperative to identify novel therapeutic targets for glioma patients^5,8,9.

Accumulating evidence from contemporary research emphatically underscores the pivotal role of mitochondrial metabolism in sustaining the aberrant proliferation and progression characteristic of glioma cells^10,11. This phenomenon, frequently termed mitochondrial hyperfunction, highlights the profound dependency of these aggressive neoplasms on augmented mitochondrial activity to fulfill their demanding requirements for energy production, macromolecular biosynthesis, and redox homeostasis^10,11. Therefore, the strategic targeting of key mitochondrial proteins represents a compelling and novel therapeutic avenue for the development of effective anti-glioma interventions^10,11.

NDUFS4 (NADH:ubiquinone oxidoreductase subunit S4) is an accessory subunit integral to the structure and function of mitochondrial respiratory Complex I^12,13. As the primary entry point for electrons derived from NADH into the electron transport chain, mitochondrial Complex I is essential for cellular energy production through oxidative phosphorylation (OXPHOS)^12,13. NDUFS4 is strategically positioned within the hydrophilic module of mitochondrial Complex I and plays a critical role in the intricate process of complex assembly, facilitating the incorporation and stabilization of other subunits¹⁴. Cryo-EM analysis of Complex I from NDUFS4-deficient mice reveals a disrupted enzyme structure with a loosely associated NADH-dehydrogenase module and the absence of mature subunit NDUFA12¹⁴. Furthermore, it significantly influences the catalytic efficiency of the holoenzyme, impacting the rate of NADH oxidation and ubiquinone reduction¹⁴. Proteomic analysis of the NDUFS4 knockout mouse brain reveals region-specific changes, including reduced Complex I subunits¹³.

Beyond its well-established structural and enzymatic contributions, accumulating evidence indicates that NDUFS4 participates in additional mitochondrial processes, including the modulation of reactive oxygen species (ROS) production and the adaptation of mitochondrial metabolism to varying cellular demands^12,13,15. Aberrant expression or genetic mutations of NDUFS4 have been linked to a spectrum of human pathologies, notably Leigh syndrome^13–18, a severe neurological disorder characterized by progressive psychomotor regression, as well as other mitochondrial diseases and an emerging role in the context of tumorigenesis¹⁹, underscoring its fundamental importance in maintaining mitochondrial integrity and cellular physiology. The precise role of NDUFS4 in human cancer, and specifically within the context of glioma, has not been subjected to extensive investigation. Given the pivotal contribution of mitochondrial hyperfunction to glioma growth and the promising prospects of targeting mitochondrial proteins as a novel therapeutic strategy, the present study aims to comprehensively evaluate the expression profile, functional implications, and underlying molecular mechanisms of NDUFS4 in glioma cells.

Results

NDUFS4 expression is elevated in glioma and correlates with tumor grade and poor overall survival in TCGA-glioma data

First, we performed a comprehensive differential expression analysis of all known mitochondrial respiratory Complex I subunits using data from The Cancer Genome Atlas (TCGA) and publicly available glioma single-cell RNA sequencing datasets to find those upregulated in glioma tissues and cells. Second, among the upregulated subunits, we prioritized those with a significant correlation to poor patient survival outcomes, indicating potential clinical relevance. Finally, we conducted a thorough literature review to exclude any subunits already well-established in the context of glioma or other cancer types. This systematic approach led us to nominate NDUFS4 as a novel candidate for detailed investigation. TCGA dataset revealed a significant upregulation of NDUFS4 expression in glioma tissues (“Tumor”) compared to normal brain tissue (Fig. 1A). Furthermore, NDUFS4 expression also showed a positive correlation with the World Health Organization (WHO) grade of glioma, with higher grades (G3 and G4) displaying significantly elevated expression compared to G2 (Fig. 1B). IDH (isocitrate dehydrogenase) mutation status is a crucial biomarker for classifying adult diffuse gliomas. Compared to the IDH-mutant (Mut) type, the IDH-wildtype (WT) is more aggressive and has significant clinical relevance in predicting worse patient survival^20,21. Gliomas with IDH mutations showed significantly lower NDUFS4 expression compared to those with wild-type (WT) IDH (Fig. 1C). We also examined the expression levels of NDUFS4 in the mutant and wild-type groups for several other genes, including EGFR, AKT, TP53, PTEN and androgen receptor (AR). TCGA analyses revealed that NDUFS4 expression is lower in the IDH1/2 mutant group but higher in the AR mutant group (Fig. 1D).

Fig. 1 — Box plot showing significantly higher *NDUFS4* expression in glioma tumor tissues (Tumor) compared to normal brain tissues (Normal) (A). Correlation of *NDUFS4* expression with glioma WHO grade, demonstrating significantly increased expression in higher grades (G3 and G4) compared to G2 (B). *NDUFS4* expression in gliomas with wild-type (WT) and mutant (Mut) *IDH*, and in gliomas with other mutations (C, D). Box plot illustrating significantly higher *NDUFS4* expression in glioma patients who experienced an overall survival (OS) event (Dead) compared to those alive (Alive) (E). Box plot showing significantly elevated *NDUFS4* expression in glioma patients who experienced a disease-specific survival (DSS) event (Yes) compared to those who did not (No) (F). Kaplan–Meier survival curve depicting significantly shorter overall survival (OS) in glioma patients with high *NDUFS4* expression (G). Kaplan–Meier survival curve showing significantly reduced disease-specific survival (DSS) in glioma patients with high *NDUFS4* expression (H). Kaplan–Meier curves show that high *NDUFS4* expression is significantly associated with poor DSS in patients with primary gliomas (I), but no significant association in patients with relapsed gliomas (J). Kaplan-Meier survival curves for patients with *IDH*-wildtype (WT) or *IDH*-mutant (Mut) gliomas, stratified by *NDUFS4* expression levels (low versus high), are shown (K, L). Data were derived from the TCGA glioma dataset. Asterisks indicate statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001.

Analysis of overall survival (OS) data from the TCGA glioma cohort revealed a significantly higher NDUFS4 expression in patients who experienced a fatal event compared to those who were alive at the last follow-up (Fig. 1E). Similarly, a statistically significant elevation of NDUFS4 expression was observed in patients who experienced a disease-specific survival (DSS) event (Yes) compared to those who did not (No) (Fig. 1F). Kaplan–Meier survival analysis demonstrated that glioma patients with high NDUFS4 expression had significantly shorter overall survival compared to those with low NDUFS4 expression [Fig. 1G, Hazard Ratio (HR) = 1.44, 95% Confidence Interval (CI): 1.13–1.83, P = 0.003]. Furthermore, high NDUFS4 expression was significantly associated with reduced DSS in glioma patients (Fig. 1H, HR = 1.39, 95% CI: 1.08–1.80, P = 0.010). We have stratified the DSS data to distinguish between deaths resulting from primary tumor progression and those caused by relapse. TCGA analysis indicates that high NDUFS4 expression is significantly associated with poor DSS in cases of primary tumor progression (Fig. 1I), but not with deaths that occurred following a relapse (Fig. 1J). The overall survival was also analyzed using Kaplan-Meier survival curves, stratified by IDH status. In the IDH-WT (Fig. 1K) or mutant (Fig. 1L) cohort, there was no significant difference in overall survival between patients with high versus low NDUFS4 expression.

Single-cell RNA sequencing reveals elevated NDUFS4 expression in glioma cells and co-expressed genes involved in cellular respiration and energy metabolism

Analysis of a single-cell RNA sequencing dataset of gliomas (GSE182109), with cell type annotations provided by the original study²² (Fig. 2A), revealed the cellular landscape and the distribution of three glioma conditions within the reduced dimensional space: LGG (lower grade glioma), ndGBM (no-differentiated GBM), and rGBM (recurrent GBM) (Fig. 2B). Expression density and dot plots (Fig. 2C, D) demonstrated that NDUFS4 exhibited higher expression levels within all glioma cell populations compared to other cell types present in the tumor microenvironment (Fig. 2C, D). To further investigate the functional relevance of NDUFS4 in glioma cells, we specifically analyzed the glioma cell cluster and performed correlation analysis on NDUFS4 expression with other genes within this population. Applying a threshold of P < 0.05 and a Pearson correlation coefficient R > 0.5 (Fig. 2E), we identified genes positively correlated with NDUFS4. Subsequent Gene Ontology (GO) biological process enrichment analysis and Reactome pathway enrichment analysis of these co-expressed genes revealed significant enrichment in pathways related to cellular respiration, mitochondrial ATP synthesis, the electron transport chain, SLIT-ROBO signaling, and metabolic pathways (Fig. 2F, G).

Elevated NDUFS4 expression in glioma tissues from locally-treated patients and various glioma cells

Next, investigation into NDUFS4 expression within glioma tissues obtained from locally-treated patients revealed a pronounced and statistically significant upregulation. This comprehensive analysis encompassed sixteen high-grade glioma (HGG) tissues (designated “T”) and their matched corresponding adjacent normal brain tissues (designated “N”). As illustrated in Fig. 3A, a statistically significant increase in NDUFS4 mRNA expression was consistently observed within the glioma tissues. Quantitative analysis, presented in Fig. 3A, unequivocally demonstrated an approximate three-fold elevation in NDUFS4 mRNA expression in glioma tissues when compared to the “N” tissues. Furthermore, corroborating these transcriptional findings, immunoblotting results (Fig. 3B) for four selected HGG patients (Patient 1# through Patient 4#) distinctly showed a consistent upregulation of NDUFS4 protein expression in their respective glioma tissues. The aggregated mean NDUFS4 protein levels, carefully derived from the quantitative immunoblotting analysis of all sixteen paired tumor and adjacent normal tissue samples, were presented in Fig. 3C, further solidifying the evidence for elevated protein expression in the tumor microenvironment.

Fig. 3 — *NDUFS4* mRNA expression in glioma tissues (“T”) was presented relative to adjacent normal brain tissues (“N”) from 16 locally-treated high-grade glioma (HGG) patients (A). Representative immunoblotting results illustrated NDUFS4 protein expression in tumor (“T”) versus normal (“N”) tissues from four selected HGG patients (Patient 1# to Patient 4#) (B). Quantified mean NDUFS4 protein levels, derived from the immunoblotting analysis of sixteen paired tumor and adjacent normal tissue samples, were shown (C). *NDUFS4* mRNA and protein expression levels were detected across primary human astrocytes (“Astrocytes1/2”), immortalized A172, U87 and T98 cells, and primary human glioma cells (“P1”, “P2”, “P3”) (D–G). All data were represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) when compared to “N” tissues or “Astrocyte1”.

Subsequent experimental endeavors were directed at evaluating the potential upregulation of NDUFS4 across various human glioma cells, thereby extending our observations beyond primary tissue samples. This included primary human glioma cells (designated “P1-P3,” derived from three distinct patients, from Dr. Cao^23–26) and the well-characterized immortalized A172 cell line. The findings conclusively demonstrated a substantial and consistent elevation in NDUFS4 expression among the examined glioma cells when contrasted with levels observed in primary human astrocytes (designated “Astrocytes1/2”, also from Dr. Cao^23–26), which served as crucial non-malignant controls. Importantly, an evident and robust upregulation of NDUFS4 mRNA was detected in both primary and A172 glioma cells (Fig. 3D). Consistent with both the transcriptional observations and the initial tissue analysis, the expression of NDUFS4 protein levels was also markedly elevated in both primary and A172 human glioma cells (Fig. 3E). In addition, the mRNA and protein levels of NDUFS4 are significantly higher in U87 and T98 glioma cell lines compared to the Astrocytes1 cells (Fig. 3F, G), providing a comprehensive picture of NDUFS4 upregulation in glioma. These data indicate a consistent and significant NDUFS4 upregulation at both mRNA and protein levels in human glioma tissues and cells.

