Skip to main content
Download PDF

A novel mitochondrial quality regulation gene signature for anticipating prognosis, TME, and therapeutic response in LUAD by multi-omics analysis and experimental verification

Cancer Cell International volume 25, Article number: 138 (2025) Cite this article

Abstract

Background

Lung adenocarcinoma (LUAD) is the predominant form of non-small cell lung cancer (NSCLC). Mitochondrial quality-related genes (MQRGs) contribute to the genesis and advancement of tumors. Despite advances in LUAD treatment and detection, early diagnostic biomarkers are still lacking, and the roles of MQRGs in LUAD are not well understood.

Methods

We extensively examined transcriptome and clinical data from TCGA and GEO databases to discover differentially expressed MQRGs. Utilizing the LASSO algorithm and multivariate COX regression, a predictive risk model was created. Kaplan-Meier study and ROC curves were implemented to predict patient prognosis, resulting in a new Mitochondrial Quality Regulation Gene Signature for accurate prognosis forecasting. R software and packages facilitated statistical, consensus cluster, survival, Cox regression, Lasso regression, and tumor microenvironment analyses. Model-related gene expression was measured using RT-qPCR, immunohistochemistry, single-cell sequencing, HPA data, and UNCAN data.

Results

We created a concise risk model using four MQRGs (STRAP, SHCBP1, PKP2, and CRTAC1) to forecast overall survival in LUAD patients. High-risk patients experienced significantly lower survival rates. Functional analysis linked these MQRGs to alpha-linolenic acid metabolism pathways. Moreover, the tumor immune microenvironment supports previous findings that higher CD8 + T cell infiltration improves LUAD outcomes. Analysis of different risk scores showed increased activated memory T-cell CD4, suggesting its activation is crucial for LUAD prognosis. Nomograms were generated with clinical data and the MQRGscore model. mRNA and IHC analysis manifested significantly upregulated STRAP, SHCBP1, and PKP2 expression and mitigated CRTAC1 expression in the LUAD contrasted with normal lung tissue. qRT-PCR and immunohistochemistry confirmed these findings, aligning with TCGA data.

Conclusions

We created a succinct MQRGs risk model to ascertain the LUAD patient’s prognosis, potentially offering a novel method for diagnosing and treating this condition.

Introduction

At over 40% of all lung cancer instances, the prevailing histological subtype of non-small cell lung cancer (NSCLC) is lung adenocarcinoma (LUAD) [1, 2]. Characterized by a median 5-year survival rate of about 15%, it remains the prevailing reason for cancer-linked fatalities worldwide [3]. LUAD patient’s prognosis remains unfavorable, mostly because of delayed diagnosis and elevated rates of metastasis and recurrence [4]. There has been limited success in improving long-term survival rates with current treatment modalities, encompassing surgery, chemotherapy, immunotherapy, and targeted therapies [5]. Identifying new markers and therapeutic targets is essential to enhance the precision of diagnosis, prognosis, and therapy in LUAD instances [6].

Moreover, it seems that mitochondria are key in many steps of tumorigenesis and disease [7,8,9,10]. For example, Mitochondrial dysfunction is thought to be closely associated with LUAD [11]. Optimal mitochondrial activity is crucial for the proper functioning of cells, and any impairment in mitochondrial function might potentially lead to a diverse range of human disorders, including cancer [8]. As cancer cells proliferate, mitochondrial oxidative phosphorylation is the primary energy source. Mutations in mitochondrial oxidative metabolism, which disrupt energy production, can facilitate tumor initiation and progression. Multiple cancer types, encompassing breast [12, 13], pancreatic [14], colon [15], prostate [16], and liver cancers [17], have been associated with mitochondrial quality. At now, there is a scarcity of studies examining the connection between mitochondrial quality-related genes (MQRGs) and LUAD. The elucidation of the processes of mitochondrial quality control (MQC) in cancer etiology is crucial for developing new treatment approaches for LUAD. Extensive research efforts have been directed toward identifying the genes and proteins that are functionally active within mitochondria to address the myriad of diseases associated with mitochondrial dysfunction. MitoCarta 3.0 provides a comprehensive mitochondrial protein inventory, resulting in MQRGs [18].

However, investigating the link between MQRGs and LUAD is of paramount importance. In this investigation, according to the specific gene group (MQRG) expression profiles, we aimed to classify LUAD patients into molecular subtypes, construct a prognostic model, and investigate the association between these subtypes and clinical characteristics, tumor microenvironment (TME), immune status, and drug sensitivity. The conclusion is as follows: our work provided a thorough description of the MQRG’s prognostic and predictive significance in LUAD. Our comprehensive analysis identified four (STRAP, SHCBP1, PKP2, and CRTAC1) biomarkers associated with LUAD prognosis, providing insights into the molecular mechanisms driving disease progression. A personalized therapeutic strategy and improved patient outcomes may be possible based on these findings.

Methods

Data acquisition

The research methodology is shown in Figure S1. RNA-seq data were acquired from 522 LUAD specimens downloaded from TCGA (https://portal.gdc.cancer.gov) and 117 LUAD specimens acquired from the GEO database (https://www.ncbi.nlm.nih.gov/geo/), comprising somatic mutation patterns and their pertinent clinical information, which included the vital condition, age, gender, tumor grade, and pathological stage of LUAD patients. In addition, the model-linked gene expression patterns were validated by collecting GSE13213 records from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). The datasets examined for this work are available on the UNCAN online platform (https://uncan.eu/) and the HPA website (https://www.proteinatlas.org/). The data were all converted and standardized. From previous publications, we retrieved twenty MQRGs [19]. In Table S1, these genes are displayed. The clinical information and data implemented in this investigation were sourced from public databases; as a result, written informed consent from patients and ethics committee approval was not necessary for this investigation.

Consensus clustering analysis of MQRG

The R package “ConsensusClusterPlus” was implemented to analyze consensus clusters. After clustering, the correlation within subtypes was enhanced, as indicated by the cumulative distribution function (CDF) curve, which exhibited a flatter slope compared to steeper slopes that typically indicate strong within-group similarities. In contrast, correlations between subtypes were notably weaker, highlighting a clear distinction between the clustered groups. Through analyzing the prognostic MQRG expression profile, specimens were categorized into several MQRG subtypes. Moreover, a principal component analysis (PCA) was implemented. The classification of several subgroups can be evaluated using PCA analysis, which also partially reflects the variations amongst subgroups. Our Kaplan-Meier survival study was conducted with the “survival” program to compare the different MQRG survival rates subtypes.

Determination of DEGs and functional enrichment analysis

With an adjusted p-value of 0.05 and a fold-change of 2.0, 660 DEGs were estimated across different MQRG subtypes with the R “limma” package. Enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG) helped to elucidate the molecular signaling pathways implicated [20]. The classification and description of gene and protein activities, such as molecular function (MF), cellular component (CC), and biological process (BP), are made possible depending on gene ontology (GO) enrichment analysis [21]. Utilizing the “clusterProfiler” program, we conducted a functional enrichment investigation of DEGs with GO and KEGG analysis. Differential analysis at the signaling system level is made possible by gene set variation analysis (GSVA). According to our investigation, the “clusterProfiler” and “GSVA” packages can be employed to implement them [22].

Create the MQRG risk score model

The MQRG score was computed to quantify tumor-specific MQRG patterns. The first stage included performing a univariate Cox regression study on the dataset to detect DEGs linked to LUAD OS. Additional unsupervised clustering was used to detect two distinct subtype groups by analyzing the prognostic MQRG gene expression, namely MQRG gene subtypes A and B. Additionally, a 1:1 randomization procedure was employed to randomly assign all patients with LUAD to either the training or testing groups. Training datasets were used to compute the prognostic scores for MQRG. The “glmnet” R package was used to perform Lasso Cox regression on the MQRG predictive gene set in order to mitigate the potential for over-fitting. A 10-fold cross-validation model was devised after an analysis was performed on each independent variable change trajectory. Multivariate Cox analysis was used to forecast MQRG statistics within the training set. A median risk score was implemented to classify patients into low-risk groups (LRG) (MQRG scores below the median value) and high-risk groups (HRG) (MQRG scores above the median value). A KM survival analysis was then conducted on these groups. Similarly, the testing set was allocated into LRG and HRG, and receiver operating characteristic (ROC) curves and KM survival studies were conducted on each group.

Establishment and verification of the nomogram

A predictive nomogram was created with the R package “rms,” encompassing clinical characteristics and risk scores. Every variable’s score contributes to a patient’s score. We obtained the AUC values for 1, 3, and 5 years based on each patient’s total score. Additionally, ROC curves were generated to ascertain this nomogram’s effectiveness during these specific time intervals. A calibration plot was created to evaluate this nomogram performance by comparing founded survival events of 1, 3, and 5 years with the anticipated outcomes.

