USP5 silencing inhibits the proliferation of bladder cancer cells and induces cell apoptosis and ferroptosis by destabilizing COL14A1 expression

Introduction

Bladder cancer (BC) represents the most common type of urinary system cancer and the most prevalent malignancy within the genitourinary system. Latest global cancer statistics indicate that in 2022, there were over 610,000 new cases and 220,000 deaths, reflecting a substantial disease burden (1). In clinical practice, the construction of multimodal comprehensive treatment regimens, including radical surgery, systemic chemotherapy, and radiotherapy, is of paramount importance for improving patient survival rates (2). However, despite early and aggressive interventions, BC is characterized by high risks of recurrence and progression; at least 50% of patients ultimately experience recurrence, resulting in a poor overall prognosis (3). Therefore, there is an urgent need to conduct in-depth research to elucidate the potential molecular mechanisms underlying the tumorigenesis and progression of BC, thereby providing a scientific basis for precise clinical management of BC.

Ferroptosis differs significantly from traditional necrosis or apoptosis in both morphological and biochemical mechanisms, with a core feature of high dependency on iron catalysis (4). The molecular basis of this process lies primarily in the disruption of the balance between the generation and degradation of intracellular reactive oxygen species (ROS), leading to a significant reduction in cellular antioxidant defense capabilities. This subsequently triggers the excessive accumulation of lipid peroxides and plasma membrane rupture, ultimately inducing ferroptosis (5,6). Comprehensive studies have indicated that ferroptosis plays a crucial regulatory role in tumor occurrence, development, and therapeutic response (7,8). Currently, extensive research has deeply explored the close relationship between ferroptosis-related genes and the pathological mechanisms as well as the prognosis of BC, identifying that certain key genes act as inhibitors of ferroptosis in BC cells and can significantly promote malignant cancer progression. For instance, phosphoglycerate dehydrogenase (PHGDH) inhibited ferroptosis by upregulating solute carrier family 7 member 11 (SLC7A11) expression, thereby driving the malignant evolution of BC (9); conversely, fibronectin leucine rich transmembrane protein 2 (FLRT2) has been confirmed to suppress BC progression by inducing ferroptosis (10). Therefore, an in-depth analysis of the molecular regulatory mechanisms associated with ferroptosis in BC not only contributes to elucidating its etiological basis but also provides a significant theoretical foundation for the development of novel therapeutic strategies.

Deubiquitinating enzymes (DUBs) represent a vital category of proteases that serve a pivotal function in preserving protein stability and modulating cellular homeostasis through the specific cleavage of ubiquitin chains from substrate proteins (11). Such precise deubiquitinating activity is indispensable for a multitude of cellular processes, consequently exerting a broad impact on a range of physiological and pathological states, such as cancer, neurodegenerative disorders, and metabolic syndromes (12-15). Ubiquitin specific peptidase 5 (USP5) stands out as a significant member of the DUB family and is capable of specifically recognizing and removing ubiquitin molecules from the proximal end of unanchored polyubiquitin chains (16). Accumulating research evidence indicates that USP5 exerts a significant oncogenic role in various malignancies, such as lung cancer (17), breast cancer (18), and liver cancer (19). Of particular note, recent studies have further revealed that USP5 similarly contributes to the malignant progression of BC (20).

Collagen, as the most abundant structural protein in the extracellular matrix (ECM), is not only essential for maintaining tissue structural integrity but also plays a core role in supporting key physiological functions such as cell proliferation and differentiation (21). The collagen type XIV alpha 1 chain (COL14A1) gene is located on chromosome 8 (8q24.12) and encodes the COL14A1 (22). As a macromolecular glycoprotein predominantly distributed in the ECM, COL14A1 is closely associated with the maturation and assembly processes of collagen fibrils. Current research suggests that abnormal alterations in ECM components are closely related to tumor malignant progression and distant metastasis (23). Furthermore, COL14A1 interacts closely with proteoglycans; the latter belong to the small leucine-rich proteoglycan family and have been increasingly confirmed to strongly regulate tumor cell growth behavior (24,25). It is particularly noteworthy that studies have explicitly indicated that COL14A1 expression levels are significantly correlated with lymph node positivity in patients with BC (26).