NDUFS4 silencing impairs mitochondrial bioenergetics and induces oxidative stress in glioma cells

To thoroughly elucidate the functional significance of NDUFS4 within the context of glioma pathophysiology, a series of carefully designed experiments were executed. We aimed to precisely assess the profound impact of NDUFS4 silencing on both mitochondrial function and the intricate balance of oxidative stress within these malignant cells. In primary human glioma cells (P1), the targeted silencing of NDUFS4, successfully achieved through the application of three distinct short hairpin RNAs (shRNAs: shNDUFS4-sh1, shNDUFS4-sh2, shNDUFS4-sh3, targeting non-overlapping sequences), consistently and significantly compromised mitochondrial respiratory function. The efficacy of NDUFS4 knockdown was rigorously confirmed at both the transcriptional (mRNA) and translational (protein) levels (Fig. 4A, B), unequivocally demonstrating a robust and sustained reduction in NDUFS4 expression when compared to the control shRNA (shC)-treated P1 cells. The mRNA and protein expression of a control gene, NDUFS1, was unchanged by the applied shRNAs (Fig. 4A, B). This profound impairment in mitochondrial function was comprehensively characterized by a suite of critical bioenergetic indicators. A statistically significant reduction in oxygen consumption rates (OCR) was universally observed across all tested shRNA constructs (Fig. 4C), providing compelling evidence of inhibited overall mitochondrial respiration. More specifically, both the basal and maximal respiration rates, indicative of cellular energy demand and reserve capacity, were significantly diminished (Fig. 4C). Furthermore, a substantial decrease in the activity of Complex I, the primary entry point for electrons into the mitochondrial electron transport chain, was detected (Fig. 4D). Concurrently, NDUFS4 silencing led to significantly decreased intracellular ATP levels (Fig. 4E), directly reflecting a profound impact on the cell’s capacity for oxidative phosphorylation and overall cellular energy production. These severe functional deficits were linked to the induction of mitochondrial membrane depolarization (Fig. 4F), as evidenced by alterations in JC-1 fluorescent dye from red polymers to green monomers. Moreover, a significant increase in overall cellular oxidative stress was detected, as compellingly demonstrated by enhanced MitoSOX red fluorescence intensity (Fig. 4G) and DCF-DA green fluorescence intensity (Fig. 4H). In addition, decreased GSH/GSSG ratio (Fig. 4I) also signified a pronounced shift towards a more oxidized cellular environment in NDUFS4-silenced P1 glioma cells. This was further corroborated by a significant elevation in lipid peroxidation, precisely quantified through TBARS activity (Fig. 4J), indicating extensive oxidative damage to vital cellular membranes in NDUFS4-silenced glioma cells. The addition of succinate can bypass the blocked complex I by donating electrons directly to complex II, allowing the electron transport chain to continue and generate ATP. We found that succinate supplementation (“SU)” partially restored compromised mitochondrial function in NDUFS4-depleted cells, as evidenced by restored ATP levels (Fig. 4K) and inhibited mitochondrial ROS production (MitoSOX staining assays, Fig. 4L). This further validates that NDUFS4 is essential for functional Complex I activity. In P1 glioma cells, the targeted lentiviral shRNA-mediated silencing of NDUFS1 (as confirmed in Fig. S2A) resulted in quantifiably less severe impairments of mitochondrial function. Specifically, the reduction in ATP content (Fig. S2B) and the degree of mitochondrial depolarization (Fig. S2C) were substantially less pronounced.

Fig. 4 — *NDUFS4* and *NDUFS1* mRNA expression in P1 primary human glioma cells following treatment with three distinct shRNAs (shNDUFS4-sh1, shNDUFS4-sh2, shNDUFS4-sh3) compared to control shRNA (shC) was shown (A). Representative immunoblotting and quantification of NDUFS4 and NDUFS1 protein expression in P1 glioma cells after the shRNA treatment was shown (B). The cells were cultivated for designated time, mitochondrial function in these primary P1 glioma cells was evaluated. Assessed parameters encompassed oxygen consumption rates (OCR) determined via Seahorse metabolic analysis (C), enzymatic activity of mitochondrial Complex I (D), and quantification of total cellular ATP levels (E). Corresponding indicators of oxidative stress measured in these P1 cells included the mitochondrial membrane potential based on JC-1 monomer fluorescence intensity (F), and cellular oxidative stress detected using MitoSOX red fluorescence intensity (G), DCF-DA green fluorescence intensity (H), the cellular glutathione redox state (GSH/GSSG ratio) (I), and quantification of lipid peroxidation products via TBARS activity (J). P1 glioma cells with shNDUFS4-sh3 were treated with succinate (SU, 10 mM) or vehicle control (PBS) for 24 h, control cells were with non-sense control shRNA (shC), ATP contents (K) and mitochondrial ROS production (MitoSOX intensity, L) were measured. *NDUFS4* and *NDUFS1* mRNA expression in additional primary human glioma cells (P2, P3) and the immortalized A172 cell line after shNDUFS4-sh3-induced silencing was shown (M, N). Cells were further cultured for designated hours, mitochondrial Complex I activity (O), total cellular ATP content (P), mitochondrial membrane potential via JC-1 staining (Q), and intracellular ROS production utilizing the MitoSOX probe (R) were measured. Non-cancerous astrocytes (Astrocytes1/2) were modified to stably express shNDUFS4-sh3 or shC. Following cultivation for indicated periods, enzymatic activity of mitochondrial Complex I (S) and total cellular ATP levels (T) were measured. “Ctrl” stands for parental control glioma cells. All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to control shRNA (shC)-treated cells. # indicates statistical significance (P < 0.05) (K, L). The abbreviation “n.s.” denotes a non-significant statistical difference (P > 0.05). Each experiment was performed in five independent replicates. The scale bar represents 100 μm.

To confirm the generalizability and inherent robustness of these findings, NDUFS4 silencing using the highly effective shNDUFS4-sh3 construct was subsequently applied to additional primary glioma cells (P2 and P3) and the immortalized A172 glioma cell line. Consistent with the initial observations in P1 cells, NDUFS4 silencing by shNDUFS4-sh3 (Fig. 4M, N, showing no effect on NDUFS1 mRNA expression) elicited strikingly similar actions across these additional glioma cell types. Specifically, shNDUFS4-sh3 consistently led to comparable impairments in mitochondrial respiratory function, including decreased Complex I activity (Fig. 4O), and significantly lower ATP levels (Fig. 4P). Furthermore, mitochondrial membrane depolarization (JC-1 green monomers accumulation, Fig. 4Q) and oxidative stress (MitoSOX intensity increasing, Fig. 4R), encompassing elevated ROS production, were consistently observed following NDUFS4 knockdown in these glioma cell models. In primary human astrocytes (“Astrocytes1/2”) with low basal NDUFS4 expression (see Fig. 3), silencing NDUFS4 with the shNDUFS4-sh3 construct resulted in only a minimal change in mitochondrial complex I activity (Fig. 4S) and had no significant effect on cellular ATP content (Fig. 4T). These collective results unequivocally demonstrate that NDUFS4 is important for maintaining robust mitochondrial function and ensuring redox homeostasis within glioma cells.

NDUFS4 silencing impairs malignant phenotypes of human glioma cells

To ascertain the functional significance of NDUFS4 in the progression of glioma, a series of in vitro studies were conducted utilizing P1 primary human glioma cells. The targeted knockdown of NDUFS4, successfully achieved through above-mentioned three different shRNAs (shNDUFS4-sh1, shNDUFS4-sh2, shNDUFS4-sh3, see Fig. 4), consistently and significantly attenuated several critical malignant phenotypes. Specifically, a marked reduction in cell viability (CCK-8 optical density/OD) was observed (Fig. 5A), alongside a profound impairment in the capacity for colony formation (Fig. 5B). NDUFS4 knockdown also led to a substantial and quantifiable decrease in cell proliferation, as evidenced by reduced nuclear EdU incorporation (Fig. 5C). Beyond anti-proliferative effects, the migratory and invasive capacities of P1 glioma cells were also significantly inhibited (Fig. 5D, E, respectively), underscoring a critical and indispensable role for NDUFS4 in mediating glioma cells’ metastatic potential. NDUFS4 silencing also induced moderate cell apoptosis, as compellingly demonstrated by modest increases in Caspase-3 activity and TUNEL-positive cells (Fig. 5F). Intriguingly, the anti-glioma effects, including viability reduction, proliferation inhibition and apoptosis activation, induced by shNDUFS4-sh3 in P1 cells were significantly abrogated by the supplementary addition of the antioxidant N-acetylcysteine (nac) or high glucose (glu) concentrations (Fig. 5G–I), suggesting that the detrimental effects of NDUFS4 silencing on glioma cells are mediated, at least in part, by the induction of oxidative stress and mitochondrial function impairment. NDUFS1 silencing using a lentiviral shRNA exhibited a milder inhibitory effect on both P1 glioma cell proliferation (Fig. S2D) and migration (Fig. S2E).

Fig. 5 — P1 primary human glioma cells were modified to stably express lentiviral shNDUFS4-sh1, shNDUFS4-sh2, shNDUFS4-sh3 (with nonoverlapping sequences), or a control scramble shRNA (shC). These cells were cultured for specified durations to evaluate viability (CCK-8 OD, A), colony formation (B), cell proliferation (EdU-incorporated nuclei percentage, C), in vitro migration (“Transwell” assays, D), in vitro invasion (“Matrigel Transwell” assays, E), and apoptosis (via measuring Caspase-3 activity and TUNEL-positive cell ratio, F). P1 cells with shNDUFS4-sh3 were treated with N-acetylcysteine (NAC, 500 μM) or high glucose (Glu, 10 mM) to assess their impact on viability (CCK-8 OD, G), proliferation (EdU-incorporated nuclei percentage, H), and apoptosis (TUNEL-positive cell ratio, I) after cultivation for indicated time periods. Primary human glioma cells derived from two other patients (P2 and P3), as well as immortalized A172 cells, were genetically modified to stably express shNDUFS4-sh3 or shC. Following cultivation for indicated periods, cell viability (CCK-8 OD, J), proliferation (EdU-incorporated nuclei percentage, K), in vitro migration (“Transwell” assays, L), and apoptosis (via measuring Caspase-3 activity and TUNEL-positive cell ratio, M) were assessed similarly. Non-cancerous astrocytes (Astrocytes1/2) were modified to stably express shNDUFS4-sh3 or shC. *NDUFS4* and *NDUFS1* mRNA expression were analyzed (N, O). Following cultivation for indicated periods, cell viability (CCK-8 OD, P), cell proliferation (EdU-incorporated nuclei percentage, Q) and apoptosis (TUNEL-positive cell ratio, R) were tested similarly. “Ctrl” stands for parental control glioma cells. All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to control shRNA (shC)-treated cells. # indicates statistical significance (P < 0.05) (G–I). The abbreviation “n.s.” denotes a non-significant statistical difference (P > 0.05). Each experiment was performed in five independent replicates. The scale bar represents 100 μm.

Importantly, targeting NDUFS4 in P1 glioma cells via siRNA transfection resulted in substantial protein downregulation (Fig. S3A). The transient loss of NDUFS4 compromised mitochondrial integrity, evidenced by reduced ATP levels (Fig. S3B) and pronounced mitochondrial depolarization (Fig. S3C). Consistent with these mitochondrial defects, NDUFS4 silencing also significantly suppressed P1 glioma cell proliferation (Fig. S3D) and migration (Fig. S3E).