Culture of cells with qRT-PCR analysis

The cell lines Beas-2B, PC9, A549, HCC827, and H1299 used in our study were purchased from Procell (Wuhan, China). The frozen batch of Beas-2B, PC9, A549, HCC827, and H1299 cells was defrosted by immersing them in a water bath at 37°C. The cells were cultured on 10 cm plates using RPMI-1640 medium enriched with 10% fetal bovine serum and 1% penicillin/streptomycin. The specimens were subsequently incubated in a 37°C humidified atmosphere containing 5% carbon dioxide. Ribosomal RNA was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Synthetic complementary DNA (cDNA) was generated by mixing total RNA with a Takara PrimeScript RT reagent kit. The RT-qPCR technique was performed on a CFX-96 instrument (Bio-Rad Laboratories, Inc. USA), utilizing Takara SYBR-Green assays. The 2−ΔΔC t approach was used to compile the data and normalize it using GAPDH. The primers for STRAP [23] were 5’-AAGGGACACTTTGGTCCTATTC-3‘(fwd), 5’-CCTACCACAGTTTGCCATAGT-3’ (rev). The primers for SHCBP1 [24] were 5’-GCTACCGTGATAAACCAGGTTC-3‘(fwd), 5’-AGGCTCTGAATCGCTCATAGA-3‘(rev). The primers for PKP2 [25] were 5’-AGATTACCAGCCAGATGACA-3‘(fwd), 5’-ATGCCACAGCCACTCCAC-3‘(rev). The primers for CRTAC1 [26] were 5’-CCCTGGCTGACTTCAACCGT-3’ (fwd) and 5’-ATGGAGAACTTGGGTGAGGC-3’ (rev). Here are the primer sequences that were used in this study’s qRT-PCR.

Immunohistochemical (IHC) staining

Experimental procedure. IHC staining was performed for tissue sections and Lung adenocarcinoma tissue microarray(TMA)(AiFang Biological). A, Dewax paraffin sections by immersing them sequentially in Xylene I, II, and III for 15 min each, then in Anhydrous Ethanol I and II for 5 min each, followed by 85% Ethanol and 75% Ethanol for 5 min each, and finally rinse with distilled water. B, For antigen retrieval, dilute 1 M Citrate buffer (100×, pH 6.0) 100 times, immerse sections in the solution in a pressure cooker, boil for 3 min, cool at room temperature, then wash in PBS (pH 7.4) with gentle shaking for 3 cycles of 5 min each. C, Block endogenous peroxidase by incubating sections in 3% hydrogen peroxide in the dark for 15 min, then wash in PBS (pH 7.4) with gentle shaking for 3 cycles of 5 min each. D, Serum blocking: Apply 3% BSA(G5001, Servicebio) to the circled slide areas to cover the tissue, incubating at room temperature for 30 min. Use rabbit serum for goat-derived primary antibodies; otherwise, use BSA. E, Primary antibody (STRAP, 18277-1-AP, Sanyi) application: Remove the blocking solution, add a 1:500 diluted antibody, and incubate overnight at 4 °C in a humidified chamber with water to prevent evaporation. F, Secondary antibody application: Wash slides three times in PBS (pH 7.4) for 5 min each on a shaker. After drying, apply the rabbit anti-HRP secondary antibody to the circled areas and incubate at room temperature for 30 min. G, DAB Staining: Shake slides in PBS (pH 7.4) for 5 min, three times. Dry sections, apply fresh DAB solution, and monitor staining under a microscope for a brown-yellow result. Rinse with tap water to stop staining. H, Hematoxylin Counterstaining: Counterstain with hematoxylin for 3 min, rinse, differentiate briefly, rinse, blue with blueing solution, and rinse again. I, Dehydration and Mounting: Dehydrate sections in 75% ethanol, 85% ethanol, Anhydrous Ethanol I and II, n-Butanol, and Xylene I, each for 5 min. Air dry and mount with neutral resin. J, Microscopic Examination and Image Analysis. Hematoxylin stains the cell nuclei blue, while the positive expression revealed by DAB appears brown-yellow.

Statistics. K. Using Visiopharm software, the entire tissue area was selected (chip circle selection: open the chip viewer, naming the tissue area from left to right: 1, 2, 3… and from top to bottom: A, B, C… directions). The target objects were then analyzed using H-Score. The H-Score was calculated as follows: H-Score = ∑(pixi) = (percentage of weak intensity cells × 1) + (percentage of moderate intensity cells × 2) + (percentage of strong intensity cells × 3), where i represents the grading of positive cells: negative (no staining) = 0 points; weak positive (light yellow) = 1 point; moderate positive (brownish yellow) = 2 points; strong positive (brown) = 3 points. Pi represents the percentage of positive cells. The H-Score ranges from 0 to 300, with a higher value indicating stronger overall positive staining intensity. The positivity rate was calculated as the number of positive cells (grades 1 + 2 + 3) divided by the total number of cells.

Assessment of immune cells infiltration and TME

The therapeutic relevance of the aforementioned categorization was established by examining the correlation between MQRG subtypes and factors such as age, sex, T and N stages, and prognosis. The ssGSEA approach was implemented to ascertain the proportional composition of 23 various immune cell types in each LUAD specimen. Additionally, every LUAD sample was assigned an ImmuneScore, StromalScore, and ESTIMATEScore with the ESTIMATE package. To ascertain the percentage of tumor-infiltrating immune cells (TIICs) in the TME, the CIBERSORT method was implemented to quantify the infiltrating immune cell number in heterogeneous specimens from both the LRG and HRG. Four genes with MQRG scores were compared with the fractions of the 19 infiltrating immune cells.

Assessment of the TMB, CSC, and mutation

Through the use of the R package “maftools“ [27], a mutation annotation format (MAF) was developed with TCGA to detect mutations with HRG and LRG in LUAD patients. Furthermore, we examined the connection between CSC, TMB, and the HRG and LRG.

Drug susceptibility analysis

The “pRRophetic“ [28] software was implemented to compute the semi-inhibitory concentrations (IC50) of chemotherapeutic medications routinely used in treating LUAD, to ascertain whether the therapeutic implications of these treatments differ between the two groups.

Data processing for Single-cell sequence and mRNA expression

Model-related gene expression was analyzed using data from the TCGA. The ULCAN database (https://ulcan.path.uab.edu/) was implemented to examine the model-linked gene protein expression. The GEO database provided the NSCLC scRNA-seq datasets GSE99254. Utilizing the R package “Seurat,” the samples were combined. Cell data satisfying the specified criteria were preserved, including gene counts ranging from 300 to 7,000 and total. The Human protein mapping tool accessible at https://www.proteinatlas.org/ assesses the expression of LUAD and DEG proteins in healthy lung tissue.

Statistical analysis

The data for this work were generated using the Perl programming language (v5.32.1). The processing, analysis, and display of data were carried out using R (version 4.3.3). A statistical difference was stated as P < 0.05 between the comparison groups.

Results

Transcriptional and genetic alterations of MQRGs in LUAD

Herein, we incorporated 20 MQRGs. The differential expression analyses of MQRGs from normal and LUAD samples revealed that five MQRGs were significantly upregulated in LUAD samples, including TFAM, ESRRA, MFN1, OPA1, and MFF. Conversely, 12 MQRGs were significantly down-regulated, including PPARGC1A, PPARG, NRF1, NFE2L2, MFN2, FIS1, MIEF, PINK1, PARK2, MAP1LC3A, MAP1LC3B and MAP1LC3B (Fig. 1A). According to our research, there is a notable difference between normal and LUAD samples in terms of MQRG expression levels and genetic landscape. These outcomes reveal that MQRGs could have a hidden role in LUAD oncogenesis.

Fig. 1
figure 1

The genetic and transcriptional modifications of 20 MQRGs in LUAD. (A) Analysis of 20 MQRGs that are differentially expressed between normal and LUAD specimens. (B) The MQRGs somatic mutation frequency in the TCGA cohort. (C) Determining the positions of CNV modifications in MQRGs on 23 chromosomes. (D) Ascertaining CNV gain, loss, and non-CNV Frequencies of MQRGs. (E) Correlation analysis for the MQRGs. (F) Interactions between MQRGs in LUAD. The MQRGs interaction is shown as a line, with the line thickness indicating the link intensity. The pink and blue lines on the graph indicate positive and negative connections, respectively. MQRGs, mitochondrial quality-related genes; LUAD, lung adenocarcinoma; TCGA, The Cancer Genome Atlas; CNV, copy number variant; *p < 0.05; **p < 0.01; ***p < 0.001

Full size image

A low-frequency mutation was detected in LUAD samples based on analysis of somatic mutations in 20 MQRGs. There were 105 mutations in MQRGs out of 616 LUAD samples (17.05%), with PPARGC1A having the greatest mutation frequency (4%), then NFE2L2 (3%) and PRKN (2%) in LUAD samples. Eight genes (PPARG, MFN1, SQSTM1, MFN2, NRF1, MFF, MAP1LC3C, and OPA1) each had a mutation rate of 1%, whereas nine genes (PPARA, MIEF1, MIEF2, PINK1, TFAM, FIS1, MAP1LC3A, ESRRA, and MAP1LC3B) showed no mutations (Fig. 1B). Figure 1C visually represents the exact locations of the CNV changes in the MQRGs on their respective chromosomes.

