Clinically relevant stratification of lung squamous carcinoma patients based on ubiquitinated proteasome genes for 3P medical approach

Abstract

Relevance

The proteasome is a crucial mechanism that regulates protein fate and eliminates misfolded proteins, playing a significant role in cellular processes. In the context of lung cancer, the proteasome’s regulatory function is closely associated with the disease’s pathophysiology, revealing multiple connections within the cell. Therefore, studying proteasome inhibitors as a means to identify potential pathways in carcinogenesis and metastatic progression is crucial in in-depth insight into its molecular mechanism and discovery of new therapeutic target to improve its therapy, and establishing effective biomarkers for patient stratification, predictive diagnosis, prognostic assessment, and personalized treatment for lung squamous carcinoma in the framework of predictive, preventive, and personalized medicine (PPPM; 3P medicine).

Methods

This study identified differentially expressed proteasome genes (DEPGs) in lung squamous carcinoma (LUSC) and developed a gene signature validated through Kaplan–Meier analysis and ROC curves. The study used WGCNA analysis to identify proteasome co-expression gene modules and their interactions with the immune system. NMF analysis delineated distinct LUSC subtypes based on proteasome gene expression patterns, while ssGSEA analysis quantified immune gene-set abundance and classified immune subtypes within LUSC samples. Furthermore, the study examined correlations between clinicopathological attributes, immune checkpoints, immune scores, immune cell composition, and mutation status across different risk score groups, NMF clusters, and immunity clusters.

Results

This study utilized DEPGs to develop an eleven-proteasome gene-signature prognostic model for LUSC, which divided samples into high-risk and low-risk groups with significant overall survival differences. NMF analysis identified six distinct LUSC clusters associated with overall survival. Additionally, ssGSEA analysis classified LUSC samples into four immune subtypes based on the abundance of immune cell infiltration with clinical relevance. A total of 145 DEGs were identified between high-risk and low-risk score groups, which had significant biological effects. Moreover, PSMD11 was found to promote LUSC progression by depending on the ubiquitin–proteasome system for degradation.

Conclusions

Ubiquitinated proteasome genes were effective in developing a prognostic model for LUSC patients. The study emphasized the critical role of proteasomes in LUSC processes, such as drug sensitivity, immune microenvironment, and mutation status. These data will contribute to the clinically relevant stratification of LUSC patients for personalized 3P medical approach. Further, we also recommend the application of the ubiquitinated proteasome system in multi-level diagnostics including multi-omics, liquid biopsy, prediction and targeted prevention of chronic inflammation and metastatic disease, and mitochondrial health-related biomarkers, for LUSC 3PM practice.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13167-024-00352-w.

Introduction

Current status of lung squamous carcinoma (LUSC) treatment

Lung cancer impacts more lives than any other forms of cancers worldwide, with a staggering 1.8 million individuals who are diagnosed annually as lung cancer. LUSC, a subtype of non-small cell lung cancer (NSCLC), is a prevalent form of lung cancer. It is responsible for approximately 25 to 30% of cases within the NSCLC category. Unfortunately, most patients are diagnosed with the advanced form [1]. Patients with LUSC have a poor prognosis compared to lung adenocarcinoma and lack available target drugs [2]. Currently, the main therapeutic approaches for LUSC encompass surgical intervention, chemotherapy, and radiation therapy. The primary method of treating early-stage LUSC is through surgery. In cases where the cancer is localized, surgical removal alone can lead to more favorable treatment outcomes. The use of chemotherapy is prevalent among patients diagnosed with intermediate and advanced LUSC, with the aim of prolonging tumor progression and enhancing survival rates. Radiation therapy is often used as adjunctive therapy after surgery or chemotherapy. Immunotherapy is a new type of LUSC treatment that attacks and kills cancer cells by activating or boosting a patient’s immune system. Immunotherapy is commonly administered using immune checkpoint inhibitors, tumor vaccines, CAR-T cell therapy, and other similar methods. Currently, several types of immunotherapy medications have received approval to manage LUSC, including but not limited to PD-1/PD-L1 inhibitors and CTLA-4 inhibitors [3]. The use of these medications can greatly enhance the longevity and well-being of patients while causing minimal adverse reactions. Overall, the management of LUSC remains challenging, and personalized treatment strategies should be developed based on the individual circumstances and health status of each patient.

The role of proteasome in LUSC

Proteasomes are an intracellular protein degradation system that mainly maintains the balance of the intracellular environment by degrading and digesting intracellular protein [4]. The crucial role of proteasomes in lung cancer cannot be underestimated. The proteasome is able to regulate lung cancer cell proliferation and plays significant roles in promoting metastasis and drug resistance [5]. According to research findings, proteasomes have the capability to degrade specific proteins, such as p27 and p21, which possess inhibitory properties against lung cancer cell proliferation. Therefore, lung cancer cell proliferation can be facilitated by proteasome. Additionally, the metastasis of lung cancer cells is remarkably impacted by proteasome. The proteasome-mediated degradation of certain proteins such as E-cadherin is observed to promote lung cancer cell metastasis by removing their proliferation-inhibiting effects [6]. They can also affect the drug resistance of lung cancer cells [4]. Indeed, research has indicated that lung cancer cells can develop a higher resistance to certain anti-tumor medications, like those targeting p53, due to the proteasome’s ability to break down proteins [7]. Ubiquitination is a modification that covalently binds ubiquitin molecules to proteins. Ubiquitin labeling of proteins facilitates their subsequent recognition and degradation in the proteasome. Ubiquitination is mainly accomplished by three enzyme systems: ubiquitin-activating enzyme (E1), ubiquitin-binding enzyme (E2), and ubiquitin ligase (E3) [8]. E1 enzyme activates ubiquitin into ubiquitin-amp complex, E2 enzyme receives ubiquitin from E1 enzyme and transfers it to E3 enzyme, and E3 enzyme connects ubiquitin to the target protein [9]. The degradation of proteins in cells can be facilitated by ubiquitin, which achieves this through a mechanism of labeling proteins for degradation by proteasomes [10]. The proteasome recognizes and degrades proteins that have been marked with ubiquitin by interacting with ubiquitin recognition proteins, namely Rpn10 and Rpn13, both of which are components of the proteasome [10]. Ubiquitination and proteasomes are closely related. Proteasomes are the primary organelles responsible for protein degradation, which is facilitated by ubiquitin. Quantitative ubiquitinomics has been utilized to identify and analyze proteasome genes in LUSCs relative to controls. Additionally, various investigations have been conducted in this field. The research shows that proteasome genes have already been proven to be tightly associated with immunity, stem cells, and cancer survival [11]. More studies are needed to examine the wider processes that cause proteasome malfunction, such as the epigenetic control and post-translational modification alteration of the proteasome’s component proteins. Due to the crucial role of proteasomes in lung cancer, researchers are currently exploring innovative therapeutic strategies, such as the utilization of proteasome inhibitors, to improve the efficacy of lung cancer treatment. This investigation aims to identify new avenues for combating lung cancer by targeting proteasomes.

The role of proteasome in immunity

The proteasome is an organelle mainly responsible for the degradation and metabolism of intracellular proteins. The immune microenvironment is greatly influenced by the proteasome, which has crucial roles in its functionality. This essential component functions by breaking down proteins into smaller peptide fragments. These fragments can then bind to MHC molecules and subsequently be identified by the immune system. This process enables effective recognition and response within the immune system [12, 13]. The identification and removal of cancerous cells is a crucial step for the immune system. Proteasomes can also participate in the regulation of immune cells. The proteasome has the ability to break down various immune cell components, including cytokines, receptors, and signaling molecules like TNF and IL. This breakdown process can ultimately impact the activity and function of immune cells [12]. In addition, proteasomes can also participate in the apoptosis and autophagy of immune cells, thus affecting the number and stability of immune cells. The proteasome also plays a role in tumor immunotherapy. Immunotherapy is a new tumor treatment method, including T cell stimulation therapy, anti-CTLA-4, PD-1, and other treatments. In these various approaches, CD8 + T cells are harnessed to effectively eradicate cancer cells. The proteasome contributes to the increased cytotoxicity of T cells that target tumor antigens. This is achieved by facilitating the display of cancer-specific antigens on MHC molecules, thereby amplifying the efficacy of cancer treatment. By improving the display of these antigens, the proteasome contributes to the overall effectiveness of therapeutic interventions against cancer [14]. The efficacy of tumor immunotherapy relies heavily on the number of neoantigens present in the tumor that are degraded by the proteasome. Immunotherapy is more likely to be effective for patients with a high tumor neoantigen load [15]. Within the tumor microenvironment, the proteasome can influence the extent of immune cell infiltration. Notably, patients who display a heightened level of immune cell infiltration tend to demonstrate a more positive response towards immunotherapy. This observation underscores the potential significance of the proteasome in modulating the efficacy of immunotherapeutic interventions [16]. Proteasome inhibitors, as a kind of drug with potential anti-tumor effect, can affect the immune microenvironment by regulating proteasome activity, thus improving the efficacy of immunotherapy. In certain types of tumors, promising therapeutic outcomes have been demonstrated through the combination of proteasome inhibitors with immune checkpoint inhibitors [17]. Thereby, the proteasome is essential in modulating the immune microenvironment and has the potential to impact the efficacy of immunotherapy through various mechanisms [4]. Studying the role of proteasomes in immunotherapy is helpful to better predict patient’s response to immunotherapy and provide more targeted guidance for clinical treatment.

For LUSC, researchers have found some genes related to ubiquitinated proteasomes, such as PSMA1 and PSMD14 [18, 19]. The dysfunction of the ubiquitinated proteasome caused by the abnormal expressions of these genes can disrupt cellular growth, differentiation, and apoptosis, ultimately facilitating the development and progression of tumors.

Working hypothesis and significance in the context of 3P medicine

We hypothesize that the genes corresponding to ubiquitinated proteome proteins play important roles in LUSC, and these genes are significantly different in LUSC compared to normal controls. For this present study, a thorough analysis was carried out on the ubiquitinated proteasome genes in LUSC. The aim was to comprehensively analyze and identify any mutations or abnormal expression patterns within genes associated with the ubiquitinated proteasome. Analysis of these elements can attain a deeper comprehension of the function and ramifications of the ubiquitinated proteasome in LUSC. These results shed light on the underlying mechanisms of LUSC and could serve as potential markers to predict and diagnose LUSC. The ubiquitin–proteasome gene is crucial in the biological mechanisms of LUSC, including functions such as cell proliferation, apoptosis, and autophagy. A thorough examination of the correlation between genes involved in the ubiquitin–proteasome system pathway and LUSC could facilitate the discovery of novel therapeutic targets and drugs, ultimately enhancing the efficacy of treatment for this type of cancer. Analysis of genomic data of individuals diagnosed with LUSC can identify specific genetic variations linked to this particular cancer type in relation to the ubiquitinated proteasome gene. Through this investigation, valuable insights can be obtained regarding the fundamental molecular mechanisms and potential genetic elements that contribute to the initiation and advancement of LUSC. The exploration of the ubiquitination proteasome gene in LUSC holds significant importance due to its potential to unveil the underlying mechanisms of the disease’s initiation, advancement, and treatment. This valuable information can be leveraged to develop personalized treatment strategies for LUSC patients. By understanding the intricate molecular aspects associated with this gene, healthcare professionals can optimize treatment plans tailored to individual patients, ultimately enhancing their overall prognosis and therapeutic outcomes. This study provides significant benefits in terms of the diagnosis, treatment, and prognosis evaluation for LUSC. Moreover, it introduces innovative approaches and strategies that have practical implications in clinical medicine. By leveraging the findings from this study, healthcare professionals can enhance their ability to accurately diagnose the disease, effectively treat patients, and assess their long-term prognosis. Additionally, the novel approaches and strategies derived from this study offer promising avenues for further advancements and improvements in clinical practice.

