Elevated NDUFAF2 expression correlates with poor prognosis and drives metabolic dependency in hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide, with limited therapeutic options for advanced and immunotherapy-resistant cases. Mitochondrial oxidative phosphorylation (OXPHOS) has emerged as a key metabolic dependency in a subset of HCCs, yet the molecular regulators of this vulnerability remain poorly understood. Here, we report that NDUFAF2, a core assembly chaperone for mitochondrial Complex I, is significantly overexpressed across multiple human cancers, particularly in HCC, where elevated expression correlates with poor overall survival, progression-free interval, and disease-specific survival in large-scale TCGA and CPTAC cohorts. At the transcriptomic and proteomic levels, NDUFAF2 upregulation was consistently observed in primary HCC tissues compared with paired normal liver samples, and correlated with advanced histological grade, pathological stage, high AFP levels, and vascular invasion. Bioinformatic pathway analyses identified OXPHOS as the dominant biological process associated with NDUFAF2 expression. Mechanistically, stable NDUFAF2 knockdown in two independent HCC cell lines (HCC-LM3 and SNU-449) resulted in profound mitochondrial dysfunction, including reduced oxygen consumption rate, impaired ATP production, increased reactive oxygen species (ROS) accumulation, and compensatory upregulation of glycolytic genes. These metabolic defects ultimately led to suppressed cell proliferation in vitro. Preliminary correlative analyses further suggested weak associations between NDUFAF2 expression and immunosuppressive features in the tumor microenvironment, although these trends remain hypothesis-generating and require experimental validation. Collectively, our findings establish NDUFAF2 as a pivotal regulator of mitochondrial bioenergetics and malignant progression in HCC, identifying it as a promising prognostic biomarker and a potential target for metabolic therapy in OXPHOS-dependent liver cancers.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13062-026-00827-y.

Introduction

Hepatocellular carcinoma (HCC) is one of the most prevalent and fatal malignancies worldwide, ranking the sixth most commonly diagnosed cancer and the third leading cause of cancer-related death globally [1]. Despite continuous advances in clinical management, HCC treatment remains hindered by a lack of specific and actionable molecular targets. In contrast to lung cancer driven by EGFR mutations and ALK rearrangements [2], or HER2-amplified breast cancer with mature targeted regimens [3], HCC lacks widely applicable molecular alterations to guide precise targeted therapy. Current mainstream strategies mainly rely on anti-angiogenic agents and immune checkpoint inhibitors [4]. Although these approaches have modestly improved patient survival, they are not tumor-specific targeted therapies. Most patients eventually develop treatment resistance and disease progression, with limited alternative therapeutic options available. This clinical bottleneck highlights the urgent demand to identify novel and specific molecular vulnerabilities for targeted intervention in HCC.

Bioinformatics has become a mature and powerful strategy for mining potential oncogenic biomarkers and therapeutic targets. Through integrating high-throughput gene expression profiles, prognostic correlation analysis and multi-omics data, bioinformatics enables systematic screening of abnormally expressed tumor genes and clarifying their clinical relevance, providing reliable preliminary evidence for basic and translational cancer research [5–9]. Such in silico analytical workflows facilitate the exploration of key molecular events driving HCC progression and lay a solid foundation for subsequent mechanistic verification.

Metabolic reprogramming is a core hallmark of cancer, and mitochondrial dysfunction acts as a crucial driver facilitating tumor progression [10, 11]. Defects in oxidative phosphorylation (OXPHOS) and electron transport chain (ETC) components reshape intracellular redox homeostasis and metabolic plasticity, thereby supporting malignant proliferation [12]. As the first and rate-limiting complex of the respiratory chain, mitochondrial Complex I is indispensable for maintaining membrane potential and efficient ATP production [13]. Accumulating evidence demonstrates that abnormal Complex I assembly and function contribute to tumorigenesis by inducing genomic instability, inhibiting apoptosis and enhancing metabolic adaptability, making it a key regulatory node in cancer metabolism [14, 15].

Mitochondrial Complex I assembly factors have recently attracted extensive attention in oncology research. NDUFAF2 (NADH: ubiquinone oxidoreductase complex assembly factor 2) is a highly conserved molecular chaperone that mediates iron–sulfur cluster incorporation and stepwise assembly of Complex I on the inner mitochondrial membrane [16]. NDUFAF2 mutations are well documented in mitochondrial metabolic disorders, especially Leigh syndrome with severe neurodegeneration [17, 18]. However, the biological function and clinical significance of NDUFAF2 in malignant diseases remain poorly characterized, especially in HCC. Biochemical evidence has confirmed that NDUFAF2 sustains Complex I stability under stress conditions, implying its potential adaptive role in the metabolically hostile tumor microenvironment.