Furthermore, it was found that 20 regulators have a prevalent CNV alteration, mostly copy number amplification, whereas PINK1, PPARA, MFF, PRKN, MIEF2, and MAP1LC3 have dispersed CNV deletions (Fig. 1D). We also performed a correlation analysis of MQRGs (Fig. 1E). The research we conducted revealed a noteworthy distinction in the MQRGs expression levels and genetic landscape between the LUAD and control samples. Our outcomes exhibit that MQRGs may have a hidden role in LUAD oncogenesis. LUAD patients with high expression of MAP1LC3A, MAP1LC3C, PARK2, PPARGC1A, PINK1, and PPARA had better OS (Figure S2). LUAD MQRG interactions and prognostic significance were systematically revealed by a network, as shown in Fig. 1F.

Identification of MQRG subtypes in LUAD

Heatmaps of the consensus matrix indicated that the classification technique with k = 2 was the most effective (Figure S3A-C). Additionally, the LUAD samples were categorized into MQRG clusters A and B (Figs. 2A). PCA exhibited significant variation in the two MQRGs subtypes transcription patterns (Fig. 2B). Based on the Kaplan-Meier curves (log-rank test, P < 0.001; Fig. 2C), subtype B patients possessed a longer OS contrasted with subtype A patients.

Fig. 2
figure 2

Comparison of clinical characteristics and immune infiltration between two consistently clustered MQRG subtypes in LUAD specimens. (A) Investigation of prognostic MQRGs by clustering analysis utilizing consensus matrix heatmaps. Two clusters discovered with a value of k = 2 were shown along with their corresponding correlation regions. (B) Based on the PCA analysis of the prognostic MQRGs, two groups of patients were found in the different subtypes of MQRG. Blue and yellow dots denote MQRGcluster A and MQRGcluster B. (C) Kaplan-Meier curves for OS of the two MQRG subtypes (chi-square test, P < 0.001). (D) Heat map illustrating the variations in clinical characteristics and expression levels of MQRGs between MQRGcluster A and B. (E) A GSVA which compares the biological pathways of two distinct subtypes, in which red denotes pathways that are activated, and blue denotes those that are suppressed. (F) GSVA of molecular function between two different subtypes, where blue and red, respectively, stand for blocked and activated pathways. (G) The prevalence of 23 distinct kinds of infiltrating immune cells in the two MQRG subtypes. MQRGs, mitochondrial quality-related genes; LUAD, lung adenocarcinoma; PCA, principal components analysis; OS, overall survival. GSVA, gene set variation analysis; *p < 0.05, **p < 0.01, ***p < 0.001; TME, tumor microenvironment

Full size image

Comparison of clinical characteristics and immune infiltration between the two gene subtypes

Comparing the clinicopathologic features of different subtypes of LUAD by TCGA and GSE13213 databases, we found significant differences in MQRGs expression and clinicopathologic features (Fig. 2D). The cluster A possessed elevated six MQRGs (TFAM, MFN1, OPA1, NRF1, NFE2L2, PPARA) expression contrasted with in cluster B. As cluster A suffers from poor overall prognosis, these genes are presumably significant molecules that influence the LUAD patient’s prognosis. In LUAD, MFN1 has been implicated in regulating mitochondrial dynamics and cellular responses to stress [29]. The downregulation of MFN1 disrupts mitochondrial fusion, leading to impaired mitochondrial function and altered cellular metabolism, which can contribute to tumor progression. Furthermore, MFN1’s interaction with other mitochondrial proteins, such as OPA1, is essential for maintaining the mitochondria integrity and functionality. The discovery indicates that MFN1 contributes to both mitochondrial dynamics and the regulation of immunological responses. This has potential consequences for advancing treatment approaches aimed at enhancing mitochondrial function in LUAD. TFAM has been shown to be significantly upregulated, suggesting its role in mitochondrial biogenesis and function [30]. Moreover, tobacco-specific nitrosamines have been found to upregulate TFAM, thereby stimulating mitochondrial redox signaling and promoting lung tumor growth [31]. Given these findings, TFAM appears to be a key player in LUAD progression, making it a possible target for therapeutic intervention. Based on GSVA enrichment analysis, pathway B was significantly enriched, including arachidonic acid metabolism, alpha-linolenic acid metabolism, neuroactive ligand-receptor interaction, linoleic acid metabolism, and complement and coagulation cascade pathways (Fig. 2E). The GO pathways differential enrichment between clusters A and B was quantified using the GSVA R program (Fig. 2F). An investigation on the involvement of alpha-linolenic acid (ALA) metabolism in cancer, including LUAD, mammary carcinomas [32], nasopharyngeal carcinoma [33], and prostate cancer [34], has attracted considerable interest because of its possible consequences on tumor development and patient prognosis.

We examined the MQRGs contribution in the TME of LUAD by analyzing the connections between the two subtypes and 23 specific human immune cell subsets in each specimen with the CIBERSORT technique. Significant disparities were revealed in most immune cell infiltration between the two subtypes (Fig. 2G). The B subtype exhibited elevated levels of 15 immune cell types, including activated B cells, activated CD8 + T cells, activated dendritic cells, CD56dim natural killer cells (NKc), Eosinophilous, Immature B cells, MDSC, macrophages, mast cells, monocytes, neutrophils, T follicular helper cells, plasmacytoid dendritic cells, Type 1 T helper cells, and Type 17 T helper cells, when compared to the A subtype. Cluster B exhibited significantly larger levels of activated CD8 + T cells, CD56dim NKc, and macrophages compared to cluster A. The immunological cells in the TME have crucial functions in regulating the development and metastasis of tumors. Specifically, macrophages (Mφ) and T cells were found to be the predominant immune cells in LUAD tissues, with macrophages showing the highest infiltration levels [35]. The existence of elevated CD8 + T cell levels, which are associated with better immune responses, was more pronounced in certain subtypes, correlating with improved patient survival [36]. Elevated levels of CD8 + T, CD56bright NK cells, and decreased CD56dim NK cells were observed in certain LUAD subtypes, suggesting a robust adaptive antitumor immune response but impaired innate immunity [37]. This aligns with previous findings that higher CD8 + T cell infiltration correlates with better clinical outcomes in LUAD [38]. Overall, our findings pave the way for personalized immunotherapy approaches that target specific immune cell populations and their regulatory pathways to improve patient outcomes in LUAD.

Generation of gene subtypes utilizing DEGs

A total of 660 DEGs linked to MQRG subtypes were discovered with the R package “limma.” Subsequently, functional enrichment analyses were conducted (Figure S4A). There was a significant enrichment of MQRG subtype-related genes in immune-linked BP (Fig. 3A). The analysis of KEGG pathways showed enrichment in immune and cancer-linked pathways (Fig. 3B), indicating that MQRG regulates the immune response within the TME. By using consensus clustering algorithms to assess this regulation mechanism further, we classified individuals into two genomic subtypes according to prognostic genes. The findings indicated that when there were two subgroups, the best grouping outcomes could be achieved (Figures S4B–E).

Fig. 3
figure 3

Generation of gene subtypes utilizing DEGs. (A-B) Enrichment analyses of DEGs among two MQRG subtypes using GO and KEGG. (C) OS curves for the two gene subtypes, as represented by Kaplan–Meier curves (log-rank tests, P < 0.001). (D) The heat map illustrates the variations in clinical characteristics and predictive DEG expression levels between geneCluster A and B. (E) Variations in the 17 MQRGs genes expression on the two gene subtypes. DEGs, differentially expressed genes; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MQRGs, mitochondrial quality-related genes

Full size image

Gene cluster B possessed a superior OS contrasted with gene cluster A, as shown by the KM curves. These curves revealed a noticeable disparity in MQRG between the two gene clusters (log-rank test, P < 0.001; Fig. 3C). There were no noticeable disparities in age between the two subtypes based on their clinical characteristics. At the same time, geneCluster A had higher expression levels of most prognostic DEGs than geneCluster B (Fig. 3D). The majority of the MQRGs had different expressions, according to our analysis of MQRG expression between these two gene clusters (Fig. 3E).