To rigorously confirm the broader applicability and generalizability of these pivotal findings, NDUFS4 silencing via the highly effective shNDUFS4-sh3 construct (see Fig. 4) was subsequently employed in other primary glioma cells (P2 and P3) and the well-established immortalized A172 glioma cell line. Crucially, shNDUFS4-sh3 consistently induced similar anti-survival (CCK-8 assays, Fig. 5J), anti-proliferative (EdU incorporation assays, Fig. 5K), anti-migratory (Fig. 5L), and pro-apoptotic (Caspase-3 and nuclear TUNEL staining assays, Fig. 5M) actions across these diverse glioma cell models. Most importantly, shNDUFS4-sh3 construct was utilized to silence NDUFS4, but not NDUFS1, in noncancerous astrocytes (Astrocytes1/2) (Fig. 5N, O). When NDUFS4 was silenced in these non-cancerous cells, the observed effects on cell viability (CCK-8 OD, Fig. 5P), proliferation (Fig. 5Q), and apoptosis (Fig. 5R) were found to be non-significant. This striking differential dependency unequivocally highlights that glioma cells exhibit a unique and profound vulnerability to NDUFS4 depletion when compared to non-malignant astrocytes. Therefore, NDUFS4 silencing led to impaired viability, growth, migration, and survival in cultured glioma cells.

NDUFS4 knockout impairs mitochondrial function and malignant phenotypes in human glioma cells

To further corroborate the role of NDUFS4 in glioma cell progression, a more robust and enduring genetic strategy was implemented: the stable knockout of NDUFS4 utilizing the CRISPR/Cas9 gene editing system in P1 primary human glioma cells. Successful ablation of NDUFS4 protein expression was confirmed at the protein levels (Fig. 6A, B), demonstrating a profound and sustained depletion of NDUFS4 protein (but not NDUFS1) when compared to the control cells (koC) through the utilization of two different single guide RNAs (sgRNAs: koNDUFS4-sg1 and koNDUFS4-sg2). NDUFS4 knockout in P1 glioma cells precipitated significant mitochondrial dysfunction, closely mirroring the detrimental effects previously observed with shRNA-mediated silencing. Specifically, NDUFS4 knockout, by koNDUFS4-sg1 or koNDUFS4-sg2, resulted in a substantial impairment in overall mitochondrial respiratory function, characterized by a significant reduction in OCR (Fig. 6C), with both basal and maximal respiration being significantly affected. It was accompanied by a significant decrease in mitochondrial Complex I activity (Fig. 6D) and significantly reduced intracellular ATP levels (Fig. 6E), supporting disruption to mitochondrial bioenergetics. NDUFS4 knockout also induced significant mitochondrial membrane depolarization (Fig. 6F), as evidenced by alterations in JC-1 fluorescence, and a substantial increase in the production of ROS (MitoSOX intensity increasing, Fig. 6G). Beyond its impact on mitochondrial integrity and function, NDUFS4 knockout also profoundly inhibited several key malignant phenotypes characteristic of P1 glioma cells. Cell proliferation was markedly suppressed in NDUFS4 knockout cells (Fig. 6H). Furthermore, both the migratory (Fig. 6I) and invasive (Fig. 6J) capacities of the cells were substantially attenuated. Importantly, NDUFS4 knockout also promoted moderate apoptosis, tested via nuclear TUNEL staining and Caspase-3 activity assays (Fig. 6K, L).

Fig. 6 — P1 primary human glioma cells were subjected to CRISPR/Cas9-mediated NDUFS4 knockout (using two different sgRNA sequences, koNDUFS4-sg1 or koNDUFS4-sg2), with control cells expressing a control non-sense sgRNA construct (koC). Representative immunoblotting (A) and quantification (B) of NDUFS4 and NDUFS1 protein expression in theP1 glioma cells were shown. Cells were cultivated for designated time, mitochondrial function in these primary P1 glioma cells was evaluated. Assessed parameters encompassed oxygen consumption rates (OCR) determined via Seahorse metabolic analysis (C), enzymatic activity of mitochondrial Complex I (D), and quantification of total cellular ATP levels (E). Corresponding indicators of oxidative stress measured in these P1 cells included the mitochondrial membrane potential based on JC-1 monomer fluorescence intensity (F), and cellular oxidative stress detected using MitoSOX red fluorescence intensity (G). Assessed malignant phenotypes included cell proliferation (EdU-incorporated nuclei percentage, H), in vitro migration (“Transwell” assays, I), in vitro invasion (“Matrigel Transwell” assays, J), and apoptosis (by measuring TUNEL-positive cells and Caspase-3 activity, K, L). NDUFS4 knockout using koNDUFS4-sg2-expressing construct was also performed in P2 primary human glioma cells, with NDUFS4 and NDUFS1 protein expression tested (M). Cells were cultivated for designated time, assessed parameters encompassed ATP levels (N), Complex I activity (N), mitochondrial depolarization (O), cell proliferation (P), migration (Q), and apoptosis (R) using the same methods. All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to control cells (koC). The abbreviation “n.s.” denotes a non-significant statistical difference (P > 0.05). Each experiment was performed in five independent replicates. The scale bar represents 100 μm.

The NDUFS4 knockout using the CRISPR/Cas9 methodology was also performed in another distinct primary glioma cell type, P2. Consistent with the compelling observations in P1 cells, the koNDUFS4-sg2-expressing construct resulted in NDUFS4 protein depletion in P2 primary cells (Fig. 6M). NDUFS4 knockout led to a significant decrease in ATP levels (Fig. 6N), impaired Complex I activity (Fig. 6N), induced mitochondrial depolarization (Fig. 6O), and significantly reduced both cell proliferation (Fig. 6P) and migration (Fig. 6Q). Moreover, NDUFS4 knockout consistently promoted modest apoptosis in P2 glioma cells (Fig. 6R). These results again underscore the critical role of NDUFS4 in maintaining mitochondrial function and, consequently, in driving the aggressive malignant phenotypes characteristic of human glioma cells.

NDUFS4 overexpression enhances mitochondrial function and promotes malignant phenotypes in human glioma cells

We next investigated the consequences of NDUFS4 overexpression in human glioma cells. A lentiviral construct engineered for NDUFS4 overexpression was stably introduced into P1 primary human glioma cells, leading to the successful establishment of two distinct and stable clonal cell selections, designated as oeNDUFS4-Sl1 and oeNDUFS4-Sl2. These cells exhibited a statistically significant increase in both NDUFS4 mRNA and protein expression (Fig. 7A, B, respectively) when compared to vector control P1 cells (Vec). Importantly, this overexpression was specific to NDUFS4, as it did not exert any significant effect on the expression levels of NDUFS1 (Fig. 7A, B). The enforced overexpression of NDUFS4 in P1 glioma cells led to a marked enhancement of mitochondrial function and a concomitant promotion of aggressive malignant phenotypes. Specifically, NDUFS4 overexpression resulted in a significant increase in mitochondrial Complex I activity (Fig. 7C) and substantially elevated intracellular ATP levels (Fig. 7D), collectively indicating a robust improvement in overall mitochondrial bioenergetics and cellular energy production. Furthermore, overexpression of NDUFS4 enhanced cellular viability (CCK-8 OD, Fig. 7E) and cell proliferation, the latter was evidenced by enhanced EdU incorporation (Fig. 7F). Moreover, both the invasive (Fig. 7G) and migratory (Fig. 7H) capacities of the cells were consistently augmented. To confirm the broader applicability and generalizability of these findings, NDUFS4 overexpression, utilizing the identical lentiviral construct, was also performed in other primary (P2, P3) and established (A172) glioma cell types. The construct led to NDUFS4 upregulation without affecting NDUFS1 (Fig. 7I, J). Consistent with the compelling observations in P1 glioma cells, NDUFS4 overexpression in these diverse glioma cell models elicited similar actions. Specifically, it enhanced Complex I activity (Fig. 7K), elevated ATP contents (Fig. 7L), increased cell proliferation (Fig. 7M), and augmented cell migration (Fig. 7N). These results unequivocally suggest that NDUFS4 overexpression enhanced mitochondrial function and enhanced the malignant phenotypes observed in human glioma cells.

Fig. 7 — P1 primary human glioma cells were treated with a lentiviral NDUFS4 overexpression construct and two stable cell selections (oeNDUFS4-Sl1, oeNDUFS4-Sl2) were established after puromycin treatment. NDUFS4 and NDUFS1 mRNA (A) and protein (B) expression in these cells and vector control cells (“Vec”) were shown. The cells were cultivated for designated time, the enzymatic activity of mitochondrial Complex I (C), and quantification of total cellular ATP levels (D) were measured. Assessed malignant phenotypes included cell viability (CCK-8 OD, E), cell proliferation (EdU-incorporated nuclei percentage, F), in vitro migration (“Transwell” assays, G), in vitro invasion (“Matrigel Transwell” assays, H), NDUFS4 overexpression was also performed in P2, P3, and A172 glioma cells, and NDUFS1/NDUFS4 mRNA tested (I, J). The cells were cultivated for designated time, assessed parameters encompassed the mitochondrial Complex I activity (K), cellular ATP contents (L), cell proliferation (M), and cell migration (N). All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to vector control cells (Vec). The abbreviation “n.s.” denotes a non-significant statistical difference (P > 0.05). Each experiment was performed in five independent replicates. The scale bar represents 100 μm.

NDUFS4 correlation gene COX5B is important for maintaining mitochondrial homeostasis and promoting malignant phenotypes in glioma cells

To precisely identify potential key downstream effectors of NDUFS4, we integrated the above single-cell RNA sequencing findings (Fig. 2) with bulk transcriptomic data from TCGA-glioma (Fig. 1) and proteomic data from DepMap. Using the same correlation thresholds (P < 0.05, R > 0.5), we identified NDUFS4-correlated genes and proteins in the TCGA-glioma and DepMap datasets, respectively. The intersection of the NDUFS4-correlated gene list from the single-cell data, the NDUFS4-correlated gene list from TCGA-glioma transcriptomics, the NDUFS4-correlated protein list from DepMap proteomics and the MitoCarta 3.0 mitochondrial gene database yielded three consistently co-expressed molecules: COX5B, GADD45GIP1, and NDUFS6 (Fig. 8A). Correlation scatter plots using TCGA glioma transcriptomic data further validated the positive correlations between NDUFS4 and each of these three co-expressed molecules: COX5B (Pearson R = 0.584, P < 0.001), GADD45GIP1 (Pearson R = 0.544, P < 0.001), and NDUFS6 (Pearson R = 0.562, P < 0.001) (Fig. 8B). These results highlight COX5B, GADD45GIP1, and NDUFS6 as potential functionally related mitochondrial genes of NDUFS4 in glioma. Our subsequent focus was directed toward COX5B, primarily due to its documented overexpression in glioma and its strong correlation with key prognostic indicators in patients²⁷. Moreover, COX5B’s pro-tumorigenic activity has been substantiated in other oncological contexts^28,29, whereas the precise roles of the remaining two candidate genes, GADD45GIP1, and NDUFS6, in tumor cell biology are not as well-defined.

