Skip to main content
NCBI home page
Discover Oncology logoLink to Discover Oncology
. 2025 Nov 22;16:2261. doi: 10.1007/s12672-025-03994-y

Cellular senescence related gene signature predicts prognosis and immune features in skin cutaneous melanoma

Mengna Li 1,#, Xintao Cen 2,#, Yan Yan 1, Li Li 1, Wei Lai 3,✉
PMCID: PMC12748409  PMID: 41275062

Abstract

Background

Skin cutaneous melanoma (SKCM) is the deadliest skin cancer, with rising global incidence. Cellular senescence plays an essential role in tumorigenesis, progression, and immune modulation in cancer, however, its role in SKCM prognosis and immunotherapy response remains unclear.

Methods

We analyzed 279 senescence-related genes (SRGs) in 469 patients with SKCM from The Cancer Genome Atlas. A cellular senescence-related signature (CSRS) was constructed using univariate and LASSO Cox regression analyses. Kaplan-Meier survival curves and receiver operating characteristic (ROC) analyses were used to evaluate its predictive performance. Consensus clustering based on SRG expression stratified patients into distinct subgroups. External validation was performed using the GSE65904 dataset. We further assessed the association between CSRS, immune cell infiltration, and immunotherapy response. Additionally, immunohistochemistry validated the expression of prognosis-related SRGs and functional assays explored the role of RuvB-like AAA ATPase 2 (RUVBL2) in SKCM cells.

Results

The CSRS effectively stratified patients with SKCM into high- and low-risk groups with significantly different survival outcomes and immune profiles. Moreover, our results suggest that higher levels of cellular senescence may enhance immunosurveillance and promote tumor suppression via a senescence-associated secretory phenotype-dependent mechanism. Based on the expression profiles of 113 SRGs, patients were classified into three distinct clusters, with Cluster 1 associated with the poorest prognosis. Among the identified SRGs, RUVBL2 was markedly upregulated in SKCM cells and its knockdown inhibited cell proliferation.

Conclusions

The CSRS is a robust prognostic and predictive biomarker in SKCM, highlighting the relevance of cellular senescence in shaping the tumor immune microenvironment and informing therapeutic strategies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-03994-y.

Keywords: Skin cutaneous melanoma, Cellular senescence, Prognostic signature, Tumor microenvironment, Immunotherapy

Introduction

Skin cutaneous melanoma (SKCM) is the most aggressive form of skin cancer, and its incidence is rising worldwide. In 2020, an estimated 325,000 individuals were diagnosed, resulting in approximately 57,000 deaths worldwide. Projections suggest new cases may reach 510,000, with 96,000 deaths annually by 2040 if current trends persist [1]. Although the American Joint Committee on Cancer (AJCC) staging system provides a framework for SKCM classification, considerable variability in clinical outcomes remains among patients with similar clinical and histopathological features. Despite advancements in targeted therapies and immunotherapies, disease relapse continues to pose a major therapeutic challenge [2–4].

The tumor microenvironment (TME) is a dynamic ecosystem comprising malignant cells, cancer-associated fibroblasts, endothelial cells, keratinocytes, adipocytes, and diverse immune populations. Over time, tumor-infiltrating immune cells often become dysfunctional, facilitating immune evasion by melanoma cells. The success of immune checkpoint inhibitors (ICIs), including PD-1/PD-L1 and CTLA-4 blockers, highlights the importance of immune regulation in the TME. Notably, combination therapies including nivolumab and ipilimumab have significantly improved survival in metastatic melanoma, with approximately 50% five-year survival [5]. Nevertheless, the relatively low response rates and frequent development of resistance to ICIs limit their broad efficacy [6]. Thus, predictive tools that can accurately assess the TME’s immune status and forecast immunotherapy responses are critically needed.

Cellular senescence, characterized by an irreversible growth arrest in response to stress or damage, plays a complex role in cancer biology [7–9]. Although transient senescence is a potent tumor-suppressive mechanism, the accumulation of senescent cells can foster a pro-tumorigenic environment [10, 11]. This duality is largely mediated by the senescence-associated secretory phenotype (SASP), a diverse array of inflammatory cytokines, chemokines, matrix-remodeling enzymes, and growth factors secreted by senescent cells. The impact of the SASP on the TME varies depending on the cell type undergoing senescence, senescence triggers, and immune pathways involved [12]. SASP factors can stimulate immunosurveillance and reinforce tumor suppression; conversely, they can promote tumor progression by supporting immune evasion, angiogenesis, and tumor growth [13, 14]. In melanoma, a cancer characterized by a high mutational burden and complex immune interactions, the role of cellular senescence in shaping tumor progression and therapeutic responses remains underexplored [15].

In this study, we systematically evaluated the expression patterns of 279 cellular senescence-related genes (SRGs) in SKCM and normal skin tissues. We constructed a cellular senescence-related signature (CSRS) to predict patient prognosis, and assessed its association with clinical features, immune cell infiltration, and immunotherapy response. Additionally, we stratified patients into three distinct molecular clusters based on SRG expression profiles, each with distinct survival outcomes. Finally, we identified RuvB-like AAA ATPase 2 (RUVBL2) as a potential oncogenic SRG in SKCM and demonstrated its role in promoting melanoma cell proliferation. To the best of our knowledge, this is the first study to establish a CSRS linked to immunotherapy outcomes in SKCM, while also uncovering the potential beneficial role of SASP-mediated senescence in promoting antitumor immunity. These insights provide a new basis for the development of individualized treatment strategies for melanoma.

Materials and methods

Data acquisition and processing

Expression data and corresponding clinical follow-up information for 469 patients with SKCM were obtained from The Cancer Genome Atlas (TCGA; https://portal.gdc.cancer.gov). The TCGA-SKCM cohort includes both primary and metastatic melanoma specimens. Normal skin tissue data were extracted from TCGA and the Genotype-Tissue Expression project (GTEx; https://gtexportal.org/home/). A total of 279 SRGs (Supplementary Table 1) were retrieved from the CellAge database [16] (https://genomics.senescence.info/cells/). The GSE65904 dataset (n = 214) was downloaded from the Gene Expression Omnibus [17] (GEO; http://www.ncbi.nlm.nih.gov/geo/). Expression profiles and clinical information from 348 patients with metastatic urothelial cancer treated with anti-PD-L1 therapy (IMvigor210 cohort) were also collected [18] (http://research-pub.gene.com/IMvigor210CoreBiologies/).

Specimens collection

Five SKCM specimens and five normal skin tissue samples were collected from the Department of Dermatology at the Third Affiliated Hospital of Sun Yat-sen University (Guangzhou, China) for experimental validation. All participants provided written informed consent, and the study protocol was approved by the Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University, in accordance with the Declaration of Helsinki and relevant institutional and national guidelines.