Discovery of prognostic genes and development of predictive MQRG scores

A predictive risk score model was constructed using MQRG subtype-linked DEGs. Initially, LUAD patients were randomized at random to either the training or testing groups. Both training and testing sets contained 312 LUAD samples, with 312 LUAD samples in each. To further determine the ideal prognostic signature for MQRG subtype-related prognostic DEGs, LASSO and multivariate Cox studies were performed. Nine OS-associated genes were found using LASSO regression analysis, as shown by the minimal partial likelihood of deviance (Figures S5A-B). Figure 4A illustrates the LUAD specimen distribution using different classification methods. After that, we assessed nine OS-linked genes using multivariate Cox regression analysis, ultimately identifying four (STRAP, SHCBP1, PKP2, and CRTAC1). The MQRG score was developed with multivariate Cox regression analysis with the subsequent approach: Risk score = (0.291816619100497* STRAP expression) + (0.245634683234024* SHCBP1 expression + (0.13014817748504* PKP2 expression) + (-0.115972961512995* CRTAC1 expression).

As a result of the differential analysis, geneCluster A and MQRGcluster A showed higher risk scores than geneCluster B and MQRGcluster B (Figs. 4B-C).

Fig. 4
figure 4

LASSO regression, MQRG score creation, and TME and checkpoint assessment across risk groups. (A) Analyzed using LASSO regression, representative potential prognostic genes were discovered together with the partial likelihood deviation on these genes. Analysis of the disparity in risk scores between the two gene subtypes (B) and between the two subtypes of MQRG (C). (D) Risk scores, survival status, and four prognostic gene expressions’ distribution in the training. (E) KM curve outcomes for LUAD patients with various MQRG scores in the training group. (F) Based on the MQRG scores, ROC curves are used to estimate the sensitivity and specificity of 1-, 3-, and 5-year survival in the training group. (G) Expression differences of MQRGs in different MQRG scores. (H) Correlation heatmap of gene models and immune cell contents. (I) The violin diagram displays TME scores in high and low-risk groups. LUAD, lung adenocarcinoma; ROC, receiver operating characteristic. *p < 0.05; **p < 0.01; ***p < 0.001

Full size image

Using the risk score function defined above, in the training group, Fig. 4D computes each patient’s risk score. Utilizing the “SurvMiner” R software package, the median critical point was ascertained, and thereafter the patients were classified into HRG and LRG. As the Risk score rose, patients’ survival periods were mitigated, and the fatalities number accordingly rose. Figure 4D displays a heatmap of four prognostic genes assigned to HRG and LRG. Analysis of the KM survival plots exhibited that LRG experienced much greater survival rates than those with higher scores (log-rank test, P < 0.001; Fig. 4E). Moreover, the AUC values obtained from this model for 0.5, 1.0, and 1.5 years were 0.703, 0.694, and 0.703, respectively (Fig. 4F). The multigene model possessed strong predictive capability for the LUAD patients’ OS.

Our evaluation of the MQRG scores’ predictive accuracy was conducted by calculating them on the testing set. Furthermore, the patients were classified into LRG and HRG according to the method implemented for the training set. Figures S6A depict the disparities in MQRG scores and patient survival status across LRG and HRG. Statistical examination of survival data showed that the LRG had much superior prognoses compared to the HRG (log-rank; P < 0.001; Figure S6B). An analysis of the classification efficiency of the MQRG scores at 1, 3, and 5 years reported that the AUC values were still relatively high despite the fact that they were derived from the prognostic prediction scores (Figure S6C). Furthermore, we examined the variance in expression levels of MQRGs across different MQRG scores. Our findings revealed that 11 genes exhibited differential expression among 17 MQRGs, with the majority of these genes showing elevated expression in the LRG (Fig. 4G).

Assessment of checkpoints and TME between HRG and LRG

We implemented the CIBERSORT approach to evaluate the connection between MQRG scores and the immune cell number. The examination of scatter plots indicated a direct link between MQRG scores and many cell types, including M0/1 macrophages, active mast cells, neutrophils, activated memory CD4 + T cells, and CD8 + T cells. MQRG scores were shown to be adversely linked to resting dendritic, active dendritic, resting mast, activated NK, and resting memory CD4 + T cells and monocytes (Figure S7). The detection of a greater percentage of active memory T-cell CD4 between immune cells suggests that the memory T-cell CD4 activation could possess a significant influence on the LUAD patient’s prognosis. The outcomes reveal a strong connection between high stromal and immune scores and low MQRG scores, as depicted in Fig. 4I. Furthermore, the correlation analyses conducted between immune cell contents and four model-related gene expression levels revealed significant associations between gene expression levels and specific types of immune cells, as illustrated in Fig. 4H.

Distinctive Features of MQRG Modification in Immunotherapy

The following outcomes indicated that PD-L1 exhibited varying levels of expression between groups with high and low MQRG scores, with the latter group demonstrating high levels of expression (P = 4.4e-08; Figure S8A), potentially indicating a favorable response to anti-PD-L1 treatment. Therefore, a study was performed to ascertain if the alteration pattern of MQRG might forecast the reaction of patients to immune checkpoint blockade (ICB). Data on the IPS of LUAD was obtained from The Cancer Immunome Atlas (TCIA) available at https://tcia.at/home. Comparative analysis exhibited that patients in the high MQRGscore group possessed superior results when administered anti-PD-L1 treatment contrasted with those in the low MQRGscore group (Figure S8B). Notably, our work further emphasized the significance of immunological checkpoints, encompassing PD-L1/-1, in regulating LUAD. Differential expression of these checkpoints among LUAD subtypes underscores their potential as therapeutic targets. High PD-L1 expression in certain subtypes suggests a mechanism of immune evasion, which can be targeted by immune checkpoint inhibitors to enhance antitumor immunity [39, 40]. It was established that LUAD patients with elevated PD-L1 expression had enhanced responses to immunotherapy, elevating survival rate [41]. Overall, our outcomes provide a thorough understanding of the immune landscape in LUAD and highlight the importance of immune cell infiltration in determining patient prognosis and therapeutic response.

Connection of MQRG scores with TMB and CSC index

Much evidence surfaces that neoantigens produced by genetically altered tumor cells can induce strong immune responses, and patients with high TMB are thought to respond well to immunotherapy. Patients exhibiting a high tumor mutational burden (TMB) may derive therapeutic advantages from immunotherapy due to elevated neoantigen levels, as supported by emerging evidence. Our examination of mutation data from the TCGA LUAD cohort reveals that HRG had a significantly greater TMB contrasted with LRG (Fig. 5A), suggesting a potential benefit from immunotherapy in this subgroup. The TMB and MQRG scores showed a positive association, as shown by the Spearman correlation analysis (Fig. 5B). Furthermore, we combined the CSC index and MQRG score values to evaluate any possible link between the index and MQRG scores in LUAD. The linear link between the CSC index and MQRG scores is displayed in Fig. 5C. After analyzing the data, we revealed that the CSC index had a positive connection with MQRG scores (R = 0.47, P < 2.2e-16). This indicates that LUAD cells exhibiting elevated MQRG scores displayed increased stem cell characteristics and reduced levels of cellular differentiation. Additionally, individuals with high TMB demonstrated a significant survival advantage (Fig. 5D). The integration of risk score and TMB in Fig. 5E resulted in the classification of patients into four distinct groups with significantly varying prognoses (P < 0.001). Subsequently, an analysis was conducted to compare the distribution of somatic mutations between two groups based on MQRG scores within the TCGA-LUAD cohort. The ten most frequently detected mutant genes in both the HRG and LRG were TP53, TTN, MUC16, CSMD3, RYR2, LRP1B, ZFHX4, USH2A, KRAS, and XIRP2 (Figs. 5F–G). The evaluation reported a significant elevation in the mutation frequency among persons with a high MQRG score contrasted with those with a low score.

Fig. 5
figure 5

Tumor mutation landscape outcomes were analyzed across different MQRG score groups, and nomograms were created for LUAD patients. (A) The TMB expression in different MQRG scores. (B) The analysis used the Spearman correlation between TMB and MQRG scores. (C) Links between the CSC index and MQRG scores. (D) Results of survival analyses for high and low TMB patients in the TCGA-LUAD cohort. (E) Variations in OS among high and low TMB from TCGA. (F) Somatic mutation features result in high MQRG scores. (G) Somatic mutation features result in low MQRG scores. Every column denoted a distinct patient. The top barplot showed the expression of TMB. The numerical value on the right corresponds to the mutations frequency in each gene. The right barplot illustrates the percentage of each kind of variation. (H) Graph of nomogram generated using clinicopathological factors and MQRG score. (I) calibration plot implemented to verify the nomogram. LUAD, lung adenocarcinoma. TMB, tumor mutation burden; CSC, cancer stem cell

Full size image

Creation of A nomogram to anticipate survival

Utilizing clinical features and MQGR scores, a prognostic nomogram was developed to accurately forecast the LUAD patient’s prognosis (Fig. 5H). The fact that the result is nearly 45° suggests that the nomogram forecast is correct. The nomogram had good predictive potential, according to the results of the Concordance Index (C-index) (Fig. 5I).