Fig. 8 — Venn diagram showing the overlap of *NDUFS4*-correlated genes from single-cell RNA sequencing, TCGA-glioma transcriptomics, *NDUFS4*-correlated proteins from DepMap proteomics and the MitoCarta 3.0 mitochondrial gene database, identifying three consistently co-expressed molecules: *COX5B*, *GADD45GIP1*, and *NDUFS6* (A). Correlation scatter plots showing the relationship between *NDUFS4* and *COX5B*, *GADD45GIP1*, and *NDUFS6* expression in the TCGA-glioma transcriptomic dataset, with corresponding Pearson correlation coefficients (R) and P values shown (B). *COX5B* mRNA (C) and COX5B-NDUFS4 protein levels (D) following shRNA-mediated COX5B silencing (shCOX5B) compared to control cells (shC) were shown in P1 glioma cells. Cells were further cultivated for indicated time periods, intracellular ATP contents were measured (E), with mitochondrial depolarization tested via JC-1 staining assays (F) and ROS production measured via CellROX staining assays (G). Malignant phenotypes were also assessed, including cellular proliferation quantified by EdU incorporation (H), migratory capacity (I), and apoptosis tested by TUNEL staining (J). All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to control shRNA (shC)-treated cells. Each experiment was performed in five independent replicates. The scale bar represents 100 μm.

To validate the functional significance of COX5B in glioma, we employed shRNA-mediated gene silencing, which successfully achieved a substantial reduction in both COX5B mRNA (Fig. 8C) and protein (Fig. 8D) levels in P1 glioma cells. COX5B silencing failed to affect NDUFS4 protein expression (Fig. 8D). This targeted depletion of COX5B expression consequently led to a pronounced compromise in mitochondrial function, unequivocally evidenced by a significant decrease in intracellular ATP content (Fig. 8E). Consistent with this mitochondrial impairment, we observed a significant increase in mitochondrial depolarization (JC-1 staining assays, Fig. 8F) and an elevated production of ROS (CellROX staining assays, Fig. 8G) in the shCOX5B-treated cellular population. Beyond its direct impact on mitochondrial integrity and function, COX5B knockdown also exerted a profound inhibitory effect on several defining malignant phenotypes. Cell proliferation was significantly suppressed, as quantitatively demonstrated by reduced EdU incorporation (Fig. 8H). Furthermore, the migratory capacity (Fig. 8I) of the P1 glioma cells were substantially attenuated. COX5B knockdown also induced modest cell apoptosis, as evidenced by a heightened percentage of apoptotic nuclei (Fig. 8J). Collectively, these results establish COX5B as a pivotal regulator governing both mitochondrial function and the malignant phenotypes observed in glioma cells.

The critical role of COX5B as a downstream effector in NDUFS4-regulated glioma progression

Next experiments were carried out to confirm a regulatory axis between NDUFS4 and COX5B in P1 glioma cells. Specifically, the genetic knockdown of NDUFS4 (using shNDUFS4-sh3, see Figs. 4 and 5) or its knockout via CRISPR/Cas9 (koNDUFS4-sg1, see Fig. 6) resulted in a significant reduction in both COX5B mRNA (Fig. 9A) and protein levels (Fig. 9B) in P1 glioma cells. Conversely, forced overexpression of NDUFS4 (oeNDUFS4-Sl1, see Fig. 7) led to a substantial upregulation of COX5B expression at both the mRNA (Fig. 9C) and protein (Fig. 9D) levels, solidifying the direct and positive regulatory influence of NDUFS4 on COX5B. The functional dependency of NDUFS4’s anti-glioma effects on COX5B was then assessed through rescue experiments. After establishing NDUFS4 knockdown (with shNDUFS4-sh3) in P1 glioma cells, we introduced a COX5B overexpression vector (oeCOX5B) to restore COX5B expression (Fig. 9E, F). Crucially, the re-expression of COX5B in NDUFS4-silenced P1 glioma cells successfully rescued mitochondrial function. This was evidenced by the restoration of ATP levels (Fig. 9G), a reduction in mitochondrial ROS production (measured by MitoSOX staining, Fig. 9H), and the attenuation of mitochondrial depolarization (assessed via JC-1 assays, Fig. 9I). Consequently, COX5B re-expression significantly ameliorated the anti-glioma cell activities that were initially induced by NDUFS4 knockdown. Specifically, the marked inhibition of cell proliferation (quantified by nuclear EdU incorporation, Fig. 9J), the substantial reduction in cell migration (Fig. 9K), and the pronounced promotion of apoptosis (measured by nuclear TUNEL staining, Fig. 9L) observed in NDUFS4-knockdown cells were all significantly attenuated upon the restoration of COX5B expression. These results, together with results in Fig. 8, provide evidence that COX5B serves as a pivotal downstream mediator through which NDUFS4 exerts its effects on the maintenance of mitochondrial function and the aggressive malignant phenotypes in glioma cells.

NDUFS4 regulates both the mRNA and protein expression of COX5B in glioma cells (Fig. 9A–D), prompting us to investigate a potential transcriptional mechanism. Given that Sp1 has been identified as a proposed transcription factor for COX5B³⁰, we performed a chromatin immunoprecipitation (ChIP) assay to test this hypothesis. Our results revealed that NDUFS4 silencing (using shNDUFS4-sh3) or knockout (by koNDUFS4-sg1) significantly decreased the amount of Sp1 bound to the COX5B promoter in P1 glioma cells (Fig. 9M). Importantly, this change was not due to altered total Sp1 protein levels, which remained consistent in the NDUFS4-depleted cells (Fig. 9M). Conversely, overexpression of NDUFS4 led to a substantial increase in Sp1 binding to the COX5B promoter (Fig. 9N), again without affecting total Sp1 protein expression (Fig. 9N). These findings suggest that NDUFS4 likely regulates the transcriptional activity of COX5B by modulating the binding of Sp1 to its promoter.

Intracranial growth of patient-derived glioma xenografts in nude mice is hindered after NDUFS4 silencing

To ascertain the role of NDUFS4 in glioma progression in vivo, P1 primary human glioma cells, engineered to stably express either the shNDUFS4-sh3 (shNDUFS4) or a control shRNA construct (shC) (see Figs. 4 and 5), were intracranially implanted into immunodeficient nude mice. Within 20 days post-implantation, the initial mouse within the shC group began to manifest discernible neurological symptoms, indicative of progressive intracranial tumor burden. Subsequent volumetric measurements of the resultant intracranial xenografts, as visually depicted in representative magnetic resonance images (Fig. 10A) and precisely quantified in Fig. 10B, revealed a statistically significant reduction in tumor size for the shNDUFS4 P1 glioma xenografts when compared to their shC counterparts. These results supported the substantial growth inhibition of patient-derived P1 glioma xenografts within the murine brain following targeted NDUFS4 silencing. The mouse body weights remained unaffected throughout the experimental period (Fig. 10C). Two xenograft tumors per group were then isolated, and NDUFS4 mRNA and protein expression levels were significantly reduced within the shNDUFS4 group compared to the shC group (Fig. 10D, E), thereby confirming the sustained and effective knockdown of NDUFS4 in vivo. This reduction was highly specific to NDUFS4, as expression levels of NDUFS1 remained entirely unchanged (Fig. 10D, E). The IHC staining in the xenograft slides further confirmed NDUFS4 silencing in shNDUFS4 group (Fig. 10F).

Fig. 10 — P1 primary human glioma cells stably expressing shNDUFS4-sh3 (shNDUFS4) or a control shRNA construct (shC) were intracranially implanted into immunodeficient nude mice. Representative magnetic resonance images (A) and quantification of intracranial xenograft volumes (B), as well as the mouse body weights (C) at day 20 post-implantation were shown. The expression levels of listed mRNAs or proteins in the described xenograft tissues were tested (D, E, K, and M). Mitochondrial function in xenograft tissues was assessed by Complex I activity (G) and intracellular ATP content (H). Indicators of oxidative stress in the xenograft tissues included analyzing the GSH/GSSG ratio (I) and TBARS activity (J). Histological/fluorescence examination of xenograft tissue slides included NDUFS4 immunohistochemical staining (F), Ki-67 immunohistochemical staining for proliferation (L), and nuclear TUNEL immunofluorescence staining for apoptosis (N). All data are represented as mean ± standard deviation (SD). An asterisk (*) indicates statistical significance (P < 0.05) compared to shC-treated xenografts. The abbreviation “n.s.” denotes a non-significant statistical difference (P > 0.05). Data for A–C reflect n = 8 mice per group. D–N Analyses utilized five distinct tissues randomly selected per xenograft (n = 5). Scale bars represent 100 µm. The scale bar represents 100 μm.

Consistent with the extensive in vitro findings, a pronounced impairment in mitochondrial function was observed within the NDUFS4-silenced xenografts. This impairment was characterized by reduced mitochondrial Complex I activity (Fig. 10G) and substantially decreased ATP contents (Fig. 10H). Furthermore, compelling evidence of increased oxidative stress was present, including significantly decreased GSH/GSSG ratio (Fig. 10I) and markedly elevated TBARS activity (Fig. 10J). Importantly, the protein expression of COX5B, the key downstream protein, was also downregulated in NDUFS4-silenced xenograft tissues (Fig. 10K). Histological examination of the xenograft tissues revealed significantly inhibited proliferation within the NDUFS4-silenced P1 xenograft tumors, as quantitatively indicated by a substantial reduction in Ki-67 immunohistochemical staining (Fig. 10L). A moderate increase in apoptosis was also observed, evidenced by elevated levels of cleaved-PARP and cleaved-Caspase-3 (Fig. 10M), and was further confirmed by increased nuclear TUNEL staining (Fig. 10N). These findings collectively demonstrated that NDUFS4 silencing effectively inhibited P1 glioma growth in vivo, impairing mitochondrial function, hindering tumor proliferation and promoting apoptotic cell death.

Discussion

Finding new therapeutic targets are important for treating glioma, an aggressive and challenging CNS cancer^1,2. Despite some progress with current treatments, patient prognoses remain grim, highlighting an urgent need for innovative strategies^1,2. Mitochondrial proteins are particularly promising candidates for targeted therapies in cancer. This is because they play central roles in cell metabolism, energy production, and regulating the processes that lead to cell death^11,31,32. Dickinson et al., showed that sustained mitochondrial DNA (mtDNA) depletion reduced proliferation, and delayed in vivo tumor formation, demonstrating mtDNA’s role in GBM tumorigenesis³². Using label-free liquid chromatography mass spectrometry (LCMS) on GBM mitochondrial fractions, Deighton and colleagues identified 117 significantly altered mitochondrial proteins³¹. In GBM, horizontal mitochondrial transfer from astrocytes to tumor cells enhanced tumorigenesis by increasing mitochondrial respiration and upregulating proliferation-linked metabolic pathways, thereby promoting cell cycle progression, self-renewal, and overall tumor aggressiveness³³.

The precise contribution of NDUFS4 to the malignant phenotype across various cancers is not comprehensively defined. Elevated NDUFS4 expression in gastric carcinoma is significantly associated with advanced tumor stage and diminished patient survival, and its functional downregulation impeded malignant cellular behaviors, including proliferation, migration, and invasion, both in vitro and in xenograft models¹⁹. These data underscore the potential of NDUFS4 as a prognostic indicator and a promising therapeutic target for gastric cancer¹⁹. The results in the current study strongly supports NDUFS4 as a promising therapeutic target for glioma. TCGA analyses consistently show elevated NDUFS4 expression in glioma tissues, correlating with more advanced tumor grade and shorter patient survival. Single-cell RNA sequencing provided further detail, pinpointing this increased expression primarily to glioma cells. NDUFS4 co-expressed genes are associated with cellular respiration and mitochondrial ATP synthesis. These findings were corroborated in patient tissues and various primary and established glioma cell types, confirming consistent NDUFS4 overexpression.