Enrichment analysis

The Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses were conducted with the “clusterProfiler” package in R. A protein-protein interaction (PPI) network of differentially expressed SRGs was constructed using Cytoscape (https://cytoscape.org/). Gene set enrichment analysis (GSEA) was performed with Java GSEA software (version 4.1.0) to compare biological processes among clusters.

Consensus clustering

To explore the biological characteristics and prognostic value of SRGs, the “ConsensusClusterPlus” package in R was used to cluster patients with SKCM into distinct subgroups [19]. The optimal number of clusters (k) was determined by analyzing an empirical cumulative distribution function (CDF) plot. Kaplan-Meier survival curves were used to assess the prognostic differences among the clusters.

Construction and validation of the CSRS

Univariate Cox regression analysis was performed to identify the prognostically relevant differentially expressed SRGs (p < 0.001) in TCGA-SKCM cohort (p < 0.001). Subsequently, the least absolute shrinkage and selection operator (LASSO) Cox regression analysis was applied to select prognostic SRGs. The GSE65904 dataset (n = 214) served as the external validation cohort. The patients in TCGA-SKCM and validation cohorts were classified either into the high-risk (risk score ≥ median) or low-risk (risk score < median) groups. Survival curves, risk score distributions, heat maps, and receiver operating characteristic (ROC) curves were generated to evaluate CSRS.

Signature genes analyses

Gene Expression Profiling Interactive Analysis (GEPIA2; http://gepia2.cancer-pku.cn/) [20] and UALCAN (http://ualcan.path.uab.edu) [21] web tools were used to analyze the CSRS genes.

Correlation between immune cell infiltration and CSRS

To analyze the relationship between CSRS and immune cell infiltration, RNA-seq-derived infiltrating immune cell populations were estimated using the TIMER, EPIC, xCELL, CIBERSORT, and quanTIseq algorithms in TIMER2.0 (http://timer.comp-genomics.org/) [22]. The Single Sample Gene Set Enrichment Analysis (ssGSEA) algorithm was used to quantify differences in immune cell infiltration between the high- and low-risk groups using the “GSVA” package in R. Tumor purity, immune score, stromal score, and ESTIMATE score were calculated for each patient with SKCM.

Assessment of CSRS and immunotherapeutic responses

Immunophenoscore (IPS) data for SKCM were downloaded from The Cancer Immunome Atlas (TCIA; https://tcia.at) to predict response to ICIs. Additionally, the IMvigor210 cohort was used to evaluate differences in immunotherapy response between high- and low-risk groups.

Tumor mutational burden (TMB) analysis

Somatic mutation data for patients with SKCM were obtained from the TCGA database. The “maftools” R package was used to calculate and compare TMB between the high- and low-risk groups.

Cell culture and transfections

Normal human epidermal melanocytes and melanoma cell lines (A375 and SK-Mel-5) were obtained from the American Type Culture Collection (ATCC) and cultured under standard conditions as previously described [23]. When cell confluence reached 30–50%, cells were transfected with RUVBL2-targeting siRNAs using the riboFECT™ CP Transfection Kit (RiboBio, China) according to the manufacturer’s instructions. The RUVBL2 siRNA sequences were as follows: si-RUVBL2-1, TCCTGATCATGGCCACCAA; and si-RUVBL2-2, GCGAGAAAGACACGAAGCA.

RNA isolation and quantitative real-time PCR (qRT-PCR)

Total RNA was extracted using the TRIzol reagent (Invitrogen, USA), and reverse transcription was performed using the RevertAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, USA). qRT-PCR was carried out with a SYBR Fast qRT-PCR Master Mix (Kapa Biosystems, USA) on a LightCycler 480 System (Roche Diagnostics, Switzerland). The primer sequences used were as follows: RUVBL2 Forward: 5′-AAGAAGATGTGGAGATGAG-3′; Reverse: 5′-CAGGAAGAGTGAGTAGAC-3′; GAPDH Forward: 5′-CTGGGCTACACTGAGCACC-3′; Reverse: 5′-AAG TGGTCGTTGAGGGCAATG-3′. The results were calculated using the relative quantification (2−ΔΔCt) method.

Cell viability assay

Melanoma cells were seeded into 96-well plates in 100 µL medium per well. To asses cell proliferation, 10 µL CCK-8 reagent (Dojindo, Japan) was added to each well and incubated for 2 h. Absorbance at 450 nm was then measured using a microplate reader (Thermo Fisher Scientific).

5-ethynyl-2’-deoxyuridine (EdU) staining

Cell proliferation was assessed using the EdU Apollo 567 kit (RiboBio, China). Briefly, after incubation with 10 µM EdU for 2 h, melanoma cells were fixed with 4% paraformaldehyde, permeabilized with 0.3% Triton X-100, and stained with Apollo fluorescent dye. DAPI was used to stain the cell nuclei for 10 min. EdU-positive cells were counted using a fluorescence microscope in five random fields.

Immunohistochemistry (IHC) staining and evaluation

Paraffin embedded tissue sections were incubated overnight at 4 °C with primary antibodies against FOXM1 (ab207298, 1:200), NOTCH3 (ab300527, 1:200), and NADPH oxidase 4 (ab133303, 1:100). Sections were then incubated with the appropriate secondary antibody, developed with DAB, and counterstained with hematoxylin. Staining was evaluated by scoring both the percentage of positive cells and the staining intensity. The staining intensity was scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). The proportion of expression was scored as 1 (0–25%), 2 (26–50%), 3 (51–75%), or 4 (76–100%). The score for each sample was calculated by multiplying the signal intensity by the proportion of expression.

Statistical analysis

All statistical analyses were performed using R software (version 4.1.3), GraphPad Prism 8.0, and SPSS Statistics V25.0. Data were presented as mean ± SD of at least three independent experiments, and differences between two groups were compared using Student’s t-test. Survival outcomes were compared using Kaplan-Meier analysis. Statistical significance was set at p < 0.05.

Results

Characterize the expression pattern of SRGs and identification of cellular senescence-related subtypes in SKCM

The overall study design is shown in Supplementary Fig. 1. We first identified 7,896 differentially expressed genes (DEGs) between normal skin samples (n = 558) and SKCM samples (n = 469) using thresholds of FDR < 0.01 and | log2FC) | >1. Among these, 3,857 DEGs were significantly upregulated in SKCM compared to normal skin, whereas 4,039 were downregulated.

Next, we identified 113 differentially expressed SRGs by intersecting the DEGs with 279 SRGs obtained from the CellAge database. Among these, 68 genes were upregulated, and 45 were downregulated (Fig. 1A; Supplementary Fig. 2A; Supplementary Table 2). GO and KEGG pathway enrichment analyses revealed that 113 SRGs were significantly involved in cellular senescence- and cell cycle-related pathways (Supplementary Fig. 2B, 2 C; Supplementary Tables 3 and 4). A PPI network further explored the interactions among the 113 DEGs (Supplementary Fig. 2D).