Study design

Proteasome plays important roles in LUSC, and ubiquitination of proteasome regulates the functions of proteasome [11, 20]. This study used the genes of ubiquitinated proteasome of LUSCs extracted from TCGA database to construct prognostic model, and further test its clinical relevance in LUSC. These data will benefit in-depth insights into molecular mechanism and discovery of therapeutic targets, establish effective ubiquitinated proteasome-based prognostic biomarker for clinically relevant stratification, predictive diagnosis, prognostic assessment, and personalized treatment of LUSC in the framework of 3P medical approach to improve and promote the healthcare of LUSC patients. The overall flow chart to identify proteasome gene-related signatures in LUSC was summarized in Fig. 1.

Fig. 1.

Flow chart for identification of proteasome gene-related signatures in lung squamous cell carcinoma

Materials and methods

Data processing

The genes responsible for forming proteasome complex can be classified as 19S regulatory particle lid (consisting of 14 genes: PSME2, PSMD7, PSMD12, PSMD8, PSMD6, PSMD14, PSMD11, PSME1, PSMD13, PSMD4, SEM1, PSMD3, PSMD9, and PSME3), 19S regulatory particle base (consisting of 10 genes: ADRM1, PSMD1, PSME4, PSMC4, PSMC6, PSMC3, PSMC2, PSMD2, PSMC1, and PSMC5), and 20S core particle (consisting of 14 genes including PSMA3, PSMB2, PSMB7, PSMB4, PSMB6, PSMA6, PSMB1, PSMA5, PSMA2, PSMB3, PSMB5, PSMA7, PSMA1, and PSMA4). In cases of immunoproteasomes, PSMB1, PSMB5, and PSMB2 are replaced with PSMB9, PSMB8, and PSMB10, respectively. However, in thymoproteasomes, PSMB5 is replaced by PSMB11 (Supplementary Table 1). The corresponding ubiquitinated proteins of proteasome identified in lung squamous cell cancers relative to controls were listed (Table 1). The Supplementary Table 2 is generated from The Cancer Genome Atlas (TCGA) website (https://portal.gdc.cancer.gov/), which contains information about mRNA expressions of genes related to the proteasome complex in LUSC.

Table 1.

Ubiquitinated proteins of proteasome identified in lung squamous cell cancers relative to controls. Ratio (T/N) ratio of tumors (T) to controls (N), K* ubiquitinated lysine residue

Accession no	Gene name	Protein name	Modified peptides	Modified positions	Modified probabilities	Modified level (T)	Modified level (N)	Ratio (T/N)	t-test p-value
Q16186	ADRM1	Proteasomal ubiquitin receptor ADRM1	MSLK*GTTVTPDKR	34	1	143,170,000	63,045,000	2.271	1.24E − 02
G3V5Z7	PSMA6	Proteasome subunit alpha type	K*VPDKLLDSSTVTHLFK	55	0.999	14,948,333	2,045,967	7.306	1.75E − 03
G3V4X1	PSMC1	26S protease regulatory subunit 4 (fragment)	TLLAK*AVANQTSATF	74	1	5,294,200
P62191	PSMC1	26S protease regulatory subunit 4	VAEEHAPSIVFIDEIDAIGTK*R	293	1	9,735,900
A0A140VK70	PSMC2	26S protease regulatory subunit 7	QIK*QVEDDIQQLLK	46	1	6,017,400	1,213,663	4.958	4.59E − 03
A0A140VK70	PSMC2	26S protease regulatory subunit 7	DFLEAVNK*VIK	415	0.758	2,390,167	364,000	6.566	2.28E − 01
A0A140VK42	PSMC3	26S protease regulatory subunit 6A	LLDSEIK*IMK	53	1	1,151,833	5,224,300	0.220	5.47E − 03
A0A140VK42	PSMC3	26S protease regulatory subunit 6A	DAFALAK*EK	276	0.993	26,500,667	5,408,667	4.900	3.50E − 02
A0A140VK42	PSMC3	26S protease regulatory subunit 6A	IKENSEK*IK	79	0.999	3,947,167	3,949,800	0.999	9.99E − 01
A0A140VK42	PSMC3	26S protease regulatory subunit 6A	K*MNVSPDVNYEELAR	372	1	1,168,267
A8K2M0	PSMC4	Proteasome (prosome, macropain) 26S subunit, ATPase, 4, isoform CRA_b	ENAPAIIFIDEIDAIATK*R	273	1	8,303,567
J3QLH6	PSMC5	26S protease regulatory subunit 8 (fragment)	VSGSELVQK*F	214	1	19,623,333
A0A087X2I1	PSMC6	26S protease regulatory subunit 10B	SENDLK*ALQSVGQIVGEVLK	62	1	4,433,833	1,867,500	2.374	4.51E − 03
A0A087X2I1	PSMC6	26S protease regulatory subunit 10B	GCLLYGPPGTGK*TLLAR	194	1	4,502,467
A0A087X2I1	PSMC6	26S protease regulatory subunit 10B	LDILK*IHAGPITK	328	0.999	15,559,000
A0A087X2I1	PSMC6	26S protease regulatory subunit 10B	AVASQLDCNFLK*VVSSSIVDK	211	0.999
O00231	PSMD11	26S proteasome non-ATPase regulatory subunit 11	EASIDILHSIVK*R	32	1	2,674,600
A0A0S2Z489	PSMD12	Proteasome (prosome, macropain) 26S subunit, non-ATPase, 12, isoform CRA_a (fragment)	LTK*TLATIK	147	1	49,735,667
Q59EG8	PSMD2	Proteasome 26S non-ATPase subunit 2 variant	EWQELDDAEK*VQR	183	1	30,879,667
O43242	PSMD3	26S proteasome non-ATPase regulatory subunit 3	LVSK*SVFPEQANNNEWAR	273	1	2,132,100
Q5VWC4	PSMD4	26S proteasome non-ATPase regulatory subunit 4	ILSK*LHTVQPK	74	1	28,412,667	5,504,733	5.161	1.79E − 02
Q5VWC4	PSMD4	26S proteasome non-ATPase regulatory subunit 4	LQAQQDAVNIVCHSK*TR	40	1	32,204,333	952,700	33.803	3.13E − 03
P48556	PSMD8	26S proteasome non-ATPase regulatory subunit 8	SPNLSK*CGEELGR	111	1

The correlation analysis and protein–protein interaction (PPI) network of the proteasome genes in proteasome complex

The correlation analysis between proteasome genes within the proteasome complex in the context of LUSC was conducted utilizing the Corrplot R package. The genes under scrutiny encompassed a diverse array, including but not limited to PSME4, PSMD12, PSMD13, PSMB10, PSMD7, PSMB3, PSMA2, PSMB6, PSMD14, PSMD2, PSMA7, PSMD11, PSMD3, PSMA1, SEM1, PSMD9, PSMD4, PSMC5, PSMB7, PSMB5, PSMA5, PSME2, PSMB4, PSME3, PSMD1, PSMB2, PSMB8, PSMD6, PSMD8, PSMA3, PSMC1, PSMB1, PSMC2, PSMC3, PSMA6, PSMB9, ADRM1, PSMA4, PSMA2, PSMB11, PSMB10, and PSMB6.

To investigate the inherent interactive associations between these proteasome genes, a constructed PPI network was used. The intricacies of this network emerged as the genes were meticulously mapped into the expansive canvas provided by the STRING database (https://string-db.org/). Striving for a stringent threshold of accuracy, only interactions with a combined score over 0.9 were embraced, ensuring the inclusion of robust and meaningful associations.

For a more comprehensive exposition of these dynamic interplays, readers are directed to consult the wealth of data encapsulated within Supplementary Table 3, which elegantly captures the essence of these protein–protein interactions and their implications within the realm of the proteasome complex in LUSC.

The correlation of the proteasome genes in proteasome complex and drug sensitivity

By leveraging the capabilities of the CellMiner tool, intricate dissection of transcriptional and drug response patterns within the repository of NCI-60 cell lines becomes feasible. Meticulously overseen and maintained by the National Cancer Institute’s Developmental Therapeutics Program (DTP), this assembly of cell lines serves as an invaluable resource for the systematic evaluation of the therapeutic efficacy of anticancer compounds. Within this repository of cellular diversity, a concerted effort was undertaken to unravel the underlying interplay. The correlation was scrutinized between proteasome complex gene expression profiles and subtle responses to anticancer drugs. This endeavor found its foundation in the integration of the Corrplot R package and the rigor of Spearman’s method, culminating in a holistic analytical framework. This pursuit of insight was underpinned by a stringent criterion, where the statistical significance threshold was judiciously set as p < 0.05. The critical threshold acted as a guardian, guaranteeing the revelation of only strong and significant connections, shedding a concentrated spotlight on the complex correlations between the expressions of proteasome genes and the sensitivity to drugs in the intricate environment of LUSC.

Relationship of proteasome gene expressions with clinical features in LUSC

Chi-square analysis explored proteasome gene associations with clinical traits in LUSC cases. Clinical data, including age, pathologic T, sex, pathologic N, stage, and pathologic M, were cataloged (Supplementary Table 5). The correlations were summarized between the expressions of proteasome genes and the pathologic stages (I, II, III, and IV) (Supplementary Table 6). An overall survival analysis of the LUSC proteasome gene was performed using the Kaplan–Meier plotting website (https://kmplot.com/analysis/), which showed statistical significance (p < 0.05) (Supplementary Fig. 1).

Estimation of immune-related scores in LUSC

The gene expression data allowed ones to estimate the stroma and immune cells infiltrating the cancer tumor tissue. The R package ESTIMATE (https://bioinformatics.mdanderson.org/estimate/rpackage.html) was used to generate four scores: stroma score (reflecting the presence of stroma), immune score (indicating immune cell infiltration), ESTIMATE score, and tumor purity (Exhibit 8). The Spearman method (p < 0.05) within the R package Corrplot (https://cran.r-project.org/web/packages/corrplot/vignettes/corrplot-intro.html) was utilized to identify proteasome gene expression with the immunity score, correlation between ESTIMATE score, stromal score, and tumor purity. The outcomes of this analysis are depicted in Supplementary Fig. 2.

The proteasome gene signature in LUSC identified with multivariate regression analysis

Differentially expressed proteasome genes underwent unpaired Student’s t-tests between tumor and normal samples. The LUSC samples were randomly divided into two groups (training set and test set) (Supplementary Table 8). A prognostic model was constructed via multiple regression analysis, incorporating genes with optimal Akaike Information Criterion (AIC) values. Receiver operating characteristic (ROC) curve analysis was used to evaluate risk score-based classification efficacy. The prognostic model’s performance in predicting overall survival was verified via Kaplan–Meier method and “survival” package (https://www.bioconductor.org/packages/devel/bioc/vignettes/survtype/inst/doc/survtype.html) in R software. The nomogram assessed LUSC patient outlook, including 1-, 2-, and 3-year survival rates. Scrutiny of high- and low-risk score group correlations employed the R package pheatmap (http://bioconductor.org/packages/3.8/bioc/html/heatmaps.html). Additionally, Cox regression analyzed the association between clinical features and overall survival in LUSC patients with univariate and multivariate models.