Our genetic manipulation studies unequivocally demonstrated NDUFS4’s functional importance in maintaining glioma cell progression. Both shRNA-mediated silencing and CRISPR/Cas9-mediated knockout of NDUFS4 led to a potent anti-glioma effect: a marked suppression of cell proliferation, migration, and invasion, coupled with a moderate increase in cell apoptosis. Importantly, this inhibitory effect was specific to malignant cells, sparing non-cancerous astrocytes. Conversely, exogenously increasing NDUFS4 expression in primary and immortalized glioma cells further exacerbated aggressive malignant characteristics. In vivo, NDUFS4 silencing effectively impeded intracranial growth of patient-derived glioma xenografts. These results established NDUFS4 as a critical driver of glioma progression.

It is noteworthy that while high NDUFS4 expression serves as a robust prognostic indicator in the general glioma cohort, this survival difference is attenuated when patients are stratified by IDH mutation status. Our data indicate that NDUFS4 expression is significantly higher in IDH-wildtype tumors compared to IDH-mutants; thus, NDUFS4 acts as a hallmark of the more malignant molecular subtype rather than an independent prognostic factor within the subtype itself. The lack of survival stratification within the IDH-wildtype cohort likely reflects a ‘ceiling effect,’ where the uniformly high metabolic demand and aggressiveness of these tumors overshadow subtle variations in NDUFS4 levels. The prognostic independence does not equate to functional redundancy. Our experimental data unequivocally demonstrate that IDH-wildtype glioma cells harbor a critical dependency on NDUFS4. The genetic depletion of NDUFS4 collapsed mitochondrial respiration, induced oxidative stress, and abrogated tumor growth in vitro and in vivo. Therefore, NDUFS4 remains biologically significant not merely as a biomarker, but as an essential driver of mitochondrial bioenergetics and a requisite vulnerability for the proliferation of IDH-wildtype gliomas, validating its potential as a therapeutic target.

Mitochondrial Complex I deficiency due to NDUFS4 mutations causes Leigh syndrome, a severe neurological disorder, and NDUFS4 knockout mouse models are being utilized to study its pathogenesis and potential treatments¹⁵. In NDUFS4-deficient mice, reduced mitochondrial respiration in the hippocampus correlated with decreased presynaptic protein synaptophysin levels, and in vitro studies suggested that NDUFS4 silencing impaired neurite outgrowth and ERK signaling, indicating its potential role in neuroplasticity¹⁷. Sonsalla et al., showed that mitochondrial dysfunction due to NDUFS4 mutations in patient-derived astrocytes impaired their direct reprogramming into neurons by triggering the unfolded protein response, which can be overcome by transiently inhibiting this stress pathway¹⁸. Mise et al., reported that overexpression of NDUFS4 in podocytes of diabetic mice improved mitochondrial structure and function, reducing albuminuria, and suggesting NDUFS4 as a potential therapeutic target for diabetic kidney disease³⁴. Myeloid-specific NDUFS4 deletion, impairing mitochondrial Complex I in macrophages, led to a persistent pro-inflammatory state, defective efferocytosis, and impaired post-myocardial infarction cardiac repair in mice, which can be rescued by reducing mitochondrial ROS³⁵.

Here, the functional importance of NDUFS4 in maintaining glioma cell mitochondrial activity and driving disease progression was demonstrated by our genetic manipulation studies. Both shRNA-mediated silencing and CRISPR/Cas9-mediated knockout of NDUFS4 consistently caused a profound impairment of mitochondrial function in glioma cells. We quantitatively evidenced this impairment by significant reductions in OCR, Complex I activity and ATP production, alongside an observable increase in oxidative stress and lipid peroxidation. Conversely, increasing NDUFS4 expression in primary and immortalized glioma cells boosted mitochondrial functions, increasing Complex I activity and ATP production. Importantly, NDUFS4 shRNA-induced anti-glioma cell activity was significantly alleviated following treatment with NAC or glucose supplement. In NDUFS4-silenced P1 xenograft tissues, compromised mitochondrial functions were also detected. These data underscore NDUFS4’s indispensable role in sustaining the mitochondrial function in glioma cells.

Further investigation, combining multi-omics data with experimental verification, identified COX5B as a critical downstream effector of NDUFS4. COX5B is an essential component of the mitochondrial electron transport chain, fundamentally contributing to ATP synthesis³⁶. COX5B is significantly downregulated in cryptorchid testes, and its reduction led to cellular senescence, impaired proliferation, and mitochondrial dysfunction³⁷. Elevated COX5B promoted breast cancer progression; its downregulation suppressed breast cancer cell growth and induced senescence via mitochondrial dysfunction and metabolic disorders^28,29. Through gene expression analysis, three oncogenic pathways (MAPK, Wnt, and ErbB) were identified as deregulated in most human gliomas, with COX5B emerging as a central gene and found to be significantly upregulated in glioma tissues, correlating with clinical stage and lymph node status²⁷.

Our findings conclusively showed that shRNA-induced silencing of COX5B mimicked the detrimental outcomes observed when NDUFS4 was depleted in primary glioma cells. Crucially, restoring COX5B in NDUFS4-deficient glioma cells effectively ameliorated mitochondrial dysfunction and the anti-glioma effects. This supported COX5B’s direct and indispensable involvement in mediating NDUFS4-promoted glioma cell progression. Complementary in vivo studies, using patient-derived glioma xenografts, further corroborated these findings, and COX5B protein expression was decreased in NDUFS4-silenced intracranial P1 xenograft tissues. Therefore, COX5B is a key molecule involved in glioma progression, acting downstream of NDUFS4.

While NDUFS4 inhibition directly compromises Complex I, our data shows it also leads to a significant downregulation of COX5B (Complex IV) in glioma cells. The electron transport chain functions as a coupled circuit; therefore, a deficiency in Complex IV will create a downstream bottleneck in the electron transport chain that forces electrons to back up, thereby causing significant indirect ROS leakage at the upstream complexes^29,38–40. Consequently, while COX5B is not a direct site of ROS generation, its restoration in NDUFS4-depleted cells can alleviate this downstream blockade. By re-establishing efficient electron flow through the terminal oxidase, the upstream accumulation of reduced intermediates is diminished, thereby mitigating the conditions that favor ROS generation.

NDUFS4 possibly regulates COX5B’s transcriptional activity by modulating the binding of the Sp1 transcription factor to the COX5B promoter. While NDUFS4 depletion significantly reduced Sp1 binding to the COX5B promoter, total Sp1 protein levels remained unchanged, suggesting a regulatory mechanism affecting Sp1 binding affinity rather than its abundance. Sp1 activity is known to be tightly regulated by post-translational modifications^41,42, which dictates its DNA-binding capacity. Since NDUFS4 silencing disrupted mitochondrial respiration and elevated oxidative stress in glioma cells, it is likely that the resultant mitochondrial retrograde signaling activates specific stress-response proteins and/or alters the cellular redox state. These intracellular changes may induce post-translational modifications of Sp1 or modulate the recruitment of co-factors, thereby diminishing its affinity for the COX5B promoter without altering its total expression. Future studies will focus on dissecting the specific signaling cascades linking mitochondrial dysfunction to Sp1 modification in this context.

The vital role of NDUFS4 in mitochondrial respiration across all cell types warrants consideration of potential systemic toxicity when targeting this protein. While a systemic NDUFS4 inhibitor could cause metabolic harm in healthy tissues, our study identified a potential therapeutic window: NDUFS4 depletion specifically impaired malignant glioma cells while sparing non-cancerous astrocytes. This suggests that glioma cells may have a heightened dependence on NDUFS4, a concept consistent with oncogene addiction. Future research should therefore focus on leveraging this selective vulnerability, including the development of targeted delivery systems and further investigation into the precise metabolic mechanisms underlying this differential sensitivity. These efforts will be crucial for optimizing a clinical strategy that maximizes anti-tumor efficacy while mitigating off-target effects.

In summary, our comprehensive preclinical investigation unequivocally establishes NDUFS4 as an essential contributor in glioma progression. Its oncogenic influence is largely mediated through its critical role in mitochondrial function and via COX5B. These compelling data collectively position NDUFS4 as a highly attractive and promising therapeutic target for this aggressive malignancy.

Methods

Reagents

Fluorescent dyes, encompassing TUNEL, DAPI, EdU, CellROX, MitoSOX, JC-1, succinate, and DCF-DA, were obtained from Thermo-Fisher Invitrogen (Suzhou, China). Antibodies were sourced from Cellular Signaling Tech (Danvers, MA) and Abcam (Cambridge, United Kingdom). Genechem (Shanghai, China) provided the validated mRNA primers, as well as shRNA, knockout and overexpression viral constructs. Serum, FBS, high-glucose medium, and antibiotics were supplied by Hyclone (Logan, UT). Various chemical reagents, including N-acetyl cysteine (NAC) and glucose were procured from Sigma-Aldrich (St. Louis, MO).

Human tissues and cells

Human tissues were procured, following informed consent, from a cohort of sixteen (n = 16) primary patients. These individuals contributed high-grade glioma (HGG) tissues and corresponding adjacent brain tissues, as previously documented^23–26. Furthermore, Dr. Cao provided the primary human glioma cells, derived from three consented patients (designated “P1”, “P2”, and “P3”), the A172, U87 and T98 immortalized glioma cell lines, and primary human astrocytes (termed “Astrocytes1/2”) originating from patients “P1” and “P2”^23–26. All the primary human glioma cells (P1-P3) used in this study are GBM cells characterized by the following molecular status: IDH wild-type, PTEN depletion, EGFR amplification and p53 wild-type. The primary human glioma cells were cultured in DMEM with 10% FBS supplemented with non-essential amino acids, EGF and bFGF at 25 ng/mL each, 2 μg/mL heparin, L-glutamine, and antibiotics (1% penicillin/streptomycin), with medium changes every 3–4 days. The primary human astrocytes and all glioma cell lines were cultured in DMEM with 10% FBS supplemented with non-essential amino acid and antibiotics. All experimental procedures involving human biological materials strictly conformed to the tenets of the Declaration of Helsinki and received approval from the Ethics Board of Soochow University (SY-2023-LL057).

Short hairpin RNA (shRNA)

For the transcriptional attenuation of NDUFS4, lentivirus (GV369 construct, Genechem, Shanghai, China)-packaged NDUFS4 shRNAs were deployed. These included shNDUFS4-sh1 (targeting: TGGAGTTCCAGAAGAGCATATAAAA), shNDUFS4-sh2 (targeting: TCTTTGTTCCTGCTCGCAATAACAT), and shNDUFS4-sh3 (targeting: ACTCAAGACACACAACTCATAACAGT). Glioma cells or astrocytes underwent viral transduction at a multiplicity of infection (MOI) of 11 for 32 h, followed by puromycin-mediated selective pressure over five passages. Control cells received transduction with a non-targeting scramble control shRNA (designated “shC”). Subsequent to selection, NDUFS4 expression in the resultant stable cells was rigorously verified at both the mRNA and protein levels. The same lentiviral construct and treatment procedure were employed to achieve COX5B (cytochrome c oxidase subunit 5B) knockdown (shCOX5B), targeting CTGCATCTGTGAAGAGGACAATACC, or NUDFS1 knockdown (shNUDFS1), targeting TCCACCTCAGCTAACTATAAA.

siRNA

For the silencing of NDUFS4, P1 glioma cells were transfected with 200 nM of a validated NDUFS4 siRNA (sequence: 5’-GAGUUGGACCUAUUAUGUU) using Lipofectamine 3000. To achieve robust knockdown, two successive rounds of transfection were performed, separated by a 24-hour incubation period. Successful depletion of NDUFS4 protein was verified via Western blotting assay.