Drug sensitivity analysis in different MQRG scores

The patient’s reaction to medication therapy may be reflected in their drug sensitivity. Furthermore, we selected a small number of drugs already implemented for treating LUAD to ascertain the sensitivity level shown by patients in the LRG and HRG towards these drugs. An intriguing observation is that patients with elevated MQRG scores possessed mitigated IC50 values for BI.2536 [42], Bleomycin [43], Cisplatin [44], Cyclopamine [45], Cytarabine [46], Docetaxel [47], Doxorubicin [48], Etoposide, Gemcitabine [49], Midostaurin [44], Paclitaxel [47], Vinblastine [49], Vinorelbine [47] and ZM.447,439 [50], whereas patients with low MQRG scores possessed significantly mitigated IC50 values for therapeutics like DMOG [51], Erlotinib [49], Imatinib [52], Metformin [53], MK.2206 [54] and Roscovitine [55]. Combining these outcomes confirmed that MQRGs were connected with drug sensitivity (Figure S9).

Confirmation of mRNA expression end genetic changes linked to four prognostic genes

LUAD adenocarcinoma tissues expressed significantly higher levels of STRAP, SHCBP1, and PKP2 than normal lung tissues, while CRTAC1 levels were significantly lower (Fig. 6A). ULCAN, available at https://ualcan.path.uab.edu/, was implemented to investigate the protein expression level. Figure S10A displays the proteomic expression data of four genes in lung cancer and healthy lung gland tissues. The findings indicated that STRAP, SHCBP1, and PKP2 had significantly elevated average expression levels in LUAD tissue contrasted with normal lung tissue, consistent with the mRNA data. However, LUAD tissue had a higher expression level of CRTAC1 (Figure S10A). Figure 6B reveals a significant link between the presence of STRAP, SHCBP1, PKP2, and CRTAC1 in LUAD tissues and the OS rate (P < 0.001) of LUAD patients. The results indicate that STRAP, SHCBP1, and PKP2 could contribute to progressing and improving the LUAD pathogenesis. An investigation of protein expression level was conducted using a human protein mapping database available at http://www.proteinatlas.org. Figure S10B displays the IHC outcomes of four genes in lung cancer and normal lung gland tissues. In line with the mRNA findings, IHC analysis revealed that STRAP, SHCBP1, and PKP2 exhibited significantly elevated average expression levels in LUAD tissue contrasted with normal lung tissue. By contrast, the normal lung tissue possessed elevated CRTAC1 expression levels contrasted with LUAD tissue (Figure S10B).

Fig. 6
figure 6

The expression of the 4 MQRG prognostic signature mRNA was significantly correlated with poor prognosis in lung adenocarcinoma (LUAD). (A) The mRNA differential expression levels in lung adenocarcinoma tissues. (B) The log-rank test was implemented to assess Kaplan–Meier curves that represented the overall survival of four genes.The levels of (C) STRAP; (D) SHCBP1; (E) PKP2; (F) CRTAC1 mRNA expression in Beas-2B, PC9, A549, HCC827 and H1299 cell lines. (G, H) IHC staining for STRAP in LUAD tissue microarray. (I) Kaplan–Meier curves for OS according to high and low STRAP levels of LUAD tissues. *p < 0.05; **p < 0.01; ***p < 0.001

Full size image

Single-cell sequencing analysis

Four MQRGs expression in TME was ascertained with the single-cell dataset NSCLC-GSE99254 from the GEO database. Download the corresponding single-cell data in.h5 format and annotation results from TISCH [56]. Use the R software MAESTRO and Seurat to process and analyze the single-cell data. Re-cluster the cells using the t-SNE method. We can clearly see that the four MQRGs were mainly expressed in CD4 + T conventional cells, with mitigated expression in Proliferative T cells, Regulatory T cells, Monocytes, and Macrophages (Figures S11A-B). In Figure S11C, it is observed that among the four genes, STRAP exhibits the highest expression level.

MQRGs validation with LUAD cells

We utilized the qRT-PCR method to explore STRAP, SHCBP1, PKP2, and CRTAC1 expression levels in the LUAD PC9, A549, HCC827, and H1299 cell lines and Beas-2B control cell line. The results revealed that STRAP, SHCBP1, and PKP2 expression were upregulated in the LUAD PC9, A549, HCC827, and H1299 cell lines, while CRTAC1 was downregulated in these cell lines, which consisted of the TCGA cohort (Fig. 6C-F). Subsequently, we used a TMA containing 80 LUAD tissues and adjacent normal lung tissues to further evaluate the expression of STRAP in LUAD. IHC staining showed that STRAP was mainly distributed in the cytoplasm and cell membrane of LUAD cells, and its expression was higher than that in normal lung tissues (Fig. 6G, H). These results are consistent with those of the public database analysis. Survival analysis revealed that patients with LUAD with high STRAP expression had significantly worse OS than those with low STRAP expression (P = 0.008, Fig. 6I). Previous studies have experimentally validated three other molecules (SHCBP1, PKP2, CRTAC1). The results showed that, compared to normal tissues, SHCBP1 [57,58,59] and PKP2 [60] are highly expressed in lung adenocarcinoma, while CRTAC1 [26] is lowly expressed in lung adenocarcinoma. Moreover, the expression levels of these molecules are significantly correlated with the prognosis of lung adenocarcinoma.

Discussion

LUAD is the prevailing NSCLC subtype, which is the main reason for cancer-linked deaths globally [2]. The high mortality linked to LUAD is attributed to its aggressive nature and late-stage diagnosis, often leading to poor prognosis and limited treatment options [61]. This condition has a profound effect on the patient’s life quality and places a considerable strain on healthcare systems globally [3]. Consequently, the establishment of predictive models for LUAD is crucial to improve patient outcomes and survival rates [5]. Liu et al. established a pattern of Gln metabolism, consisting of five independent immunotherapy cohorts, to predict LUAD prognosis and verify the model accuracy in anticipating immunotherapy effectiveness, specifically examining the involvement of EPHB2 in the development of LUAD [62]. This newly published study utilizes machine learning to construct programmed cell death-based signals for predicting the prognosis of LUAD. An in-depth examination of bulk RNA, single-cell RNA transcriptomics, and relevant clinicopathological information obtained from TCGA-LUAD and six GEO datasets was conducted to develop risk score signatures for 10 genes. These signatures will be used to guide the selection of targeted therapies [63]. Furthermore, there have been publications in recent years that investigate the apoptosis-linked genes and their correlation with the LUAD prognosis [64,65,66,67]. In contrast to previous investigations, our analysis incorporates data from the TCGA and GEO datasets after eliminating batch effects. Implementing this method guarantees a more thorough consolidation of data and strengthens the reliability of our results compared to research that relies on separate databases. Through internal validation, quantitative real-time PCR (qRT-PCR) and Immunohistochemistry our model has exhibited high predictive accuracy. Furthermore, we analyzed the immune cell infiltration at various risk levels and explored the existence of MQRGs in the tumor immune microenvironment and their implication existence the effectiveness of treatment.

The critical importance of mitochondria in cancer cell proliferation and viability is undeniable. Moreover, the preservation of MQC is crucial for the cancer cells’ survival, and the fundamental role of mitochondrial dynamics and the autophagy pathway in this mechanism cannot be overstated for the development of targeted cancer therapies focusing on specific facets of mitochondrial dynamics and autophagy [68]. Normal cells require mitochondrial metabolism, which is controlled by MQC [69]. MQC function may enhance chemotherapy sensitivity or suppress tumor growth, according to several studies [70]. Nevertheless, the MQC disruption may have contributed to the development of LUAD. As our comprehension of aberrant mitochondrial metabolism expands, it unveils opportunities for developing new and enhanced therapeutic strategies for LUAD. Presently, there is a notable scarcity of prognostic models that are based on genes related to mitochondrial quality. Consequently, our study is of considerable significance.