Fig. 1.

Fig. 1

Open in a new tab

Consensus clustering analysis of 279 SRGs. A Venn diagram showing 113 overlapping genes between DEGs and SRGs. B Consensus clustering matrix at k = 3. (C) CDF curves for k = 2–8. D The relative change in area under the CDF curves for k = 2–8. E PCA plot of the three clusters. F Kaplan–Meier survival analysis for SKCM samples stratified into the three clusters. G Heatmap of SRG expression and the associated clinical parameters across the three clusters. *p < 0.05, **p < 0.01, ***p < 0.001

We performed a consensus clustering analysis on patients with SKCM using the expression patterns of 113 SRGs. An optimal k = 3 was determined from the consensus matrix (Fig. 1B), consensus CDF curves (Fig. 1C), and relative change in the area under the CDF curves (Fig. 1D). Principal component analysis (PCA) confirmed three distinct clusters (Fig. 1E), and prognostic analysis revealed Cluster 1 had the poorest survival compared with Clusters 2 and 3 (Fig. 1F). Moreover, Chi-squared analysis revealed significant differences in age, T stage, and Breslow thickness among the clusters (Fig. 1G).

Development and validation of the CSRS for SKCM

Univariate Cox proportional hazards regression analysis identified 15 of the 113 SRGs significantly associated with overall survival (OS) (p < 0.01; Supplementary Table 5). Among these, ITSN2, CBX7, MAP4K1, ABI3, BCL6, HK3, NDRG1, and NOX4 were protective factors with a hazard ratio HR < 1, whereas PSMB5, MVK, AKT1, NOTCH3, FOXM1, SFN, and RUVBL2 were risk factors HR > 1 (Fig. 2A). Correlations among the 15 potential prognostic DEGs showed that most of these genes were positively correlated (Fig. 2B).

Fig. 2.

Fig. 2

Open in a new tab

Identification of prognostic cellular senescence-related DEGs. A Forest plot of univariate Cox analysis of 15 candidate prognostic genes in patients with SKCM. B Correlation network of the 15 candidate prognostic genes. C LASSO coefficient profiles and cross validation for tuning parameter selection. D Forest plot of multivariate Cox analysis of the 12 signature genes. E The expression levels of the 15 candidate prognostic genes in SKCM versus normal skin tissues. *p < 0.05, **p < 0.01, ***p < 0.001

Afterwards, LASSO Cox regression was applied to the 15 candidate SRGs, and the penalty parameter λ = 12 was selected by the minimum-criteria rule, resulting in a 12-gene prognostic signature (Fig. 2C). High expression of MVK, NOTCH3, FOXM1, SFN, and RUVBL2 cohort was consistently indicative of poor prognosis, while high expression of ITSN2, CBX7, ABI3, BCL6, HK3, NDRG1, and NOX4 correlated with longer OS (Supplementary Fig. 3). Moreover, multivariate Cox regression analysis indicated that NOTCH3 was an independent risk factor and NDRG1 a protective factor in SKCM (Fig. 2D). We then established a risk score formula based on the expression of these 12 SRGs: Risk score = (0.0271 ×expression value of MVK) + (-0.0459 × expression value of ITSN2) + (-0.0749 × expression value of CBX7) + (0.2481 ×expression value of NOTCH3) + (0.1034 × expression value of FOXM1)+ (-0.0861 × expression value of ABI3)+ (0.0823 × expression value of SFN)+ (0.0498 × expression value of RUVBL2)+ (-0.1105 × expression value of BCL6)+ (-0.2307 × expression value of HK3)+ (-0.1952 × expression value of NDRG1)+ (-0.1708 × expression value of NOX4).

The risk score for each patient was calculated and cases were divided into low- and high-risk groups at the median. Analysis of the TCGA mRNA expression levels for the 15 candidates showed that eight genes (PSMB5, MAP4K1, AKT1, FOXM1, ABI3, RUVBL2, HK3, and NOX4) were upregulated, and seven (MVK, ITSN2, CBX7, NOTCH3, SFN, BCL6, and NDRG1) were downregulated (p < 0.001) (Fig. 2E).

The distribution of the CSRS risk scores, survival status, and heatmap showing the expression profiles of the 12 prognosis-related SRGs in the high- and low-risk groups are presented in Fig. 3A. Kaplan-Meier survival analysis demonstrated that the high-risk group had a shorter OS than the low-risk group in the TCGA-SKCM dataset (Fig. 3B, HR = 2.261, 95% CI1.712–2.985, log-rank p < 0.0001). The five-year survival rate in the high-risk group (24.23%) was significantly lower than that of the low-risk group (41.85%). Time-dependent ROC analysis evaluated the sensitivity and specificity of CSRS, yielding area under the curve (AUC) values of 0.68, 0.64, and 0.65 at 2, 3, and 5 years, respectively (Fig. 3C). In addition, significant OS differences were observed in both early stage (HR = 2.691, 95% CI 1.859–3.894, log-rank p < 0.0001) and advanced-stage SKCM (Fig. 3D, HR = 2.5, 95% CI 1.773–3.526, log-rank p < 0.0001).

Fig. 3.

Fig. 3

Open in a new tab

Development of CSRS in TCGA dataset. A The distribution of the risk score, survival status and heatmap of signature genes in TCGA dataset. B Kaplan-Meier curves of overall survival (OS) in total SKCM patients of the TCGA dataset based on risk score. C The receiver operating characteristic (ROC) curves were used to measure the predictive value of the CSRS at 2, 3, and 5 years in TCGA dataset. D Kaplan-Meier curves of OS in patients with early-stage and advanced-stage SKCM based on risk score. E Univariable and multivariable Cox regression analysis of CSRS and OS in TCGA dataset. F The distribution of the risk score, survival status and heatmap of signature genes in GSE65904 dataset. G Kaplan-Meier curves of OS in total SKCM patients of the GSE65904 dataset based on risk score. H The ROC curves were used to measure the predictive value of the CSRS at 2, 3, and 5 years in GSE65904 dataset

Subsequently, univariate and multivariate Cox regression analyses were performed to examine whether the risk score was an independent prognostic variable for SKCM. In univariate analysis, age, T stage, N stage, M stage, TNM stage, Breslow depth, Clark stage, ulceration status, tumor location, and risk score were all correlated with OS. Multivariate Cox regression analyses further confirmed that the risk score was an independent prognostic factor (HR = 2.136, 95% CI 1.534–2.973, p < 0.001) for patients with SKCM (Fig. 3E).

