We obtained 174 lung clinical samples surgically excised and pathologically confirmed to be of LUAD from Sir Run Run Shaw Hospital, Affiliated with Zhejiang University, China, over 2018–2021. The 174 clinical surgically resected tissue samples include both tumor and paracancerous regions. These samples were from retrospectively enrolled patients who had not received any radiotherapy or chemotherapy presurgically. All patients were adults aged 34–86 years. We determined the clinical characteristics for analysis based on the research direction and the patient information available, which include Sex, Age, Smoking history, Visceral pleural invasion, Intravascular tumor embolism, TNM stages, and levels of LUAD-related biomarkers (CA125, CEA, CYFRA-21-1, Pro-GRP). To ensure the completeness of the analysis, we selected patients who met all the clinical characteristics mentioned above, patients who were missing any of the criteria will be excluded. This process was entirely random and involved no subjectivity. All experiments on the clinical samples were approved by the Research Ethics Committee of Sir Run Run Shaw Hospital, Affiliated with Zhejiang University.

The 174 lung clinical samples obtained in section 2.1 are paraffin-embedded sections of surgically resected LUAD tissues. The sections were sequentially placed in xylene I for 8 min, xylene II for 8 min, xylene III for 8 min, absolute ethanol I for 5 min, absolute ethanol II for 5 min, 85% ethanol for 5 min, 75% ethanol for 5 min, and washed with water for 2 min. The sections were stained with hematoxylin for 5 min, differentiated in hydrochloric acid solution for 2 s, then blued in ammonia water solution for 15–30 s, followed by rinsing with water. Afterward, the sections were immersed in 95% ethanol for dehydration, then stained in eosin solution for 5–8 s. The sections were sequentially placed in absolute ethanol I for 30 s, absolute ethanol II for 2.5 min, absolute ethanol III for 2.5 min, xylene I for 2.5 min, xylene II for 2.5 min, and finally mounted with neutral gum. The sections were examined under a microscope, and images were captured for analysis.

The paraffin-embedded sections were deparaffinized, rehydrated, and subjected to antigen retrieval. The tissue sections were placed in a retrieval box filled with sodium citrate buffer (pH 6.0). Once the buffer inside the pressure cooker reached boiling, the retrieval box was carefully placed inside. Antigen retrieval was performed at 121°C under high pressure for 3 minutes. Afterward, the pressure was released, and the sections were allowed to cool naturally to room temperature. The slides were then transferred to PBS and washed on a decolorizing shaker three times for 5 minutes each. Next, they were incubated in 3% hydrogen peroxide at room temperature for 25 min, followed by washing with 1× phosphate-buffered saline (PBS). The tissue sections were evenly covered with 3% bovine serum albumin and incubated at room temperature for 30 min. The sections were then incubated with the rabbit EFTUD2 polyclonal antibody (Abcam Cat# ab72456, RRID: AB_1268731) at 4°C overnight. The dilution of the EFTUD2 antibody for IHC staining is 1:200. After three PBS washes, the sections were incubated with a goat antirabbit secondary antibody conjugated with poly-horseradish peroxidase at room temperature for 50 min and visualized using a 3,3’-diaminobenzidine chromogen substrate. The sections were counterstained with hematoxylin for 3 min, dehydrated, air-dried, and finally, sealed with a neutral resin. Two independent pathologists (GX Fu and YF Xu) specializing in LUAD evaluated all sections.

EFTUD2 expression was calculated according to the tumor cell staining intensity and staining positivity rate. Staining intensities were rated as follows: 0, negative staining; 1, weak staining; 2, moderate staining; and 3, strong staining. Subsequently, the final IHC score for each sample was obtained by summing the range of positive cells and the staining intensity of the EFTUD2 protein. Moreover, positivity rates were scored from 0 to 4 as follows: 0, 0%; 1, 1%–25%; 2, 26%–50%; 3, 51%–75%; and 4, >75%. The final scores were a sum of the two aforementioned scores and categorized as follows: negative, <3; weak, 3; moderate, 4; and strong, ≥5 (25).

The GSE32863 (26), GSE43458 (27), and GSE75037 (28) datasets from the Gene Expression Omnibus (GEO) database (29) were downloaded to analyze EFTUD2 expression in LUAD and normal samples. The criteria for selecting the GSE32863, GSE43458, and GSE75037 datasets is as follows: (1) LUAD tissue derived from human species; (2) Like the TCGA dataset, LUAD patients are from the USA; (3) The number of cancer samples is greater than 50, and the combined number of cancer and adjacent normal samples is greater than 100. EFTUD2 expression was also assessed in TCGA datasets (30) by using the web-based tool Gene Expression Profiling Interactive Analysis (GEPIA; version 2.0). Clinical factors, including age, sex, and pathologic (tumor: T, node: N, and metastasis: M) stages et al., were obtained from TCGA and used to group the samples. In the TNM staging system, T represents the size of the tumor and the extent of the primary tumor, N indicates the involvement of regional lymph nodes, and M refers to the presence of distant metastasis. The Kruskal–Wallis test was used to compare EFTUD2 expression among different clinical factor groups in R (version 4.1.2; RRID: SCR_001905). The diagnostic and prognostic value of EFTUD2 in patients with LUAD was assessed using receiver operating characteristic (ROC) curve and Kaplan–Meier (K–M) survival analyses. The mutation MAF data for LUAD samples were downloaded from the Genomic Data Commons (GDC, version 2.15.5). The Tumor Mutation Burden (TMB) and Fraction of Genome Altered (FGA) for these samples were then calculated using maftools (version 2.6.05) in R (version 4.3.1). Subsequently, the cor function (version 4.1.2) in R was used to calculate the correlations between TMB, FGA, and the expression levels of EFTUD2.