NDUFS4 knockout (KO)

NDUFS4 knockout (KO) was achieved in glioma cells through the utilization of a CRISPR-associated protein 9 (Cas9)-expressing construct (from Dr. Cao^23–25). Lentivirus-packaged CRISPR/Cas9-NDUFS4-KO constructs, harboring sgRNAs directed against human NDUFS4 (targeting: TCGGAACCCTGGAAACGGAA, PAM: AGG/koNDUFS4-sg1, or GGCGGCGGTGTCAATGTCAG, PAM: TGG/koNDUFS4-sg2), were employed for cellular infection. This was succeeded by five passages of puromycin selection. Cells were subsequently aliquoted into 96-well plates, and the successful NDUFS4 knockout was corroborated via Western blotting assays, leading to the establishment of NDUFS4 knockout glioma cells (“koNDUFS4”). Control glioma cells were stably transduced with a lentiviral CRISPR/Cas9-control construct incorporating a non-targeting scramble sgRNA (designated “koC”).

Gene overexpression

To facilitate the ectopic NDUFS4 or COX5B expression, glioma cells were subjected to lentiviral infection with a construct encoding NDUFS4 (hNDUFS4[NM_002495.4]) or COX5B (hCOX5B[NM_001862.3]) cDNA sequence within a GV248 vector (tag-free, Genechem, Shanghai, China). Following puromycin (3.5 μg/mL) selection across five passages, stable cells, denoted “oeNDUFS4” or “oeCOX5B”, were successfully generated. Control glioma cells received transduction with the control lentiviral empty vector (termed “Vec”). The augmented NDUFS4/COX5B expression in these stable cells was subsequently confirmed at both the mRNA and protein levels.

Western blotting

Following protein extraction from cultured cells or tissues, quantification was performed using a Thermo-Fisher Scientific protein assay kit. Equivalent protein amounts (20–30 μg lysates per sample) were separated by 8–12% SDS-PAGE, then transferred to a PVDF blot at 4 °C overnight. Proteins were sequentially incubated with primary and secondary antibodies, also at 4 °C. Immunoreaction signals were visualized via the Amersham ECL plus Western blotting system. All uncropped blotting images are in Fig. S1.

Quantitative polymerase chain reaction (qPCR)

After cell or tissue collection, total RNA was extracted using TRIzol, with concentration assessed by a Nanodrop 2000 spectrophotometer. Reverse transcription, utilizing the Promega M kit, produced cDNA. qPCR then commenced with the SYBR Green PCR kit (Thermo Fisher Scientific). Gene quantification, employing the 2-ΔΔCq method, used GAPDH as the internal control.

Assessment of mitochondrial complex I activity and ATP levels

We precisely quantified the mitochondrial Complex I enzymatic activity using a commercial Sigma kit^43–45. This method involved spectrophotometrically quantifying NADH’s oxidation to NAD + , a direct catalytic function of Complex I. Activity was quantitatively gauged by the concomitant diminution in absorbance at 435 nm. ATP concentrations were quantified with a commercial Sigma colorimetric kit, strictly following its prescribed procedural mandates. Each assay used a 30 µL aliquot of cellular or tissue homogenates (30 µg total protein).

GSH to GSSG ratio

The cellular redox state was quantified by the ratio of reduced glutathione (GSH) to oxidized glutathione (GSSG) using a Thermo-Fisher Scientific kit (Suzhou, China). Lysates were incubated with 5,5’-Dithio-bis (2-nitrobenzoic acid) (DTNB), glutathione reductase, and NADPH in a defined reaction milieu. We spectrophotometrically monitored the absorbance at 435 nm over a 4–5-minute interval. A standard curve, generated from calibrated GSH and GSSG standards, enabled precise quantification of their respective concentrations. Each analysis utilized 30 µL of cellular or tissue homogenates (30 µg total protein).

Thiobarbituric acid reactive substances (TBARS) assay

A TBARS assay kit, procured from Thermo-Fisher Scientific (Suzhou, China), was employed for the assessment of lipid peroxidation. Tissue or cellular protein lysates reacted with thiobarbituric acid (TBA) to engender the TBARS complex. Following this conjugation and subsequent expurgation of precipitates via refrigeration and centrifugation, the optical density at 535 nm was spectrophotometrically ascertained. Each assay analyzed a 30 µL aliquot of cellular or tissue homogenates (30 µg total protein).

Oxygen consumption rate (OCR)

OCR was measured using an Agilent Seahorse XF24 Extracellular Flux Analyzer, strictly following previously established protocols⁴⁶. Cellular respiration was assessed by sequentially perfusing cells with metabolic modulators: 1 µM oligomycin, followed by 0.5 µM FCCP (carbonyl cyanide-p-trifluoromethoxyphenylhydrazone), and concluding with a combined application of 0.5 µM of antimycin A and 0.5 µM of rotenone. This systematic pharmacological perturbation allowed for the quantification of basal, ATP-linked, maximal, and non-mitochondrial OCR components. All derived OCR values were normalized to the corresponding intracellular protein content.

CCK-8 and colony formation assays

For cell viability assays, genetically modified glioma cells or astrocytes were seeded in 96-well plates at a density of 4000 cells per well and incubated for 96 h. Following this, a CCK-8 mixture was added, and after an additional 75-minute incubation, absorbance was measured at 450 nm using a microplate reader. For the colony formation assay, a diluted single-cell suspension was prepared and seeded into 6-well plates containing complete growth medium. The plates were gently swirled to ensure even cell distribution across the well surface. They were then incubated for ten days at 37 °C in a humidified atmosphere with 5% CO₂, after which colonies were manually counted.

EdU staining

The Apollo-567 EdU Kit (RiboBio, Wuxi, China) was employed. Cells were seeded into 24-well plates (5×10⁴ cells/well) for specified durations, treated with 10 mM EdU for 2 h, and then fixed with 4% paraformaldehyde. Concurrently, DAPI was utilized to stain cell nuclei for 5–6 minutes. EdU-positive nuclei were visualized using a ZEISS LSM fluorescence microscope. The EdU ratio (EdU/DAPI × 100%) was then determined.

TUNEL staining

Cells were seeded into 24-well plates (5×10⁴ cells/well) for specified durations. After 15-minute fixation in 4% paraformaldehyde and 5-minute Triton treatment, cells were exposed to TUNEL detection solution (Beyotime Biotechnology, Wuxi, China) at 37°C for 1 h. DAPI then stained nuclei for 5–6 min. TUNEL-positive nuclei were visualized via a ZEISS LSM fluorescence microscope.

Caspase-3 activity assay

The caspase-3 activity was measured in cultured glioma cells using a fluorometric assay kit (e.g., specific manufacturer, Cat. No.). Cells were seeded in a 96-well plate and cultured for 72 h. A fixed amount of lysate protein (30 µg per sample) was incubated with the caspase-3-specific fluorogenic substrate (Ac-DEVD-AMC, Sigma-Aldrich) in the attached reaction buffer. The cleavage of the substrate by activated caspase-3 releases a fluorescent product, detected using a microplate reader at an excitation wavelength of 380 nm and an emission wavelength of 460 nm. Relative caspase-3 activity was normalized to that of control cells.

Other cellular fluorescence staining protocols

Cells were seeded into 24-well plates (5×10⁴ cells/well) for specified durations. Following a pre-specified incubation period, cells were chemically fixed with 4% paraformaldehyde and thoroughly washed. Subsequently, cells were incubated with specific fluorescence probes and subjected to additional washes. Visualization was conducted using a Leica fluorescence microscope, and fluorescence emission intensity was quantitatively measured with a Hitachi F-7000 spectrophotometer.

Migration/invasion assays

Glioma cells, subjected to the described genetic treatments, were re-suspended in serum-free medium and seeded onto “Transwell” chamber surfaces (Corning, Suzhou, China). The lower chamber compartment received medium with 10% FBS. After 20 h, glioma cells on the upper surface were removed. Membranes on the lower chamber were then fixed with 4% paraformaldehyde and stained with crystal violet to quantify migrated cells. For invasion assays, the chambers were always coated with Matrigel.

Chromatin immunoprecipitation (ChIP)

To investigate the interaction between the transcription factor SP1 and the COX5B promoter, a ChIP assay was performed based on established protocols^23,47. The process began by cross-linking and fragmenting genomic DNA from cell lysates, effectively capturing protein-DNA complexes. An anti-Sp1 antibody was then used to specifically immunoprecipitate DNA fragments bound to Sp1. The resulting enriched DNA was subsequently quantified using quantitative PCR to measure the association between Sp1 and the proposed target NDUFS4 promoter sequence (CCGAAAG)³⁰. Final results were standardized by normalizing them to control samples.

Xenograft studies

The glioma xenograft experiments utilized 5–6-week-old athymic nude mice (18.1–18.5 g), equally mixed male and female. Mice were housed in ventilated cages under controlled conditions with ad libitum access to food and water. For glioma cell implantation, mice were anesthetized via intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). To establish the patient-derived glioma xenograft (PDX) model, P1 glioma cells were directly administered into the brains of nude mice following established protocols^23–25,48. Post-injection, mice recovered for at least 2 h until fully conscious. Intracranial P1 glioma xenografts subsequently formed, visualized by Magnetic Resonance Imaging (MRI). Mice were randomized to treatment groups, and data collection was blinded to minimize bias. Xenograft tissues were collected after humane euthanasia by CO₂ asphyxiation followed by cervical dislocation. All animal procedures received ethical clearance from both the Institutional Animal Care and Use Committee (IACUC) and the Animal Ethics Review Board of Soochow University.

Immunohistochemistry (IHC)

The protocols were described in an early study⁴⁹. Briefly, paraffin-embedded tissue slides underwent heat incubation, dewaxing, and rehydration. Tissues were then treated with citric acid buffer and hydrogen peroxide. Slides were exposed to the primary antibody for 6–8 h at 4°C, then heated, washed, and incubated with a biotin-labeled secondary antibody for 1 h followed by horseradish-labeled streptavidin. The diaminobenzidine (DAB) process was performed, and IHC images captured.

Tissue TUNEL fluorescence staining

Paraffin-embedded tissue slides were heated, dewaxed, and rehydrated. They then received citric acid buffer treatment at 95°C for 12 min, followed by washes with cold PBS. Subsequently, slides were blocked with goat serum, exposed to specific TUNEL and DAPI fluorescence dyes, washed again, and finally visualized under a ZEISS confocal microscope.

Statistical analysis

In vitro experiments were replicated five times (biological repeats). Data, conforming to normal distributions, were always presented as mean ± standard deviation (SD). We assessed statistical differences between two distinct groups using a two-tailed Student’s t-test (Excel 2010). For comparisons among multiple groups, we employed ANOVA analysis with a Student-Newman-Keuls post hoc test (SPSS 23.0). P values < 0.05 were utilized to indicate statistical significance.

Supplementary information

Supplementary Information^{(2.8MB, pdf)}

Acknowledgements

This work is supported by National Natural Science Foundation of China (82273055) and Scientific Research Project of Jiangsu Provincial Health Commission (Z2022044).

Author contributions

J.W., J.L., Y.L., and L.J. participated in the conception and design of the study. J.W., J.L., L.X., and L.J. performed experiments. J.W., J.L., L.X., Y.L., and L.J. drafted the manuscript and contributed to its critical revision. All listed authors provided substantial intellectual input and granted final approval of the submitted version to the journal.

Data availability

Data are provided within the manuscript or supplementary information files.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Jiang Wu, Juan Li, Li Xu.