To mitigate overfitting, the validity of proteasome signature as a prognostic indicator was confirmed through internal validation of the IMvigor210 database’s cohort (Supplementary Table 9). The IMvigor210 cohort’s prognostic model, based on overall survival data, classified samples into low- and high-risk groups via median risk score division. Kaplan–Meier analysis was executed with the “survival” package (https://www.bioconductor.org/packages/devel/bioc/vignettes/survtype/inst/doc/survtype.html) in R software. The Maftools R package calculated TMB scores (Supplementary Table 10). Spearman method within the R package Corrplot (https://www.rdocumentation.org/packages/corrplot/versions/0.92) assessed TMB and risk score correlation (p < 0.05). Mutation status for low- and high-risk score groups was found in Supplementary Table 11. Maftools R package enabled tumor mutation distribution analysis, yielding a mutation gene waterfall plot.

GSEA analysis between high- and low-risk score group

Gene Set Enrichment Analysis (GSEA) is a valuable resource for dissecting genome microarray data, encompassing diverse functional gene sets. This tool enhances the precision and depth of gene expression analysis, yielding more nuanced and meaningful outcomes. In the LUSC cohort, patients were stratified into low- and high-risk groups based on their risk scores. The GSEA enrichment analysis was executed on two distinct TCGA datasets to unearth significant gene sets within the high- and low-risk categories (Supplementary Table 12; Supplementary Fig. 3).

The proportions of different immune cells in LUSC

The LM22 gene signature-based CIBERSORT algorithm employed in this study effectively determined immune cell ratios in LUSC tissue samples. Operating through the CIBERSORT web portal, the algorithm sensitively and specifically differentiated 22 human immune cell phenotypes, leveraging standard annotation files and the LM22 gene signature with 1000 permutations (Supplementary Table 13). Unpaired Student’s t-tests evaluated immune cell distribution disparities between low- and high-risk score groups.

Risk scoring was carried out using the R package Corrplot with a variety of immune cells (including T cells CD4-naïve, T cells CD8, T cells CD4-memory activated, T cells CD4-memory resting, B-cell-naïve, follicular helper T cells, B-cell-memory, plasma cells, NK cells in the resting state, regulatory T cells, gamma-delta T cells, monocytes, activated state NK cells, M0-type, M1-type, M2-type, activated-state dendritic cells, resting-state dendritic cells, activated-state mast cells, resting-state mast cells, neutrophils, and eosinophils), and the correlation analysis between the visualizations was achieved with the circlize package (https://cran.r-project.org/web/packages/circlize/index.html).

Furthermore, immune-related checkpoints, namely VTCN1, CTLA4, PDCD1LG2, PDCD1, CD274, CD276, CD80, and CD86 (Supplementary Table 14), underwent unpaired Student’s t-tests for expression level comparisons between low- and high-risk groups.

GSVA analysis between high- and low-risk score groups

Utilizing non-parametric, unsupervised techniques, GSVA quantifies gene set enrichment variations across an expression dataset’s samples. In LUSC patients, risk-based categorization yielded distinct high- and low-risk groups. Significant Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were obtained using TCGA datasets (p < 0.05; FDR < 0.05). Subsequent GSVA analysis of these identified pathways discerned differences between high- and low-risk groups (Supplementary Table 15).

Differentially expressed genes (DEGs) between low- and high-risk score groups

A detailed analysis of DEGs was performed to reveal the different expression patterns between low- and high-risk groups. This analysis harnessed the empirical Bayesian approach facilitated by the well-regarded R package limma (http://www.bioconductor.org/packages/release/bioc/html/limma.html) (Supplementary Table 16). The stringent criteria were used to assess significance, identifying DEGs by setting a p-value threshold of less than 0.05. These effectively distinguished genes showed significant expression differences between low- and high-risk score groups. This judicious approach served as a cornerstone for unraveling the intricate molecular underpinnings that may be pivotal in the context of LUSC progression.

Functional and pathway enrichment analyses of the identified DEGs between low- and high-risk score groups

The KEGG pathways associated with DEGs were subjected to comprehensive analysis utilizing the DAVID Functional Annotation Bioinformatics Microarray tool (https://david.ncifcrf.gov/home.jsp) with a statistical threshold set at p < 0.05 (Supplementary Table 17; Supplementary Fig. 4).

Further enrichment analysis of gene annotations for DEGs was conducted using Cytoscape ClueGO plugin (Supplementary Table 18). This analysis was conducted through a two-sided hypergeometric test, and the statistical significance threshold was set at an adjusted p-value of less than 0.05, with the Benjamini–Hochberg method employed for correction. The comprehensive gene-annotation enrichment assessment encompassed biological process (BP), molecular function (MF), and cellular component (CC).

The intricate interplay and relationships among DEGs were unraveled with PPI network. The STRING database (https://cn.string-db.org/) was utilized to assist in mapping this network, and only interactions with a composite score above 0.7 were considered. The exhaustive list of DEGs integrated into the PPI network were summarized in Supplementary Table 19. This comprehensive analysis offers profound insights into the potential roles and interconnectedness of these genes, shedding light on their potential implications in the context of LUSC.

WCGNA analysis between DEGs and immune-related scores

The goal of WGCNA is to detect modules of genes that are co-expressed and investigate the associations between gene networks and a particular phenotype. Additionally, it enables the exploration of key genes within a network. In this study, the “WGCNA” package was used to perform the WGCNA analysis. The assessment of correlations between immune-related scores and distinct modules involved the computation of Module Significance (MS). This procedure facilitated the identification of genes encompassed within individual modules. The MS values were leveraged to categorize the module exhibiting the least MS as the negative module, whereas the module exhibiting the greatest MS was designated as the positive module. Once the module of interest was chosen per its MS value, a roster of the genes identified within this module was compiled (Supplementary Table 20; Supplementary Fig. 5).

LUSC subclasses identified with NMF

To cluster the proteasome genes, the non-negative matrix factorization (NMF) was adopted. Prior to commencing NMF, a filtering process was executed, excluding genes with a median absolute deviation (MAD) value of ≤ 0.5 across lung cancer patients, thereby narrowing down the pool of candidate genes. Subsequently, the unsupervised NMF clustering methods were employed on the metadata set utilizing the NMF R package (https://www.rdocumentation.org/packages/NMF/versions/0.23.0). Based on the coexistence correlation coefficient K value, a cluster number of 6 was found to be optimal (Supplementary Fig. 6). The Kaplan–Meier method and “survival” package (https://www.bioconductor.org/packages/devel/bioc/vignettes/survtype/inst/doc/survtype.html) within the R software were used to generate the overall survival curve. Furthermore, an analysis of varied immune elements, encompassing immune-related cells, checkpoints, and immune-related scores, was carried out across distinct NMF clusters. To illustrate different distribution of NMF clustering between the low- and high- risk score groups, the ggalluvial R package (https://www.rdocumentation.org/packages/ggalluvial/versions/0.12.3/topics/ggalluvial-package) was employed for visualization.

Low- and high-immunity clusters in LUSC

The TME immune cell infiltration was quantified using the single-sample Gene Set Enrichment Analysis (ssGSEA) approach. This was accomplished with the R package GSEABase and leveraging RNA-seq gene expression data. This analysis encompassed 28 distinct human TME immune cell subtypes. These included diverse categories such as Tcm, Tgd cells, activated T cells, Tem CD4 + /CD8 + T cells, Treg cells, Th1 cells, Th2 cells, Th17 cells, Tfh cells, as well as various B cell subsets including activated, immature, and memory B cells. Additionally, innate immunity-associated cell types were explored such as monocytes, NKT cells, mast cells, NK cells, macrophages, eosinophils, neutrophils, activated plasmacytoid and immature DCs, and MDSCs. The comprehensive analysis was underpinned with the R package GSEABase, employing RNA-seq gene expression data to provide insights into the abundance of these diverse TME immune cell subtypes.

The hierarchical agglomerative consensus clustering on LUSC tissues was analyzed to exhibit varying degrees of immune cell infiltration within the TME. This clustering was performed utilizing Ward’s linkage and Euclidean distance as measures of similarity. To ascertain the optimal K value for patient grouping in preparation for subsequent analysis, utilizing the proportion of ambiguous clustering in unsupervised methods is a practical and effective technique, particularly when it comes to K-means analysis. Cluster analysis was carried out using the ConsensuClusterPlus R package, involving 1000 cycles of computation to ensure the robustness and consistency of classification. A consensus clustering analysis, guided by ssGSEA scores, was conducted on cancer tissue samples to identify the most stable groups. The LUSC tissue samples were grouped with varying levels of immune cell infiltration within the TME based on their abundance (Immunity_A, Immunity_B, Immunity_C, and Immunity_D) (Supplementary Table 21).

Additionally, the distribution of diverse immune components was investigated, including immune cells, immune checkpoints, and immune scores, across various immunity clusters. The disparity in immunity clustering between high- and low-risk score groups was represented with R package ggalluvial (https://www.rdocumentation.org/packages/ggalluvial/versions/0.12.3/topics/ggalluvial-package). This visualization allowed for a clear depiction of the distinct distribution patterns.

Cell function analysis

Following transfection with siRNA-PSMD11 and the corresponding control sequence, the cells were separately plated into 96-well plates. Subsequently, 10 µL CCK8 solution was introduced into each well at intervals ranging from 1 to 4 days, followed by a 2-h incubation period. The absorbance values (OD) were measured at a 450-nm wavelength. The growth trajectories of the cells (H226 and H520) were established through cell counting.

For the cell invasion assay, a 24-well plate and transwell chamber were employed. Cells transfected with siRNA-PSMD11 and control sequences were seeded into the chamber using serum-free medium. Following a 24-h incubation period (37 °C; 5% CO₂), the number of invaded cells was manually counted.

Cells that underwent transfection with siRNA-PSMD11 and control sequences were introduced into 6-well plates containing serum-free medium. Subsequently, a wound healing experiment was conducted. Once the cells reached an approximate confluence of 90%, an artificial wound was generated using a 10-µL pipette tip. To facilitate the observation of wound healing progression, images were captured at both the 0-h and 48-h time points. Subsequent to image acquisition, the relative percentage of wound healing was computed to quantitatively assess the extent of healing.

Cycloheximide (CHX) is an inhibitor of protein synthesis in eukaryotes, which inhibits protein synthesis and RNA synthesis in vivo. MG-132 is a commonly used proteasome inhibitor, which can penetrate cells and selectively inhibit proteasome. Chloroquine (CQ) is an autophagic lysosome inhibitor. In order to test the protein degradation pathways of PSMD11, its protein expression was detected with western blotting after treating with CHX, MG132, and CQ for 0 h, 3 h, 6 h, and 9 h.

Statistical analysis

Statistical significance evaluation of variables following normal distribution entailed unpaired t-tests with a p-value less than 0.05. In the context of gene ontology (GO) and KEGG analyses, the FDR and Benjamini–Hochberg methodologies were adopted to ensure appropriate adjustments for multiple testing. Survival curve generation and significance assessment were accomplished via the Kaplan–Meier method. For significance testing, the Log-rank (Mantel-Cox) test was utilized, with a p-value less than 0.05. To determine the hazard ratio, multivariate and univariate Cox proportional hazard regression models were utilized, with a p-value less than 0.05.