Based on the median staining scores, immunohistochemical (IHC) staining results of the all patient clinical samples were categorized into high and low EFTUD2 expression groups (n = 91 and 83, respectively). Subsequently, two clinical characteristic data categorical variables were obtained, using the high and low expression levels of EFTUD2 as the classification criteria. The correlation between clinical characteristics and EFTUD2 expression was analyzed by the univariate chi-square test. The clinical characteristics of the study include: Sex, Age, Smoking history, Visceral pleural invasion, Intravascular tumor embolism, TNM stages, and levels of LUAD-related biomarkers (CA125: carbohydrate antigen 125; CEA: carcinoembryonic antigen; CYFRA-21-1: cytokeratin-19 fragment antigen 21-1; Pro-GRP: pro-gastrin releasing peptide precursor). Moreover, and TCGA samples were divided into high- and low-expression groups based on their median EFTUD2 expression. Then, we used Fisher’s exact test to compare each clinical factor’s distribution between the two groups in R (version 4.1.2).

TCGA gene expression profiles based on the Illumina HiSeq 2000 platform were downloaded from the UCSC-Xena database (31), which yielded 559 samples, including 501 LUAD tumor and 58 normal control samples. Differential expression analysis was then performed on the LUAD and control groups using the R package limma (version 3.56.2). Genes with a false discovery rate (FDR) of <0.05 and |log₂ fold change (FC)| of >1 were considered significantly differentially expressed.

GSEA of LUAD genes was performed using GSEA (version 4.3.2) (32) based on the median EFTUD2 expression, with the screening thresholds set to FDR (q) < 0.05 and p < 0.001.

We constructed a protein–protein interaction (PPI) network of EFTUD2 with the screened differentially expressed genes (DEGs), cGAS-STING pathway–related genes, and m6A modification–related genes by using the STRING database (version 12.0) (33). The network was visualized using Cytoscape (version 3.9.0). This EFTUD2 PPI was then analyzed for hub genes by using the cytoHubba plugin (version 0.1) of Cytoscape (3.9.0), based on these 12 topological analysis algorithms: maximal clique centrality, degree-based maximal neighborhood centrality, maximal neighborhood centrality, edge percolated centrality, Degree, BottleNeck, Closeness, Eccentricity, Betweenness, Radiality, Stress, and Clustering Coefficient. The genes commonly selected by at least nine of the algorithms were retained as the final hub genes. The comparison results were displayed using the R package UpSetR (version 1.4.0).

We performed correlation analysis to evaluate the relationships of EFTUD2 with the hub genes, immune cells, ESTIMATE scores, immune-related genes, cGAS-STING pathway–related genes, and m6A modification–related genes using the cor function in R. The Gene-Corr module of TIMER (version 2.0) (34) was used to assess correlations between EFTUD2 and glycolytic genes in LUAD.

ROC analysis (35) and Cox analysis (36) are widely used and recognized as common single-gene independent diagnostic methods, making them highly suitable for assessing the diagnostic and prognostic value of EFTUD2 in LUAD. The diagnostic value of EFTUD2 and the hub genes was assessed using ROC curve analysis in R package pROC (version 1.12.1). The survival of EFTUD2, immune cells, immune scores, immune genes, cGAS-STING pathway–related genes, and m6A modification–related genes was analyzed using the R package survival (version 2.41-1). Based on the clinical information from the TCGA dataset samples and the median grouping of EFTUD2 gene expression levels, Cox univariate and multivariate regression analyses were performed using the survival package (Version 2.41-1) in R (version 4.1.2).

Immune cell proportions in the TCGA tumor samples were assessed using CIBERSORT (RRID: SCR_016955) (37), followed by the calculation of ESTIMATE scores, immune scores, stromal scores, and tumor purity by using the R package estimate (version 1.0.13). Finally, the distributions of immune cells, ESTIMATE scores, immune scores, stromal scores, and tumor purity were compared between the high- and low-expression groups.