Contributor Information

Yuanyuan Liu, Email: liuyuanyuansz@hotmail.com.

Li Jiang, Email: jiangli@hospital.westlake.edu.cn.

Supplementary information

The online version contains supplementary material available at 10.1038/s41698-026-01281-9.

References

1.Weller, M. et al. Glioma. Nat. Rev. Dis. Prim.10, 33 (2024). [DOI] [PubMed] [Google Scholar]
2.Weller, M. et al. Glioma. Nat. Rev. Dis. Prim.1, 15017 (2015). [DOI] [PubMed] [Google Scholar]
3.Siegel, R. L., Miller, K. D., Wagle, N. S. & Jemal, A. Cancer statistics, 2023. CA Cancer J. Clin.73, 17–48 (2023). [DOI] [PubMed] [Google Scholar]
4.Omuro, A. & DeAngelis, L. M. Glioblastoma and other malignant gliomas: a clinical review. JAMA310, 1842–1850 (2013). [DOI] [PubMed] [Google Scholar]
5.Reifenberger, G., Wirsching, H. G., Knobbe-Thomsen, C. B. & Weller, M. Advances in the molecular genetics of gliomas - implications for classification and therapy. Nat. Rev. Clin. Oncol.14, 434–452 (2017). [DOI] [PubMed] [Google Scholar]
6.Lim, M., Xia, Y., Bettegowda, C. & Weller, M. Current state of immunotherapy for glioblastoma. Nat. Rev. Clin. Oncol.15, 422–442 (2018). [DOI] [PubMed] [Google Scholar]
7.Hotchkiss, K. M. et al. A brave new framework for glioma drug development. Lancet Oncol.25, e512–e519 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Yang, K. et al. Glioma targeted therapy: insight into future of molecular approaches. Mol. Cancer21, 39 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Xu, S., Tang, L., Li, X., Fan, F. & Liu, Z. Immunotherapy for glioma: current management and future application. Cancer Lett.476, 1–12 (2020). [DOI] [PubMed] [Google Scholar]
10.Chen, L., Zhang, H., Shang, C. & Hong, Y. The role and applied value of mitochondria in glioma-related research. CNS Neurosci. Ther.30, e70121 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Ordys, B. B., Launay, S., Deighton, R. F., McCulloch, J. & Whittle, I. R. The role of mitochondria in glioma pathophysiology. Mol. Neurobiol.42, 64–75 (2010). [DOI] [PubMed] [Google Scholar]
12.Iuso, A. et al. Dysfunctions of cellular oxidative metabolism in patients with mutations in the NDUFS1 and NDUFS4 genes of complex I. J. Biol. Chem.281, 10374–10380 (2006). [DOI] [PubMed] [Google Scholar]
13.van de Wal, M. A. E. et al. Ndufs4 knockout mice with isolated complex I deficiency engage a futile adaptive brain response. Biochim. Biophys. Acta Proteins Proteom.1873, 141055 (2025). [DOI] [PubMed] [Google Scholar]
14.Yin, Z., Agip, A. A., Bridges, H. R. & Hirst, J. Structural insights into respiratory complex I deficiency and assembly from the mitochondrial disease-related ndufs4(-/-) mouse. EMBO J.43, 225–249 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.van de Wal, M. A. E. et al. Ndufs4 knockout mouse models of Leigh syndrome: pathophysiology and intervention. Brain145, 45–63 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Ortigoza-Escobar, J. D. et al. Ndufs4 related Leigh syndrome: a case report and review of the literature. Mitochondrion28, 73–78 (2016). [DOI] [PubMed] [Google Scholar]
17.Shil, S. K. et al. Ndufs4 ablation decreases synaptophysin expression in hippocampus. Sci. Rep.11, 10969 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Sonsalla, G. et al. Direct neuronal reprogramming of NDUFS4 patient cells identifies the unfolded protein response as a novel general reprogramming hurdle. Neuron112, 1117–1132 e1119 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Cheng, T. et al. NDUFS4 promotes tumor progression and predicts prognosis in gastric cancer. Carcinogenesis43, 980–987 (2022). [DOI] [PubMed] [Google Scholar]
20.Kloosterhof, N. K., Bralten, L. B., Dubbink, H. J., French, P. J. & van den Bent, M. J. Isocitrate dehydrogenase-1 mutations: a fundamentally new understanding of diffuse glioma?. Lancet Oncol.12, 83–91 (2011). [DOI] [PubMed] [Google Scholar]
21.Han, S. et al. IDH mutation in glioma: molecular mechanisms and potential therapeutic targets. Br. J. Cancer122, 1580–1589 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Abdelfattah, N. et al. Single-cell analysis of human glioma and immune cells identifies S100A4 as an immunotherapy target. Nat. Commun.13, 767 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Wang, Y. et al. G protein inhibitory alpha subunit 2 is a molecular oncotarget of human glioma. Int. J. Biol. Sci.19, 865–879 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Liu, F. et al. YME1L overexpression exerts pro-tumorigenic activity in glioma by promoting Galphai1 expression and Akt activation. Protein Cell14, 223–229 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Guo, Y. Z. et al. TIMM44 is a potential therapeutic target of human glioma. Theranostics12, 7586–7602 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Wang, Y. et al. Neuronal-driven glioma growth requires Galphai1 and Galphai3. Theranostics11, 8535–8549 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Hu, T. & Xi, J. Identification of COX5B as a novel biomarker in high-grade glioma patients. Onco Targets Ther.10, 5463–5470 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Gao, S. P. et al. High expression of COX5B is associated with poor prognosis in breast cancer. Future Oncol.13, 1711–1719 (2017). [DOI] [PubMed] [Google Scholar]
29.Gao, S. P. et al. Loss of COX5B inhibits proliferation and promotes senescence via mitochondrial dysfunction in breast cancer. Oncotarget6, 43363–43374 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Bachman, N. J., Yang, T. L., Dasen, J. S., Ernst, R. E. & Lomax, M. I. Phylogenetic footprinting of the human cytochrome c oxidase subunit VB promoter. Arch. Biochem. Biophys.333, 152–162 (1996). [DOI] [PubMed] [Google Scholar]
31.Deighton, R. F. et al. Interactions among mitochondrial proteins altered in glioblastoma. J. Neurooncol.118, 247–256 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Dickinson, A. et al. The regulation of mitochondrial DNA copy number in glioblastoma cells. Cell Death Differ.20, 1644–1653 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Watson, D. C. et al. GAP43-dependent mitochondria transfer from astrocytes enhances glioblastoma tumorigenicity. Nat. Cancer4, 648–664 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Mise, K. et al. NDUFS4 regulates cristae remodeling in diabetic kidney disease. Nat. Commun.15, 1965 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Cai, S. et al. Mitochondrial dysfunction in macrophages promotes inflammation and suppresses repair after myocardial infarction. J. Clin. Invest.133, e159498 (2023). [DOI] [PMC free article] [PubMed]
36.Chu, Y. D., Lim, S. N., Yeh, C. T. & Lin, W. R. COX5B-mediated bioenergetic alterations modulate cell growth and anticancer drug susceptibility by orchestrating claudin-2 expression in colorectal cancers. Biomedicines10, 60 (2021). [DOI] [PMC free article] [PubMed]
37.Wang, X., Miao, D., Ye, S., Xian, H. & Ge, W. Abnormal expression of COX5B gene and disorder of mitochondrial function in cryptorchid rats. J. Cell Mol. Med.28, e70234 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Mansilla, N., Racca, S., Gras, D. E., Gonzalez, D. H. & Welchen, E. The complexity of mitochondrial complex IV: an update of cytochrome c oxidase biogenesis in plants. Int. J. Mol. Sci.19, 662 (2018). [DOI] [PMC free article] [PubMed]
39.Hinkelbein, J. et al. Decreased tissue COX5B expression and mitochondrial dysfunction during sepsis-induced kidney injury in rats. Oxid. Med. Cell Longev.2017, 8498510 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Bourens, M., Fontanesi, F., Soto, I. C., Liu, J. & Barrientos, A. Redox and reactive oxygen species regulation of mitochondrial cytochrome C oxidase biogenesis. Antioxid. Redox Signal19, 1940–1952 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Sun, X. et al. Role of post-translational modifications of Sp1 in cancer: state of the art. Front. Cell Dev. Biol.12, 1412461 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Chang, W. C. & Hung, J. J. Functional role of post-translational modifications of Sp1 in tumorigenesis. J. Biomed. Sci.19, 94 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Zhao, Y. et al. Mitochondrial carrier homolog 2 is important for mitochondrial functionality and non-small cell lung cancer cell growth. Cell Death Dis.16, 95 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Zha, J., Li, J., Yin, H., Shen, M. & Xia, Y. TIMM23 overexpression drives NSCLC cell growth and survival by enhancing mitochondrial function. Cell Death Dis.16, 174 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Cheng, F., Huang, H., Yin, S., Liu, J. S. & Sun, P. Expression and functional implications of YME1L in nasopharyngeal carcinoma. Cell Death Dis.15, 423 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Ding, M. et al. Mfn2-mediated mitochondrial fusion alleviates doxorubicin-induced cardiotoxicity with enhancing its anticancer activity through metabolic switch. Redox Biol.52, 102311 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Yao, J. et al. The requirement of phosphoenolpyruvate carboxykinase 1 for angiogenesis in vitro and in vivo. Sci. Adv.8, eabn6928 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Liu, Y. Y. et al. microRNA-200a downregulation in human glioma leads to Galphai1 over-expression, Akt activation, and cell proliferation. Oncogene37, 2890–2902 (2018). [DOI] [PubMed] [Google Scholar]
49.Qian, J. et al. COL8A1 overexpression promotes glioma cell growth by activating focal adhesion kinase signaling cascade. NPJ Precis. Oncol.8, 273 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information^{(2.8MB, pdf)}

Data Availability Statement

Data are provided within the manuscript or supplementary information files.