Current studies regarding NDUFAF2 in cancers are limited and fragmented. In lung adenocarcinoma, elevated NDUFAF2 expression correlates with advanced tumor stage and poor prognosis, and promotes cell proliferation via ROS-mediated MAPK signaling [19]. Upregulated NDUFAF2 also accelerates invasion and metastasis through JNK pathway activation in esophageal squamous cell carcinoma [20]. Knockdown of Ndufaf2 significantly suppresses cell growth in neuroblastoma cells [21]. In ovarian cancer, NDUFAF2 amplification is preliminarily associated with platinum chemotherapy resistance, while pan-cancer validation and mechanistic exploration are still lacking [22]. Collectively, existing studies suggest that NDUFAF2 exerts broad oncogenic effects across multiple human cancers.

To fill the research gap of NDUFAF2 in HCC, we first performed pan-cancer expression and prognostic analyses based on TCGA and CPTAC databases to screen cancer types with clinical significance. HCC was identified as the primary research focus. We further verified NDUFAF2 expression patterns using multiple public datasets, and analyzed its correlation with clinicopathological features, immune infiltration and patient survival. Two independent enrichment strategies consistently confirmed the close association between NDUFAF2 and oxidative phosphorylation. Furthermore, in vitro knockdown models were applied to verify the regulatory effects of NDUFAF2 on mitochondrial function and HCC cell proliferation (Fig. 1). Through this integrated bioinformatic and experimental workflow, the present study aims to clarify the biological function of NDUFAF2 in HCC progression and explore its potential value as a novel prognostic biomarker and therapeutic target.

Fig. 1.

Flowchart of this study

Materials and methods

Data collection and NDUFAF2 expression analysis

RNA sequencing data and related clinical information, encompassing 11,123 samples from 32 cancer types, including Adrenocortical carcinoma (ACC), Bladder Urothelial Carcinoma (BLCA), Breast invasive carcinoma (BRCA), Cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), Cholangiocarcinoma (CHOL), Colon adenocarcinoma (COAD), Lymphoid Neoplasm Diffuse Large B-cell Lymphoma (DLBC), Esophageal carcinoma (ESCA), Glioblastoma multiforme (GBM), Head and Neck squamous cell carcinoma (HNSC), Kidney Chromophobe (KICH), Kidney renal clear cell carcinoma (KIRC), Kidney renal papillary cell carcinoma (KIRP), Acute Myeloid Leukemia (LAML), Brain Lower Grade Glioma (LGG), Liver hepatocellular carcinoma (LIHC), Lung adenocarcinoma (LUAD), Lung squamous cell carcinoma (LUSC), Mesothelioma (MESO), Ovarian serous cystadenocarcinoma (OV), Pancreatic adenocarcinoma (PAAD), Pheochromocytoma and Paraganglioma (PCPG), Prostate adenocarcinoma (PRAD), Rectum adenocarcinoma (READ), Sarcoma (SRC), Skin Cutaneous Melanoma (SKCM), Stomach adenocarcinoma (STAD), Testicular Germ Cell Tumors (TGCT), Thyroid carcinoma (THCA), Thymoma (THYM), Uterine Corpus Endometrial Carcinoma (UCEC), Uterine Carcinosarcoma (UCS), and Uveal Melanoma (UVM)), were obtained from the TCGA database in Level-3 HTSeq-FPKM format. This comprehensive dataset was subsequently converted to Transcripts Per Million reads (TPM) format, and then underwent a log₂(value + 1) transformation to facilitate normalization and comparison across samples. The processed data were then analyzed for expression differences and paired analysis to identify significant patterns and correlations across the diverse cancer types. Additionally, the GSE112790 dataset was downloaded from the Gene Expression Omnibus (GEO) database for external validation.

Differences in gene expression between tumor and normal tissues were compared using student’s t-test. All bioinformatic analyses were conducted using R software (v4.2.1). The R package “ggplot2” (v3.4.4) was used for visualized box plots and correlation graphics.

Clinical proteomic data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) were analyzed via the UALCAN online portal. Immunochemistry images reflecting NDUFAF2 protein expression in HCC tissues were acquired from the Human Protein Atlas (HPA) database (https://www.proteinatlas.org/).

Prognostic analysis

Prognostic evaluation and survival visualization were performed in R software (v4.2.1). The survival package (v3.3.1) was used for proportional hazards assumption verification and Cox regression analysis, and results were visualized with the ggplot2 package (v3.4.4). Kaplan–Meier survival analysis was conducted using the survminer package (v0.4.9). The optimal cutoff value for stratifying high‑ and low‑NDUFAF2 expression subgroups was determined via the surv_cutpoint function.

Construction of the ROC curve

The diagnostic capability of NDUFAF2 in LIHC was assessed by receiver operating characteristic (ROC) curve analysis. The pROC package (v1.18.0) was applied to calculate the area under the curve (AUC) and 95% confidence intervals (CIs) based on 2,000 bootstrap replicates. Final ROC visualization was generated using ggplot2 (v3.4.4).