Immune scores with differential EFTUD2 distribution were obtained on the basis of EFTUD2 expression levels. Subsequently, the correlation between these differential associated immune scores and the abovementioned DEG was assessed using the cor function in R (version 4.1.2). All selected immune-related genes associated with EFTUD2 had p < 0.05 and |cor| > 0.3. These immune-related genes were then subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis based on DAVID (version 6.86) (38, 39), with FDR < 0.05 as the selection threshold.

Expression data for genes related to m6A modification and the cGAS-STING pathway were extracted from the TCGA LUAD expression profile. The included genes were related to methylation (METTL3, METTL14, METTL15, WTAP, VIRMA, RBM15, RBM15B, and ZC3H13), demethylation (FTO and ALKBH5), m6A regulation (RBMX, YTHDC1, YTHDC2, IGF2BP1, IGF2BP2, IGF2BP3, YTHDF1, YTHDF2, YTHDF3, HNRNPA2B1, and HNRNPC), and the cGAS-STING pathway (SAMHD1, DDX41, IRF3, PRKDC, XRCC5, XRCC6, TRIM21, TBK1, DTX4, STAT6, IFI16, and NLRC3). The limma package (version 3.56.2) was used to compare the gene expression levels between the LUAD and normal control groups. To assess whether genes associated with EFTUD2 expression are influenced by their own copy number variations, mutation analysis, including CNV and SNV, was performed on the EFTUD2-related differentially expressed genes in the TCGA-LUAD dataset using the cBioPortal online platform (version 6.0.23).

MRC-5 (RRID: CVCL_0440), A549 (RRID: CVCL_0023), H1299 (RRID: CVCL_0060), and PC9 (RRID: CVCL_B260) cell lines (all from American Type Culture Collection) were cultured in Dulbecco’s modified Eagle medium (C11995500BT; Gibco, USA) supplemented with 10% fetal bovine serum (FBS; 164210; Pricella, China) and penicillin–streptomycin (BL505A; Labgic, China) at 37°C in a humidified incubator containing 5% CO₂ and 95% air. Small interfering RNA (siRNA) targeting EFTUD2 (GenePharma, China) and Lipofectamine 2000 (11668019; Invitrogen, USA) were separately diluted in OPTI-MEM (31985070; Gibco, USA), followed by incubation for 5 min. Then, the two solutions were mixed, followed by incubation for an additional 15 min. This mixture was added to the cells in a six-well plate, followed by incubation for 6–8 h. The final concentration of siRNA used for transfection was 100 nM. Finally, the transfection medium was replaced with the regular culture medium. Subsequently, the transfected cells were cultured in a 37°C incubator with 5% CO2 for 48 h. All EFTUD2-targeting siRNA sequences are listed in the Supplementary Table S2 .

Total RNA was extracted using a TRIzol reagent kit (9109; Takara Bio, Japan). The extracted RNA was then reverse-transcribed into cDNA using a reverse transcription kit (RR047A; Takara Bio, Japan). Reverse transcription real-time polymerase chain reaction (qRT-PCR) was then performed using TB Green qPCR mix (RR420; Takara Bio, Japan), according to the manufacturer’s instructions. All primers used here, synthesized by Tsingke (China), are presented in the Supplementary Table S1 . The qRT-PCR results were analyzed using the 2^−ΔΔCT method.

Prepared cells were lysed in a cell lysis buffer (Cell Signaling Technology, Danvers, USA) at 4°C. Protein quantification was performed using a bicinchoninic acid protein assay kit (P1511; Applygen, China). Western blotting was performed in a sodium dodecyl sulfate–polyacrylamide gel electrophoresis system. The following primary antibodies were used: rabbit EFTUD2 polyclonal antibody (Abcam Cat# ab72456, RRID: AB_1268731, dilution factor: 1: 2000), rabbit anti-GAPDH polyclonal antibody (Cell Signaling Technology Cat# 2118, RRID: AB_561053, dilution factor: 1: 1000), rabbit anti-PKM2-specific antibody (Proteintech Cat# 15822-1-AP, RRID: AB_1851537, dilution factor: 1: 2000), rabbit anti-ALDOB polyclonal antibody (Cohesion Biosciences Cat# CPA3123, RRID: AB_3106870, dilution factor: 1: 1000), and rabbit anti-β-actin polyclonal antibody (Cell Signaling Technology Cat# 4970, RRID: AB_2223172, dilution factor: 1: 1000). TBST (Tris Buffered Saline with Tween-20) supplemented with 5% fetal bovine serum (FBS) was used as the antibody diluent. The dilution of the antibodies was determined according to the manufacturer’s instructions.

Flow cytometry was used to analyze the cell cycle (40). Cells were prepared and digested with 0.25% trypsin without ethylenediaminetetraacetic acid to obtain single-cell suspensions at concentrations of 1 × 10⁵ to 5 × 10⁵ cells/mL. The cells were then washed twice with PBS, and the supernatant was discarded. Cell-cycle analysis was conducted using a Cell Cycle kit (C543; Dojindo, Japan), according to the manufacturer’s instructions. The flow cytometry results for the cell cycle were analyzed and plotted using Flowjo (version 7.6).