Our research has illustrated the crucial involvement of MQRGs in controlling mitochondrial activity and their ability to distinguish LUAD patients. This study focuses on integrating multi-omics data and bioinformatics to explore the molecular mechanisms associated with LUAD prognosis and construct a predictive model. Through a comparison of the expression of MQRGs across tumor and normal tissues in The Cancer Genome Atlas Lung Adenocarcinoma (TCGA-LUAD) and GSE13213 datasets, we have discovered 660 DEGs. The LUAD prognosis was discovered to be related to nine MQRGs. In order to construct a regression model with the Least Absolute Shrinkage and Selection Operator (LASSO) technique, the TCGA-LUAD and GSE13213 datasets were merged and then partitioned into training and test cohorts. The penalty coefficient λ was ascertained with the cross-validation method. Next, a Cox proportional hazards model was developed to estimate the coefficients and risk scores of the listed signature genes, which were then verified using the data from the test cohort. Subsequently, a prognostic prediction model for LUAD was developed using STRAP, SHCBP1, PKP2, and CRTAC1. Prior investigations have shown a conclusive connection between these genes and cancer advancement. Specifically, Sanguinarine has been shown to dephosphorylate STRAP and MELK, thereby disrupting their interaction and inducing intrinsic apoptosis. The overexpression of STRAP and MELK could operate as possible biomarkers for colorectal cancer (CRC), and their dissociation might be indicative of therapeutic effectiveness [23]. CRTAC1 upregulation markedly reduced LUAD cell proliferation, invasion, and migration in cell studies [26]. SHCBP1 knockdown significantly suppressed tumor growth in both in vivo and in vitro assays [24]. Similarly, the PKP2 knockdown hindered the lung cancer cells proliferation and invasion in vitro and impeded the formation of xenograft lung tumors in vivo [25]. The significant variation in genetic expression levels of most MQRGs between cancerous and healthy tissues, coupled with numerous instances of copy number alterations encompassing amplifications and deletions, highlight the critical role of these genes. Moreover, the observed positive and adverse connections between MQRG expression and the number of immune cells, as well as the significant positive link present between most MQRGs, suggest that aberrant MQRG expression might be involved in the LUAD pathogenesis. An investigation of immune cell infiltration at various risk levels exhibited a greater percentage of activated memory T-cell CD4 between immune cells, indicating that the memory T-cell CD4 activation could be essential in determining the LUAD patient’s prognosis. The nomograms were generated via the synthesis of clinical data and the MQRGscore model. We demonstrate that the Concordance Index (C-index) confirms the risk score as an independent predictive factor for LUAD patients. Furthermore, the risk score-based nomograms constructed using MQRGs demonstrated reliability in anticipating the clinical prognosis of LUAD patients. Moreover, we examined the connection between these four MQRGs and the immunological microenvironment of the tumor.

The LUAD patients were stratified into cohorts using the high or low MQRGs-linked gene expression. Notably, the distinct subgroups exhibited significant differences in both the expression profiles and prognostic outcomes of MQRGs. Moreover, the risk score has been identified as a strong independent predictive predictor for LUAD. Our findings suggest that the four identified signature genes can serve as reliable prognostic indicators, thereby assisting clinicians in formulating tailored treatment strategies for patients.

Despite the comprehensive approach and significant findings, several constraints of this research should be recognized. First and foremost, the study lacks sufficient wet-lab experiments. For instance, drug sensitivity analysis would be meaningful, as it could provide practical value for the treatment response in lung adenocarcinoma and offer additional validation and mechanistic insights for molecular discoveries. Besides, although the sample size is adequate for initial analysis, it may still be considered relatively small. Additionally, biases inherent in the use of a limited dataset may constrain the generalizability of the results. However, we intend to substantiate our findings on lung cancer characteristics through additional experiments and research. While further validation and clinical trials are necessary to confirm its utility in real-world settings, we believe that with proper implementation, this model could become an important part of precision medicine in the future.

Conclusion

In conclusion, this study successfully identified distinct MQRG subtypes of LUAD, revealing significant variations in clinical characteristics and TME among these subtypes. A robust MQRG prognostic scoring model was created, with the ability to differentiate between HRG and LRG and forecast patient survival rates. Moreover, significant disparities in immunological state, CSC index, and somatic mutations were noted between HRG and LRG. Drug sensitivity analysis indicated differential responses to common chemotherapeutic agents, providing a basis for personalized treatment strategies. The constructed nomogram demonstrated a high predictive accuracy for 1-, 3-, and 5-year survival rates of LUAD patients. These outcomes present new perspectives into the LUAD molecular heterogeneity and open up possibilities for more personalized and effective therapeutic approaches in the next years.

Data availability

Data is provided within the manuscript or supplementary information files.

Abbreviations

LUAD:

Lung adenocarcinoma

MQRG:

Mitochondrial quality-related genes

HRG:

High-risk groups

LRG:

Low-risk groups

TCGA:

The Cancer Genome Atlas

GEO:

Gene Expression Omnibus

PCA:

Principal component analysis

DEGs:

Differentially Expressed Genes

GSVA:

Gene Set Variation Analysis

GO:

Gene Ontology analysis

KEGG:

Kyoto Encyclopedia of Genes and Genomes

ssGSEA:

Single sample gene set enrichment analysis

LASSO:

Least absolute shrinkage and selection operator

ROC:

Receiver operating characteristic

AUC:

Area under the curve

TMB:

Tumor mutation burden

qRT-PCR:

Quantitative real-time PCR

IHC:

Immunohistochemistry

TME:

Tumor microenvironment

References

  1. Naranjo S, Cabana CM, LaFave LM, Romero R, Shanahan SL, Bhutkar A, Westcott P, Schenkel JM, Ghosh A, Liao LZ, et al. Modeling diverse genetic subtypes of lung adenocarcinoma with a next-generation alveolar type 2 organoid platform. GENE DEV. 2022;36(15–16):936–49.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Kishikawa S, Hayashi T, Saito T, Takamochi K, Sasa K, Suehara Y, Takahashi F, Sasahara N, Kohsaka S, Suzuki K, et al. Distinct properties of pure- and mixed-type high-grade fetal lung adenocarcinomas by genetic profiling and transcription factor expression. VIRCHOWS ARCH. 2022;480(3):609–19.

    Article  PubMed  Google Scholar 

  3. Li D, Shi J, Dong X, Liang D, Jin J, He Y. Epidemiological characteristics and risk factors of lung adenocarcinoma: A retrospective observational study from North China. FRONT ONCOL. 2022;12:892571.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Lv D, Luo X, Chen Z, Liu J, Liu M, Li Y, Huang S, Tang M, Yang L, Lu L, et al. Apelin/APJ signaling activates autophagy to promote human lung adenocarcinoma cell migration. LIFE SCI. 2021;281:119763.

    Article  PubMed  Google Scholar 

  5. Zhou B, Zhu W, Yuan S, Wang Y, Zhang Q, Zheng H, Zhu L, Xu J. HighGNG4 expression is associated with poor prognosis in patients with lung adenocarcinoma. THORAC CANCER. 2022;13(3):369–79.

    Article  PubMed  Google Scholar 

  6. Xie Y, Chen H, Tian M, Wang Z, Wang L, Zhang J, Wang X, Lian C. Integrating multi-omics and machine learning survival frameworks to build a prognostic model based on immune function and cell death patterns in a lung adenocarcinoma cohort. FRONT IMMUNOL 2024, 15.

  7. Caldeira D, Weiss DJ, Rocco P, Silva PL, Cruz FF. Mitochondria in focus: from function to therapeutic strategies in chronic lung diseases. FRONT IMMUNOL. 2021;12:782074.

    Article  PubMed  PubMed Central  Google Scholar 

  8. Song J, Herrmann JM, Becker T. Quality control of the mitochondrial proteome. Nat Rev Mol Cell Biol. 2021;22(1):54–70.

    Article  PubMed  Google Scholar 

  9. Bock FJ, Tait S. Mitochondria as multifaceted regulators of cell death. NAT REV MOL CELL BIO. 2020;21(2):85–100.

    Article  Google Scholar 

  10. Luo Y, Ma J, Lu W. The significance of mitochondrial dysfunction in cancer. INT J MOL SCI. 2020;21(16):5598.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Gorbunova AS, Zamaraev AV, Yapryntseva MA, Kovaleva OV, Tchevkina EM, Turkina MV, Zhivotovsky B, Kopeina GS. Prognostic signature based on mitochondria quality control proteins for the prediction of lung adenocarcinoma patients survival. CELL DEATH DISCOV. 2023;9(1):352.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Boyer-Guittaut M, Poillet L, Liang Q, Bole-Richard E, Ouyang X, Benavides GA, Chakrama FZ, Fraichard A, Darley-Usmar VM, Despouy G et al. The role of GABARAPL1/GEC1 in autophagic flux and mitochondrial quality control in MDA-MB-436 breast cancer cells. AUTOPHAGY. 2014;10(6):986–1003.

  13. Gaowa S, Futamura M, Tsuneki M, Kamino H, Tajima JY, Mori R, Arakawa H, Yoshida K. Possible role of p53/Mieap-regulated mitochondrial quality control as a tumor suppressor in human breast cancer. CANCER SCI. 2018;109(12):3910–20.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Kang R, Xie Y, Zeh HJ, Klionsky DJ, Tang D. Mitochondrial quality control mediated by PINK1 and PRKN: links to iron metabolism and tumor immunity. AUTOPHAGY. 2019;15(1):172–3.