In this study, ferroptosis-related differentially expressed genes (DEGs) between BC tissues and normal bladder tissues were analyzed. By integrating weighted gene co-expression network analysis (WGCNA) and machine learning algorithms, COL14A1 was identified as a key target. Further analysis revealed USP5 as the upstream regulator of COL14A1; therefore, the USP5/COL14A1 axis was constructed to elucidate the ferroptosis-related mechanisms in BC, aiming to provide new insights for the treatment of this disease. We present this article in accordance with the MDAR and ARRIVE reporting checklists (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0311/rc).

Methods

Transcriptome analysis

RNA-seq transcriptomic data (HTSeq-Counts format) of bladder urothelial carcinoma were downloaded from The Cancer Genome Atlas (TCGA) database. The raw count data were retrieved using the R package TCGAbiolinks, and lowly expressed genes were removed (retaining genes with a count ≥10 in ≥50% of the samples). Normalization was performed using DESeq2 (via the median of ratios method) to obtain normalized counts for subsequent differential expression analysis. Batch effects were corrected using the ComBat function from the sva package. The normalized data were then log2-transformed to meet the assumptions for subsequent statistical analyses. All data processing and downstream analyses were conducted in the R environment (version 4.x), primarily utilizing the TCGAbiolinks, DESeq2, dplyr, and glmnet packages. BC tissues and normal bladder tissues were downloaded the GSE13507 dataset (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507). The TCGA database comprised 408 BC tissues and 19 normal bladder tissues, while the GSE13507 dataset included 165 primary BC samples and 10 normal bladder mucosa tissues. DEGs were identified with the criteria of P<0.05 and |log₂fold change (log₂FC)| >1.2. The screened DEGs were visualized using volcano plots.

Protein-protein interaction (PPI) network

The DEGs identified from the above analysis were intersected with ferroptosis-related genes retrieved from the GeneCards database (https://www.genecards.org/) to identify common DEGs that met both criteria. Subsequently, the PPI network of these intersecting genes was constructed using the STRING database (https://string-db.org/), ultimately generating a PPI network map to visually present the molecular functional associations of the intersecting genes.

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis

The intersecting genes between the DEGs identified above and ferroptosis-related genes were submitted to the DAVID 6.8 database (https://david.ncifcrf.gov/), a platform widely used for gene functional annotation and enrichment analysis, to perform GO functional annotation and KEGG pathway enrichment analysis. Subsequently, the SangerBox biomedical data analysis platform (http://sangerbox.com/home.html) and the Bioinformatics platform (http://www.bioinformatics.com.cn/) were utilized to visualize the enrichment results. Finally, based on gene count ranking, the top 5 terms in GO analysis and the top 6 pathways in KEGG pathway analysis were displayed, respectively.

WGCNA

To deeply mine core genes closely associated with the BC phenotype, a gene co-expression network was constructed using the WGCNA algorithm based on expression profile data integrated from the TCGA database and the GSE13507 dataset. During network construction, the scale-free topology fit index and the coefficient of determination (R²) were calculated under different powers to ensure the stability of the network structure. Based on gene expression correlation, a gene hierarchical clustering dendrogram was constructed, and genes with similar expression patterns were classified by combining dynamic tree cutting algorithms, finally identifying multiple co-expression gene modules. Subsequently, the correlation between module eigengenes and the BC phenotype was calculated, and key modules highly associated with the disease were screened out based on correlation strength for subsequent analysis.

Machine learning algorithm analysis

To further screen key feature genes from the candidate genes, the “glmnet 4.1-8” package in R software was utilized to perform least absolute shrinkage and selection operator (LASSO) logistic regression analysis on the sample data. This method effectively avoids model overfitting by introducing a penalty coefficient, and genes were strictly screened according to the optimal value of lambda.min to identify gene features with the most diagnostic value. In addition, to validate the robustness of gene screening, the “e1071 1.7-16” package was employed to execute the support vector machine recursive feature elimination (SVM-RFE) algorithm. This algorithm eliminates irrelevant features through stepwise iterative calculation, ultimately selecting the gene combination with the highest classification accuracy or the lowest error as the optimal feature subset for subsequent analysis.