Prepared single cells were seeded into a six-well plate at a density of 1,000 cells per well and cultured for 10–12 days. Cell culture was terminated when visible clones appeared, and the supernatant was discarded. The cells were then gently washed twice with PBS and fixed in 4% paraformaldehyde for 15 min, followed by three PBS washes. Subsequently, the cells were exposed to an appropriate amount of 0.1% crystal violet for 3 min. The plates were photographed, and clone numbers were calculated using ImageJ (version 1.54h).

The treated cell suspensions were seeded at a density of 5,000 cells in 100 μL per well in a 96-well plate. The CCK-8 solution (CCK-8; C543; Dojindo, Japan) was mixed with a serum-free medium at a 1:10 ratio and used to replace the cell culture medium, followed by incubation at 37°C for 1–4 h. The optical density was measured at 450 nm on an MRX II microplate reader (Dynex, USA). Each experiment was performed independently in three replicates.

Matrigel (1567ML005; Neofroxx, Germany) was first diluted and then added to the upper chamber of a transwell plate. Next, a single-cell suspension was added to the upper chamber of the transwell plate, whereas a medium containing 10% FBS was added to the lower chamber. This was followed by incubation at 37°C for 48 h, fixing with 4% paraformaldehyde for 30 min, and finally, staining with 0.1% crystal violet for 3 min. Cells that did not pass through the Matrigel in the upper chamber were gently wiped away with a wet cotton swab. Images were captured under a fluorescence microscope (Olympus, Tokyo, Japan).

Cells were seeded at a density of approximately 3 × 10⁵ cells per well in a 6-well plate, with 2 mL of DMEM medium containing 5% FBS added to each well. After incubation for 12 h in a 37°C incubator with 5% CO2, scratches were made in each well using a 10 μL pipette tip along a sterile ruler. The detached cells were removed by washing with sterile 1 × PBS, and 2 mL of serum-free medium was added to each well. The plates were then placed back in a 37°C incubator with 5% CO2. Microscopic images were taken at the same location at 0 and 48 h. The scratch width was measured using ImageJ (version 1.54h).

For the tumor xenograft experiments, 4-week-old male Balb/c nude mice weighing 18 ± 2 g were selected. The mice were housed in a specific pathogen-free (SPF) sterile environment, maintained at a constant temperature of 25°C with a 12-h light-dark cycle, and provided with ample food and water. The nude mice were randomly divided into two groups: a control group (si-NC) and an EFTUD2 knockdown group (si-EFTUD2), with 5 mice in each group, totaling 10 mice. A549 cells were used for modeling in the animal experiments, and the cell culture conditions were the same as described in section 2.14. A549 cells were digested with trypsin, collected by centrifugation at 1000 rpm for 5 min, and resuspended in PBS. The cells were washed twice by centrifugation and resuspension. Cell counting was performed, and the cell density was adjusted to 1×10⁶ cells/mL. The cell suspension was divided into 100 μL portions, and each portion was injected into the right axilla of the forelimb of each mouse. After cell inoculation, the mice were returned to normal housing. Once the tumor xenograft masses became palpable, the diameter of the tumors was recorded every 3 days, and growth curves for body weight and tumor size were plotted.

When the tumor xenograft diameter reached 5-7 mm, the knockdown group mice were treated with a mixture of 15 µg EFTUD2 siRNA and 5 μL Lipofectamine 2000 (11668019; Invitrogen, USA) every 3 days via an intratumoral multi-point injection method. The EFTUD2 siRNA used in the animal experiments was modified with 3’Cholesterol and 2’OMe, and the nucleotide sequence was the same as shown in Supplementary Table S2 . The control group was injected with negative siRNA and 5 µL Lipofectamine 2000, with all other treatment conditions being the same as those for the knockdown group. Before euthanasia, blood samples were taken from the mice. At the end of the experiment, the mice were euthanized by cervical dislocation, and their appearance was photographed for records. Tumor xenografts were then excised, photographed, measured, and recorded.

The Lactate Assay Kit (BC2235; Solarbio, China) was used to measure lactate levels in serum and tumor xenografts. The specific steps were performed according to the protocol provided with the kit.

The mouse tumor xenografts were placed into dehydration cassettes and dehydrated using a dehydration machine (Donatello; DIAPATH, China) according to the following program: 75% ethanol for 4 h, 85% ethanol for 2 h, 90% ethanol for 2 h, 95% ethanol for 1 h, anhydrous ethanol I for 30 min, anhydrous ethanol II for 30 min, ethanol-xylene for 5 min, xylene I for 5 min, xylene II for 5 min, melted paraffin I at 65°C for 1 h, melted paraffin II at 65°C for 1 h, and melted paraffin III at 65°C for 1 h. The tumor xenografts were then placed in embedding molds with the melted wax, and before the wax solidified, the embedding molds were placed on a -20°C cold plate to cool. After the wax solidified, the wax blocks were removed from the molds and sectioned. The IHC and HE staining procedures for the mouse tumor xenograft specimens are described in sections 2.2 and 2.3. The IHC results from the animal experiments are expressed as the percentage of positive cells. The antibodies used for IHC staining of the mouse tumor xenografts, including EFTUD2, GAPDH, PKM2, and ALDOB, are the same as those used in the section 2.16.