Results

Ubiquitinated proteins of proteasome identified in LUSCs relative to controls

The proteasome complex comprises different subunits categorized into 3 groups: the 19S regulatory particle lid (consisting of 14 genes: PSME3, PSMD7, PSME2, PSME1, PSMD4, PSMD3, PSMD12, PSMD11, PSMD9, PSMD8, SEM1, PSMD6, PSMD14, and PSMD13), the 19S regulatory particle base (consisting of 10 genes: ADRM1, PSMD1, PSME4, PSMC4, PSMC6, PSMC3, PSMC2, PSMD2, PSMC1, and PSMC5), and 20S core particle (consisting of 14 genes: PSMA3, PSMB2, PSMB7, PSMB4, PSMB6, PSMA6, PSMB1, PSMA5, PSMA2, PSMB3, PSMB5, PSMA7, PSMA1, and PSMA4). The immunoproteasome has a modified composition, replacing PSMB10, PSMB2, and PSMB9 with PSMB8, PSMB5, and PSMB1, respectively. The replacement of PSMB5 with PSMB11 occurred in the thymoproteasome (Fig. 2A; Supplementary Table 1). Among those proteasome genes in proteasome complex, a total of 9 ubiquitinated proteins with 23 ubiquitinated sites were identified in LUSCs relative to controls (Table 1), including PSMD12, ADRM1, PSMA6, PSMC1, PSMD4, PSMC2, PSMD2, PSMC4, PSMD11, PSMC3, PSMC5, PSMC6, PSMD3, and PSMD8.

Fig. 2.

The correlation of proteasome genes in proteasome complex and clinical features (drug sensitivity, clinical stage, survival rate, or immune-related score). A The proteasome genes in proteasome complex. B The correlation of proteasome genes in proteasome complex with each other. C The PPI network of proteasome genes in proteasome complex. D The correlation of proteasome genes in proteasome complex and drug sensitivity. E The correlation of proteasome genes in proteasome complex and clinical stage. F The correlation of proteasome genes in proteasome complex and survival rate. G The correlation of proteasome genes in proteasome complex and immune-related score

The correlation, PPI, drug sensitivity, clinical feature, and immune-related analyses of proteasome genes in LUSC

The correlation analysis of proteasome genes in proteasome complex showed that some proteasome genes had high co-expression coefficients with each other; for example, PSMD11 and PSMD2, PSMD11 and PSME3, PSMD11 and PSMD3, PSMB3 and PSMD3, PSMB4 and PSMA4, PSMB4 and PSMB3, and PSMD4 and PSMB4 (Fig. 2B; Supplementary Table 2). PPI network showed some proteasome genes had both combined_score and high co-expression coefficients with each other; for example, PSMA2 and PSMA4, PSMA2 and PSMA3, PSMA3 and PSMA2, PSMA3 and PSMA6, PSMA4 and PSMA2, PSMA6 and PSMA3, PSMA3 and PSMA4, PSMA4 and PSMA6, PSMA4 and PSMA3, and PSMA6 and PSMA4 (Fig. 2C; Supplementary Table 3). Some proteasome genes showed substantial correlations with drug sensitivity (p < 0.05) (Fig. 2D; Supplementary Table 4). Some showed |correlation coefficient|> 0.5, including PSMB10 and Melphalan, PSMB10 and Cyclophosphamide, PSMB10 and Triethylenemelamine, PSMB10 and Thiotepa, PSMB10 and Uracil mustard, PSMB10 and Pipobroman, PSMB10 and Asparaginase, PSMB10 and Chlorambucil, PSMA6 and Chelerythrine, and PSMB10 and Hydroxyurea. In addition, the relationship between proteasome genes and pathologic stages (I, II, III, and IV) was examined in LUSCs (Fig. 2E). Proteasome genes exhibited notable variations in their expression levels across different pathological stages (Supplementary Table 5 and 6); for example, PSMD9, PSMC6, PSMC5, PSMA5, PSMB6, PSMA3, PSMC2, PSMD14, PSMC3, PSMB5, and PSMA7. The survival rates of LUSC patients were analyzed for each proteasome gene (Fig. 2F), and a significant correlation was found between many of them and overall survival (Supplementary Fig. 1), including PSMB5, PSMC3, PSMD1, PSMD2, PSMD12, PSME1, PSME3, PSME4, and SEM1. The strong correlation was demonstrated between the expression of proteasome genes and immune-related scores, suggesting a possible link to the immune microenvironment (Fig. 2G; Supplementary Table 5; Supplementary Fig. 2), including stromal score, immune score, ESTIMATE score, and tumor purity.

Construction of risk score model based on DEGs of proteasome genes between LUSC and normal tissues

The Student t-test was employed to determine DEGs of proteasome genes in LUSCs vs. controls (p < 0.05). A total of 41 DEGs were identified, including PSMB10, PSMB7, PSMB9, PSMB4, PSMB6, PSMB3, PSMB1, PSMB2, PSMA6, PSMA7, PSMA4, PSMB5, PSMA3, PSMA1,PSMA2, PSME4, PSMC6, PSMC4, PSMC5, PSMC3, PSMC1, PSMC2, PSMB8, PSMD9, ADRM1, PSMD2, PSMD1, PSME3, PSME1, PSMD6, PSMA5, SEM1, PSMD14, PSMD12, PSMD13, PSMD11, PSMD7, PSMD8, PSME2, and PSMD4 (Fig. 3A).

Fig. 3.

Multiple regression identified the prognostic model in LUSC. A The DEGs of proteasome genes between tumor and normal tissues in LUSC. B Multiple regression identified the prognostic model in LUSC. C Receiver operating characteristic curve in training group. D Receiver operating characteristic curve in test group. E Overall survival analysis between high- and low-risk score groups in training group. F Overall survival analysis between high- and low-risk score groups in test group. G The risk score assessment nomogram to evaluate prognosis in LUSC (1-, 2-, and 3-year survival rates). *p < 0.05, **p < 0.01, and ***p < 0.001

The TCGA LUSC specimens were randomly grouped into training and testing groups. A total of 41 DEGs were analyzed with multiple regression analysis in train group. Of them, 11 genes were obtained for further study, including PSMD3, PSMD4, PSMD6, PSME2, PSMD1, PSMC5, PSMA1, PSMB3, PSMB4, PSMB6, and PSMB10 (Fig. 3B; Supplementary Table 8). The ROC curve displayed an area under the curve (AUC) of 0.711 in the training group (Fig. 3C). Similarly, the testing group had an AUC of 0.626 through the ROC curve (Fig. 3D). The Kaplan–Meier method was used to compare overall survival rates between the low- and high-risk score groups in the training and testing sets, respectively (Fig. 3E–F), which revealed a noteworthy disparity in overall survival rates. To provide a more user-friendly and easily interpretable prediction model, a nomogram was constructed. This nomogram integrates crucial clinical attributes and risk scores to offer a simplified means of estimating a patient’s likelihood of survival (Fig. 3G).

The clinical heatmap illustrated the association between the risk group and clinical attributes, including clinical M stage and pathological stage (Fig. 4A). Significant correlations between overall survival and various factors such as age, cancer status, pathologic stage, pathologic M, pathologic T, and risk score were unveiled with univariate Cox regression analysis (Fig. 4B). Multivariate Cox regression revealed that age, cancer status, and risk score were all independent factors associated with the risk of LUSC (Fig. 4C). To mitigate potential overfitting effects, a validation of the prognostic value of the proteasome signature was carried out using an internal validation cohort derived from the IMvigor210 database (Supplementary Table 9). The IMvigor210 database cohort was divided into two groups based on their risk scores, namely high- and low-risk groups. The division was made with the median value as the demarcation point. The assessment of survival rates for these distinct high- and low-risk categories within the IMvigor210 dataset cohort was performed with the Kaplan–Meier method (Fig. 4D).

Fig. 4.

The clinical correlation between high- and low-risk score groups. A Heat map of clinical correlation between high- and low-risk score groups. B The univariate Cox regression analysis of risk factors. C The multivariate Cox regression analysis of risk factors. D The samples in IMvigor210 database cohort were divided into high- vs. low-risk groups based on the median value of all risk scores and do survival curve. E The correlation of TMB and risk score. F The mutation status in the high-risk score group. G The mutation status in the low-risk score group. H The GSEA between high- and low-risk score groups

The distribution of gene mutation status between low- and high-risk score groups

An examination was conducted to assess the correlation between risk score and tumor mutational burden (TMB) (Supplementary Table 10), which revealed a statistically significant correlation (Fig. 4E). The mutation gene statuses of LUSC within the high-risk score and low-risk score groups were summarized (Supplementary Table 11). These statuses encompassed various types such as 3-prime UTR variant, synonymous variant, stop gained, missense variant, intron variant, noncoding transcript exon variant, intergenic variant, 5-prime UTR variant, frameshift variant, splice region variant, downstream gene variant, stop retained variant, upstream gene variant, inframe deletion, splice donor variant, stop lost, start lost, inframe insertion, splice acceptor variant, and regulatory region variant. The Maftools R package was employed to further explore the distribution of tumor mutations, culminating in the creation of a waterfall plot showcasing the mutated genes. The top 10 gene mutations in the high-risk score group included TTN, CSMD3, MUC16, RYR2, TP53, USH2A, LRP1B, SYNE1, ZFHX4, and XIRP2 (Fig. 4F). Similarly, the top 10 gene mutations in the low-risk score group included TTN, LRP1B, ZFHX4, RYR2, MUC16, USH2A, TP53, SYNE1, CSMD3, and XIRP2 (Fig. 4G).

The different GSEA enrichment between high- and low-risk score groups

To compare the gene sets between low- and high-risk score groups, GSEA was performed between the low- and high-risk score groups, a total of 22 gene sets were identified as statistical significance (Fig. 4H; Supplementary Table 12; and Supplementary Fig. 3), including bidus metastasis up, mitsiades response to Aplidin down, ramalho stemness up, Fujii YBX1 targets down, Horiuchi WTAP targets down, PID ATR pathway, vantveer breast cancer metastasis down, Pujana BRCA centered network, Winnepenninckx melanoma metastasis up, reactome metabolism of noncoding RNA, Nakamura cancer microenvironment down, reactome cell cycle mitotic, Zhang TLX targets up, reactome mitotic G2_G2_M phases, MANALO hypoxia down, reactome mitotic M_M_G1 phases, Schuhmacher MYC targets up, GARY CD5 targets down, reactome DNA replication, Kang doxorubicin resistance DN, Odonnell targets of TFRC and MYCdown, and Kauffmann melanoma relapse up.

Correlation between risk score and immune

An investigation was conducted to examine the correlation between immune system factors [such as immune cells (Supplementary Table 13), immune checkpoint proteins (Supplementary Table 14), and immune-related indicators (Supplementary Table 7)] and risk score. Distinguishable variations emerged in numerous immune cell populations when the high-risk score group was comparted to low-risk score group. These immune cells included plasma cells, Tregs cells, monocytes, macrophages M0, and resting dendritic cells (Fig. 5A). The relationship between risk score and immune-related cells was carried out using the Corrplot technique with the Spearman method (p < 0.05), which unveiled notable correlations with significant correlation coefficients (Fig. 5B). Further disparities were observed in certain immune checkpoints between low- and high-risk score groups, such as PDCD1 and CD274 (Fig. 5C). Moreover, the high-risk score group exhibited elevated levels of immune score and ESTIMATE score, accompanied with reduced levels of tumor purity in relation to immune-related scores, compared to the low-risk score group (Fig. 5D–G).

Fig. 5.