Gene set enrichment analysis (GSEA)

To elucidate the potential molecular mechanisms of core genes and identify biological pathways closely associated with them, GSEA was conducted. Specifically, based on the full cohort data, Spearman correlation coefficients were calculated between the expression levels of all genes and the core gene, and a ranked gene list was generated accordingly as input for GSEA. The resulting plots detailed the main enriched pathways significantly associated with the expression of each core gene, with the top ten key pathways selected based on the lowest adjusted P values (adj.P). A positive enrichment score (ES) indicates that the pathway is significantly enriched in the gene set positively correlated with the expression of the core genes, suggesting that the core genes may participate in disease occurrence and development by regulating these pathways.

Cell culture

Human BC cells (5637 and HT1376; Procell, Wuhan, China) and immortalized cells of human bladder epithelium (SV-HUC-1; EK-Bioscience, Shanghai, China) were cultured in Roswell Park Memorial Institute-1640 (RPMI-1640; Procell), NEAA-containing MEM (Procell), and F12K medium (Procell). To maintain normal growth, the cells were cultured in medium supplemented with 10% fetal bovine serum (FBS; Procell) and 1% penicillin/streptomycin (Procell) at 37 ℃ in a 5% CO₂ atmosphere.

Cell transfection

GenePharma Co., Ltd. (Shanghai, China) supplied the small hairpin RNA (shRNA) targeting USP5 (sh-USP5) and the COL14A1 overexpression plasmid (COL14A1), along with their corresponding negative controls (sh-NC; vector pCDH-U6-MCS-EF1-GreenPuro) and protruding clustered DNA (pcDNA) 3.1(+) vector (pcDNA). BC cells were seeded in six-well plates at a density of 2×10⁵ per well and cultured until they reached 70–80% confluence before transfection. For transfection in each well, 4.0 µg of plasmid DNA [or 100 pmol small interfering RNA (siRNA)/shRNA] was diluted in 250 µL of Opti-MEM (Solarbio, Beijing, China). Meanwhile, 10 µL of Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) was diluted in 250 µL of Opti-MEM. The diluted DNA and Lipofectamine 3000 were then mixed and incubated for 15 minutes at room temperature before being added dropwise to the cells for further culture.

Western blotting

Total protein lysates were prepared utilizing radioimmunoprecipitation assay (RIPA) lysis buffer (Beyotime, Shanghai, China). Subsequently, protein concentration was quantified employing a bicinchoninic acid (BCA) protein assay kit (Beyotime). The protein samples were then subjected to separation via gel electrophoresis and transferred onto polyvinylidene fluoride (PVDF) membranes. Following the blocking step, the membranes were incubated with primary antibodies targeting COL14A1 (61964; 1:1,000; CST, Shanghai, China), USP5 (A301-543A; 1:2,000; Thermo Fisher, Waltham, MA, USA), and β-actin (4967; 1:1,000; CST) at 4 ℃ overnight. Afterward, the blots underwent incubation with a secondary antibody (7074; 1:2,000; CST). Finally, protein signals were detected using an enhanced chemiluminescence (ECL) detection kit (Beyotime).

Co-immunoprecipitation (Co-IP) assay

5637 and HT1376 cells were collected and subsequently lysed using immunoprecipitation (IP) lysis buffer (Beyotime) for a duration of 30 min. Following this, specific primary antibodies for IP (anti-USP5, A301-543A, Thermo Fisher; anti-COL14A1, sc-517033, SantaCruz, Santa Cruz, CA, USA) or control immunoglobulin G (IgG; ab109489; Abcam, Cambridge, UK) were introduced into the protein samples, which were then gently rotated for 2 h at 4 ℃. Subsequently, the mixtures were incubated overnight with protein-A/G agarose (SantaCruz), after which the beads were rinsed four times with IP lysis buffer and finally resuspended in sodium dodecyl sulfate (SDS) sample buffer before being heated at 95 ℃ for 10 min. Ultimately, the resulting samples were analyzed via immunoblotting.

Ubiquitination assay

Following the transfection of 5637 and HT1376 cells with either sh-USP5 or sh-NC utilizing Lipofectamine 3000 (Invitrogen), the cells underwent lysis in IP lysis buffer for a duration of 30 min. Subsequently, a Co-IP assay was performed as detailed previously by employing an anti-COL14A1 antibody (sc-517033; SantaCruz). The extent of ubiquitination was then evaluated via immunoblotting using an anti-ubiquitination polyclonal antibody (58395; CST). Additionally, untreated cell lysates served as the input controls for the experiment.