On sections of the mouse tumor xenograft, 100 µL of proteinase K solution (E00492; Fermentas, Canada) at a concentration of 20 µg/mL was added and incubated at room temperature for 20 min. Afterward, the sections were washed with PBS for 3 min, repeating this step 3 times. The treated samples were placed in a humidified box to maintain moisture. Then, 100 µL of 3% H₂O₂ solution was applied to completely cover each sample and incubated at room temperature for 5 min. The samples were then rinsed with PBS and air-dried. The TUNEL reaction mixture (G1507; Servicebio, China) was prepared, and TUNEL staining was performed according to the instructions provided with the kit. After completing the TUNEL staining, the samples were counterstained with hematoxylin. The samples were immersed in hematoxylin (G1004; Servicebio, China) for 5 min, rinsed with distilled water, treated with hematoxylin differentiator (from the hematoxylin kit) for 2 s, rinsed again with distilled water, and then treated with hematoxylin bluing solution (from the hematoxylin kit) for 5 s, followed by another rinse with distilled water. The samples were immersed in absolute ethanol for dehydration 4 times, each for 5 min; immersed in xylene for 5 min, then transferred to fresh xylene for another 5 min. Finally, the slices were sealed with neutral gum. The samples were examined under a brightfield microscope. The nuclei of apoptotic cells appeared brownish-yellow, while the nuclei of normal cells appeared blue.

Public data from the TCGA and GEO databases were statistically processed and analyzed using R (version 4.1.2). Graphs and charts were created using both GraphPad Prism (version 9.0) and R (version 4.1.2). The overall survival (OS) was analyzed using the log-rank test. Relationships between two continuous variables were assessed using Pearson or Spearman analysis. Moreover, between-group comparisons were performed using Student’s t test or one-way analysis of variance, and categorical data were analyzed using the chi-square test. Cox univariate and multivariate analyses were used to investigate the independent prognostic value of EFTUD2 in LUAD. ImageJ (version 1.54h) was used for quantitative analysis of protein band intensities in Western blotting. All quantitative data are expressed as means ± standard deviations (SDs). Each experiment was independently performed at least three times to ensure reproducibility. Significant differences were defined based on a p value or FDR of <0.05.

To determine EFTUD2 mRNA expression profiles, we analyzed the GEO datasets GSE32863, GSE43458, and GSE75037. Compared with the normal tissues in GEO datasets, the LUAD tissues demonstrated significantly higher EFTUD2 expression (GSE32863: p < 0.001, GSE43458: p < 0.001, and GSE75037: p < 001; Figure 1A ). Elevated EFTUD2 expression was also verified in TCGA LUAD dataset ( Figure 1B ). Among clinical samples, the HE staining results demonstrated that LUAD tumor cells exhibited an irregular glandular invasive growth pattern, with weakly eosinophilic or clear cytoplasm ( Figure 1C ). Their nuclei were enlarged and irregular with uneven chromatin distribution and had visible mitotic figures ( Figure 1C ). Their stroma was accompanied by a desmoplastic fibrotic tissue reaction and showed chronic inflammatory cell infiltration ( Figure 1C ). Next, EFTUD2 expression in 174 paraffin-embedded primary LUAD tissues was assessed through IHC staining. The heatmap in Figure 1D illustrates differences in EFTUD2 expression between LUAD tissues and adjacent normal tissues in every sample: EFTUD2 expression was significantly higher in the LUAD tissues than in the adjacent tissues ( Figure 1E ), and the high expression group demonstrated significantly higher median EFTUD2 expression than the low-expression group ( Figure 1F ). The distribution of EFTUD2 expression in LUAD patients harboring EFTUD2 mutations was assessed. The findings revealed a very low frequency of EFTUD2 SNVs in LUAD patients (n = 4), and no significant differences in EFTUD2 expression were observed between the missense and non-mutated groups ( Supplementary Figure S1A ). In contrast, EFTUD2 mRNA expression was more markedly influenced by CNVs, with significantly elevated expression levels in the Gain and Amplification groups compared to the diploid group, and the Amplification group exhibiting the most pronounced increase ( Supplementary Figure S1B ). While a few Shallow Deletion mutations were detected in LUAD patients, the number of affected individuals was minimal. Furthermore, EFTUD2 expression in LUAD demonstrated a significant positive correlation with the TMB ( Supplementary Figure S1C ) and FGA ( Supplementary Figure S1D ).