Gene association analysis and KEGG pathway enrichment analysis

Gene co-expression analysis of NDUFAF2 in HCC was performed using the LinkedOmics database based on RNA-seq data of 371 TCGA-LIHC patients (Firehose_RSEM_log2 normalized, HiSeq platform). Pearson correlation analysis was conducted across 19,922 genes, and 17,330 uniquely mapped Entrez IDs were included for subsequent enrichment screening.

KEGG pathway enrichment analysis was performed using the WEB-based Gene Set Analysis Toolkit (WebGestalt). Analytical parameters were set as follows: gene category size range of 5–2,000 genes, top 10 enriched functional terms, and 500 permutation iterations for false discovery rate correction. Hallmark gene set enrichment analysis was conducted using the BEST online platform for systematic functional annotation (https://rookieutopia.hiplot.com.cn/app_direct/BEST/).

Cell culture

Human HCC cell lines HCC-LM3 and SNU-449 were purchased from Abiowell Biotechnology Inc. (Changsha, China). HCC-LM3 cells were cultured in high-glucose DMEM medium, and SNU-449 cells were maintained in RPMI 1640 medium. Both culture media were supplemented with 10% heat-inactivated fetal bovine serum (FBS). All cells were incubated at 37 °C in a humidified incubator containing 5% CO₂.

Lentivirus infection

Recombinant lentiviruses carrying shRNA targeting NDUFAF2 (LV-shNDUFAF2: 5′-GCCTTGTGGAGATCGCTGTCA-3′) and negative control shRNA (LV-shNC) were synthesized by GenePharma (Shanghai, China). Lentiviral transduction was performed following the manufacturer’s protocols and our previously published methods [23]. Briefly, cells were seeded in 24-well plates at a density of 1 × 10⁵ cells/well and cultured overnight. After 72 h of transfection, cells were treated with 5 µg/mL polybrene and screened with puromycin to establish stable knockdown cell lines. The knockdown efficiency was verified before subsequent functional experiments.

Real time PCR

Total cellular RNA was extracted using TRIzol reagent (Life Technologies, #15596-026). Complementary DNA was reverse-transcribed from 0.5 to 1 µg total RNA using RT Master Mix for qPCR II. Quantitative PCR was performed with SYBR Green Master Mix on a Bio-Rad CFX96 system. Relative NDUFAF2 mRNA expression was normalized to β‑actin.

Primer sequences were listed as follows:

NDUFAF2: Forward 5′-CCTTGTGGAGATCGCTGTCAA-3′, Reverse 5′-AGTTCTTGTACTGCGGGATGTA-3′.

β-actin: Forward 5′-CATGTACGTTGCTATCCAGGC-3′, Reverse 5′-CTCCTTAATGTCACGCACGAT-3′.

Western blotting

Western blotting was performed as previously described [24, 25]. Total cellular protein was extracted using RIPA lysis buffer with protease inhibitors. Protein concentration was quantified by BCA assay. Equal amounts of protein (20–30 µg) were separated by 10% SDS‑PAGE and transferred onto PVDF membranes. After blocking with 5% non‑fat milk for 1 h at room temperature, membranes were incubated overnight at 4 °C with primary antibodies: anti‑NDUFAF2 (1:1,000; 13891‑1‑AP, Proteintech) and anti-β-actin (1:10,000; #4967, Cell Signaling Technology). After TBST washing, membranes were incubated with HRP‑conjugated secondary antibodies for 1 h. Protein bands were visualized by ECL chemiluminescence and quantified with ImageJ software.

Immunohistochemistry (IHC) staining

Four pairs of HCC and matched adjacent non-tumor tissues were obtained from Outdo Biotech (Shanghai, China). IHC staining was performed according to standard protocols. Paraffin sections were dewaxed, rehydrated, and subjected to high-pressure antigen retrieval. After H₂O₂ blocking and serum sealing, sections were incubated with anti-NDUFAF2 antibody (1:100; 13891-1-AP, Proteintech) overnight at 4℃. Subsequently, secondary antibody incubation, DAB color development, hematoxylin counterstaining, dehydration and mounting were performed for microscopic observation and image collection.

Mitochondrial stress assay

Mitochondrial respiratory function was detected using the Seahorse XF‑96 Extracellular Flux Analyzer. Briefly, 1.5 × 10⁴ cells were seeded into 96-well plates and cultured overnight. After medium replacement and temperature equilibration, cells were incubated in a non-CO₂ environment for 1 h. Mitochondrial inhibitors, including oligomycin (1.5 µM), FCCP (0.5 µM), and rotenone/antimycin A (0.5 µM), were sequentially added to detect real-time oxygen consumption rate (OCR). Data acquisition and normalization were performed in accordance with the official manufacturer’s guidelines.