    Article  PubMed  Google Scholar 

  15. Chen YJ, Hong WF, Liu ML, Guo X, Yu YY, Cui YH, Liu TS, Liang L. An integrated bioinformatic investigation of mitochondrial solute carrier family 25 (SLC25) in colon cancer followed by preliminary validation of member 5 (SLC25A5) in tumorigenesis. CELL DEATH DIS. 2022;13(3):237.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Lindberg J, Mills IG, Klevebring D, Liu W, Neiman M, Xu J, Wikstrom P, Wiklund P, Wiklund F, Egevad L, et al. The mitochondrial and autosomal mutation landscapes of prostate cancer. EUR UROL. 2013;63(4):702–8.

    Article  PubMed  Google Scholar 

  17. Ma X, Niu M, Ni H, Ding W. Mitochondrial dynamics, quality control and mtDNA in Alcohol-associated liver disease and liver cancer. HEPATOLOGY. 2024.

  18. Rath S, Sharma R, Gupta R, Ast T, Chan C, Durham TJ, Goodman RP, Grabarek Z, Haas ME, Hung WHW, et al. MitoCarta3.0: an updated mitochondrial proteome now with sub-organelle localization and pathway annotations. NUCLEIC ACIDS RES. 2021;49(D1):D1541–7.

    Article  PubMed  Google Scholar 

  19. Chen T, Wang Y, Chen X, Zheng W, Guo W, Liang Q, Wang J, Chen Z, Zhou Y, Xiao L. The MQRG score: a novel prognostic tool for adrenocortical carcinoma patients based on mitochondrial quality. Front Endocrinol (Lausanne). 2024;15:1222281.

    Article  PubMed  Google Scholar 

  20. Kanehisa M. KEGG: Kyoto encyclopedia of genes and genomes. GS. 2000;28(1):27–30.

    Google Scholar 

  21. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, et al. Gene ontology: tool for the unification of biology. The gene ontology consortium. NAT GENET. 2000;25(1):25–9.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Hanzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics. 2013;14:7.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Gong X, Chen Z, Han Q, Chen C, Jing L, Liu Y, Zhao L, Yao X, Sun X. Sanguinarine triggers intrinsic apoptosis to suppress colorectal cancer growth through disassociation between STRAP and MELK. BMC Cancer. 2018;18(1):578.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Zhou M, Duan L, Chen J, Li Y, Yin Z, Song S, Cao Y, Luo P, Hu F, Yang G, et al. The dynamic role of nucleoprotein SHCBP1 in the cancer cell cycle and its potential as a synergistic target for DNA-damaging agents in cancer therapy. CELL COMMUN SIGNAL. 2024;22(1):131.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Wu Y, Liu L, Shen X, Liu W, Ma R. Plakophilin-2 promotes lung adenocarcinoma development via enhancing focal adhesion and Epithelial-Mesenchymal transition. CANCER MANAG RES. 2021;13:559–70.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Tan L, Zhang H, Ding Y, Huang Y, Sun D. CRTAC1 identified as a promising diagnosis and prognostic biomarker in lung adenocarcinoma. SCI REP-UK 2024, 14(1).

  27. Mayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. GENOME RES. 2018;28(11):1747–56.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Yan C, Niu Y, Ma L, Tian L, Ma J. System analysis based on the cuproptosis-related genes identifies LIPT1 as a novel therapy target for liver hepatocellular carcinoma. J TRANSL MED 2022, 20(1).

  29. Yu X, Jiang H, Li J, Ding J, Chen K, Ding Z, Xu X. NDV inhibited IFN-beta secretion through impeding CHCHD10-mediated mitochondrial fusion to promote viral proliferation. VET MICROBIOL. 2024;290:109973.

    Article  PubMed  Google Scholar 

  30. Yu J, Wang Q, Chen N, Sun Y, Wang X, Wu L, Chen S, Yuan H, Xu A, Wang J. Mitochondrial transcription factor A regulated ionizing radiation-induced mitochondrial biogenesis in human lung adenocarcinoma A549 cells. J RADIAT RES. 2013;54(6):998–1004.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Romero-Garcia S, Prado-Garcia H, Valencia-Camargo AD, Alvarez-Pulido A. Lactic acidosis promotes mitochondrial biogenesis in lung adenocarcinoma cells, supporting proliferation under normoxia or survival under hypoxia. FRONT ONCOL 2019, 9.

  32. Roy S, Rawat AK, Sammi SR, Devi U, Singh M, Gautam S, Yadav RK, Rawat JK, Singh L, Ansari MN, et al. Alpha-linolenic acid stabilizes HIF-1 alpha and downregulates FASN to promote mitochondrial apoptosis for mammary gland chemoprevention. Oncotarget. 2017;8(41):70049–71.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Fang Z, Huang H, Wang L, Lin Z. Identification of the alpha linolenic acid metabolism-related signature associated with prognosis and the immune microenvironment in nasopharyngeal carcinoma. FRONT ENDOCRINOL. 2022;13:968984.

    Article  Google Scholar 

  34. Chavarro JE, Stampfer MJ, Li H, Campos H, Kurth T, Ma J. A prospective study of polyunsaturated fatty acid levels in blood and prostate cancer risk. Cancer Epidemiol Biomarkers Prev. 2007;16(7):1364–70.

    Article  PubMed  Google Scholar 

  35. Sun BZH. A study on immune cell infiltration in lung adenocarcinoma. COMB CHEM HIGH T SCR. 2022;25(12):2082–8.

    Google Scholar 

  36. Zhang C, Wang H, Wang X, Zhao C, Wang H. CD44, a marker of cancer stem cells, is positively correlated with PD-L1 expression and immune cells infiltration in lung adenocarcinoma. CANCER CELL INT. 2020;20(1):583.

    Article  PubMed  PubMed Central  Google Scholar 

  37. Xiao R, Ma Y, Li H, Li X, Sun Z, Qi Q, Yin P, Yang F, Qiu M. Lung adenocarcinoma manifesting as subsolid nodule potentially represents tumour in the equilibrium phase of immunoediting. Immunology. 2023;168(2):290–301.

    Article  PubMed  Google Scholar 

  38. Zhang L, Jiang B, Lan Z, Yang C, Yao Y, Lin J, Wei Q. Immune infiltration landscape on prognosis and therapeutic response and relevant epigenetic and transcriptomic mechanisms in lung adenocarcinoma. FRONT IMMUNOL 2022, 13.

  39. Deng B, Xiang J, Liang Z, Luo L. Identification and validation of a ferroptosis-related gene to predict survival outcomes and the immune microenvironment in lung adenocarcinoma. CANCER CELL INT. 2022;22(1):292.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Chen C, Guo Q, Tang Y, Qu W, Zuo J, Ke X, Song Y. Screening and evaluation of the role of immune genes of brain metastasis in lung adenocarcinoma progression based on the TCGA and GEO databases. J THORAC DIS. 2021;13(8):5016–34.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Liu Y, Wu Q, Li J, Jia W, Zhai X, Yu J, Zhu H. Comprehensive analysis and validation of competing endogenous RNA network and Tumor-infiltrating immune cells in lung adenocarcinoma. COMB CHEM HIGH T SCR. 2022;25(13):2240–54.

    Google Scholar 

  42. Li C, Tian C, Zeng Y, Liang J, Yang Q, Gu F, Hu Y, Liu L. Integrated analysis of MATH-Based subtypes reveals a novel screening strategy for Early-Stage lung adenocarcinoma. FRONT CELL DEV BIOL. 2022;10:769711.

    Article  PubMed  PubMed Central  Google Scholar 

  43. Zhai K, Jiang N, Wen JF, Zhang X, Liu T, Long KJ, Ke XX, Xu G, Chen C. Overexpression of TWF1 promotes lung adenocarcinoma progression and is associated with poor prognosis in cancer patients through the MMP1 signaling pathway. J THORAC DIS. 2023;15(5):2644–58.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Liu M, Xiao Q, Yu X, Zhao Y, Qu C. Characterization of lung adenocarcinoma based on immunophenotyping and constructing an immune scoring model to predict prognosis. FRONT PHARMACOL. 2022;13:1081244.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Zhang Y, Li T, Liu H, Wang L. Function and prognostic value of basement membrane -related genes in lung adenocarcinoma. FRONT PHARMACOL 2023, 14.

  46. Hong S, Zhang Y, Wang D, Wang H, Zhang H, Jiang J, Chen L. Disulfidptosis-related LncRNAs signature predicting prognosis and immunotherapy effect in lung adenocarcinoma. Aging 2024, 16.

  47. Zhang W, Wei C, Huang F, Huang W, Xu X, Zhu X. A tumor mutational burden-derived immune computational framework selects sensitive immunotherapy/chemotherapy for lung adenocarcinoma populations with different prognoses. FRONT ONCOL. 2023;13:1104137.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Jiang X, Qin N, Hua T, Wei X, Li Y, Chen C, Gong L, Liu S, Wang C, Yin R, et al. Functional characterization and clinical significance of super-enhancers in lung adenocarcinoma. MOL CARCINOGEN. 2022;61(8):776–86.