To explore whether high EFTUD2 expression is correlated with clinical characteristics and tumor biomarkers (20), we divided the collected patient samples into two groups according to median EFTUD2 expression and found that age (p = 0.0093), N classification (p = 0.0163), visceral pleural invasion (VPI; p = 0.0270), and intravascular tumor embolism (p = 0.0092) were significantly associated with EFTUD2 upregulation ( Table 1 ). Moreover, EFTUD2 upregulation was also closely associated with cytokeratin-19 fragment antigen 21-1 (CYFRA-21-1), an independent prognostic factor in patients with LUAD (p = 0.0319; Table 2 ) (41). In addition to our clinical study, the TCGA results also demonstrated significant differences in EFTUD2 expression across the subgroups of pathologic N stage (N1 vs. N0), cancer stage (III vs. I), and smoking history (current smoker vs. previous smoker; Figure 1G ). Significant distribution differences were noted in age, N stage, and smoking history between groups with high and low EFTUD2 expression ( Table 3 )—consistent with results presented in Table 1 . The ROC curve for EFTUD2 from the TCGA-LUAD dataset demonstrated that EFTUD2 had an area under the ROC curve (AUC) of 0.873 ( Figure 1H ). The K–M survival analysis revealed that high EFTUD2 expression is associated with poor OS ( Figure 1I ). Cox analysis was performed for both univariate and multivariate analyses based on EFTUD2 expression levels and clinical characteristics to evaluate the independent prognostic value of EFTUD2 in LUAD. As shown in Table 4 , the univariate analysis revealed that EFTUD2, TNM staging, and pathological staging exhibited good prognostic performance with HR > 1. However, the multivariate analysis results for EFTUD2 were not significant, suggesting that the prognostic value of EFTUD2 in LUAD may be related to the progression of the disease. Thus, EFTUD2 may be a LUAD biomarker, and its upregulation may predict poor prognosis.

To determine the molecular mechanism associated with EFTUD2, we screened DEGs between 58 normal donors and 501 patients with LUAD from the TCGA dataset. The volcano plot results included 344 downregulated and 354 upregulated DEGs; one of these upregulated DEGs was EFTUD2 ( Figure 2A ). Then, the interactions between EFTUD2 and the screened DEGs were analyzed by using the STRING database; the results demonstrated 51 (red) and 13 (blue) proteins having strong positive and negative interactions with EFTUD2, respectively ( Figure 2B ). Hub genes related to EFTUD2 in the PPI network were further identified using 12 topology analysis algorithms, whereby the genes simultaneously identified by nine or more algorithms were retained as hub genes. In total, 16 hub genes were found to meet the screening criteria ( Figure 2C ). We next performed univariate Cox regression analysis on these 16 hub genes. The hazard ratios for all 16 hub genes were >1 ( Figure 2D ), indicating that they can be independent prognostic factors for LUAD. Moreover, EFTUD2 was significantly and positively correlated with all 16 hub genes (all p < 0.05; Figure 2E ). The five genes most positively related to EFTUD2 were KIF23, BUB1, BUB1B, ANLN, and MYBL2 ( Figure 2F ). CNV analysis was performed on 16 hub genes to investigate whether the expression levels of EFTUD2 and the hub genes are influenced by genetic mutations. The results indicated that, among the 16 hub genes, the expression of BUB1 (Spearman: 0.15) and TRIP13 (Spearman: 0.13) may be somewhat affected by copy number variations, while the correlation coefficients were comparatively weak ( Supplementary Figure S1E ). The survival analysis results demonstrated that strong expression of these five genes suggested poor OS in patients with LUAD ( Figure 2G ). The AUCs of all these five genes were >0.9 ( Figure 2H ), suggesting the significant diagnostic value of EFTUD2 in LUAD. The diagnostic and prognostic value of EFTUD2-related hub genes in LUAD indicated the crucial role of EFTUD2 in LUAD diagnosis and prognosis.