ROS detection

Mitochondrial reactive oxygen species levels were measured using the MitoBright ROS Deep Red Detection Kit (Dojindo, MT16). Cell samples were stained strictly following the manufacturer’s protocols and detected by a Beckman CytoFlex flow cytometer. Flow cytometry data were analyzed using CytExpert 2.4 software.

Immune infiltration analysis

The tumor immune microenvironment landscape of LIHC was evaluated via single-sample gene set enrichment analysis (ssGSEA) using the GSVA package (v1.46.0) in R. Twenty-four types of tumor-infiltrating immune cells were quantified. The correlations between NDUFAF2 and key immune checkpoint genes were analyzed and visualized by lollipop charts. Differences in immune cell infiltration, immune score and stromal score between high‑ and low‑NDUFAF2 groups were compared. The Wilcoxon rank-sum test and Levene’s test (car package, v3.1-0) were used for intergroup comparison. The ESTIMATE algorithm (v1.0.13) was applied to calculate immune and stromal scores.

Statistical analysis

All bioinformatic and statistical analyses were performed uniformly using R (v4.2.1) to guarantee reproducibility, with all packages run under the identical R environment. SPSS 24.0 was used for additional statistical analysis. The Shapiro–Wilk test was applied to verify data normality. Paired t-test was used for parametric comparisons, while Wilcoxon signed-rank test and Mann–Whitney U test were adopted for paired and unpaired non-parametric data, respectively. The Kruskal–Wallis H test was used for multiple-group comparisons. Spearman’s correlation analysis was applied to evaluate relationships between NDUFAF2 expression, clinical features, immune infiltration and immune checkpoint molecules.

Kaplan–Meier survival analysis with the log-rank test was used to assess prognostic differences. The Schoenfeld residual test validated the proportional hazards assumption for all Cox regression models, with no obvious violation observed. Hazard ratios with 95% confidence intervals were calculated for prognostic evaluation. ROC analysis was conducted to assess the diagnostic value of NDUFAF2. The Benjamini–Hochberg method was used for FDR correction in omics and enrichment analyses. All tests were two-tailed, and p < 0.05 was defined as statistically significant.

Results

Gene expression and prognostic value of NDUFAF2 in cancers

Integrated analysis of NDUFAF2 expression and prognostic significance revealed distinct dysregulation patterns across human cancers. Transcriptomic profiling of the TCGA dataset demonstrated significant NDUFAF2 mRNA upregulation in cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), kidney renal clear cell carcinoma (KIRC), liver hepatocellular carcinoma (LIHC), and lung adenocarcinoma (LUAD) versus matched normal tissues (all p < 0.01, Wilcoxon rank-sum test), whereas downregulation was observed in breast carcinoma (BRCA), kidney chromophobe (KICH), thyroid carcinoma (THCA), and uterine corpus endometrial carcinoma (UCEC) (p < 0.05) (Fig. 2A, B). Proteomic analysis using the Clinical Proteomic Tumor Analysis Consortium (CPTAC) database corroborated these tissue-specific expression dynamics, showing elevated NDUFAF2 protein levels in colorectal adenocarcinoma, ovarian cancer, lung cancer, and LIHC, contrasting with suppression in clear cell renal cell carcinoma, pancreatic adenocarcinoma (PAAD), head and neck cancer, and glioblastoma (Fig. 2C). Collectively, gene and protein expression analyses identified significantly elevated NDUFAF2 expression primarily in lung and liver cancers, particularly LIHC.

Fig. 2.

Pan-cancer dysregulation of NDUFAF2 and its clinical implications. (A) Comparative NDUFAF2 mRNA expression across 33 malignancies (TCGA; tumor versus matched normal tissues). (B) Verification of NDUFAF2 overexpression in paired tumor-adjacent specimens from principal cancer types. (C) Quantitative proteomic confirmation of NDUFAF2 dysregulation via Clinical Proteomic Tumor Analysis Consortium (CPTAC) data. (D) Prognostic heatmap delineating NDUFAF2 expression associations with survival outcomes: overall survival (OS), disease-specific survival (DSS), and progression-free interval (PFI). Data presented as mean ± SEM; statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001. Heatmap color scale: red = poor prognosis, blue = favorable prognosis

To ascertain whether aberrant NDUFAF2 expression correlates with oncological prognoses, we performed comprehensive pan-cancer survival analyses. As illustrated in Fig. 2D, heightened NDUFAF2 expression correlated with significantly diminished overall survival (OS) in adrenocortical carcinoma (ACC), head and neck squamous carcinoma (HNSC), KICH, kidney renal papillary carcinoma (KIRP), LIHC, LUAD, skin cutaneous melanoma (SKCM), uterine carcinosarcoma (UCS), and uveal melanoma (UVM) (p < 0.05); it also portended abbreviated progression-free intervals (PFI) in KICH, KIRP, and LIHC (p < 0.05); and predicted unfavorable disease-specific survival (DSS) in LIHC, lung squamous carcinoma (LUSC), and KIRP (p < 0.05). Notably, LIHC demonstrated consistent detrimental prognostic associations across all three survival endpoints (OS, PFI, DSS). Collectively, these findings establish NDUFAF2 as a clinically relevant prognostic biomarker in hepatocellular carcinoma, warranting further functional investigation in this malignancy.