To validate the predictive reliability of the CSRS, we calculated the risk scores for samples in the GSE65904 cohort using the same formula and then classified them into high- and low-risk groups. The risk-score distribution, survival status, and heatmap of the CSRS expression profile in the GSE65904 cohort are shown in Fig. 3F. Patients with high-risk scores had greater mortality than those in the low-risk group (HR = 1.895, 95% CI 1.29–2.783, p = 0.00111; Fig. 3G). The AUC values at two, three, and five years were 0.63, 0.63, and 0.58, respectively (Fig. 3H).

Correlations between the CSRS and clinicopathological factors

We compared clinicopathological factors between the high- and low-risk groups, including age, sex, T stage, N stage, M stage, TNM stage, Breslow depth, Clark stage, ulceration status, and tumor location. As shown in Fig. 4, patients with Breslow tumor thickness > 3 mm had significantly higher risk scores than those with ≤ 3 mm (p = 0.0048). Additionally, patients with lymphatic metastasis exhibited markedly higher risk scores than patients with primary tumors (p < 0.001).

Fig. 4.

Fig. 4

Open in a new tab

The correlation between CSRS and patients’ clinicopathological parameters, including age, gender, T stage, N stage, M stage, TNM stage, breslow depth, clark stage, location and ulceration

Patients were stratified by various clinical variables, including age (≤ 60 vs. >60), gender (female vs. male), T stage (T0–2 vs. T3–4), N stage (N0 vs. N1–3), M stage (M0 vs. M1), TNM stage (stage I–II vs. stage III–IV), Breslow depth (> 3 mm vs. ≤3 mm), Clark stage (I–III vs. IV–V), ulceration status (yes vs. no), and tumor location (primary tumor vs. metastasis). Kaplan–Meier survival analysis demonstrated that across all subgroups, patients in the high-risk group consistently had poorer survival than those in the low-risk group (Supplementary Fig. 4A-J). These findings indicate that the CSRS is a robust and reliable predictor of prognosis in patients with SKCM.

Analysis of signature genes

We analyzed the expression of the 12 signature genes using the GEPIA2 database. We found that MVK, ITSN2, CBX7, NOTCH3, SFN, BCL6, and NDRG1 were significantly downregulated in SKCM tissues, whereas FOXM1, RUVBL2, and HK3 were substantially upregulated compared to normal skin (p < 0.05; Supplementary Fig. 5A). Pan-cancer analysis revealed that FOXM1, SFN, and RUVBL2 were risk factors across multiple tumor types, including adrenocortical carcinoma, brain lower-grade glioma, liver hepatocellular carcinoma, and lung adenocarcinoma (Supplementary Fig. 5B). We then analyzed the relationship between promoter methylation and gene expression, finding significantly lower methylation of the FOXM1 promoter in SKCM samples versus normal skin, whereas BCL6, NDRG1, and HK3 exhibited higher promoter methylation in SKCM (Supplementary Fig. 5C). In addition, we retrieved and visualized the 3D protein structures of a 12 signature genes from the PDB databank (Supplementary Fig. 6).

Biological processes analysis of signature genes

To explore the biological differences between high- and low-risk groups, we identified 4,788 DEGs (FDR < 0.01, | log2FC | ≥ 1), including 1,961 downregulated and 2,827 upregulated genes in the high-risk group (Supplementary Table 6). GO and KEGG enrichment analyses indicated that these DEGs were predominantly involved in immune-related processes, such as immune response, T cell activation, and B cell activation (Supplementary Figs. 7 A, 7B). Functional annotation using GSEA showed that HALLMARK gene sets enriched in the high-risk group were mainly involved in tumor-related pathways, including oxidative phosphorylation, IFN-γ response, MYC targets, IFN-α response and E2F targets, which are closely related to malignant proliferation and immune microenvironment (Supplementary Fig. 7C). These results suggest that the CSRS is closely associated with immune regulation and tumor proliferation pathways.

CSRS is associated with alterations in SASP

Senescent cells secrete a variety of factors, collectively known as SASP, which can modulate the TME. We examined the correlation between the CSRS and SASP factors and found that key SASPs—including interleukins (IL-1 A, IL-1B, IL-6, IL-7, IL13 and IL-15), soluble or shed receptors or ligands (FAS, ICAM1,ICAM3,IL6ST, PLAUR, TNFRSF1A, TNFRSF1B, TNFRSF10C and TNFRSF11B), chemokines (CCL1, CCL3, CCL8, CCL13, CCL25, CCL26, CXCL5, and CXCL11), growth factors and regulators (ANG, FGF2, FGF7, HGF, IGFBP2, IGFBP3, IGFBP7, and VEGFA), and proteases (CTSB, MMP12 and SERPINE1) were all significantly lower in the high-risk group (Fig. 5A). Given that SASP factors facilitate the immune clearance of damaged cells, these findings suggest that patients with high CSRS scores may experience impaired immune surveillance mediated by reduced SASP activity.

Fig. 5.

Fig. 5

Open in a new tab

Association between CSRS and TME. A Differential expression of SASP components, including interleukins, soluble/shed receptors or ligands, chemokines, growth factors and regulators, and proteases, between high- and low-risk groups. B Correlation between CSRS risk scores and estimated abundances of various immune cell types using TIMER, CIBERSORT-ABS, QUANTISEQ, xCELL and EPIC. C Heatmap of tumor purity and TME scores in high- versus low-risk groups. D Comparison of the proportions of 16 immune cell populations between high- and low-risk groups. E Comparison of enrichment scores for 13 immune-related pathways between high- and low-risk groups. F Distributions of the ImmuneScore, ESTIMATEScore, StromalScore, and tumor purity across risk groups. *p < 0.05, **p < 0.01, ***p < 0.001

Association between CSRS and the tumor immune microenvironment (TIME)

We analyzed differences in immune cell infiltration between the high- and low-risk groups using TIMER, EPIC, xCELL, CIBERSORT, and quanTIseq in TIMER2.0. Compared with the low-risk group, the high-risk group exhibited significantly reduced infiltration of B cells and CD8 + T cells, while cancer-associated fibroblasts (CAFs) increased (Fig. 5B). We then assessed 16 immune cells types and 13 immune-related pathways. ssGSEA revealed that the high-risk group displayed lower immune cell infiltration and suppressed immune-related pathways (Fig. 5C-E).

The ESTIMATE algorithm confirmed that patients in the high-risk group had lower ImmuneScore, StromalScore, and ESTIMATEScore but higher tumor purity than those in the low-risk group (Fig. 5F). Survival analysis showed that lower immune scores, lower stromal scores, and higher tumor purity were associated with worse overall survival (Supplementary Fig. 8). Taken together, these results suggest that CSRS is significantly correlated with the immunosuppressive TIME.