Considering the close relationship of EFTUD2 with the immune system (17, 42), we assessed the association of EFTUD2 with LUAD TIME. First, we analyzed the association between EFTUD2 and immune cell subtypes based on LUAD expression profiles in the TCGA. Compared with patients with low EFTUD2 expression, patients with high EFTUD2 expression demonstrated significantly lower abundance of memory B cells, resting memory CD4+T cells, resting myeloid dendritic cells, and activated mast cells but significantly higher abundance of activated memory CD4+T cells, M0 macrophages, M1 macrophages, and resting mast cells ( Figure 3A ). This suggests that EFTUD2 may regulate the infiltration levels of memory B cells, T cells, macrophages, and mast cells in LUAD tumors. However, EFTUD2 expression was noted to be correlated negatively with activated mast cell, resting memory CD4+T cell, memory B cell, and resting myeloid dendritic cell abundance but positively with resting mast cell, macrophage M0, activated memory CD4+ T cell, and M1 macrophage abundance ( Figure 3C ). In terms of the TIME scores, EFTUD2 was correlated significantly and positively with TumorPurity but negatively with StromalScore, ImmuneScore, and ESTIMATEScore ( Figure 3C ). However, StromalScore, ImmuneScore, and ESTIMATEScore values were significantly lower but TumorPurity scores were higher in patients with high EFTUD2 expression than in patients with low EFTUD2 expression ( Figure 3B ). Survival analysis revealed that LUAD patients with low ESTIMATEScore and ImmuneScore values or high TumorPurity scores had shorter OS ( Figure 3D ). Moreover, patients with low memory B cell abundance had a worse prognosis ( Figure 3D ). High expression of EFTUD2 reduces the abundance of memory B cells in LUAD tissues, suggesting that EFTUD2 may lower patient prognosis by affecting the infiltration levels of memory B cells. The GO enrichment analysis results demonstrated that EFTUD2 expression was strongly associated with immune pathways, including those related to cell adhesion, chemotaxis, immune response, and cell surface receptor signaling. KEGG enrichment analysis results showed that EFTUD2 was strongly associated with the PI3K-Akt signaling, chemokine signaling, extracellular matrix receptor binding, and focal adhesion pathways ( Figure 3E ). The five immune-related genes most strongly associated with EFTUD2 are depicted in Figure 3E ; of them, CHRDL1, SLIT3, and DNASE1L3 were negatively correlated with EFTUD2, whereas MTO19 was positively correlated with EFTUD2 ( Figure 3F ). Survival analysis revealed that patients with lower CHRDL1, SLIT3, and DNASE1L3 expression or higher MTO19 expression had significantly poorer prognosis ( Figure 3G ). These data suggest that EFTUD2 upregulation is strongly associated with an immune suppressive TIME.

EFTUD2 is involved in innate immune responses (43). Given the complex relationship between the cGAS-STING pathway and innate immunity, we explored whether EFTUD2 is associated with the cGAS-STING pathway. We first analyzed the expression of m6A-related genes between normal and LUAD tissues and found that SAMHD1, DTX4, TRIM21, and STAT6 expression was significantly lower and TBK1, IFI16, XRCC6, XRCC5, DDX41, IRF3, and PRKDC expression significantly higher in LUAD tissues ( Figure 4A ). Correlation analysis revealed that EFTUD2 was correlated positively with IFI16, SAMHD1, TBK1, DDX41, XRCC6, XRCC5, and PRKDC expression and negatively with STAT6 and DTX4 expression ( Figure 4B ). In addition, based on CNV analysis, we found that the expression levels of EFTUD2 were weakly positively correlated with the copy number variations of TBK1(Spearman: 0.15) and SAMHD1(Spearman: 0.18), whereas other genes did not show the same trend ( Supplementary Figure S1F ). This suggests that the elevated expression of TBK1 and SAMHD1 observed when EFTUD2 is upregulated may be partially influenced by copy number variations. The PPI demonstrated the presence of significant interactions between EFTUD2 and the aforementioned genes ( Figure 4C ). Moreover, our K–M survival analysis on these EFTUD2-related genes revealed that patients with high XRCC6 and XRCC5 expression had poor prognoses ( Figure 4D ). Thus, EFTUD2 may be involved in cGAS-STING pathway regulation.

Because m6A modification orchestrates mRNA precursor processing, EFTUD2 is involved in spliceosome activation (8). Thus, we explored associations between EFTUD2 and m6A modification and compared the differential expression of m6A-related genes in normal and LUAD tissues. The results demonstrated that METTL14, RBM15B, METTL15, YTHDF2, RBM15, IGF2BP2, METTL3, RBMX, HNRNPA2B1, YTHDF1, HNRNPC, and IGF2BP3 were significantly overexpressed in LUAD tissues, whereas FTO, ZC3H13, and WTAP were significantly underexpressed ( Figure 4E ). We then analyzed the correlations between EFTUD2 and the aforementioned genes in patients with LUAD, and the results demonstrated that EFTUD2 expression was correlated positively with RBM15, WTAP, RBMX, ZC3H13, HNRNPC, YTHDF1, HNRNPA2B1, RBM15B, IGF2BP3, and IGF2BP2 expression and negatively with METTL15 expression ( Figure 4F ). Similarly, to assess whether changes in the expression levels of m6A genes were influenced by genetic mutations, we conducted CNV analysis. The results indicated that, among the m6A genes potentially affected by EFTUD2 levels, the mRNA levels of IGFBP2 (Spearman: 0.19), WTAP (Spearman: 0.12), and METTL15 (Spearman: -0.10) may be influenced by copy number variations to some extent; however, the correlation coefficients were relatively low ( Supplementary Figure S1G ). We also analyzed the interactions between EFTUD2 and the aforementioned m6A-associated genes and noted a robust correlation of EFTUD2 with the genes ( Figure 4G ). The K–M survival analysis revealed that IGF2BP2, HNRNPC, and IGF2BP3 expression—positively correlated with EFTUD2—were associated with poor prognosis in patients with LUAD ( Figure 4H ). These results indicated that m6A modification is strongly associated with EFTUD2.