Overexpression of NDUFAF2 and its associations with clinicopathological characteristics in HCC

To evaluate NDUFAF2 expression in patients with liver hepatocellular carcinoma (LIHC), we first analyzed the TCGA-LIHC dataset. The results revealed significantly higher NDUFAF2 mRNA expression in LIHC tissues compared to adjacent normal tissues (p < 0.001; Fig. 3A). This finding was further validated by paired analysis of matched tumor and normal samples, confirming elevated NDUFAF2 expression in tumors (p < 0.001; Fig. 3B). Consistently, the external validation dataset GSE112790 also demonstrated significantly increased NDUFAF2 expression in LIHC tissues (p < 0.001; Fig. 3C). To assess the diagnostic value of NDUFAF2, we performed receiver operating characteristic (ROC) curve analysis. The resulting area under the curve (AUC) was 0.950 (95% CI: 0.929–0.970; Fig. 3D), indicating that NDUFAF2 has excellent diagnostic performance for distinguishing HCC from normal liver tissue.

Fig. 3.

NDUFAF2 expression and clinical significance in HCC. (A) Comparative NDUFAF2 mRNA expression in TCGA-LIHC tumor (n = 374) and normal (n = 50) tissues. (B) Paired analysis of NDUFAF2 in 50 pairs of tumors and adjacent non-tumorous tissues. (C) Validation in GSE112790. (D) ROC curve analysis. (E, F) Associations with histological grade and pathological stage. (G) Correlation with T stage. (H) Stratification by AFP levels. (I) Association with vascular invasion. (J) Ethnic distribution. (K) Protein expression in CPTAC. (L) IHC staining in HCC and adjacent normal tissues (20× magnification). Data are mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001

We next explored the correlation between NDUFAF2 expression and key clinicopathological features using the TCGA-LIHC dataset. Our results showed that elevated NDUFAF2 expression was significantly associated with advanced histologic grade (Fig. 3E), pathologic stage (Fig. 3F), T stage (Fig. 3G), high AFP levels (> 400 ng/mL; Fig. 3H), presence of vascular invasion (Fig. 3I), and Asian ethnicity (Fig. 3J).

To confirm the findings at the protein level, we analyzed the CPTAC database, which showed significantly higher NDUFAF2 protein expression in primary HCC tissues compared to normal controls (p < 0.001; Fig. 3K). To further validate these findings, we performed immunohistochemical (IHC) staining on four pairs of clinical HCC and adjacent normal liver tissues. Consistent with our database results, IHC images showed stronger NDUFAF2 protein staining in HCC tissues compared with matched adjacent tissues (Fig. 3L). Collectively, these multi-level analyses consistently demonstrate that NDUFAF2 is significantly overexpressed in HCC, both at the mRNA and protein levels.

High expression of NDUFAF2 predicts adverse outcome in HCC patients

Kaplan–Meier survival analysis revealed significant prognostic stratification by NDUFAF2 expression in liver hepatocellular carcinoma (LIHC) (Fig. 4). Patients with high NDUFAF2 expression exhibited substantially reduced median overall survival (OS) (HR = 1.90, 95% CI: 1.34–2.70, p < 0.001; Fig. 4A) and progression-free interval (PFI) (HR = 1.58, 95% CI: 1.18–2.12, p = 0.002; Fig. 4B) compared to the low-expression group. Disease-specific survival (DSS) analysis further confirmed that high NDUFAF2 expression was associated with significantly worse prognosis (HR = 1.66, 95% CI: 1.06–2.59, p = 0.026; Fig. 4C).

Fig. 4.

Prognostic significance of NDUFAF2 expression in hepatocellular carcinoma. (A–C) Kaplan-Meier curves for overall survival (OS), progression-free interval (PFI), and disease-specific survival (DSS) in the total cohort (TCGA-LIHC, n = 526), stratified by NDUFAF2 expression (high vs. low). (D, E) Stage-stratified survival: early-stage (Stage I & II; D) versus advanced-stage (Stage III & IV; E). (F, K) Gender-specific survival: male (F) and female (K) subgroups. (G, H) T-stage subgroup analysis: T1–T2 (G) versus T3–T4 (H). (I, J) Prognostic stratification by serum alpha-fetoprotein (AFP) levels: ≤400 ng/mL (I) versus > 400 ng/mL (J). High- and low-expression groups are depicted in red and blue, respectively. Statistical significance was determined by the log-rank test

Subgroup analyses demonstrated consistent prognostic associations across pathological stages (Stage I & II: HR = 1.87, p = 0.013; Stage III & IV: HR = 2.30, p = 0.006; Fig. 4D, E) and T classifications (T1 & T2: HR = 1.89, p = 0.007; T3 & T4: HR = 1.86, p = 0.027; Fig. 4G, H). Notably, the prognostic power of NDUFAF2 was significant in male patients (HR = 2.29, 95% CI: 1.44–3.64, p < 0.001; Fig. 4F) and those with low α-fetoprotein levels (AFP ≤ 400 ng/mL: HR = 2.09, p = 0.006; Fig. 4I), whereas no significant correlation was observed in female patients (HR = 1.06, p = 0.843; Fig. 4K) or cases with AFP > 400 ng/mL (HR = 2.13, p = 0.089; Fig. 4J).