Association of CSRS with immunotherapy efficacy

Given the association between CSRS and immune infiltration, we compared expression patterns of immune checkpoint genes between the high- and low-risk groups and found that most were significantly upregulated in the low-risk group (Fig. 6A). In the GSE65904 cohort, PD-L1, PD-1, CTLA4, LAG3, and TIM3 were overexpressed in the low-risk group (Fig. 6B) and PD-1 and CTLA-4 levels were negatively correlated with CSRS scores (Fig. 6C). Survival analysis stratified by CSRS and checkpoint expression revealed that across subgroups stratified by PD-L1,PD1, and CTLA4 expression, low-risk patients had longer OS than high-risk patients in the TCGA cohort (Fig. 6D). IPS analysis indicated that low-risk patients were more likely to benefit from anti-CTLA4, anti-PD1, and combined anti-PD1–CTLA4 therapies (Fig. 6E). Validation in the IMvigor210 cohort confirmed that low CSRS scores were associated with better overall survival, regardless of PD-L1 expression (Fig. 6F and G). Moreover, high CSRS scores predicted poor outcomes independent of TMB status in both the TCGA and IMvigor210 datasets (Fig. 6H and I). Together, these findings suggest that the CSRS may serve as a promising biomarker for predicting immunotherapy response.

Fig. 6.

Fig. 6

Open in a new tab

Analysis of immunotherapeutic responses between risk groups. A Expression levels of immune checkpoint genes in high- versus low-risk patients (TCGA cohort). B Expression of PD-L1, PD-1, CTLA-4, LAG3 and TIM3 in high- versus low-risk patients (GSE65904 cohort). C Correlation between CSRS risk score and PD-1 or CTLA4 expression. D Kaplan-Meier overall survival curves for four patient groups defined by CSRS and PD-L1/PD-1/CTLA-4 expression (TCGA cohort). E Distribution of IPS in high- versus low-risk patients (TCGA cohort). F Kaplan-Meier survival curves for high- and low-risk patients (IMvigor210 cohort). G Kaplan-Meier survival curves for four subgroups defined by CSRS and PD-L1 expression (IMvigor210 cohort). H TMB distributions in high- versus low-risk patients. I Kaplan-Meier overall survival curves for four subgroups defined by CSRS and TMB. *p < 0.05, **p < 0.01, ***p < 0.001

Validation of signature gene expressions in SKCM tissues

To validate the expression of CSRS genes in SKCM tissues, we selected FOXM1, NOX4, and NOTCH3 for IHC analysis, as they ranked among the top genes in our CSRS and exhibited both strong prognostic relevance and consistent differential expression between high- and low-risk groups. The results showed that FOXM1 and NOX4 were significantly upregulated in SKCM tissues, whereas NOTCH3 expression was markedly downregulated (Fig. 7A).

Fig. 7.

Fig. 7

Open in a new tab

RUVBL2 promotes tumorigenesis in SKCM. A Representative IHC images showing FOXM1, NOX4, and NOTCH3 expression in SKCM versus normal skin tissues. B Relative mRNA levels in normal human epidermal melanocytes (MC) and SKCM cell lines. C RUVBL2 mRNA expression in A375 and SK-Mel-5 cells following transfection with si-RUVBL2 versus control. D CCK-8 cell viability assay in A375 and SK-Mel-5 cells after si-RUVBL2 knockdown. E EdU incorporation assay showing proliferation of A375 and SK-Mel-5 cells post–si-RUVBL2 transfection. *p < 0.05, **p < 0.01, ***p < 0.001

RUVBL2 promotes tumorigenesis in SKCM

RUVBL2 has been previously implicated in oncogenic processes in multiple cancer types but has not yet been reported in melanoma. In this study, we explored the biological role of RUVBL2 in SKCM cells. First, we verified that RUVBL2 mRNA levels were significantly higher in SK-Mel-5 and A375 melanoma cells than in normal human melanocytes (Fig. 7B). We used two individual siRNAs to knockdown RUVBL2 mRNA expression (Fig. 7C). The CCK-8 assay demonstrated that RUVBL2 knockdown significantly inhibited cell proliferation at 48, 72, and 96 h (Fig. 7D). EdU staining confirmed that RUVBL2 silencing suppressed the proliferation of both SK-Mel-5 and A375 cells (Fig. 7E).

Discussion

Cutaneous melanoma, which originates from the malignant transformation of melanocytes, is the most lethal type of skin cancer. However, the current therapeutic landscape remains limited, as most patients with SKCM receive similar treatment regimens due to the lack of reliable and effective prognostic tools. Therefore, identifying accurate biomarkers for prognostic signature is critical for improving clinical decision-making. In the present study, we analyzed mRNA expression patterns of 279 SRGs in SKCM and successfully developed a CSRS, which was validated in the GSE65904 cohort. The CSRS not only served as a prognostic indicator but also correlated significantly with TIME and immunotherapy response, providing novel insights into the role of cellular senescence in SKCM.

The CSRS model comprised 12 SRGs, including MVK, NOTCH3, FOXM1, SFN, and RUVBL2, as risk factors and ITSN2, CBX7, ABI3, BCL6, HK3, NDRG1, and NOX4 as protective factors. Several of these genes are implicated in tumorigenesis and immune regulation. RUVBL2, a highly conserved ATPase of the AAA + superfamily, regulates chromatin remodeling, DNA repair, and cell cycle progression [24–26]. To date, its role in SKCM development and progression has not been described. Here, we demonstrated that RUVBL2 expression is significantly upregulated in SKCM tissues and its knockdown suppresses proliferation in A375 and SK-Mel-5 melanoma cell lines, suggesting that RUVBL2 may contribute to SKCM cell proliferation and progression. However, additional in vivo validation is required to substantiate these findings. Forkhead Box M1 (FOXM1) overexpression has been associated with melanoma progression and suppression of senescence, whereas NADPH oxidase 4 (NOX4) upregulation has been reported in melanoma cell lines [27–31]. NOTCH3, conversely, was significantly downregulated, which is consistent with its known tumor-suppressive effects [32, 33]. In this study, IHC staining revealed that FOXM1 and NADPH oxidase 4 were significantly upregulated, whereas NOTCH3 expression was reduced in SKCM tissues. Hexokinase 3 (HK3) may function as an oncogene in various cancers [34]. Sulforaphane (SFN) induces cell differentiation and melanogenesis and inhibits melanoma cell proliferation [35]. The transcription factor B-cell lymphoma 6 (Bcl6) is essential for maintaining Treg lineage stability in the TME [36, 37]. N-Myc downstream regulated gene 1 (NDRG1), an oncogenic signaling disruptor involved in epithelial cell differentiation, plays a key role in multiple cancers [38, 39]. Taken together, these studies support the rationale and validity of the CSRS in SKCM tumorigenesis and progression. Further studies are needed to elucidate the roles of these SRGs in SKCM. With the growing awareness of the importance of the TME in cancer biology, cancer research has shifted from a cancer-centric model to one considering the TME as a whole [40]. This paradigm shift is particularly evident in the field of immunotherapy, where the dynamic interactions between tumor cells and immune components within the TME critically determine treatment efficacy [41, 42].