To elucidate the biological functions of EFTUD2 in LUAD, we first detected EFTUD2 mRNA and protein expression in the LUAD cell lines A549, H1299, and PC9 and the normal lung fibroblasts cell line MRC-5. The results demonstrated that EFTUD2 expression was significantly higher in the LUAD cells than in the normal cells ( Figure 5A ). We knocked down EFTUD2 expression in the LUAD cells by using three siRNAs targeting distinct sequences and then detected EFTUD2 protein levels ( Figure 5B ). The results demonstrated that both si-EFTUD2#1 and si-EFTUD2#3 significantly interfered with EFTUD2 expression. Compared to si-EFTUD2#1, si-EFTUD2#3 achieved the most effective knockdown in A549 cells while exhibiting relatively consistent knockdown efficiency in PC9 and H1299 cells ( Figure 5C ). Therefore, we selected si-EFTUD2#3 for subsequent experiments. CCK-8 assays revealed that the proliferation of the three LUAD cell lines significantly decreased after EFTUD2 knockdown ( Figure 5E ). The transwell assay results demonstrated that in all three cell lines, compared with the negative control, si-EFTUD2#3 reduced cell invasion considerably ( Figures 5 D, G ). The wound healing assay results demonstrated that EFTUD2 knockdown significantly attenuated wound healing in LUAD cells compared with the control group. ( Figures 5F, H ). Moreover, PC9 and A549 cells transfected with si-EFTUD2#3 demonstrated significant decreases in their colony formation ability, indicating that EFTUD2 downregulation significantly weakened the tumorigenicity of LUAD cells ( Figures 5I, J ). Flow cytometry assays revealed that EFTUD2 knockdown led to cell-cycle arrest at the G0/G1 phase in H1299 cells ( Figure 5K ) and A549 cells ( Figure 5L ). Thus, EFTUD2 knockdown can inhibit proliferation, migration, invasion, tumorigenicity, and cell-cycle progression in LUAD cells.

To investigate the possible molecular mechanisms of EFTUD2 in LUAD, we performed GSEA on genes in the low and high EFTUD2 expression groups of patients with LUAD from TCGA. The results demonstrated that EFTUD2 upregulation is correlated with a prevalent enrichment of gene signatures across several pathways: the cell cycle, glycolysis, insulin signaling, MAPK, RIG-I-like receptor, wnt signaling, RNA degradation, and autophagy regulation ( Figure 6A ). Because glycolysis is a crucial metabolic process, providing energy to cancer cells, we examined the correlation between EFTUD2 and key enzymes involved in glycolysis in LUAD by using TIMER and identified significant associations between EFTUD2 and various regulatory enzymes including ENO1, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), GPI, HK2, LDHA, PFKM, pyruvate kinase M2 (PKM2), PGAM1, PGK1, TPI1, and aldolase B (ALDOB; Figure 6B ). We further investigated whether EFTUD2 is involved in glycolysis in the LUAD cell line A549. The results demonstrated that EFTUD2 knockdown led to a significant decrease in PKM2 and GAPDH protein expression but a significant increase in ALDOB protein expression ( Figures 6C, D ).

We established a tumor xenograft model to investigate the role of EFTUD2 in mediating LUAD tumor growth in vivo. There was no significant difference in body weight between the control and knockdown groups throughout the mice’s growth period ( Figure 7C ). Over time, the growth of tumor volume in the EFTUD2 knockdown group gradually slowed down ( Figure 7D ). By day 16, the tumor volume and weight in the knockdown group were significantly lower than those in the control group ( Figures 7 A, B ). HE staining of the tumor xenografts revealed abundant microvessels and active tumor cell proliferation in the control group. In contrast, the EFTUD2 knockdown group displayed tumor necrosis foci, a reduced number of microvessels, and diminished tumor cell proliferation ( Figure 7E ).

We measured lactate levels in the serum and tumor tissues of the collected mouse samples. The results revealed that, compared to the control group, lactate levels in both serum and tumor tissues of the EFTUD2 knockdown mice were significantly reduced ( Figures 7F, G ). We further evaluated the effect of EFTUD2 on key glycolytic enzymes using IHC staining. Following EFTUD2 knockdown, the protein expression of the glycolytic enzymes GAPDH and PKM2 decreased, while the expression of the negative regulator ALDOB increased ( Figures 7H–K ). These findings are consistent with the results from the cell experiments shown in Figure 6C . Additionally, TUNEL assays on tumor samples revealed that, compared to tumors from control mice, the tumor regions in the EFTUD2 knockdown group exhibited increased apoptosis ( Figure 7L ). These data suggest that knocking down EFTUD2 inhibits LUAD tumor growth in mice and suppresses glycolytic activity. Conversely, overactivation of EFTUD2 may promote LUAD tumor proliferation and elevate lactate metabolism levels in both the tumors and the host.