Exploring the relationship between NDUFAF2 expression and immune cell infiltration

We next investigated the association between NDUFAF2 expression and the tumor immune microenvironment in hepatocellular carcinoma. Correlation analysis revealed that NDUFAF2 expression exhibited weak but statistically significant inverse associations with several antitumor effector immune cell populations, including NK cells (R = − 0.269, p < 0.001), eosinophils (R = − 0.269, p < 0.001), Th17 cells (R = − 0.256, p < 0.001), neutrophils (R = − 0.200, p < 0.001), and CD8⁺ T cells (R = − 0.189, p < 0.01). In contrast, NDUFAF2 expression showed a weak positive correlation with Th2 cells (R = 0.297, p < 0.001) (Fig. 5A).

Fig. 5.

Association of NDUFAF2 expression with the tumor immune microenvironment in hepatocellular carcinoma. (A) Correlation between NDUFAF2 mRNA levels and infiltration abundances of 24 tumor-infiltrating immune cell subsets (Spearman’s ρ; P-values adjusted by FDR). (B) Differential immune cell infiltration in NDUFAF2-high versus NDUFAF2-low tumors (ssGSEA; Wilcoxon rank-sum test, *P < 0.05, **P < 0.01, ***P < 0.001). (C) Correlations between NDUFAF2 expression and key immune checkpoint genes (CTLA4, CD276, HAVCR2, PDCD1, LAG3, TIGIT, CSF1R, IDO1, CD274) in the TCGA-LIHC cohort. (D) Comparative tumor microenvironment scores (Stromal, Immune, and combined ESTIMATE scores) between NDUFAF2 expression groups, calculated using the ESTIMATE algorithm (*P < 0.05, **P < 0.01, ***P < 0.001). All groups were stratified by the median NDUFAF2 expression threshold

Comparison of ssGSEA-derived immune cell scores between NDUFAF2-high and NDUFAF2-low subgroups further supported these findings. The NDUFAF2-high group displayed significantly lower infiltration of CD8⁺ T cells, NK cells, eosinophils, Th17 cells, neutrophils, and mast cells. Conversely, this group showed higher enrichment of Th2 cells, NK CD56bright cells, and aDCs, which are commonly associated with an immunosuppressive phenotype (Fig. 5B).

We further examined the correlation between NDUFAF2 expression and key immune checkpoint genes. NDUFAF2 expression was weakly but significantly positively correlated with multiple immune inhibitory molecules, including CTLA4, CD276, PDCD1, LAG3, and TIGIT (Fig. 5C). ESTIMATE analysis demonstrated that stromal scores were significantly lower in the NDUFAF2-high subgroup, whereas immune scores and ESTIMATE scores showed no significant differences between groups (Fig. 5D).

Collectively, these results demonstrate that high NDUFAF2 expression is associated with decreased antitumor immune cell infiltration, increased immunosuppressive cell subsets, and upregulated immune checkpoint expression. Importantly, all these findings are weak correlative observations and do not imply direct or causal regulatory relationships.

Co-expression and functional enrichment analyses

First, we identified genes significantly correlated with NDUFAF2 using the LinkedOmics database. The distribution of correlated genes is shown in the volcano plot (Fig. 6A), with the top 50 positively and top 50 negatively correlated genes displayed in heatmaps (Fig. 6B and C).

Fig. 6.

NDUFAF2 co-expression analysis and pathway enrichment in hepatocellular carcinoma (LIHC). (A) Volcano plot showing genes significantly correlated with NDUFAF2 expression (Pearson correlation). (B, C) Heatmaps of the top 50 positively and negatively co-expressed genes with NDUFAF2, respectively. (D) KEGG pathway enrichment analysis of NDUFAF2-correlated genes, ranked by normalized enrichment score. (E) Ridge plot presenting GSEA Hallmark enrichment score distributions, highlighting the most significantly associated biological processes

To infer the functional relevance of these correlated genes, we conducted two independent pathway enrichment analyses. KEGG pathway analysis revealed that oxidative phosphorylation was among the top-ranking significantly enriched pathways, alongside other metabolism-related processes (Fig. 6D). Consistently, Hallmark gene set enrichment analysis (GSEA) also identified oxidative phosphorylation as one of the most highly enriched pathways associated with NDUFAF2 expression (Fig. 6E). The convergence of these two independent enrichment analyses strongly supports a role for NDUFAF2 in the regulation of oxidative phosphorylation, consistent with its known function as a mitochondrial respiratory chain assembly factor.