Our results also showed that the CSRS is significantly associated with the composition of the TIME. Specifically, patients with low CSRS scores exhibited enrichment of B and CD8 + T cells, which are key mediators of anti-tumor immunity, whereas high CSRS scores correlated with an immunosuppressive microenvironment characterized by increased CAF infiltration and reduced immune cell activation. Previous prognostic models in SKCM have primarily focused on single dimensions—such as immune infiltration or mutational burden—without integrating senescence biology. By contrast, our CSRS combines cellular SRGs with prognosis, TIME composition and immunotherapy response, filling a gap in existing predictive tools.

Previous studies have emphasized the importance of immune checkpoint genes in modulating immune infiltration [43]. We compared the expression patterns of CSRS genes between high- and low-risk patients with SKCM and found that low-risk patients displayed higher expression of immune checkpoint genes, which may indicate a greater likelihood of benefit from ICIs. These observations are associative and require prospective validation; therefore, while the CSRS shows potential as a prognostic tool and as an indicator of possible immunotherapy responsiveness, further studies are needed to confirm its predictive value.

Our study also explored the role of the SASP in shaping tumor immunity. SASP factors, including cytokines, chemokines, and matrix metalloproteinases (MMPs) have complex effects depending on tumor context. Some of these secreted molecules may exert autocrine effects that enforce cellular senescence, whereas others may exert non-cell-autonomous effects that favor tumorigenesis in nearby non-senescent cells. Thus, the SASP can have both beneficial and detrimental effects depending on the senescence signal, tissue context, and specific secreted factors. In our study, lower levels of SASP factors, and reduced immune cell infiltration were observed in the high-risk group, suggesting that senescence-associated inflammation may promote immunosurveillance and tumor suppression in SKCM. Notably, some upregulated SASP factors in high-risk group, including MMP3, MMP14, and PLAU, may not only influence melanoma metastasis by extracellular matrix degradation, but also through regulation of genes involved in several pro-tumorigenic functions, such as tumor cell growth and motility. Consequently, we identified distinct SASP that affect tumorigenesis and immune modulation as potential mechanisms underlying immune escape and tumor progression in SKCM. Collectively, our results imply that high levels of cellular senescence stimulate immunosurveillance mechanisms and potentiate tumor-suppression in SKCM via SASP-dependent mechanisms.

A major strength of our study lies in the comprehensive integration of cellular senescence biology with TIME profiling across multiple large cohorts, including TCGA and the independent GSE65904 validation cohort. By combining bioinformatic analyses with experimental validation, we further demonstrated the functional role of RUVBL2 in melanoma cell proliferation. Nevertheless, several limitations should be acknowledged. First, the CSRS model was derived from retrospective bulk transcriptomic data, which represent mixed signals from tumor, stromal, and immune compartments. This inherent heterogeneity constrains cell type–specific interpretation and may obscure the mechanistic contributions of individual SRGs. Although we incorporated complementary approaches, such as immune infiltration estimation and validation in independent datasets, which consistently confirmed the robustness and prognostic value of the CSRS, additional studies are warranted. In particular, integrating single-cell RNA-seq data and employing patient-derived or in vivo models would enable a more precise dissection of the roles of SRGs in melanoma progression and therapeutic response. Future work using in vitro and in vivo systems will be critical to clarify the underlying mechanisms and to strengthen the link between the CSRS, cellular senescence, and immune regulation in SKCM.

Conclusions

In summary, we established and validated a CSRS that can be applied for prognostic prediction in SKCM. Our findings also uncover novel insights into the interplay between SRGs and the TIME in SKCM. These discoveries may provide potential molecular targets for clinical diagnosis and therapeutic intervention.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1. (3.5MB, tiff)
Supplementary Material 2. (123.5KB, pdf)
Supplementary Material 3. (10.2MB, tif)
Supplementary Material 4. (17.2MB, tif)
Supplementary Material 5. (27.6MB, tif)
Supplementary Material 6. (13MB, pdf)
Supplementary Material 7. (6.8MB, tif)
Supplementary Material 8. (8.2MB, tif)
Supplementary Material 9. (1.2MB, xlsx)

Acknowledgements

We acknowledge the contributions from TCGA and GEO databases.This study was supported by the National Natural Science Foundation of China (Nos. 82203901, 82373497).

Author contributions

LM designed the study, performed the data analysis and drafted the manuscript. LM and CX performed the in vitro experiments. LL and YY provided statistical advice. LW supervised the acquisition of the data. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (Nos. 82203901, 82373497).

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The use of human tissue samples was approved by the Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University. All procedures were performed in accordance with the ethical standards of the Declaration of Helsinki.

Informed consent

Written informed consent was obtained from all participants or their legal guardians prior to sample collection.

Consent for publication

Written informed consent for publication was obtained from all participants.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

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

Mengna Li and Xintao Cen are contributed equally to this work.