    Article  Google Scholar 

  49. Shang S, Wang M, Xing Z, He N, Li S. Lactate regulators contribute to tumor microenvironment and predict prognosis in lung adenocarcinoma. FRONT IMMUNOL. 2022;13:1024925.

    Article  PubMed  PubMed Central  Google Scholar 

  50. Wang G, Zheng H, Zhao X, Wang Y, Zeng Y, Du J. The prognostic model and drug sensitivity of LKB1-Mutant lung adenocarcinoma based on immune landscape. FRONT MOL BIOSCI 2022, 9.

  51. Tao Q, Li X, Zhu T, Ge X, Gong S, Guo J, Ma R. Lactate transporter SLC16A3 (MCT4) as an Onco-Immunological biomarker associating tumor microenvironment and immune responses in lung cancer. INT J GEN MED. 2022;15:4465–74.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Xie L, Dai R, Wang X, Xie G, Gao Z, Xu X. Comprehensive analysis revealed the potential implications of m6A regulators in lung adenocarcinoma. FRONT MOL BIOSCI. 2022;9:806780.

    Article  PubMed  PubMed Central  Google Scholar 

  53. Sui QHZLJ. Targeting TAM secreted S100A9 effect source cancer Lett SO 2023. CANCER LETT. 2024;581:216497.

    Article  PubMed  Google Scholar 

  54. Zhang N, Cao S, Sun R, Wang Y, Liu L, Wang W, Meng X. Signal peptidase 21 suppresses cell proliferation, migration, and invasion via the PTEN-PI3K/Akt signaling pathway in lung adenocarcinoma. PeerJ (San Francisco CA). 2022;10:e14206.

    Google Scholar 

  55. Lin W, Wang X, Xu Z, Wang Z, Liu T, Cao Z, Feng X, Gao Y, He J. Identification and validation of cellular senescence patterns to predict clinical outcomes and immunotherapeutic responses in lung adenocarcinoma. CANCER CELL INT. 2021;21(1):652.

    Article  PubMed  PubMed Central  Google Scholar 

  56. Han Y, Wang Y, Dong X, Sun D, Liu Z, Yue J, Wang H, Li T, Wang C. TISCH2: expanded datasets and new tools for single-cell transcriptome analyses of the tumor microenvironment. NUCLEIC ACIDS RES. 2023;51(D1):D1425–31.

    Article  PubMed  Google Scholar 

  57. Wang N, Zhu L, Wang L, Shen Z, Huang X. Identification of SHCBP1 as a potential biomarker involving diagnosis, prognosis, and tumor immune microenvironment across multiple cancers. COMPUT STRUCT BIOTEC. 2022;20:3106–19.

    Article  Google Scholar 

  58. Dai Y, Hu C, Zhou H, Liu W, Lai W, Xu R, Liao J, Wang J, Li G, Zhang R. Rucaparib inhibits lung adenocarcinoma cell proliferation and migration via the SHCBP1/CDK1 pathway. FEBS J. 2023;290(24):5720–43.

    Article  PubMed  Google Scholar 

  59. Yanli Luo SZHX. Prognosis and immunotherapy signiffcances of a cancer-associated ffbroblasts-related gene signature in lung adenocarcinoma. CELL MOL BIOL 2023.

  60. Shu J, Jiang J, Zhao G. Identification of novel gene signature for lung adenocarcinoma by machine learning to predict immunotherapy and prognosis. FRONT IMMUNOL 2023, 14.

  61. Tan J, Sidhu G, Greco MA, Ballard H, Wieczorek R. Villin, cytokeratin 7, and cytokeratin 20 expression in pulmonary adenocarcinoma with ultrastructural evidence of microvilli with rootlets. HUM PATHOL. 1998;29(4):390–6.

    Article  PubMed  Google Scholar 

  62. Liu J, Shen H, Gu W, Zheng H, Wang Y, Ma G, Du J. Prediction of prognosis, immunogenicity and efficacy of immunotherapy based on glutamine metabolism in lung adenocarcinoma. FRONT IMMUNOL. 2022;13:960738.

    Article  PubMed  PubMed Central  Google Scholar 

  63. Wang S, Wang R, Hu D, Zhang C, Cao P, Huang J. Machine learning reveals diverse cell death patterns in lung adenocarcinoma prognosis and therapy. NPJ PRECIS ONCOL. 2024;8(1):49.

    Article  PubMed  PubMed Central  Google Scholar 

  64. Gao J, Qiu X, Xi G, Liu H, Zhang F, Lv T, Song Y. Downregulation of GSDMD attenuates tumor proliferation via the intrinsic mitochondrial apoptotic pathway and Inhibition of EGFR/Akt signaling and predicts a good prognosis in non–small cell lung cancer. ONCOL REP. 2018;40(4):1971–84.

    PubMed  PubMed Central  Google Scholar 

  65. Liu Y, Ouyang Y, Feng Z, Jiang Z, Ma J, Zhou X, Cai C, Han Y, Zeng S, Liu S, et al. RASGRP2 is a potential immune-related biomarker and regulates mitochondrial-dependent apoptosis in lung adenocarcinoma. FRONT IMMUNOL. 2023;14:1100231.

    Article  PubMed  PubMed Central  Google Scholar 

  66. Ma C, Li F, Gu Z, Yang Y, Qi Y. A novel defined risk signature of cuproptosis-related long non-coding RNA for predicting prognosis, immune infiltration, and immunotherapy response in lung adenocarcinoma. FRONT PHARMACOL. 2023;14:1146840.

    Article  PubMed  PubMed Central  Google Scholar 

  67. Wang J, Li P, Sun L, Zhang J, Yue K, Wang Y, Wu X. FAM83B regulates mitochondrial metabolism and anti-apoptotic activity in pulmonary adenocarcinoma. Apoptosis. 2024;29(5–6):743–56.

    Article  PubMed  Google Scholar 

  68. Hu Q, Xu J, Wang L, Yuan Y, Luo R, Gan M, Wang K, Zhao T, Wang Y, Han T, et al. SUCLG2 regulates mitochondrial dysfunction through succinylation in lung adenocarcinoma. ADV SCI. 2023;10(35):e2303535.

    Article  Google Scholar 

  69. Deepak KRPDC. Mitophagy at the crossroads of cancer development: exploring the role of mitophagy in tumor progression and therapy resistance. BBA-MOL CELL RES. 2024;1871(6):119752.

    Google Scholar 

  70. Tang Y, Wang L, Yi T, Xu J, Wang J, Qin JJ, Chen Q, Yip KM, Pan Y, Hong P, et al. Synergistic effects of autophagy/mitophagy inhibitors and Magnolol promote apoptosis and antitumor efficacy. ACTA PHARM SIN B. 2021;11(12):3966–82.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We extend our gratitude to GEO and TCGA database, and all contributors who generously shared their data on these platforms.

Funding

This research was funded by Scientific Research Fund Project of Hunan Provincial Health Commission (B202304029220), Special fund for clinical research of Wu Jieping Medical Foundation (No.320.6750.2023-05-38) and Clinical Medical Research “4310” Program of the University of South China(20224310NHYCG07).

Author information

Author notes

  1. Lijun Zeng, Sixuan Wu and Zhimin Li contributed equally to this work.

Authors and Affiliations

  1. Department of Oncology, the First Affiliated Hospital, Hengyang Medical School, University of South China, Hengyang, 421001, China

    Lijun Zeng, Sixuan Wu, Zhimin Li, Yuanbin Tang, Yeru Tan & Yuehua Li

  2. Department of Thoracic Surgery, the First Affiliated Hospital, Hengyang Medical School, University of South China, Hengyang, 421001, China

    Renji Liang

Authors
  1. Lijun Zeng
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Sixuan Wu
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Zhimin Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Yuanbin Tang
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Yeru Tan
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Renji Liang
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Yuehua Li
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

LYH and LRJ designed the whole study. ZLJ finished drafted and revised the manuscript. ZLJ, WSX and LZM completed the bioinformatics. ZLJ and WSX performed qRT-PCR experiments. TYB and TYR researched on background of the study. All authors approved this manuscript.

Corresponding authors

Correspondence to Renji Liang or Yuehua Li.

Ethics declarations

Ethics approval and consent to participate

The clinical information and data implemented in this investigation were sourced from public databases; as a result, written informed consent from patients and ethics committee approval was not necessary for this investigation.

Consent for publication

All authors have agreed to publish this manuscript.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

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

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zeng, L., Wu, S., Li, Z. et al. A novel mitochondrial quality regulation gene signature for anticipating prognosis, TME, and therapeutic response in LUAD by multi-omics analysis and experimental verification. Cancer Cell Int 25, 138 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12935-025-03764-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12935-025-03764-4

Keywords

Cancer Cell International

ISSN: 1475-2867

Contact us