In vitro functional characterization of NDUFAF2 in hepatocellular carcinoma

To investigate the biological role of NDUFAF2 in HCC, we established stable NDUFAF2-knockdown models in both HCC-LM3 and SNU-449 cell lines using shRNA-mediated silencing. Successful knockdown was confirmed at the mRNA level by qPCR (p < 0.001; Fig. 7A) and at the protein level by Western blotting (Fig. 7B), showing significantly reduced NDUFAF2 expression compared with scramble shRNA controls.

Fig. 7.

NDUFAF2 knockdown induces mitochondrial dysfunction and suppresses growth in hepatocellular carcinoma (HCC). (A, B) NDUFAF2 knockdown validation in HCC-LM3 and SNU-449 cells by qRT-PCR (A) and immunoblotting (B). (C, E) Mitochondrial respiration profiles (oxygen consumption rate, OCR) measured via Seahorse XF96 analyzer following sequential inhibitor exposure: oligomycin (1 µM) → carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone (FCCP, 2 µM) → rotenone/antimycin A (1 µM). (D, F) Quantitative analysis of OCR parameters, including basal respiration, maximal respiration, proton leak, ATP production, and spare respiratory capacity. (G, H) Mitochondrial ROS detection via MitoBright™ staining in HCC-LM3 and SNU-449 cells. (I, J) Expression levels of key glycolytic pathway genes (GLUT1, HK2, PDK1, PKM2, LDHA, LDHB) measured by qRT-PCR. (K, L) Cell growth curves comparing non-targeting shRNA control (shNC) versus NDUFAF2-targeting shRNA (shNDUFAF2) cells. Statistical significance: *P < 0.05, **P < 0.01, ***P < 0.001

To assess the impact of NDUFAF2 loss on mitochondrial function, we performed Seahorse XF Mito Stress Tests. In HCC-LM3 cells, knockdown of NDUFAF2 resulted in a marked reduction in oxygen consumption rate (OCR), including decreased basal respiration, maximal respiration, ATP production, and spare respiratory capacity (Fig. 7C, D). Consistently, similar impairments in mitochondrial respiration were observed in the SNU-449 NDUFAF2-knockdown cells (Fig. 7E, F).

Given that impaired electron transport chain function often leads to reactive oxygen species (ROS) accumulation, we measured mitochondrial ROS levels using MitoBright ROS Deep Red staining. Flow cytometry analysis showed a clear shift toward higher fluorescence intensity in NDUFAF2-knockdown cells, indicating increased mitochondrial ROS production (Fig. 7G, H).

To further examine the metabolic adaptation following mitochondrial dysfunction, we measured the expression of key glycolytic genes. qPCR analysis revealed significant upregulation of GLUT1 and PDK1 in both NDUFAF2-knockdown cell lines (Fig. 7I, J), consistent with compensatory activation of glycolysis. This pattern indicates that tumor cells undergo a metabolic shift toward aerobic glycolysis (the Warburg effect) to compensate for impaired oxidative phosphorylation.

Functionally, the combined defects in mitochondrial respiration and increased oxidative stress led to a significant reduction in cell proliferation, as demonstrated by the impaired growth curves of both NDUFAF2-knockdown cell lines over 5 days (p < 0.01; Fig. 7K, L). Collectively, these results identify NDUFAF2 as a pivotal regulator of mitochondrial integrity, cellular bioenergetics, and tumor cell proliferation.

Discussion

Mitochondrial dysfunction, particularly defects in Complex I of the electron transport chain (ETC), is a recognized driver of tumorigenesis through disrupted redox homeostasis and metabolic reprogramming [26, 27]. This study comprehensively characterizes the oncogenic potential of NDUFAF2, a conserved Complex I assembly chaperone, across human malignancies, with focused mechanistic investigation in HCC. Transcriptomic and proteomic pan-cancer analyses demonstrated that NDUFAF2 is broadly overexpressed in multiple cancer types. Elevated NDUFAF2 expression further correlates with unfavorable clinical outcomes in LIHC and LUAD, supporting its potential value as a prognostic biomarker.

The widespread upregulation of NDUFAF2 across tumor types indicates that this Complex I assembly factor may exert biological functions beyond its canonical mitochondrial chaperone activity. Unlike most solid tumors that predominantly rely on the Warburg effect, a subset of HCCs retains robust oxidative phosphorylation capacity to sustain rapid proliferation and adapt to metabolic stress, a feature termed metabolic duality [28–32]. Therefore, NDUFAF2 upregulation may confer a selective tumor advantage in HCC by maintaining stable OXPHOS function. Our pathway enrichment analyses further confirmed that oxidative phosphorylation is the core biological pathway consistently associated with NDUFAF2 expression. Consistent with this notion, our loss-of-function experiments in two independent HCC cell lines demonstrated that NDUFAF2 knockdown triggers significant mitochondrial dysfunction, including reduced ATP production, abnormal ROS accumulation, and impaired cell proliferation. These functional results confirm that NDUFAF2 is essential for maintaining mitochondrial bioenergetic homeostasis and malignant progression in HCC cells.

Beyond metabolic regulation, our analyses revealed weak associations between NDUFAF2 expression and tumor immune microenvironment remodeling in HCC. High NDUFAF2 expression showed marginal correlations with decreased antitumor immune subsets and increased immunosuppressive cell infiltration, along with slightly elevated immune checkpoint gene expression. These correlative findings suggest that NDUFAF2-high HCC may exhibit an immune-suppressive, “cold” tumor microenvironment that facilitates tumor immune escape. Notably, all immune-related observations in the present study are derived from bioinformatic analyses and represent correlative trends rather than experimentally validated causal relationships. The potential crosstalk between NDUFAF2-mediated metabolic reprogramming and immune microenvironment remodeling requires further mechanistic verification.

These findings present compelling therapeutic opportunities. The dependence of HCC cells on intact Complex I function reveals a profound metabolic vulnerability. Pharmacological inhibition of Complex I assembly or function, using agents such as IACS-010759, represents a promising targeted strategy, with NDUFAF2-expressing HCC potentially serving as a predictive biomarker [33–36]. Furthermore, the lethal ROS accumulation induced by NDUFAF2 depletion suggests potential synergy between Complex I inhibition and pro-oxidant therapies. Most intriguingly, although the immune correlations are weak and non-causal, the observed trends provide a hypothetical basis for combining metabolic targeting with immune checkpoint inhibitors to improve immunotherapy response in resistant HCC.

This study has several limitations that should be acknowledged. First, all clinical and immune-related results were obtained from public databases and retrospective bioinformatic analyses, which may be affected by potential heterogeneity and cannot fully reflect real-world clinical conditions. Second, the immune-related findings are purely correlative and do not imply a direct causal relationship between NDUFAF2 and immunosuppression; further in vitro and in vivo mechanistic studies are needed to verify these associations. Third, our functional validation is limited to loss-of-function knockdown experiments in two HCC cell lines. Although consistent phenotypic changes were observed, complementary gain-of-function validation is still required to fully establish causality. Notably, as a structural chaperone, NDUFAF2 may exhibit functional saturation under overexpression conditions, which may limit phenotypic detection in overexpression models. Finally, the therapeutic potential of targeting NDUFAF2 remains to be tested in preclinical models.

For future research, multiple directions can be pursued. First, rescue experiments and overexpression models will be constructed to further validate the regulatory role of NDUFAF2 in mitochondrial metabolism. Second, immune co-culture systems and immunocompetent animal models will be applied to explore whether NDUFAF2-mediated metabolic changes indirectly remodel the HCC immune microenvironment. Third, independent clinical cohorts will be used to prospectively validate the prognostic value of NDUFAF2. Finally, the therapeutic efficacy of Complex I inhibitors combined with immunotherapy will be tested to evaluate clinical translational potential.

Despite these limitations, our findings provide novel evidence that NDUFAF2 upregulation contributes to HCC progression by supporting mitochondrial respiration and metabolic reprogramming. This study systematically identifies NDUFAF2 as a promising prognostic biomarker and metabolic target in HCC, warranting further preclinical and clinical investigation.

Conclusion

This study identifies NDUFAF2 as a critical regulator of mitochondrial metabolism and a promising prognostic biomarker in hepatocellular carcinoma. NDUFAF2 maintains mitochondrial respiratory function and oxidative phosphorylation to support HCC cell proliferation and malignant progression. Weak correlative evidence suggests that high NDUFAF2 expression is associated with an immunosuppressive tumor microenvironment, potentially linking mitochondrial metabolism to immune phenotypes in HCC. Collectively, NDUFAF2 serves as a potential prognostic and therapeutic target for HCC. Targeting NDUFAF2-dependent mitochondrial metabolism, alone or combined with immunotherapy, may represent a novel strategy for refractory HCC treatment.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

A. H. designed the study. P. Z., W. L. and Z. L. performed in vitro experiments. A. H. and J. W. performed bioinformatic analyses, data interpretation, and manuscript writing. All authors contributed to and approved the final version.

Funding

This work was supported by National Natural Science Foundation of China (No.82300179); Natural Science Foundation of Guangdong Province (No.2024A1515012228); Guangzhou Municipal Science and Technology Project (No.2023A04J2105).

Data availability

The datasets and materials of this study are derived from the following public databases. NCBI Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/), LinkedOmics (http://www.linkedomics.org/login.php). Other data will be obtained from the corresponding author, Amin Huang, upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.