References

  • 1.Arnold M, Singh D, Laversanne M, Vignat J, Vaccarella S, Meheus F, Cust AE, de Vries E, Whiteman DC, Bray F. Global burden of cutaneous melanoma in 2020 and projections to 2040. JAMA Dermatol. 2022;158(5):495–503. 10.1001/jamadermatol.2022.0160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Long GV, Swetter SM, Menzies AM, Gershenwald JE, Scolyer RA. Cutan Melanoma Lancet. 2023;402(10400):485–502. 10.1016/S0140-6736(23)00821-8. [DOI] [PubMed] [Google Scholar]
  • 3.Perez M, Chakraborty A, Lau LS, Mohammed NBB, Dimitroff CJ. Melanoma-associated glycosyltransferase GCNT2 as an emerging biomarker and therapeutic target. Br J Dermatol. 2021;185(2):294–301. 10.1111/bjd.19891. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Curti BD, Faries MB. Recent advances in the treatment of melanoma. N Engl J Med. 2021;384(23):2229–40. 10.1056/NEJMra2034861. [DOI] [PubMed] [Google Scholar]
  • 5.Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Rutkowski P, Lao CD, Cowey CL, Schadendorf D, Wagstaff J, Dummer R, Ferrucci PF, Smylie M, Hogg D, Hill A, Márquez-Rodas I, Haanen J, Guidoboni M, Maio M, Schöffski P, Carlino MS, Lebbé C, McArthur G, Ascierto PA, Daniels GA, Long GV, Bastholt L, Rizzo JI, Balogh A, Moshyk A, Hodi FS, Wolchok JD. Five-year survival with combined nivolumab and ipilimumab in advanced melanoma. N Engl J Med. 2019;381(16):1535–46. 10.1056/NEJMoa1910836. [DOI] [PubMed] [Google Scholar]
  • 6.Brenner E, Röcken M. A commotion in the skin: developing melanoma immunotherapies. J Invest Dermatol. 2022;142(8):2055–60. 10.1016/j.jid.2022.01.025. [DOI] [PubMed] [Google Scholar]
  • 7.Zhang L, Pitcher LE, Yousefzadeh MJ, Niedernhofer LJ, Robbins PD, Zhu Y. Cellular senescence: a key therapeutic target in aging and diseases. J Clin Invest. 2022;132(15):e158450. 10.1172/JCI158450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Li MN, Li L, Zhang XF, Zhao HJ, Wei M, Zhai WY, Wang BX, Yan Y. LncRNA RP11-670E13.6, interacted with hnRNPH, delays cellular senescence by sponging microRNA-663a in UVB damaged dermal fibroblasts. Aging. 2019;11(16):5992–6013. 10.18632/aging.102159. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Li MN, Li L, Zhang XF, Yan Y, Wang BX. LncRNA RP11-670E13.6 regulates cell cycle progression in UVB damaged human dermal fibroblasts. Photochem Photobiol. 2018;94(3):589–97. 10.1111/php.12858. [DOI] [PubMed] [Google Scholar]
  • 10.D’Ambrosio M, Gil J. Reshaping of the tumor microenvironment by cellular senescence: an opportunity for senotherapies. Dev Cell. 2023;58(12):1007–21. 10.1016/j.devcel.2023.05.010. [DOI] [PubMed] [Google Scholar]
  • 11.Marin I, Boix O, Garcia-Garijo A, Sirois I, Caballe A, Zarzuela E, Ruano I, Attolini CS, Prats N, López-Domínguez JA, Kovatcheva M, Garralda E, Muñoz J, Caron E, Abad M, Gros A, Pietrocola F, Serrano M. Cellular senescence is Immunogenic and promotes antitumor immunity. Cancer Discov. 2023;13(2):410–31. 10.1158/2159-8290.CD-22-0523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Takasugi M, Yoshida Y, Ohtani N. Cellular senescence and the tumour microenvironment. Mol Oncol. 2022;16(18):3333–51. 10.1002/1878-0261.13268. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Takasugi M, Yoshida Y, Hara E, Ohtani N. The role of cellular senescence and SASP in tumour microenvironment. FEBS J. 2023;290(5):1348–61. 10.1111/febs.16381. [DOI] [PubMed] [Google Scholar]
  • 14.Birch J, Gil J. Senescence and the SASP: many therapeutic avenues. Genes Dev. 2020;34(23–24):1565–76. 10.1101/gad.343129.120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Chen HA, Ho YJ, Mezzadra R, Adrover JM, Smolkin R, Zhu C, Woess K, Bernstein N, Schmitt G, Fong L, Luan W, Wuest A, Tian S, Li X, Broderick C, Hendrickson RC, Egeblad M, Chen Z, Alonso-Curbelo D, Lowe SW. Senescence rewires microenvironment sensing to facilitate antitumor immunity. Cancer Discov. 2023;13(2):432–53. 10.1158/2159-8290.CD-22-0528. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Avelar RA, Ortega JG, Tacutu R, Tyler EJ, Bennett D, Binetti P, Budovsky A, Chatsirisupachai K, Johnson E, Murray A, Shields S, Tejada-Martinez D, Thornton D, Fraifeld VE, de Bishop CL. Magalhães JP. A multidimensional systems biology analysis of cellular senescence in aging and disease. Genome Biol. 2020;91. 10.1186/s13059-020-01990-9. [DOI] [PMC free article] [PubMed]
  • 17.Cabrita R, Lauss M, Sanna A, Donia M, Skaarup Larsen M, Mitra S, Johansson I, Phung B, Harbst K, Vallon-Christersson J, van Schoiack A, Lövgren K, Warren S, Jirström K, Olsson H, Pietras K, Ingvar C, Isaksson K, Schadendorf D, Schmidt H, Bastholt L, Carneiro A, Wargo JA, Svane IM, Jönsson G. Tertiary lymphoid structures improve immunotherapy and survival in melanoma. Nature. 2020;577(7791):561–5. 10.1038/s41586-019-1914-8. [DOI] [PubMed] [Google Scholar]
  • 18.Mariathasan S, Turley SJ, Nickles D, Castiglioni A, Yuen K, Wang Y, Kadel EE III, Koeppen H, Astarita JL, Cubas R, Jhunjhunwala S, Banchereau R, Yang Y, Guan Y, Chalouni C, Ziai J, Şenbabaoğlu Y, Santoro S, Sheinson D, Hung J, Giltnane JM, Pierce AA, Mesh K, Lianoglou S, Riegler J, Carano RAD, Eriksson P, Höglund M, Somarriba L, Halligan DL, van der Heijden MS, Loriot Y, Rosenberg JE, Fong L, Mellman I, Chen DS, Green M, Derleth C, Fine GD, Hegde PS, Bourgon R, Powles T. TGFβ attenuates tumour response to PD-L1 Blockade by contributing to exclusion of T cells. Nature. 2018;554(7693):544–8. 10.1038/nature25501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Wilkerson MD, Hayes DN, ConsensusClusterPlus:. A class discovery tool with confidence assessments and item tracking. Bioinformatics. 2010;26(12):1572–3. 10.1093/bioinformatics/btq170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Tang Z, Kang B, Li C, Chen T, Zhang Z. Gepia2: an enhanced web server for Large-Scale expression profiling and interactive analysis. Nucleic Acids Res. 2019;47(W1):W556–60. 10.1093/nar/gkz430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Chandrashekar DS, Bashel B, Balasubramanya SAH, Creighton CJ, Ponce-Rodriguez I, Chakravarthi BVSK, Varambally S. UALCAN: A portal for facilitating tumor subgroup gene expression and survival analyses. Neoplasia. 2017;19(8):649–58. 10.1016/j.neo.2017.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Li T, Fu J, Zeng Z, Cohen D, Li J, Chen Q, Li B, Liu XS. Timer2.0 for analysis of Tumor-Infiltrating immune cells. Nucleic Acids Res. 2020;48(W1):W509–14. 10.1093/nar/gkaa407. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Cen X, Li M, Yao A, Zheng Y, Lai W. Immune infiltration and clinical significance analyses of the cancer-associated fibroblast-related signature in skin cutaneous melanoma. J Gene Med. 2024;26(1):e3614. 10.1002/jgm.3614. [DOI] [PubMed] [Google Scholar]
  • 24.Dauden MI, López-Perrote A, Llorca O. RUVBL1-RUVBL2 AAA-ATPase: a versatile scaffold for multiple complexes and functions. Curr Opin Struct Biol. 2021;67:78–85. 10.1016/j.sbi.2020.08.010. [DOI] [PubMed] [Google Scholar]
  • 25.Nano N, Ugwu F, Seraphim TV, Li T, Azer G, Isaac M, Prakesch M, Barbosa LRS, Ramos CHI, Datti A, Houry WA. Sorafenib as an inhibitor of RUVBL2. Biomolecules. 2020;10(4):605. 10.3390/biom10040605. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Wang H, Li B, Zuo L, Wang B, Yan Y, Tian K, Zhou R, Wang C, Chen X, Jiang Y, Zheng H, Qin F, Zhang B, Yu Y, Liu CP, Xu Y, Gao J, Qi Z, Deng W, Ji X. The transcriptional coactivator RUVBL2 regulates pol II clustering with diverse transcription factors. Nat Commun. 2022;13(1):5703. 10.1038/s41467-022-33433-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Khan MA, Khan P, Ahmad A, Fatima M, Nasser MW. FOXM1: A small Fox that makes more tracks for cancer progression and metastasis. Semin Cancer Biol. 2023;92:1–15. 10.1016/j.semcancer.2023.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Raghuwanshi S, Gartel AL. Small-molecule inhibitors targeting FOXM1: current challenges and future perspectives in cancer treatments. Biochim Biophys Acta Rev Cancer. 2023;1878(6):189015. 10.1016/j.bbcan.2023.189015. [DOI] [PubMed] [Google Scholar]
  • 29.Gong S, Wang S, Shao M. NADPH oxidase 4: A potential therapeutic target of malignancy. Front Cell Dev Biol. 2022;10:884412. 10.3389/fcell.2022.884412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Szanto I. NADPH oxidase 4 (NOX4) in cancer: linking redox signals to oncogenic metabolic adaptation. Int J Mol Sci. 2022;23(5):2702. 10.3390/ijms23052702. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Meitzler JL, Makhlouf HR, Antony S, Wu Y, Butcher D, Jiang G, Juhasz A, Lu J, Dahan I, Jansen-Dürr P, Pircher H, Shah AM, Roy K, Doroshow JH. Decoding NADPH oxidase 4 expression in human tumors. Redox Biol. 2017;13:182–95. 10.1016/j.redox.2017.05.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Cui H, Kong Y, Xu M, Zhang H. Notch3 functions as a tumor suppressor by controlling cellular senescence. Cancer Res. 2013;73(11):3451–9. 10.1158/0008-5472.CAN-12-3902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Aburjania Z, Jang S, Whitt J, Jaskula-Stzul R, Chen H, Rose JB. The role of Notch3 in cancer. Oncologist. 2018;23(8):900–11. 10.1634/theoncologist.2017-0677. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Seiler K, Humbert M, Minder P, Mashimo I, Schläfli AM, Krauer D, Federzoni EA, Vu B, Moresco JJ, Yates JR 3rd, Sadowski MC, Radpour R, Kaufmann T, Sarry JE, Dengjel J, Tschan MP, Torbett BE. Hexokinase 3 enhances myeloid cell survival via non-glycolytic functions. Cell Death Dis. 2022;13(5):448. 10.1038/s41419-022-04891-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Eom YS, Shah FH, Kim SJ. Sulforaphane induces cell differentiation, melanogenesis and also inhibit the proliferation of melanoma cells. Eur J Pharmacol. 2022;921:174894. 10.1016/j.ejphar.2022.174894. [DOI] [PubMed] [Google Scholar]
  • 36.Czerwinska P, Rucinski M, Wlodarczyk N, Jaworska A, Grzadzielewska I, Gryska K, Galus L, Mackiewicz J, Mackiewicz A. Therapeutic melanoma vaccine with cancer stem cell phenotype represses exhaustion and maintains antigen-specific T cell stemness by up-regulating BCL6. Oncoimmunology. 2020;9(1):1710063. 10.1080/2162402X.2019.1710063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Li Y, Wang Z, Lin H, Wang L, Chen X, Liu Q, Zuo Q, Hu J, Wang H, Guo J, Xie L, Tang J, Li Z, Hu L, Xu L, Zhou X, Ye L, Huang Q, Xu L. Bcl6 preserves the suppressive function of regulatory T cells during tumorigenesis. Front Immunol. 2020;11:806. 10.3389/fimmu.2020.00806. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Joshi V, Lakhani SR, McCart Reed AE. NDRG1 in cancer: A Suppressor, Promoter, or both? Cancers (Basel). 2022;14(23):5739. 10.3390/cancers14235739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Chekmarev J, Azad MG, Richardson DR. The oncogenic signaling Disruptor, NDRG1: molecular and cellular mechanisms of activity. Cells. 2021;10(9):2382. 10.3390/cells10092382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Elhanani O, Ben-Uri R, Keren L. Spatial profiling technologies illuminate the tumor microenvironment. Cancer Cell. 2023;41(3):404–20. 10.1016/j.ccell.2023.01.010. [DOI] [PubMed] [Google Scholar]
  • 41.Zhang P, Wen B, Gong J, Liu Z, Zhang M, Zhou G, Zhang L, Zhang Z. Clinical prognostication and immunotherapy response prediction in esophageal squamous cell carcinoma using the DNA damage repair-associated signature. Environ Toxicol. 2024;39(5):2803–16. [DOI] [PubMed] [Google Scholar]
  • 42.Zhang P, Zhang M, Liu J, Zhou Z, Zhang L, Luo P, Zhang Z. Mitochondrial pathway signature (MitoPS) predicts immunotherapy response and reveals NDUFB10 as a key immune regulator in lung adenocarcinoma. J Immunother Cancer. 2025;13(7):e012069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Tang T, Huang X, Zhang G, Hong Z, Bai X, Liang T. Advantages of targeting the tumor immune microenvironment over blocking immune checkpoint in cancer immunotherapy. Signal Transduct Target Ther. 2021;6(1):72. 10.1038/s41392-020-00449-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (3.5MB, tiff)
Supplementary Material 2. (123.5KB, pdf)
Supplementary Material 3. (10.2MB, tif)
Supplementary Material 4. (17.2MB, tif)
Supplementary Material 5. (27.6MB, tif)
Supplementary Material 6. (13MB, pdf)
Supplementary Material 7. (6.8MB, tif)
Supplementary Material 8. (8.2MB, tif)
Supplementary Material 9. (1.2MB, xlsx)

Data Availability Statement

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Articles from Discover Oncology are provided here courtesy of Springer

RESOURCES