The immunity and GSVA between high- and low-risk score groups. A The different immune cells between high- and low-risk score groups. B The correlation of immune cells and risk score. C The different immune checkpoints between high- and low-risk score groups. D–G The different immune-related scores between high- and low-risk score groups. H GSVA between high- and low-risk score groups

The different GSVA enrichment between low- and high-risk score groups

The contrasting gene sets between the high- and low-risk score groups were analyzed with GSVA method, which found a total of 43 statistically significant pathways between high- and low-risk score (Fig. 5H; Supplementary Table 15), such as olfactory transduction, cell cycle, sulfur metabolism, ubiquitin mediated proteolysis, DNA replication, basal transcription factors, homologous recombination, mismatch repair, oocyte meiosis, RNA degradation, endocytosis, nucleotide excision repair, ascorbate and aldarate metabolism, aminoacyl tRNA biosynthesis, lysosome, complement and coagulation cascades, pentose and glucuronate interconversions, vascular smooth muscle contraction, spliceosome, snare interactions in vesicular transport, glycosphingolipid biosynthesis ganglio series, one carbon pool by folate, vasopressin regulated water reabsorption, valine leucine and isoleucine biosynthesis, progesterone mediated oocyte maturation, proteasome, PPAR signaling pathway, primary bile acid biosynthesis, leukocyte transendothelial migration, tight junction, asthma, cell adhesion molecules (CAMs), lysine degradation, citrate cycle (TCA cycle), base excision repair, pentose phosphate pathway, starch and sucrose metabolism, Fc epsilon RI signaling pathway, viral myocarditis, graft versus host disease, other glycan degradation, pyrimidine metabolism, and porphyrin and chlorophyll metabolism.

KEGG, GO, and PPI analyses of DEGs between low- and high-risk score groups

Between low- and high-risk score groups, a total of 145 statistically significant DEGs were identified (Supplementary Table 16). These DEGs were subsequently subjected to KEGG analysis, which identified 19 statistically significant signaling pathways (Fig. 6A, Supplementary Table 17, and Supplementary Fig. 4), including PI3K-Akt signaling pathway, ECM-receptor interaction, hypertrophic cardiomyopathy, arrhythmogenic right ventricular cardiomyopathy, dilated cardiomyopathy, focal adhesion, human papillomavirus infection, regulation of actin cytoskeleton, small cell lung cancer, glycosaminoglycan biosynthesis-keratan sulfate, hematopoietic cell lineage, AGE-RAGE signaling pathway in diabetic complications, protein digestion and absorption, amoebiasis, calcium signaling pathway, relaxin signaling pathway, spinocerebellar ataxia, cell adhesion molecules, and pancreatic secretion.

Fig. 6.

The signaling analysis between high- and low-risk score groups. A The EMT signaling pathway of DEGs between high- and low-risk score groups. B The BP analysis of DEGs. C The CC analysis of DEGs. D. The MF analysis of DEGs. E The PPI analysis of DEGs. F Sample dendrogram and trait heatmap were plotted based on DEGs expression data and immune-related scores. G The heatmap of WGCNA

The gene ontology (GO) analysis revealed significant biological processes (BPs), cellular components (CCs), and molecular functions (MFs). These BPs, CCs, and MFs further underwent a hierarchical categorization into distinct clusters. For BP analysis, 20 distinct BPs were identified in LUSCs (Supplementary Table 18; Fig. 6C). These BPs encompassed a diverse range of functions including formation of primary germ layer, collagen fibril organization, gastrulation, SCF-dependent proteasomal ubiquitin-dependent protein catabolic process, negative regulation of protein binding, collagen-activated signaling pathway, integrin-mediated signaling pathway, keratan sulfate metabolic process, keratan sulfate biosynthetic process, endoderm development, endoderm formation, endodermal cell differentiation, glucosamine-containing compound metabolic process, N-acetylglucosamine metabolic process, amino sugar metabolic process, aminoglycan metabolic process, glycosaminoglycan metabolic process, aminoglycan biosynthetic process, glycosaminoglycan biosynthetic process, and mucopolysaccharide metabolic process. For CC analysis, 8 significant CCs were enriched in LUSC (Supplementary Table 18; Fig. 6D), including integrin complex, SCF ubiquitin ligase complex, complex of collagen trimers, basement membrane, basement membrane collagen trimer, network-forming collagen trimer, collagen network, and collagen type IV trimer. For MF analysis, 15 significant MFs were enriched in LUSC (Supplementary Table 18; Fig. 6D), including calmodulin-dependent protein kinase activity, negative regulation of protein binding, extracellular matrix structural constituent conferring tensile strength, regulation of receptor binding, collagen binding, laminin binding, transferase activity, transferring sulfur-containing groups, sulfotransferase activity, N-acetylglucosamine 6-O-sulfotransferase activity, P-type proton-exporting transporter activity, P-type transmembrane transporter activity, P-type ion transporter activity, ATPase-coupled cation transmembrane transporter activity, P-type calcium transporter activity, and pyrophosphate hydrolysis-driven proton transmembrane transporter activity.

For PPI analysis, the DEGs underwent examination within the STRING platform. The node total scores displayed a range from 0.700 to 0.999 (Supplementary Table 19; Fig. 6E). Notably, some DEGs exhibited the elevated combined scores and co-expression scores; for example, COL4A1 and COL4A2 and COL4A2 and COL4A1.

WGCNA analysis of DEGs and immune-related scores

The expression data of DEGs and immune-related scores were utilized to generate a dendrogram and heatmap of traits (Fig. 6F–G). In LUSC, the parameter that most significantly influenced the average connectivity degree and independence of each co-expression module was found to be the power value (β), closely associated with immune-related scores. By setting β to 3, the scale attained a heightened average connectivity degree (Supplementary Fig. 5). Consequently, the β value was leveraged to delineate distinctive gene co-expression modules in LUSC. Through the application of the adjacency matrix, a dendrogram encompassing all selected genes was subjected to clustering for identification of co-expression modules (Fig. 6G; Supplementary Table 20). These modules encompassed a blue module containing 25 DEGs, a brown module containing 23 DEGs, and a turquoise module containing 31 DEGs. These co-expression modules were designated different colors with a distinct p-value.

LUSC subtypes based on proteasome genes with NMF

A total of 42 proteasome genes were used to do NMF clustering. The NMF method has proven to be a successful technique for reducing the dimensions of data in cancer subtype identification. In this study, the factoextra package was used to determine the ideal number of clusters (K). Utilizing the NMF method with K set to 6, researchers observed the division of LUSC samples into four distinct subtypes (Fig. 7A; Supplementary Table 21; Supplementary Fig. 6) and those patients with different clusters displayed significant prognosis (Fig. 7B). In terms of immune-related scores and TumorPurity, they also obtained significant results (Fig. 7C–F). Several immune cells displayed notable dissimilarities among four NMF clusters. These cells included plasma cells, T cells CD4 memory resting, Tfh cells, Tregs cells, T cells CD8, M1, T cells CD4 memory activated, dendritic cells activated, mast cells quiescent, and mast cells stimulated (Fig. 7G). Significant disparities were noted in the expression levels of different immune checkpoint genes between low- and high-risk score groups. These genes included CD86, CD80, PDCD1, CD276, CD274, CTLA4, and PDCD1LG2 (Fig. 7I). The ggalluvial plot showed NMF clustering and risk score groups had a complicated relationship (Fig. 7H).

Fig. 7.

The NMF clusters analysis of LUSCs. A The heatmap of NMF cluster. B Overall survival analysis among NMF clusters. C–F The different immune-related scores among different clusters. G The different immune cells among NMF clusters. H The ggalluvial plot between NMF clusters and different risk score groups. I The immune checkpoints among NMF clusters

LUSC subtypes based on immune-related cells with ssGSEA

The gene expression data were used with ssGSEA analysis to compute the extent of TME cell infiltration in LUSC (Fig. 8A, B). This comprehensive assessment encompassed various immune cell types, namely activated T cells, Tem CD4 + and CD8 + T cells, Treg cells, Tgd cells, Th17 cells, Th2 cells, Tfh cells, Th1 cells, immature, as well as activated, Tcm, and memory B cells. Moreover, innate immunity-related cell types, including monocytes, neutrophils, macrophages, mast cells, NKT cells, NK cells, eosinophils, activated plasmacytoid and immature DCs, and MDSCs, were also evaluated. The characterization of immune cell information within LUSC tissue samples led to the stratification of these samples into distinct categories, denoted as Immunity-A, Immunity-B, Immunity-C, and Immunity-D (Supplementary Table 21). Significant results were also achieved in relation to immune-related scores and tumor purity (Fig. 8C–F).

Fig. 8.

The ssGSEA clusters analysis of LUSCs. A, B The heatmap of ssGSEA cluster. C–F The different immune-related scores among different clusters. G The different immune cells among ssGSEA clusters. H The ggalluvial plot between ssGSEA clusters and different risk score groups

Among the four immunity clusters, notable variations were observed in certain immune cell populations, which encompassed B cells memory, T cells CD4 memory resting, T cells CD4 naïve, plasma cells, T cells CD4 memory activated, resting mast cells, Tfh cells, resting NK cells, activated NK cells, T cells CD8, M0, M2, M1, activated mast cells, activated dendritic cells, and neutrophils (Fig. 8G). Some immune checkpoints were significantly different among groups, including CD86, CD80, PDCD1, CD276, CD274, CTLA4, VTCN1, and PDCD1LG2 (Fig. 8H). The ggalluvial plot showed immunity clustering, and risk score group had a complicated relationship (Fig. 8I).

PSMD11 facilitated LUSC progression and depended on ubiquitin–proteasome system for degradation

For an in-depth exploration into the role of PSMD11 in lung cancer progression, loss-of-function experiments were conducted using transient transfection of siRNA-PSMD11 in LUSC cell lines. These investigations unveiled a significant suppression of proliferative potential (Fig. 9A, B), migratory capacity (Fig. 9C), and invasive characteristics (Fig. 9D) within the H226 and H520 cell lines upon downregulation of PSMD11.

Fig. 9.

PSMD11 facilitated LUSC progression and depended ubiquitin–proteasome system to degradation. A CCK8 assay in H520 between siRNA-PSMD11 and control groups. B CCK8 assay in H226 between siRNA-PSMD11 and control groups. C Transwell CCK8 assay in H520 and H226 between siRNA-PSMD11 and control groups. D Wound healing assay in H520 and H226 between siRNA-PSMD11 and control groups. E The protein expression of PSMD11 in H226 was detected using western blotting after treating with CHX and MG132 after 0 h, 3 h, 6 h, and 9 h. F The protein expression of PSMD11 in H520 was detected using western blotting after treating with CHX and MG132 after 0 h, 3 h, 6 h, and 9 h. G The protein expression of PSMD11 in H226 was detected using western blotting after treating with CHX, MG132, and CQ after 0 h, 3 h, 6 h, and 9 h. H The protein expression of PSMD11 in H520 was detected using western blotting after treating with CHX, MG132, and CQ after 0 h, 3 h, 6 h, and 9 h

In eukaryotic cells, protein degradation can occur through two distinct pathways. The first pathway is an ATP-independent process that mainly degrades cell derived proteins, membrane proteins, and long-lived intracellular proteins in lysosomes. The second pathway is ATP-dependent ubiquitin–proteasome system, which is carried out in the cytoplasm and mainly degrades abnormal proteins and short-lived proteins. In order to test pathways for protein degradation of PSMD11, PSMD11 protein expression was detected using western blotting after treating with CHX, MG132, and CQ for 0 h, 3 h, 6 h, and 9 h. When treated with CHX and MG132, the protein expression of PSMD11 had no significant change, but protein degradation of PSMD11 was significantly changed when treated with MG132 both in H226 (Fig. 9E) and H520 (Fig. 9F). Moreover, protein degradation of PSMD11 was not significantly changed when treated with MG132 both in H226 (Fig. 9G) and H520 (Fig. 9H). Those results showed that protein degradation of PSMD11 was dependent on the ubiquitin–proteasome system.

Discussion

The importance of proteasome in the regulation of ubiquitination level and protein degradation in lung cancer

Proteasome and ubiquitination are important intracellular regulatory mechanisms that play a role in lung carcinogenesis. Proteasomes can regulate a range of signaling pathways associated with tumors and influence tumor cell proliferation and invasion. For example, proteasomes can regulate signaling pathways such as EGFR, Akt, and NF-κB, thereby promoting tumor cell proliferation and invasion. The functionality of proteasomes extends to governing the processing and presentation of antigens within tumor cells. This intricate process possesses the potential to exert a direct influence on the immune response mounted by antigen-specific T cells. In tumor cells, the proteasome can regulate the expression and processing of biomarkers such as MHC molecules and tumor-associated antigens, thereby influencing T cell immune surveillance and response. The proteasome possesses the capacity to modulate drug resistance in different cancer cell types. Proteasomes possess the capacity to influence the vulnerability of cancer cells towards chemotherapy medications and targeted therapies, thereby potentially impacting the efficacy of tumor treatment. This exemplifies how the effectiveness of tumor treatment can be ultimately affected. Ubiquitination is a crucial process that can regulate various signaling pathways linked to the development and advancement of tumors, including the Wnt and Hedgehog signaling pathways, among others. By modifying the ubiquitination of specific proteins, the activity of these signaling pathways can be promoted or inhibited, affecting the growth, proliferation, and metastasis of tumors. Ubiquitination can also regulate the responsiveness of cancerous cells to chemotherapy agents and targeted therapeutic agents. By regulating the degree of ubiquitination of proteins, tumor cells have the ability to evade the effects of chemotherapeutic drugs, which can ultimately result in the tumor developing drug resistance. Ubiquitination can also affect antigen processing and presentation by tumor cells, thereby affecting immune surveillance and response by T cells. The ubiquitination of biomarkers, such as MHC molecules and tumor-associated antigens, can be modulated by tumor cells to the immune response of T cells towards tumors. Ubiquitination is an essential step in the growth and advancement of pulmonary carcinoma. As a result, extensive research is currently being conducted to develop innovative drugs that can target the ubiquitinated metabolic pathways of tumor cells. For instance, therapeutic efficacy has been demonstrated in lung cancer for proteasome inhibitors and ubiquitin kinase inhibitors.

The way in which the proteasome modifies and regulates the level of ubiquitination also occupies an extremely important position in lung cancer. The proteasome is a crucial component in the process of degrading ubiquitinated proteins. It plays a vital role in the protein degradation system for a wide range of ubiquitinated proteins. This degradation removes proteins and their metabolites that have been ubiquitinated, thereby preventing these proteins from accumulating in lung cancer cells and affecting normal cell function. The level of ubiquitination can be regulated by the proteasome through controlling the expression and function of enzymes like ubiquitinases and deubiquitinases. This regulation affects the progress of ubiquitination reactions and the quantity of ubiquitinated proteins. In this way, the proteasome can control the expression of specific ubiquitinated proteins in lung cancer cells, thereby influencing cell fate, signaling pathways, and more. Some specific ubiquitinated proteins are considered to be targets in lung cancer cells because of their important biological functions. In this case, the proteasome can specifically degrade these ubiquitinated proteins and, in this way, inhibit signaling pathways such as proliferation, metastasis, and drug tolerance in lung cancer cells. The way in which the proteasome modulates the level of ubiquitination is of great importance in lung cancer, and they interact and collectively influence aspects of lung cancer cell growth, metabolism, and signaling pathways. To develop a more comprehensive comprehension of lung cancer pathogenesis, it is imperative to explore the correlation between ubiquitination and proteasomes through extensive studies. By further exploring this association, ones can gain fresh perspectives on the underlying mechanisms that propel the progression and development of pulmonary carcinoma. Furthermore, the outcomes of this study could potentially uncover new therapeutic targets that could significantly enhance the therapeutic strategies for lung cancer, ultimately leading to better patient outcomes and overall quality of life.

Importance of proteasome in the regulation of lung cancer immune microenvironment

In immunotherapy for lung cancer, proteasomes and the process of ubiquitination also contribute significantly to various biological processes. In the immune system, proteasomes have a critical role in regulating immune molecules. Proteasomes can degrade immune molecules such as cytokines, receptors, and signaling molecules, thereby regulating their biological activity and stability. Proteasomes can also regulate the expression and secretion of immune molecules. For example, proteasomes can degrade the inhibitory protein IκB of the transcription factor NF-κB, thereby promoting the nuclear translocation of NF-κB and initiating the transcription of immune molecules [21]. In addition, proteasomes can regulate the secretion pathway of immune molecules, such as promoting the secretion of immune molecules by degrading secretion pathway-related proteins on the cell membrane [13]. Our investigation has illuminated a substantial contribution of proteasome regulation to the immune system across various immune molecules. One such cytokine is VTCN1, also recognized as B7-H4, which assumes a pivotal function in immune system modulation, thereby fostering immune evasion within tumor contexts [22]. Studies have shown that the stability of VTCN1 can be regulated through ubiquitination, which can affect the immunogenicity of tumors [23]. CD86 is a molecule that provides co-stimulation and is present on the surface of immune cells. It has the ability to interact with the CD28 receptor found on T cells, thereby activating T cell immune responses [24, 25]. In the context of lung cancer, a frequently observed phenomenon is the attenuation of CD86 expression, leading to the suppression of T cell functionality. This phenomenon enables cancer cells to effectively elude immune surveillance [25]. Emerging studies have unveiled a correlation between diminished CD86 expression on dendritic cell surfaces and compromised antigen presentation, culminating in dampened T cell activation [26]. This modulation is believed to stem from the degradation of CD86 due to downregulation. Proteasomes, pivotal in immune cell functions, intricately govern a spectrum of immune responses, encompassing antigen processing, cytokine production, and T cell activation. CMTM6, as indicated by multiple investigations, serves as a pivotal regulator of PD-L1 expression [27, 28]. Through these studies, it has been revealed that CMTM6 interacts with PD-L1 at both the plasma membrane and recycling endosomes, finely regulating the expression of PD-L1 within these specific cellular compartments. Notably, CMTM6 intervenes in the ubiquitination process via STUB3, the E1 ubiquitin ligase, consequently prolonging PD-L1 protein half-life [27, 28]. In a mouse model of syngeneic melanoma, the depletion of CMTM6 led to a decrease in PD-L1 expression and a simultaneous increase in the activity of tumor-specific T cells. Additionally, studies have shown that the enzyme CSN5 acts as a deubiquitinase, breaking down proteins and removing polyubiquitination from PD-L1. This process is essential in inhibiting anti-tumor immune responses in macrophages that invade tumors [29]. The activation of NF-κB and the emergence of CSN5 in cancer cells can be prompted by TNF-α, which originates from macrophages. Consequently, this cascade obstructs the ubiquitination and subsequent breakdown of PD-L1 via CSN5. This phenomenon fosters a heightened interaction between PD-L1 and PD-1, ultimately culminating in the evasion of immune recognition by T cells within the cancer cell context [29]. Studies have shown that by augmenting ubiquitination and degradation through proteasomes, the suppression of PD-L1 expression could potentially curtail tumor proliferation. This modulation has the potential to amplify the efficacy of cytotoxic T cells in combating tumors, as evidenced in patients with NSCLC subjected to PD-1-targeted monoclonal antibody treatment. These observations accentuate the viability of a therapeutic avenue aimed at countering tumor-induced immune evasion [30]. Immunotherapy is an important treatment modality for lung cancer, but the immune escape mechanism of tumor cells is one of the main factors limiting the effectiveness of immunotherapy. Studies have shown that proteasomes and ubiquitination modification can regulate the degradation of immune escape proteins in tumor cells, thereby enhancing the effectiveness of immunotherapy. Investigations into the function of proteasomes and ubiquitination modifications in tumor and lung cancer immunotherapy are anticipated to offer novel concepts and tactics for the management of cancer.

Significance of DEGs between low- and high-risk score groups in LUSC

In the quest to identify genes with disparate expression in LUSC, a viable approach involves contrasting gene expression profiles between healthy lung tissues and those afflicted with LUSC. CHST7 is an acyltransferase that participates in glycosylation reactions by transferring sulfate groups to sugar molecules. Research has demonstrated that there is a notable rise in the expression of CHST7 in lung cancer, which correlates with the onset, progression, and metastasis of this disease. Furthermore, the variation in CHST7 expression between the serum of lung cancer patients and those afflicted with non-malignant lung inflammation is statistically significant. This discrepancy underscores the potential utility of CHST7 as a diagnostic biomarker in serum [31]. GAP43 plays a pivotal role in neuronal development, regeneration, and synaptic plasticity, which stands out as a neuron-specific protein of significance [32]. Study found the significant role of GAP43 in lung cancer, particularly its crucial involvement in brain metastasis, a consequential outcome of NSCLC [33]. Acting through the Rac1/F-actin pathway, GAP43 showcases its potential as an independent prognostic indicator for brain metastasis among NSCLC patients [33]. NCBP1 is a protein that binds to single-stranded RNA and plays an essential role in the nucleus [34]. Emerging findings underscore the noteworthy influence of NCBP1 on the advancement and evolution of lung cancer [35]. Through the action of NCBP1, the oncoprotein axis involving NCBP1-NCBP3-CUL4B takes center stage in driving the proliferation, migration, epithelial-mesenchymal transition, and wound healing capability of lung cancer cells [35]. This may be a possible target for treating lung adenocarcinoma. ITGA8 is an important integrin protein, which acts as a bridge between the extracellular matrix and cell membrane. Recent research has evidenced the significant role of ITGA8 in the progression of lung cancer. The analysis of survival rates among patients with lung adenocarcinoma revealed that increased expression of ITGA8 correlates with improved prognosis. These findings suggest that ITGA8 could have clinical applications for treating lung cancer [36]. The study concluded that ITGA8 had a positive correlation with immune checkpoints and immunomodulators. This finding showed that ITGA8 might be engaged in the immune escape mechanism of lung cancer [36]. Simultaneously, the analogous correlation observed in the infiltration of diverse immune cells—such as CD8 + T cells, B cells, neutrophils, macrophages, CD4 + T cells, and dendritic cells—suggests that ITGA8 might modulate lung cancer progression by controlling immune cell infiltration [36]. Demonstrating a converse relationship, ITGA8 displays an inverse association with tumor stem cells, suggesting its potential involvement in the modulation of lung cancer stem cells [36]. In a word, ITGA8 is significant in the onset, progression, and management of lung carcinoma. Future research will further reveal its mechanism and provide fresh approaches and novel techniques for managing lung cancer. OPA1 is a common mitochondrial fusion protein, which plays a vital role in many cellular processes. Recently study found that OPA1 might have a crucial function in the development of lung cancer, and OPA1 displays significant upregulation within lung adenocarcinoma, showing a robust link to an adverse prognosis [37]. The deficiency of OPA1 disrupts the dynamics of mitochondria and impairs respiratory function, thereby enhancing the susceptibility of tumor epithelial cells to CD8 + T cells in cases of NSCLC [37]. The identification of these DEGs not only enhances our understanding of the advancement and growth of LUSC but also presents novel targets and a strategic foundation for identification and management of this disease. In addition, by studying DEGs, molecular subtypes of LUSC can be identified and the diagnostic accuracy can be improved.

Proteasome as the target of drug sensitivity and its significance in LUSC

There is a certain correlation between drug sensitivity and proteasome. Many drugs rely on proteasomes to play their role. Bortezomib is a type of drug that inhibits the activity of proteasomes, and this protein complex is of utmost importance in facilitating the growth and viability of cancerous cells. This drug is effective in preventing the proliferation and survival of cancerous cells. Although proteasome inhibitors have shown remarkable efficacy, drug resistance is an important problem because it is almost inevitable. According to clinical research, patients diagnosed with LUSC tend to exhibit low drug sensitivity and are susceptible to developing drug resistance. Many factors may affect the drug sensitivity of LUSC, including tumor cell type, gene mutation, and protein expression. Therefore, understanding the resistance mechanism of proteasome inhibitors has become a research field of great concern. This study focused on drug sensitivity related to proteasome genes and screened out a series of potentially effective drugs. This study found that Nelarabine might be effective on the proteasome genes PSME1, PSMA6, and PSMB10. Primarily indicated for the management of T cell lymphoma and acute lymphoblastic leukemia, nelarabine represents a novel chemotherapeutic agent. The way it works is by blocking the synthesis of DNA and preventing the growth of cells, which ultimately leads to the inhibition of proliferation and spread of cancer cells [38]. This study found that hydroxyurea might be effective on proteasome genes PSMB1, PSME1, PSMD9, PSMB9, and PSMB10. Although hydroxyurea is commonly used to treat leukemia and other blood disorders, it is not frequently used for treating lung cancer. The current research emphasis lies in evaluating the safety and efficacy of combining hydroxyurea with other therapeutic agents for lung cancer. Research has demonstrated that the utilization of hydroxyurea and CHK inhibitors in combination is more effective than gemcitabine for treating lung cancer while also decreasing the damage to surrounding tissues [39]. This study found that Chlorambucil may have potential effects on the PSMB9 and PSMB10 subunits of proteasome. The primary mechanism of action of Chlorambucil, a chemotherapy medication, is to impede the replication and proliferation of cancerous cells by disrupting their ability to synthesize and repair DNA. Some studies show that Chlorambucil may have certain potential in treating lung cancer. Recent research findings have demonstrated the effective inhibition of lung cancer cell proliferation and induction of apoptosis by Chlorambucil, leading to the successful suppression of lung cancer growth. A recent research has demonstrated that the administration of Chlorambucil in combination with other chemotherapy drugs can significantly enhance the survival rate of patients suffering from lung cancer. A different research shows that Chlorambucil is a medication that selectively attacks cells lacking BRCA2 and exhibits a greater therapeutic ratio than cisplatin in treating tumors with BRCA2 deficiency [40]. Generally speaking, Chlorambucil is a potential drug for lung cancer treatment, but further research is needed to determine its safety and effectiveness in clinical application. Chelerythrine is a natural alkaloid with many biological activities, including anti-inflammatory, anti-tumor, and anti-oxidation. Research has indicated that Chelerythrine can impede the proliferation and metastasis of lung cancer cells through various pathways, suggesting that it may have a promising therapeutic impact on lung cancer. This study shows that Chelerythrine may have potential effects on proteasome PSMA7, PSMD7, PSMB1, PSME1, PSMB7, PSMB10, and PSMA6. Research has demonstrated that Chelerythrine possesses the ability to decelerate cancer progression by restraining the cancer-initiating potential of lung cancer stem cells. Moreover, its targeting specificity can be modulated by altering the concentration, making it adaptable to different cell lines [41]. This study not only found the abnormal expression of the proteasome in lung cancer cells but also further screened out some drugs, which may have drug sensitivity to lung cancer cells, thus having potential therapeutic effects. The development and research of these medications are anticipated to introduce fresh choices for treating lung cancer patients, thereby enhancing the efficacy and longevity of lung cancer therapy. However, the research and development of these drugs need further experimental verification and clinical trials to finally determine their efficacy and safety.

Significance of differentially ubiquitinated proteasome gene signature model in LUSC

Ubiquitin and proteasome are important metabolic pathways in cells, and they are involved in many biological processes, including protein synthesis, modification, folding, and degradation. The statistically significant signaling pathways that were found in this study have also been demonstrated to be modulated by ubiquitination or proteasome, which can impact the onset and progression of tumors. The cell cycle is an important stage in the cell life cycle, and its regulation process is extremely complicated. Ubiquitin, as an important regulatory mechanism, participates in several key regulatory processes in the cell cycle. For example, after the end of the S phase of the cell cycle, ubiquitination can mark the CDK inhibitor p27Kip1 and promote cells to enter the G2 phase, thus promoting the process of the cell cycle. In addition, ubiquitination can also regulate chromosome separation and mitosis in the cell cycle. Before mitosis, ubiquitination can mark proteins on chromosomes, such as CENP-E and BubR1, thus regulating the alignment and separation of chromosomes and ensuring normal mitosis. In addition, recent studies have also found that the ubiquitination and proteasome degradation of cyclin D are regulated by AMBRA1, thus affecting the development of lung adenocarcinoma. These observations show the vital role of ubiquitination in governing the cell cycle and its indispensable contribution to the smooth progression of this crucial biological process [42]. Ubiquitin–proteasome is a protein degradation mechanism under the combined action of intracellular ubiquitination and proteasome. In this pathway, ubiquitin is linked to the protein, and then the ubiquitinated protein is sent to the proteasome for digestion. This process requires a variety of ubiquitin ligase, ubiquitin-removing enzyme, and proteasome functions. The ubiquitin–proteasome pathway can either facilitate or impede the growth and progression of tumors depending on whether the expression of proteins or enzymes involved in this pathway is elevated or reduced [43]. Ubiquitin-specific protease regulates the ubiquitin–proteasome pathway by preventing the breakdown of specific proteins via ubiquitination, which can result in either carcinogenic or anticancer effects [43]. Endocytosis is a process in which substances on the surface of the cell membrane are “swallowed” into cells for decomposition and reconstruction. Receptors located on the cell membrane typically facilitate this process, while intracellular enzymes such as autophagy and lysosomes are responsible for its degradation. Studies have shown that the expression of melatonin receptors is diminished in lung adenocarcinoma cells, and targeting ubiquitin-specific protease 8 can augment the process of endocytosis. This, in turn, restrains the proliferation of lung adenocarcinoma cells and enhances their receptiveness to melatonin [44]. Studies have shown that silencing ubiquitin-specific protease 8 can enhance the anti-tumor effect of melatonin [44]. Autophagy is an important intracellular protein degradation mechanism, which degrades intracellular protein through lysosomes. The autophagy pathway includes many enzymes, such as autophagy-related protein (Atg). The autophagy pathway interacts with the ubiquitination and proteasome pathways, and they participate in intracellular protein metabolism. Recent studies have demonstrated that the autophagy pathway can enhance the migration and invasion of lung cancer cells triggered by TLR3 and TLR6 by promoting the ubiquitination of TRAF4 [45]. Therefore, inhibiting autophagy may be a viable approach in treating lung cancer [45]. This discovery presents a fresh approach and orientation for addressing lung cancer. More investigation is required to examine the participation of the autophagy pathway in the onset and progression of lung cancer, as well as to determine optimal strategies for utilizing the autophagy pathway in the treatment of lung cancer. The TCA cycle (tricarboxylic acid cycle) is a process of converting glucose and other metabolites into energy. LUSC represents a common type of lung malignancy with significant prevalence, which originated from smooth epithelial cells, and these cells have oxidase needed for oxidation reaction. Research indicates the significant involvement of the TCA cycle in the context of LUSC. Metabolic analysis of LUSC cells shows that they promote tumor growth and progress by increasing the production of TCA cycle metabolites and increasing NADPH levels to adapt to antioxidant pressure [46]. In addition, the study also found that some patients with LUSC have key gene mutations in the TCA cycle, which leads to abnormal metabolism [47]. Vital for cell survival and growth, the PI3K-AKT pathway stands as a pivotal cellular signaling route. Stimulating this signaling pathway will enhance the proliferation and viability of cancerous cells and can impact numerous other signaling pathways. AKT is the final effector of the PI3K signaling pathway, which controls biological processes such as metabolism, apoptosis, cell cycle, and DNA damage repair by regulating various downstream pathways (such as mTORC1, FOXO, and Bad). Recent research found that CERS1 has been identified as a crucial regulatory factor that can suppress brain metastasis of NSCLC. This is achieved through its interaction with USP14 and subsequent decrease in expression of the PI3K/AKT/mTOR signaling pathway [48]. Consequently, the targeting of CERS1 has emerged as a prospective therapeutic approach for the management of brain metastasis of NSCLC. The development and advancement of lung cancer entail complex mechanisms characterized by irregularities in numerous signaling pathways. These abnormal signal pathways include cell proliferation, apoptosis, angiogenesis, and immune escape. Therefore, the treatment strategy for these abnormal signal pathways is one of the important directions of lung cancer treatment.

Significance of ubiquitin–proteasome in regulating important signaling pathways in LUSC

LUSC stands as a prevalent lung cancer subtype, distinguished by distinct mechanisms and therapeutic approaches from other subtypes. Within this context, this investigation discerned 11 proteasome gene signature patterns closely intertwined with LUSC progression. These gene profiles wield significant influence on the advancement of squamous cell carcinoma in the lungs. These findings introduce novel avenues for addressing and thwarting LUSC, potentially refining early diagnostic and treatment methodologies. An essential cellular enzyme, PSMB3, participates in protein breakdown and processing within cells. Elevated PSMB3 levels correspond to an adverse prognosis among individuals diagnosed with lung adenocarcinoma [49]. PSMB10 is a gene that encodes a protein called 20S proteasome subunit beta-1. There is a significant upregulation of PSMB10 expression in LUSC patients [50]. PSMB6 is a protease found in humans that plays a role in breaking down and processing proteins within cells. Downregulation of PSMB6 inhibited the growth of lung cancer [51]. As a constituent of the proteasome complex, PSMD4 holds a pivotal function in orchestrating protein degradation processes. In the context of lung adenocarcinoma, PSMD4 expression experiences significant upregulation, with heightened PSMD4 levels notably intertwined with the severity and malignancy attributes of this ailment [18]. The protein encoded by the PSMC5 gene contributes to the degradation of various proteins. Studies have demonstrated that PSMC5 expression, serving as a radiation-sensitive biomarker, exerts influence over the radio-resistance capabilities of lung cancer cells [52]. The prognostic value of PSMC5 status in radiotherapy shows that it can be used as an indicator for predicting treatment outcomes. In cases where resistance to radiotherapy is significant, targeting PSMC5 to restore its activity could be a potential alternative approach for cancer treatment [52]. The genes that have been identified exhibit a robust association with the advancement of lung cancer. By constructing a proteasome gene model that can distinguish between different expression levels, ones can pinpoint the crucial genes that are linked to the initiation and advancement of lung cancer, thereby, shedding light on the underlying mechanisms of this disease. The findings of this study offer crucial biomarkers and therapeutic targets that can aid in the timely detection, treatment, and assessment of lung cancer prognosis. Hence, the findings of this study provide innovative perspectives and strategies for prevention and management of lung cancer. Consequently, this study holds immense importance in advancing the field and has far-reaching implications.

Roles of the proteasome in the regulation of TMB and common mutation sites in lung cancer

Tumor mutation burden (TMB) can affect the occurrence, progression, and treatment efficacy of lung cancer. The high mutational rate of lung cancer has led to significant interest in exploring the correlation between TMB and this type of cancer. Studies have found that proteasome regulation has a certain impact on tumor mutation burden. Proteasome regulation has demonstrated the capability to decrease the mutation burden within tumor cells and restrain malignancy-associated metastasis. Notably, through a focused examination, the study has revealed the potential of selective proteasome inhibitors in impeding the growth of NSCLC cells triggered by EGFR mutations. This discovery underscores the potential effectiveness of these inhibitors in inhibiting the proliferation of cancer cells associated with EGFR mutations, which could pave the way for targeted therapeutic interventions in the treatment of NSCLC. However, in other cases, proteasome regulation may increase the mutation burden of tumors and facilitate the spread of malignant cells. Several studies have revealed that the regulation of proteasomes can facilitate the advancement and spread of tumors in lung cancer and other types of cancer. Additionally, excessive activity of proteasomes may lead to a growth in the number of mutations and initiate the development of cancer. Detection of tumor mutation burden levels can provide personalized guidance for treating lung cancer. There is a close relationship between common mutation sites in tumors and proteasome regulation. The mutations typically affect several pathways, including those involved in regulating the cell cycle, repairing DNA damage, and transmitting signals. Moreover, the critical proteins within these pathways are frequently subject to degradation by the proteasome. Among them, P53 is one of the most extensively studied tumor suppressor genes. The steady-state regulation of P53 is crucial for its tumor suppressor function. This study found that TP53 is highly mutated in both high- and low-risk core groups (Fig. 4F, G). Covalent modification of P53 has been demonstrated to stabilize the protein by counteracting its ubiquitination and subsequent proteasome degradation, according to research findings [7]. In vivo, the inhibition of cell growth and tumor formation can be achieved by maintaining the stability of P53 [7]. Nonetheless, the emergence of P53 mutations in NSCLC has contributed to erlotinib resistance, an anti-tumor drug. Encouragingly, the concurrent application of proteasome inhibitors and anti-tumor medications presents a promising approach to significantly mitigate resistance concerns associated with NSCLC [53]. Moreover, research has demonstrated that proteasome inhibition leads to the accumulation of P53, and long-term proteasome inhibition can lead to the degradation of mutant P53. Our study discovered that, apart from TP53, there are several tumor mutation loci that exhibit significant associations with the susceptibility of lung cancer patients. These findings highlight the complexity of genetic alterations in lung cancer and emphasize the importance of comprehensive molecular profiling in identifying high-risk individuals. The mutation sites hold significant value in tailoring treatment and assessing the prognosis of lung cancer on an individual basis.

Conclusion and expert recommendation in the framework of 3P medicine

In the realm of LUSC, this investigation embarked on an exhaustive exploration of the ubiquitin–proteasome system, weaving together a comprehensive tapestry that unfurls a novel prognostic paradigm. This paradigm’s overarching aim resides in catalyzing the active advancement of diagnosis, prediction, and prognostication for LUSC. By delving deep into the labyrinthine interplay of the ubiquitin–proteasome system, drug responsiveness, immune microenvironment, tumor mutation load, and the intricate web of tumor mutation sites, this study’s findings reverberate with the revelation of the ubiquitinated proteasome system’s resounding significance in shaping the trajectory of LUSC’s evolution. The unveiled insights of this study unfurl like a roadmap, paving the way for the potent integration of the ubiquitin–proteasome system within the terrain of tumor treatment. In essence, this investigation unfolds as a beacon of scientific significance, poised to propel the ubiquitin–proteasome system into the forefront of innovative strategies for tackling the intricate landscape of LUSC.

We strongly recommend the emphasis on the research and practice of ubiquitin–proteasome genes in LUSC. Ubiquitin–proteasome system is the main pathway of protein degradation, which plays important roles in the entire pathophysiological process of LUSC. Proteasome is a protein complex machine with multiple protein components, its own ubiquitination has direct impact on the structure and functions of a proteasome. In-depth study of these ubiquitin–proteasome genes in LUSC biological system and applying them in multi-level diagnostics including multi-omics, liquid biopsy, prediction and targeted prevention of chronic inflammation and metastatic disease, and mitochondrial health-related biomarkers will significantly innovate PPPM of LUSC in the following three aspects:

• i.Predictive approach: Ubiquitinated proteasome genes-based signature model in combination with clinical data of LUSC can stratify LUSC patients for predictive diagnosis, and prognostic assessment to realize its predictive approach towards PPPM in LUSC. The model based on ubiquitin–proteasome genes can be closely combined with liquid biopsy to perform non-invasive predictive diagnosis for LUSC patients. This approach can facilitate early detection and diagnosis, providing more accurate and comprehensive information for personalized prophylactic treatment of LUSC [54].
• ii.Targeted prevention: E3 ligases and deubiquitination enzymes of ubiquitinated proteasome are potential therapeutic targets, and their enzyme inhibitors can serve as effective therapeutic agents towards targeted prevention to inhibit or stop occurrence and development of LUSC. The vast majority of deubiquitinases play a significant role in promoting tumors, and developing their inhibitors may provide new research direction for anti-tumor drugs. Moreover, the ubiquitin–proteasome system is associated with concurrent chronic inflammation and late metastasis of LUSC, affecting the human immune response [55, 56]. Related immune factors can serve as predictive indicators for personalized prevention and treatment. Additionally, changes in enzymes in the ubiquitin–proteasome system and other gene methylation differences can also serve as predictive indicators [57]. When mitochondrial function is impaired, ROS production can affect protein degradation, and some drugs can protect mitochondria by inhibiting ubiquitin–proteasome system activity [58]. The evaluation of mitochondrial health can also be performed through the ubiquitin–proteasome system, providing new strategies for the prevention, prediction, and treatment of LUSC [59]. Thereby, in-depth research on the role of the ubiquitin–proteasome system in lung cancer can help develop more effective personalized treatment plans, improving the diagnostic accuracy and survival rates of lung cancer patients.
• iii.Personalization of medical services: Ubiquitinated proteasome gene-based prognostic model is closely associated with clinical characteristics such as drug sensitivity, immune microenvironment, and mutation status, which can stratify LUSC patients for personalization of medical services, including personalized prognostic assessment, personalized predictive diagnosis, and personalized drug therapy in the framework of PPPM. In addition, by combining multi-omics and ubiquitinated proteasome gene analysis, ones can more effectively identify patients with high-risk LUSC and provide them with early intervention and treatment [55–57], including tobacco smoking as the high-risk factor for LUSC [60]. This can help reduce the progression and mortality of the disease, and promote personalized medical care for LUSC patients.

In summary, in-depth analysis of ubiquitinated proteasome genes enables to identify and quantify LUSC-related ubiquitinated proteasome genes to clarify molecular mechanisms and discover effective therapeutic targets, construct clinically relevant prognostic model for patient stratification, predictive diagnosis, and prognostic assessment for personalized medical services and targeted drug therapy of LUSC patients, which created a new area in the research and clinical practice of LUSC. It established an innovative state of the art contributing to the paradigm shift from reactive medicine to PPPM in LUSC.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ADRM1

ADRM1 26S proteasome ubiquitin receptor

AIC

Akaike Information Criterion

AUC

Area under the curve

BP

Biological process

CAR-T

Chimeric antigen receptor T cell

CC

Cellular component

CD274

Programmed cell death 1 ligand 1

CD276

CD276 antigen

CD80

T-lymphocyte activation antigen CD80

CD86

T-lymphocyte activation antigen CD86

CHST7

Carbohydrate sulfotransferase 7

CMTM6

CKLF-like MARVEL transmembrane domain-containing protein 6

CSN5

COP9 signalosome complex subunit 5

CTLA4

Cytotoxic T-lymphocyte protein 4

DEGs

Differentially expressed genes

DTP

Developmental Therapeutics Program

GAP43

Gap junction alpha-1 protein

GO

Gene ontology

GSEA

Gene Set Enrichment Analysis

IL

Interleukin

ITGA8

Integrin alpha-8

KEGG

Kyoto Encyclopedia of Genes and Genomes

LUSC

Lung squamous carcinomas

MAD

Median absolute deviation

MF

Molecular function

MHC

Major Histocompatibility Complex

NCBP1

Nuclear cap-binding protein subunit 1

NCI

National Cancer Institute

NMF

Nonnegative matrix factorization method

OPA1

Dynamin-like 120 kDa protein

PD-1

Programmed cell death protein 1

PDCD1

Programmed cell death protein 1

PDCD1LG2

Programmed cell death 1 ligand 2

PD-L1

Programmed cell death 1 ligand 1

PPI

Protein-protein interaction

PSMA1

Proteasome subunit alpha type-1

PSMA2

Proteasome subunit alpha type-2

PSMA3

Proteasome subunit alpha type-3

PSMA4

Proteasome subunit alpha type-4

PSMA5

Proteasome subunit alpha type-5

PSMA6

Proteasome subunit alpha type-6

PSMA7

Proteasome subunit alpha type-7

PSMB1

Proteasome subunit beta type-1

PSMB10

Proteasome subunit beta type-10

PSMB11

Proteasome subunit beta type-11

PSMB2

Proteasome subunit beta type-2

PSMB3

Proteasome subunit beta type-3

PSMB4

Proteasome subunit beta type-4

PSMB5

Proteasome subunit beta type-5

PSMB6

Proteasome subunit beta type-6

PSMB7

Proteasome subunit beta type-7

PSMB8

Proteasome subunit beta type-8

PSMB9

Proteasome subunit beta type-9

PSMC1

26S proteasome regulatory subunit 4

PSMC2

26S proteasome regulatory subunit 7

PSMC3

26S proteasome regulatory subunit 6A

PSMC4

26S proteasome regulatory subunit 6B

PSMC5

26S proteasome regulatory subunit 8

PSMC6

26S proteasome regulatory subunit 10B

PSMD1

26S proteasome non-ATPase regulatory subunit 1

PSMD11

26S proteasome non-ATPase regulatory subunit 11

PSMD12

26S proteasome non-ATPase regulatory subunit 12

PSMD13

26S proteasome non-ATPase regulatory subunit 13

PSMD14

26S proteasome non-ATPase regulatory subunit 14

PSMD2

26S proteasome non-ATPase regulatory subunit 2

PSMD3

26S proteasome non-ATPase regulatory subunit 3

PSMD4

26S proteasome non-ATPase regulatory subunit 4

PSMD6

26S proteasome non-ATPase regulatory subunit 6

PSMD7

26S proteasome non-ATPase regulatory subunit 7

PSMD8

26S proteasome non-ATPase regulatory subunit 8

PSMD9

26S proteasome non-ATPase regulatory subunit 9

PSME1

Proteasome activator complex subunit 1

PSME2

Proteasome activator complex subunit 2

PSME3

Proteasome activator complex subunit 3

PSME4

Proteasome activator complex subunit 4

ROC

Receiver operating characteristic

Rpn10

26S proteasome non-ATPase regulatory subunit 4

Rpn13

26S proteasome regulatory subunit RPN13

SEM1

26S proteasome complex subunit SEM1

ssGSEA

Single-sample Gene Set Enrichment Analysis

TCA

Tricarboxylic acid

TCGA

The Cancer Genome Atlas

TMB

Tumor mutation burden

TME

Tumor microenvironment

TNF

Tumor necrosis factor

UPS

Ubiquitin-proteasome system

VTCN1

V-set domain-containing T cell activation inhibitor 1

WGCNA

Weighted gene co-expression network analysis

Author contributions

J.Y. analyzed the data and wrote the manuscript. S.Y.O. edited and critically revised the manuscript. J.W., Z.L., X.F. and Z.Y. participated in partial data analysis. S.Z. carried out partial experiments. X.Z. and N.L. conceived the concept, designed the manuscript, coordinated and critically revised the manuscript, and was responsible for its financial support and the corresponding works. All authors approved the final manuscript.

Funding

This work was supported by the Shandong Provincial Taishan Scholar Engineering Project Special Funds (NO.tstp20221143 to X.Z.), the Shandong Provincial Natural Science Foundation (ZR2021MH156 to X.Z.; ZR2022QH112 to N.L.), the Shandong First Medical University Talent Introduction Funds (to X.Z.), and China National Nature Scientific Funds (82203592 to N.L.).

Data availability

All the data used in this study were collected in this article and supplemental materials.

Code availability

All protein and gene accession codes can be available in the Swiss-Prot and Genbank databases.

Declarations

Ethical approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.