[CR1] 1.Weller, M. et al. Glioma. Nat. Rev. Dis. Prim.10, 33 (2024). [DOI] [PubMed] [Google Scholar]

[CR2] 2.Weller, M. et al. Glioma. Nat. Rev. Dis. Prim.1, 15017 (2015). [DOI] [PubMed] [Google Scholar]

[CR3] 3.Siegel, R. L., Miller, K. D., Wagle, N. S. & Jemal, A. Cancer statistics, 2023. CA Cancer J. Clin.73, 17–48 (2023). [DOI] [PubMed] [Google Scholar]

[CR4] 4.Omuro, A. & DeAngelis, L. M. Glioblastoma and other malignant gliomas: a clinical review. JAMA310, 1842–1850 (2013). [DOI] [PubMed] [Google Scholar]

[CR5] 5.Reifenberger, G., Wirsching, H. G., Knobbe-Thomsen, C. B. & Weller, M. Advances in the molecular genetics of gliomas - implications for classification and therapy. Nat. Rev. Clin. Oncol.14, 434–452 (2017). [DOI] [PubMed] [Google Scholar]

[CR6] 6.Lim, M., Xia, Y., Bettegowda, C. & Weller, M. Current state of immunotherapy for glioblastoma. Nat. Rev. Clin. Oncol.15, 422–442 (2018). [DOI] [PubMed] [Google Scholar]

[CR7] 7.Hotchkiss, K. M. et al. A brave new framework for glioma drug development. Lancet Oncol.25, e512–e519 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Yang, K. et al. Glioma targeted therapy: insight into future of molecular approaches. Mol. Cancer21, 39 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Xu, S., Tang, L., Li, X., Fan, F. & Liu, Z. Immunotherapy for glioma: current management and future application. Cancer Lett.476, 1–12 (2020). [DOI] [PubMed] [Google Scholar]

[CR10] 10.Chen, L., Zhang, H., Shang, C. & Hong, Y. The role and applied value of mitochondria in glioma-related research. CNS Neurosci. Ther.30, e70121 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Ordys, B. B., Launay, S., Deighton, R. F., McCulloch, J. & Whittle, I. R. The role of mitochondria in glioma pathophysiology. Mol. Neurobiol.42, 64–75 (2010). [DOI] [PubMed] [Google Scholar]

[CR12] 12.Iuso, A. et al. Dysfunctions of cellular oxidative metabolism in patients with mutations in the NDUFS1 and NDUFS4 genes of complex I. J. Biol. Chem.281, 10374–10380 (2006). [DOI] [PubMed] [Google Scholar]

[CR13] 13.van de Wal, M. A. E. et al. Ndufs4 knockout mice with isolated complex I deficiency engage a futile adaptive brain response. Biochim. Biophys. Acta Proteins Proteom.1873, 141055 (2025). [DOI] [PubMed] [Google Scholar]

[CR14] 14.Yin, Z., Agip, A. A., Bridges, H. R. & Hirst, J. Structural insights into respiratory complex I deficiency and assembly from the mitochondrial disease-related ndufs4(-/-) mouse. EMBO J.43, 225–249 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.van de Wal, M. A. E. et al. Ndufs4 knockout mouse models of Leigh syndrome: pathophysiology and intervention. Brain145, 45–63 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Ortigoza-Escobar, J. D. et al. Ndufs4 related Leigh syndrome: a case report and review of the literature. Mitochondrion28, 73–78 (2016). [DOI] [PubMed] [Google Scholar]

[CR17] 17.Shil, S. K. et al. Ndufs4 ablation decreases synaptophysin expression in hippocampus. Sci. Rep.11, 10969 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Sonsalla, G. et al. Direct neuronal reprogramming of NDUFS4 patient cells identifies the unfolded protein response as a novel general reprogramming hurdle. Neuron112, 1117–1132 e1119 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.Cheng, T. et al. NDUFS4 promotes tumor progression and predicts prognosis in gastric cancer. Carcinogenesis43, 980–987 (2022). [DOI] [PubMed] [Google Scholar]

[CR20] 20.Kloosterhof, N. K., Bralten, L. B., Dubbink, H. J., French, P. J. & van den Bent, M. J. Isocitrate dehydrogenase-1 mutations: a fundamentally new understanding of diffuse glioma?. Lancet Oncol.12, 83–91 (2011). [DOI] [PubMed] [Google Scholar]

[CR21] 21.Han, S. et al. IDH mutation in glioma: molecular mechanisms and potential therapeutic targets. Br. J. Cancer122, 1580–1589 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Abdelfattah, N. et al. Single-cell analysis of human glioma and immune cells identifies S100A4 as an immunotherapy target. Nat. Commun.13, 767 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Wang, Y. et al. G protein inhibitory alpha subunit 2 is a molecular oncotarget of human glioma. Int. J. Biol. Sci.19, 865–879 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Liu, F. et al. YME1L overexpression exerts pro-tumorigenic activity in glioma by promoting Galphai1 expression and Akt activation. Protein Cell14, 223–229 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR25] 25.Guo, Y. Z. et al. TIMM44 is a potential therapeutic target of human glioma. Theranostics12, 7586–7602 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Wang, Y. et al. Neuronal-driven glioma growth requires Galphai1 and Galphai3. Theranostics11, 8535–8549 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR27] 27.Hu, T. & Xi, J. Identification of COX5B as a novel biomarker in high-grade glioma patients. Onco Targets Ther.10, 5463–5470 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Gao, S. P. et al. High expression of COX5B is associated with poor prognosis in breast cancer. Future Oncol.13, 1711–1719 (2017). [DOI] [PubMed] [Google Scholar]

[CR29] 29.Gao, S. P. et al. Loss of COX5B inhibits proliferation and promotes senescence via mitochondrial dysfunction in breast cancer. Oncotarget6, 43363–43374 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Bachman, N. J., Yang, T. L., Dasen, J. S., Ernst, R. E. & Lomax, M. I. Phylogenetic footprinting of the human cytochrome c oxidase subunit VB promoter. Arch. Biochem. Biophys.333, 152–162 (1996). [DOI] [PubMed] [Google Scholar]

[CR31] 31.Deighton, R. F. et al. Interactions among mitochondrial proteins altered in glioblastoma. J. Neurooncol.118, 247–256 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Dickinson, A. et al. The regulation of mitochondrial DNA copy number in glioblastoma cells. Cell Death Differ.20, 1644–1653 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Watson, D. C. et al. GAP43-dependent mitochondria transfer from astrocytes enhances glioblastoma tumorigenicity. Nat. Cancer4, 648–664 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Mise, K. et al. NDUFS4 regulates cristae remodeling in diabetic kidney disease. Nat. Commun.15, 1965 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Cai, S. et al. Mitochondrial dysfunction in macrophages promotes inflammation and suppresses repair after myocardial infarction. J. Clin. Invest.133, e159498 (2023). [DOI] [PMC free article] [PubMed]

[CR36] 36.Chu, Y. D., Lim, S. N., Yeh, C. T. & Lin, W. R. COX5B-mediated bioenergetic alterations modulate cell growth and anticancer drug susceptibility by orchestrating claudin-2 expression in colorectal cancers. Biomedicines10, 60 (2021). [DOI] [PMC free article] [PubMed]

[CR37] 37.Wang, X., Miao, D., Ye, S., Xian, H. & Ge, W. Abnormal expression of COX5B gene and disorder of mitochondrial function in cryptorchid rats. J. Cell Mol. Med.28, e70234 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR38] 38.Mansilla, N., Racca, S., Gras, D. E., Gonzalez, D. H. & Welchen, E. The complexity of mitochondrial complex IV: an update of cytochrome c oxidase biogenesis in plants. Int. J. Mol. Sci.19, 662 (2018). [DOI] [PMC free article] [PubMed]

[CR39] 39.Hinkelbein, J. et al. Decreased tissue COX5B expression and mitochondrial dysfunction during sepsis-induced kidney injury in rats. Oxid. Med. Cell Longev.2017, 8498510 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Bourens, M., Fontanesi, F., Soto, I. C., Liu, J. & Barrientos, A. Redox and reactive oxygen species regulation of mitochondrial cytochrome C oxidase biogenesis. Antioxid. Redox Signal19, 1940–1952 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Sun, X. et al. Role of post-translational modifications of Sp1 in cancer: state of the art. Front. Cell Dev. Biol.12, 1412461 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR42] 42.Chang, W. C. & Hung, J. J. Functional role of post-translational modifications of Sp1 in tumorigenesis. J. Biomed. Sci.19, 94 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Zhao, Y. et al. Mitochondrial carrier homolog 2 is important for mitochondrial functionality and non-small cell lung cancer cell growth. Cell Death Dis.16, 95 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Zha, J., Li, J., Yin, H., Shen, M. & Xia, Y. TIMM23 overexpression drives NSCLC cell growth and survival by enhancing mitochondrial function. Cell Death Dis.16, 174 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Cheng, F., Huang, H., Yin, S., Liu, J. S. & Sun, P. Expression and functional implications of YME1L in nasopharyngeal carcinoma. Cell Death Dis.15, 423 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Ding, M. et al. Mfn2-mediated mitochondrial fusion alleviates doxorubicin-induced cardiotoxicity with enhancing its anticancer activity through metabolic switch. Redox Biol.52, 102311 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Yao, J. et al. The requirement of phosphoenolpyruvate carboxykinase 1 for angiogenesis in vitro and in vivo. Sci. Adv.8, eabn6928 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Liu, Y. Y. et al. microRNA-200a downregulation in human glioma leads to Galphai1 over-expression, Akt activation, and cell proliferation. Oncogene37, 2890–2902 (2018). [DOI] [PubMed] [Google Scholar]

[CR49] 49.Qian, J. et al. COL8A1 overexpression promotes glioma cell growth by activating focal adhesion kinase signaling cascade. NPJ Precis. Oncol.8, 273 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Mitochondrial complex I subunit NDUFS4 overexpression drives glioma progression by regulating mitochondrial function and COX5B

Jiang Wu

Juan Li

Li Xu

Yuanyuan Liu

Li Jiang

Abstract

Introduction

Results

NDUFS4 expression is elevated in glioma and correlates with tumor grade and poor overall survival in TCGA-glioma data

Fig. 1. NDUFS4 expression is elevated in glioma and correlates with tumor grade and poor overall survival in TCGA-glioma data.

Single-cell RNA sequencing reveals elevated NDUFS4 expression in glioma cells and co-expressed genes involved in cellular respiration and energy metabolism

Fig. 2. Single-cell RNA sequencing reveals elevated NDUFS4 expression in glioma cells and co-expressed genes involved in cellular respiration and energy metabolism.

Elevated NDUFS4 expression in glioma tissues from locally-treated patients and various glioma cells

Fig. 3. Elevated NDUFS4 expression in glioma tissues from locally-treated patients and various glioma cells.

NDUFS4 silencing impairs mitochondrial bioenergetics and induces oxidative stress in glioma cells

Fig. 4. NDUFS4 silencing critically impairs mitochondrial bioenergetics and induces oxidative stress in glioma cells.

NDUFS4 silencing impairs malignant phenotypes of human glioma cells

Fig. 5. NDUFS4 silencing impairs malignant phenotypes of human glioma cells.

NDUFS4 knockout impairs mitochondrial function and malignant phenotypes in human glioma cells

Fig. 6. NDUFS4 knockout impairs mitochondrial function and malignant phenotypes in human glioma cells.

NDUFS4 overexpression enhances mitochondrial function and promotes malignant phenotypes in human glioma cells

Fig. 7. NDUFS4 overexpression enhances mitochondrial function and promotes malignant phenotypes in human glioma cells.

NDUFS4 correlation gene COX5B is important for maintaining mitochondrial homeostasis and promoting malignant phenotypes in glioma cells

Fig. 8. NDUFS4 correlation gene COX5B is important for maintaining mitochondrial homeostasis and promoting malignant phenotypes in glioma cells.

The critical role of COX5B as a downstream effector in NDUFS4-regulated glioma progression

Fig. 9. The critical role of COX5B as a downstream effector in NDUFS4-regulated glioma progression.

Intracranial growth of patient-derived glioma xenografts in nude mice is hindered after NDUFS4 silencing

Fig. 10. Intracranial growth of patient-derived glioma xenografts in nude mice is hindered after NDUFS4 silencing.

Discussion

Methods

Reagents

Human tissues and cells

Short hairpin RNA (shRNA)

siRNA

NDUFS4 knockout (KO)

Gene overexpression

Western blotting

Quantitative polymerase chain reaction (qPCR)

Assessment of mitochondrial complex I activity and ATP levels

GSH to GSSG ratio

Thiobarbituric acid reactive substances (TBARS) assay

Oxygen consumption rate (OCR)

CCK-8 and colony formation assays

EdU staining

TUNEL staining

Caspase-3 activity assay

Other cellular fluorescence staining protocols

Migration/invasion assays

Chromatin immunoprecipitation (ChIP)

Xenograft studies

Immunohistochemistry (IHC)

Tissue TUNEL fluorescence staining

Statistical analysis

Supplementary information

Acknowledgements

Author contributions

Data availability

Competing interests

Footnotes

Contributor Information

Supplementary information

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases