RNA sequencing (RNA-seq) expression matrices (HTSeq-FPKM) and corresponding clinicopathological annotations (including survival outcomes, immune subtype labels, and stemness indices) for 33 The Cancer Genome Atlas (TCGA) cancer types were downloaded from the UCSC Xena platform (https://xena.ucsc.edu/), which aggregates harmonized TCGA datasets (12,13). The expression values of CHCHD family members (CHCHD1-7 and CHCHD10) were extracted for downstream analyses, whereas CHCHD8 was excluded because of annotation and availability constraints in the downloaded expression matrix (14). For tumor types lacking matched normal samples in the TCGA, normal RNA-seq data from the Genotype-Tissue Expression (GTEx) project (https://gtexportal.org/) were incorporated to strengthen tumor-normal comparisons (15). Group comparisons between tumor and normal tissues were performed using the Wilcoxon rank-sum test in R (version 4.3.1; R Foundation for Statistical Computing; http://www.r-project.org/). The results were visualized as boxplots and heatmaps using the ‘ggpubr’ and ‘pheatmap’ packages.

Genetic alteration profiles (including somatic mutations and copy number alterations) of the CHCHD gene family across TCGA pan-cancer cohorts were characterized using cBioPortal (16), which provides integrative visualization and summary statistics for TCGA multi-omics and clinical datasets. To examine clinicopathological stage-associated expression patterns, CHCHD2 mRNA levels across pathological stages (I–IV) were analyzed using the ‘Stage Plot’ module of GEPIA2 (http://gepia2.cancer-pku.cn). Log2(TPM+1) values were compared among stages by one-way ANOVA, and P<0.05 was considered to indicate statistical significance (17).

Protein-level validation was performed using CPTAC proteomic datasets accessed via the UALCAN portal (http://ualcan.path.uab.edu/), which provides normalized protein abundance values for tumor and normal tissues. Differences between the tumor and normal groups were assessed using Student's t-test (P<0.05), and the results were visualized as boxplots (18). Immunohistochemistry (IHC) images of CHCHD2 in tumor and normal tissues were retrieved from the Human Protein Atlas (HPA; http://www.proteinatlas.org/), and staining intensity and distribution patterns were visually compared across cancer types (19).

The prognostic value of CHCHD2 expression across cancers was evaluated using the Kaplan-Meier plotter platform (https://kmplot.com/analysis/). OS was compared between the high- and low-expression groups using the platform's default settings, and hazard ratios (HRs) with log-rank P-values were reported (20).

To elucidate the LUAD-specific transcriptomic features associated with CHCHD2, co-expression analysis was first performed using TCGA-LUAD RNA-seq data and genes significantly correlated with CHCHD2 expression were identified under stringent criteria (|r|>0.4, FDR <0.05); significant correlations were visualized using Circos plots. TCGA-LUAD samples were then stratified into high- and low-expression groups, and differentially expressed genes (DEGs) were identified using thresholds of |log2FC|>1 and FDR <0.05. The functional annotation of the DEGs was subsequently conducted via Gene Ontology (GO) enrichment analysis (21) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (22) to characterize the biological processes (BP) and molecular pathways associated with CHCHD2 in LUAD.

TME infiltration levels were quantified from TCGA-LUAD RNA-seq data using the ESTIMATE algorithm to calculate the stromal score, immune score and ESTIMATE score (and infer tumor purity where applicable) (23). Stemness indices [RNA-based stemness score (RNAss) and DNA methylation-based stemness score (DNAss)] were obtained from previously published OCLR-based stemness models and/or corresponding TCGA PanCanAtlas stemness resources, and their associations with CHCHD2 expression were evaluated by Spearman correlation (24).

TCGA immune subtype annotations (C1-C6) were used to compare CHCHD2 expression across immune subtypes, following the pan-cancer immune landscape framework (25). In addition, correlations between CHCHD2 and immune checkpoint genes were assessed using Spearman correlation, with multiple-testing correction applied when appropriate. Relative abundances of tumor-infiltrating immune cells were estimated using CIBERSORT (LM22 signature matrix), and differences in immune cell fractions between the CHCHD2-high and CHCHD2-low groups were compared using Wilcoxon rank-sum tests (26). Finally, single-cell RNA-seq-based validation of CHCHD2 expression across malignant and immune/stromal cell populations was performed using the Tumor Immune Single-Cell Hub (TISCH/TISCH2) database (26).

Drug sensitivity data were downloaded from the CellMiner™ database (27,28) (version: 2020.3; database: 2.4.2; http://discover.nci.nih.gov/cellminer/home.do).

A total of 9 matched pairs of LUAD tissues and corresponding adjacent nontumorous tissues were obtained from the Department of Cardiothoracic Surgery at the Affiliated Hospital of Nantong University. All specimens were preserved at −80°C for subsequent western blotting and reverse transcription-quantitative PCR (RT-qPCR) analyses. The present study was approved by the Ethics Committee of the Affiliated Hospital of Nantong University (approval no. 2023-K010-01; Nantong, China). Written informed consent was obtained from each participant in compliance with the Declaration of Helsinki.

Human LUAD cell lines (A549, H1975, H1299, PC9 and H1650) were obtained from the Cell Bank of the Chinese Academy of Sciences. All the lines were authenticated and confirmed to be mycoplasma free. Cells were cultured in appropriate basal medium supplemented with 10% fetal bovine serum and 1% penicillin at 37°C in a humidified incubator with 5% CO₂. Standard procedures for revival, subculture and cryopreservation were performed.

CHCHD2 knockdown and overexpression plasmids were obtained from Corues Biotechnology (https://www.corues.cn/). Lipofectamine 2000 was purchased from Invitrogen (Thermo Fisher Scientific, Inc.). TRIzol™ Reagent was obtained from Invitrogen; Thermo Fisher Scientific, Inc., and HiScript III RT SuperMix and ChamQ SYBR qPCR Master Mix were obtained from Vazyme Biotech Co., Ltd. Primary antibodies against CHCHD2 (1:1,000; cat. no. 19424-1-AP; Proteintech Group, Inc.) and GAPDH (1:10,000; cat. no. AB0037; Abways Technology) were used for western blotting. For pathway validation, primary antibodies against phospho-AKT (Ser473) (1:3,000; cat. no. 28731-1-AP; Proteintech Group, Inc.), total AKT (1:10,000; cat. no. 60203-2-Ig; Proteintech Group, Inc.), phospho-ERK1/2 (Thr202/Tyr204) (1:3,000; cat. no. 28733-1-AP; Proteintech Group, Inc.) and total ERK1/2 (1:8,000; cat. no. 11257-1-AP; Proteintech Group, Inc.) were used for western blotting. HRP-conjugated anti-rabbit IgG (1:3,000; cat. no. 7074) and anti-mouse IgG (1:3,000; cat. no. 7076) secondary antibodies were obtained from Cell Signaling Technology, Inc. Western blotting consumables, including RIPA buffer, PMSF, PVDF membranes, and ECL reagents, were obtained from Beyotime Institute of Biotechnology, MilliporeSigma and Thermo Fisher Scientific, Inc. Reagents used for functional assays, including Cell Counting Kit-8, Annexin V-FITC, and Transwell chambers, were obtained from the aforementioned suppliers.

For CHCHD2 knockdown, PC9 and H1975 cells were transfected with 2.5 µg/well short hairpin RNA (shRNA) plasmids targeting CHCHD2 (Corues Biotechnology; http://www.corues.cn/) using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) in 6-well plates according to the manufacturer's instructions. The shRNA oligonucleotides were subsequently cloned and inserted into the pRNA3-SV40-EGFP(2A)Puro-U6 vector, which contains a U6 promoter, an EGFP reporter, and a puromycin resistance cassette. A total of 3 CHCHD2-targeting shRNAs and a non-targeting negative-control shRNA (shNC) were used. The target sequences were as follows: shCHCHD2-1, 5′-CAGTGGAGGAAGTAATGCTGA-3′; shCHCHD2-2, 5′-CCTCTATGAGATCAAACAGTT-3′; shCHCHD2-3, 5′-GAAACAGTGCCGACTTGCAAA-3′; and shNC, 5′-GGAGTACTTTATCCGTGTATT-3′. Knockdown efficiency was quantitatively evaluated at both the mRNA and protein levels. Among the three shRNAs tested, shCHCHD2-1 produced the strongest and most consistent reduction in CHCHD2 expression across PC9 and H1975 cells and was therefore selected for subsequent functional experiments. For CHCHD2 overexpression, H1299 and H1650 cells were transfected with 2.5 µg/well CHCHD2 expression plasmids (Corues Biotechnology) using Lipofectamine 2000. The knockdown and overexpression efficiencies were confirmed by RT-qPCR and western blotting, respectively, at 48 h post-transfection, and the cells were collected at 48–72 h post-transfection for functional experiments.

Total RNA was extracted with TRIzol and quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Inc.) by measuring absorbance at 260 nm; RNA purity was assessed using the A260/A280 ratio. Reverse transcription was performed with HiScript III RT SuperMix (Vazyme Biotech Co., Ltd.) according to the manufacturer's instructions. RT-qPCR was conducted on an ABI 7500 system with ChamQ SYBR Master Mix (Vazyme Biotech Co., Ltd.) according to the manufacturer's instructions. The thermocycling conditions were as follows: Initial denaturation at 95°C for 30 sec, followed by 40 cycles of denaturation at 95°C for 10 sec and annealing/extension at 60°C for 30 sec. GAPDH served as the internal reference gene, and relative CHCHD2 expression was calculated using the 2^−ΔΔCq method (29). The sequences of the primers used were as follows: CHCHD2 forward, 5′-ACCAGTCGCTCAGCCACCAG-3′ and reverse, 5′-CTCGCAGGCTCAGCATTACTTCC-3′; and GAPDH forward, 5′-CAGGAGGCATTGCTGATGAT-3′ and reverse, 5′-GAAGGCTGGGGCTCATTT-3′.

Proteins were extracted using RIPA lysis buffer supplemented with PMSF and phosphatase inhibitors and quantified using a BCA assay. Equal amounts of protein (20 µg per lane) were separated by 10% SDS-PAGE and transferred onto PVDF membranes. After blocking with 5% non-fat milk in TBST for 1 h at room temperature, the membranes were incubated with the indicated primary antibodies overnight at 4°C, followed by incubation with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using ECL reagents (Thermo Fisher Scientific, Inc.) and quantified using ImageJ software (version 1.53; National Institutes of Health). GAPDH was used as the loading control for CHCHD2 expression. AKT and ERK pathway activation was evaluated using the p-AKT/total AKT and p-ERK1/2/total ERK1/2 ratios, respectively.

Cells were plated on 96-well culture plates at a density of 2,000 cells per well in 100-µl complete medium with three replicate wells per condition. After a 12-h incubation period to ensure cell attachment, 10 µl of Cell Counting Kit-8 (CCK-8) solution (Corning, Inc.) was added to each well at the indicated time points (24, 48, 72 and 96 h). The plates were then incubated at 37°C for an additional 2 h, after which the absorbance at 450 nm was recorded using a Bio-Rad microplate reader. Cell viability was assessed by constructing growth curves based on OD450 measurements over time.

After 24 h, the cells were seeded in 6-well plates to 90–100% confluence. Sterile 200-µl pipette tips were used to create linear scratches (0.5–1 cm intervals). The wells were then washed three times with PBS, serum-free medium was replenished, and the wells were incubated at 37°C with 5% CO₂. Images of the scratches were captured at 0 and 24 h (Olympus inverted microscope), and closure was quantified via ImageJ software (National Institutes of Health).

Trypsinized cells were centrifuged at 138 × g for 5 min and resuspended at 2–5×10⁴ cells/well. The upper chambers received 100 µl of the suspension, and the lower chambers received 600 µl of the complete medium. After 1–2 days, the contents of the upper chambers were removed, non-migratory cells were swabbed, and the membranes were fixed in 4% paraformaldehyde for 30 min, air-dried, stained with 0.1% crystal violet for 20 min, washed three times, and air-dried. The migratory cells were microscopically imaged and counted using ImageJ software (version 1.53; National Institutes of Health).

Cells (5×106/sample) were trypsinized, incubated with Annexin V-FITC (20 µg/ml) at 4°C in the dark for 20 min, resuspended in 400 µl of PBS, stained with PI (50 µg/ml, 1 µl) for 2 min, and analyzed on a CytoFLEX S cytometer (Beckman Coulter, Inc.; 488 nm excitation, FITC at 515 nm, PI >560 nm). Flow cytometry data were analysed using CytExpert software (version 2.4; Beckman Coulter, Inc.). Apoptotic rates were determined by summing the early and late populations.

Male BALB/c nude mice (4–6 weeks old; 18–22 g) were randomly assigned to the shNC and shCHCHD2-1 groups (n=4 per group; 8 mice in total). The mice were housed under specific pathogen-free conditions at 22±2°C and 50–60% relative humidity under a 12-h light/dark cycle, with ad libitum access to food and water. PC9 cells transfected with shCHCHD2-1 or shNC were subcutaneously injected into the mice. Tumor dimensions were measured every 3 days using callipers, and tumor volume was calculated as V=(length × width²)/2. Animal health and behavior were monitored daily, including general condition, activity, food and water intake, and tumor-associated abnormalities. Humane endpoints included marked deterioration in body condition, inability to access food or water, severe impairment of activity, tumor ulceration or necrosis, or excessive tumor burden according to the institutional animal-welfare protocol. Animals meeting these criteria would have been euthanized immediately.

The experimental duration was 30 days. No animals succumbed or required early euthanasia. All mice were euthanized on Day 30 by CO₂ inhalation. CO₂ was gradually introduced from a compressed-gas cylinder into a non-pre-charged chamber at a displacement rate of 30% of the chamber volume per min and maintained for 3 min after respiratory arrest. Death was confirmed by irreversible cessation of respiration and cardiac activity. No anesthetic agent was used for cell inoculation or tumor measurements.

Xenograft tumors were harvested, fixed in 4% paraformaldehyde at room temperature for 24 h, paraffin-embedded, sectioned at a thickness of 4 µm, and subjected to IHC staining using an anti-CHCHD2 primary antibody. DAB and hematoxylin were used for visualization and counterstaining, respectively, and CHCHD2 staining was quantified by average optical density (AOD) using ImageJ. All animal procedures were approved by the Animal Ethics Committee of Nantong University (approval no. P20250307-034; Nantong, China).

Data are presented as the mean ± standard deviation (SD). Comparisons between two groups were performed using paired Student's t-test for matched tumor and adjacent non-tumorous tissues and unpaired Student's t-test for independent experimental groups, whereas multiple-group comparisons were performed using one-way ANOVA followed by Tukey's post hoc test. Time-course data were analyzed using two-way ANOVA or repeated-measures two-way ANOVA, as appropriate, followed by Sidak's multiple-comparisons test. Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software, Inc.; Dotmatics). P<0.05 was considered to indicate a statistically significant difference. Unless otherwise stated, in vitro experiments were conducted in triplicate and independently repeated at least three times.

The present analysis revealed distinct expression patterns among the CHCHD family members in the pan-cancer dataset. CHCHD2 was consistently expressed at high levels across a broad range of cancers, whereas CHCHD1, CHCHD3 and CHCHD10 were expressed at moderate levels, and CHCHD4-CHCHD7 were expressed at relatively low levels (Fig. 1A). Further stratification indicated that CHCHD1 was most highly expressed in uterine corpus endometrial carcinoma (UCEC), bladder urothelial carcinoma (BLCA) and cholangiocarcinoma (CHOL). CHCHD2, CHCHD3 and CHCHD5 expression levels peaked in glioblastoma multiforme, whereas the highest expression of CHCHD6 was detected in CHOL (Fig. 1B). Correlation analysis revealed heterogeneity among gene associations: CHCHD5 and CHCHD1 had the strongest positive correlation (r=0.44), whereas CHCHD3 and CHCHD1 had the weakest correlation (r=−0.01) (Fig. 1C).

Next, differential expression analysis comparing tumor and normal tissues revealed that CHCHD2 expression was significantly upregulated in multiple cancer types, including BLCA, breast invasive carcinoma (BRCA) and LUAD (Fig. 1D). The expression levels of other members, such as CHCHD1, CHCHD3-CHCHD7 and CHCHD10, were also elevated in most tumor types (Fig. S1A-G).

To characterize the genomic alteration landscape of the CHCHD gene family across cancers, alterations in these genes were analyzed using cBioPortal based on TCGA data. Across 10,967 tumor samples, the average alteration frequency of CHCHD family genes was ~1.5%. Among them, CHCHD2 (2.4%) and CHCHD7 (2.7%) presented relatively high genomic alteration frequencies (Fig. 1I). The 2.4% frequency for CHCHD2 represents its overall genomic alteration frequency, including mutations and copy number alterations, rather than evidence of loss-of-function mutations. Overall, 11% of tumors harbored at least one CHCHD family alteration, with copy-number amplification being the predominant alteration type. Uterine carcinosarcoma had the highest overall alteration frequency (26%), whereas thyroid carcinoma had the lowest (1.5%) (Fig. 1J). Alteration details for individual CHCHD genes are provided in Figs. S2 and S3.

To further investigate CHCHD2, its relationship with pathological stage was examined across cancer types using GEPIA. A positive correlation between CHCHD2 expression and pathological stage was observed in LUAD (Fig. 1E), liver hepatocellular carcinoma (LIHC) (Fig. 1F), head and neck squamous cell carcinoma (HNSC) (Fig. 1G) and kidney chromophobe (Fig. 1H). At the transcript level, CHCHD2 is consistently overexpressed in numerous cancers. To assess whether this trend extended to the protein level, proteomic data from UALCAN were analyzed. CHCHD2 protein levels were significantly elevated in LIHC, UCEC, BRCA, ovarian serous cystadenocarcinoma, colon adenocarcinoma, lung squamous cell carcinoma and LUAD (Fig. S4A). IHC staining from the HPA further confirmed these findings (Fig. S4B).

The prognostic significance of CHCHD2 expression across diverse cancer types was evaluated using the KM Plotter database. Increased CHCHD2 expression was significantly correlated with reduced OS in patients with LIHC, HNSC, LUAD, BLCA and ESCA, as evidenced by HRs uniformly exceeding 1 (Fig. 2A-E; P<0.05). By contrast, higher CHCHD2 expression in thymoma was associated with a favorable OS outcome (Fig. 2F, P<0.05). Collectively, these findings suggest that CHCHD2 is overexpressed in diverse malignancies and is associated with an unfavorable prognosis in several tumor types, particularly LUAD. This highlights its potential clinical utility as both a diagnostic biomarker and a therapeutic biomarker. Given the strong prognostic association in LUAD, focus was further addressed on this cancer to investigate CHCHD2-related genes, pathways and cellular functions, aiming to delineate its oncogenic mechanisms in greater detail.

A co-expression analysis was performed using RNA sequencing data from TCGA and 9,775 genes that were significantly correlated with CHCHD2 expression were identified, including 5,261 negatively correlated genes and 4,514 positively correlated genes. The potential interaction network among the 11 genes most strongly correlated with CHCHD2 is shown in Fig. S5A.

The samples were then stratified into CHCHD2 high- and low-expression groups, revealing 1,150 DEGs, which comprised 156 significantly upregulated genes and 994 significantly downregulated genes. A heatmap was generated to visualize the top 100 DEGs (Fig. S5B).

Functional annotation via GO enrichment analysis revealed that in the BP category there was significant enrichment in terms such as potassium ion transport and potassium ion transmembrane transport (Fig. S5C). In the cellular component category, the enriched terms included collagen-containing extracellular matrix, neuronal cell body, and glutamatergic synapse. In the molecular function category, the most enriched terms were metal ion transmembrane transporter activity, voltage-gated monoatomic cation channel activity, and potassium ion transmembrane transporter activity. KEGG pathway analysis revealed significant enrichment in pathways such as PI3K-Akt signaling, neuroactive ligand-receptor interactions, cAMP signaling and MAPK signaling (Fig. S5D), suggesting that these pathways may be associated with CHCHD2-related transcriptional changes in LUAD. However, these enrichment results should be interpreted as hypothesis-generating rather than direct mechanistic evidence. These pathway associations were subsequently evaluated experimentally by assessing AKT and ERK phosphorylation following CHCHD2 knockdown.

Although the role of CHCHD2 in tumor immunity has not been extensively studied, mitochondria are increasingly recognized as key regulators of the TME and play important roles in tumor progression, immune evasion and therapeutic responses (30). To further elucidate the potential association between CHCHD2 expression levels and TME characteristics, as well as stemness scores, Spearman's rank correlation coefficient was used for nonparametric correlation analysis. The results revealed a significant positive correlation between CHCHD2 expression and both the RNAss (R=0.37, P<0.001) and the DNAss (R=0.13, P<0.001) (Fig. S6A), suggesting that CHCHD2 may play a role in maintaining or enhancing the stemness phenotype. On the other hand, CHCHD2 expression was negatively correlated with several TME-related scoring metrics, including the stromal score (R=−0.19, P<0.001), immune score (R=−0.17, P<0.001) and the composite ESTIMATE score (R=−0.19, P<0.001) (Fig. S6A), indicating that increased CHCHD2 expression may be associated with reduced immune infiltration and stromal content, further suggesting a potential immunosuppressive role within the TME.

To evaluate the differential expression of CHCHD2 across various immune subtypes, the Kruskal-Wallis H test was applied to LUAD samples, and a non-parametric comparison of CHCHD2 expression among six predefined immune subtypes was conducted (that is, C1, wound healing; C2, IFN-γ dominant; C3, inflammatory; C4, lymphocyte-depleted; C5, immunologically quiet; and C6, TGF-β dominant) (13,25). The statistical results indicated that CHCHD2 expression was relatively higher in the C1 subtype, whereas the lowest expression was observed in the C3 subtype (Fig. S6B), suggesting that CHCHD2 may perform distinct regulatory functions in different immune-inflammatory microenvironmental contexts.

The relationship between CHCHD2 expression and 50 immune checkpoint-related genes was first analyzed to assess its potential association with immune checkpoint inhibitors. The results revealed that the expression of CHCHD2 was significantly correlated with that of 10 immune checkpoint-related genes (P<0.001). Specifically, CHCHD2 expression was negatively correlated with CD40LG, TNFSF15, TNFSF14, CD28, BTLA, ADORA2A, CD160, TIGIT and NRP1 expression but positively correlated with TNFSF18 expression (Fig. S6C). Given the observed correlations with immune checkpoint-related genes, it was next evaluated whether CHCHD2 expression was also associated with differences in tumor-infiltrating immune cell composition.

To further investigate the association between CHCHD2 and the tumor immune microenvironment, the CIBERSORT algorithm was used to evaluate immune cell subpopulations in LUAD samples with different CHCHD2 expression levels. A comparison of the high- and low-expression groups revealed that CHCHD2 expression was positively correlated with the numbers of follicular helper T cells, activated CD4 memory T cells and gamma delta T cells but negatively correlated with the numbers of resting CD4 memory T cells and resting mast cells (Fig. S6D and E).

Single-cell RNA sequencing analysis was performed using multiple TISCH2 datasets, with NSCLC_GSE131907 as the primary LUAD-derived validation cohort. Although categorized under NSCLC in TISCH2, NSCLC_GSE131907 was derived from patients with pathologically diagnosed LUAD. In this dataset, CHCHD2 expression was detectable across multiple cellular compartments, including epithelial cells, T cells, B cells, fibroblasts and plasma cells (Fig. S7A-D). Other NSCLC datasets, including NSCLC_GSE117570, NSCLC_GSE127465, and NSCLC_GSE99254, were used only as supportive references and were not interpreted as LUAD-specific evidence.

To investigate the potential association between CHCHD2 expression and drug sensitivity in human cancer cell lines, correlation analysis was performed using the CellMiner™ database. The results (Fig. S8A-P) revealed a significant negative correlation between CHCHD2 expression and sensitivity to Irofulven. By contrast, CHCHD2 expression was positively correlated with sensitivity to multiple agents, including uracil mustard, dexamethasone (Decadron), asparaginase, acrichine, fluphenazine, vorinostat, PX-316, chelerythrine, nelarabine, imexon, fenretinide, hydroxyurea, carmustine, lomustine and ifosfamide (P<0.05).

Based on these multi-omics and database-derived findings, LUAD-specific experimental validation was next performed at the tissue, cellular and animal levels. Tumor tissues and matched adjacent normal tissues from nine patients with stage II LUAD were used to assess CHCHD2 expression levels. Western blotting and RT-qPCR results revealed that the protein and mRNA levels of CHCHD2 were significantly higher in LUAD tissues than in adjacent normal tissues (Fig. 3A and B), which is consistent with the findings of the aforementioned bioinformatics analysis. A total of five LUAD cell lines were selected to examine CHCHD2 expression levels, and the results revealed that the protein and mRNA levels of CHCHD2 were markedly higher in PC9 and H1975 cells than in H1299 and H1650 cells (Fig. 3C and D).

Two LUAD cell lines with high CHCHD2 expression, PC9 and H1975, were chosen to establish transient knockdown models using shRNA. Western blotting and RT-qPCR analyses showed that all three CHCHD2-targeting shRNAs reduced CHCHD2 expression, with shCHCHD2-1 producing the strongest and most consistent knockdown across both cell lines (Fig. 4A-D). Direct comparisons showed significantly greater protein suppression by shCHCHD2-1 than by shCHCHD2-2 or shCHCHD2-3 in both PC9 and H1975 cells. At the mRNA level, shCHCHD2-1 also demonstrated greater knockdown in PC9 cells and was significantly more effective than shCHCHD2-2 in H1975 cells, whereas the difference from shCHCHD2-3 in H1975 cells was not significant. Therefore, shCHCHD2-1 was selected for subsequent functional assays.

CCK-8 assays revealed that CHCHD2 knockdown significantly reduced cell proliferation in PC9 and H1975 cells (Fig. 4E and F). Flow cytometry revealed significantly increased apoptosis following CHCHD2 knockdown (Fig. 4G). Transwell and wound-healing assays further demonstrated impaired migration in both cell lines (Fig. 4H-J).

CHCHD2-overexpressing cell models were established in H1299 and H1650 cells, which exhibit low endogenous CHCHD2 expression. Western blotting and RT-qPCR analyses confirmed a significant increase in CHCHD2 expression levels in the overexpression group (Fig. 5A and B). CCK-8 assays revealed that CHCHD2 overexpression significantly increased the proliferation rates of H1299 and H1650 cells (Fig. 5C and D). Flow cytometric analysis revealed that CHCHD2 overexpression significantly reduced both early and late apoptosis in H1299 and H1650 cells (Fig. 5E). Wound healing and Transwell assays revealed that CHCHD2 overexpression significantly promoted cell migration (Fig. 5F-H). These results indicate that both knockdown and overexpression of CHCHD2 can alter LUAD cell proliferation, migration and apoptosis, suggesting a potential oncogenic role for CHCHD2 in LUAD.

To further evaluate the signaling pathways suggested by the KEGG enrichment analysis, AKT and ERK phosphorylation were examined following CHCHD2 knockdown. Western blot analysis showed that CHCHD2 knockdown reduced the levels of phosphorylated AKT and ERK in both PC9 and H1975 cells, as reflected by decreased p-AKT/t-AKT and p-ERK/t-ERK levels (Fig. 6A and B). These findings provide experimental support for the involvement of PI3K-AKT and MAPK signaling in CHCHD2-associated LUAD phenotypes.

To further evaluate the effect of CHCHD2 knockdown in vivo, PC9-shCHCHD2-1 and PC9-shNC cells were subcutaneously injected into nude mice. Tumor growth was monitored every 3 days. At Day 30, xenograft tumors were excised, and representative tumor images are shown in Fig. 6C. Tumor growth curves demonstrated that CHCHD2 knockdown significantly reduced xenograft tumor volume compared with the shNC group (Fig. 6D). IHC staining further showed reduced CHCHD2 expression in xenograft tumors from the shCHCHD2-1 group (Fig. 6E), which was confirmed by AOD quantification (Fig. 6F). These findings indicate that CHCHD2 knockdown suppresses LUAD tumor growth in vivo and are consistent with the inhibitory effects observed in vitro.