Cell colony formation assay

Transfected 5637 and HT1376 cells, at a density of roughly 1,000 cells per well, were plated into six-well plates and allowed to grow for a period of 10 days. Following the culture period, the cells underwent fixation with 4% paraformaldehyde (PFA; Maokangbio, Shanghai, China) for 30 min, followed by staining using a 0.1% crystal violet solution (Maokangbio) for another 30 min. Finally, the observable colonies were enumerated and captured under a microscope.

Flow cytometry

For the purpose of evaluating cellular apoptosis, the Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) apoptosis detection kit (Simubiotech, Tianjin, China) was employed. A total number of 1×10⁵ cells were collected and then subjected to incubation with 100 µL of Annexin V binding buffer that had been supplemented with 5 µL of FITC and 5 µL of PI reagent in a dark environment for a duration of 30 min. Following the conclusion of the incubation period, 400 µL of binding buffer was introduced to terminate the reaction, and the prepared samples were maintained on ice until they were analyzed using flow cytometry.

ROS

To assess intracellular ROS levels, 2',7'-dichlorodi-hydrofluorescein diacetate (DCFH-DA; MedChemExpress, Princeton, NJ, USA) was employed as the fluorescent probe. Briefly, approximately 1×10⁵ cells were resuspended in 1 mL of phosphate-buffered saline (PBS) containing 1 µM DCFH-DA and subsequently incubated in the dark for 30 min. Following the incubation period, the samples were washed twice, resuspended in 1 mL of PBS, and kept on ice for further analysis using a fluorescence microscope.

Determination of intracellular glutathione (GSH), Fe²⁺, and malondialdehyde (MDA) levels

5637 and HT1376 cells were plated into six-well plates at a density of 1×10⁵ cells per well and cultured for 24 h in culture medium. Following transfection with the specified shRNAs or plasmids, the cells were lysed utilizing the lysis buffer supplied within the reduced glutathione assay kit (Abcam) and centrifuged at 14,000 ×g for 10 min. The GSH in the supernatant reacted with 5,5'-dithiobis-(2-nitrobenzoic acid) (DTNB) to generate yellow 2-nitro-5-thiobenzoic acid, and its absorbance was determined using a microplate reader.

Regarding iron detection, the harvested cells were immediately homogenized in PBS. After centrifugation, the Fe²⁺ level in the supernatant was quantified using an iron assay kit (Abcam) in accordance with the manufacturer’s instructions.

MDA levels were measured using an MDA detection kit (Beyotime). In brief, TBA solution was added to the samples and standards, followed by incubation at 95 ℃ for 60 min and cooling in an ice bath for 10 min. The mixture was transferred to the wells of a microplate, and the absorbance was measured at 532 nm using a microplate reader.

Xenograft mouse model assay

Male BALB/c nude mice (5-week-old, 18–20 g, n=10), obtained from SPF (Beijing) Biotechnology Co., Ltd. (Beijing, China), were randomized into different experimental groups (n=5 per group) using a random number table. The animals were housed in a specific pathogen-free (SPF) barrier facility under controlled environmental conditions, maintaining a constant temperature (22±2 ℃), relative humidity (50%±10%), and a 12-hour light/dark cycle. A suspension containing 5×10⁶ sh-NC or sh-USP5 5637 cells was subcutaneously injected into the mice. The dimensions of the xenografts were recorded every 10 days, and the tumor volume was determined by applying the formula V = (length × width²)/2. All mice were euthanized 30 days after the xenograft procedure, after which the weight and volume of the tumors were documented, and the protein expression levels of USP5 and COL14A1 within the tumor tissues were evaluated via Western blotting. Experiments were performed under a project license (No. 2018035) granted by the Animal Care Committee of Shantou Longhu People’s Hospital, in compliance with institutional guidelines for the care and use of animals. Inclusion criteria were: (I) mice were enrolled in the treatment phase only upon successful tumor implantation, defined as the establishment of a palpable tumor with a volume ≥50 mm³ within 7 to 10 days post-injection. (II) Only animals demonstrating normal grooming, activity, and posture, without signs of non-tumor-related distress or infection, were included. Exclusion criteria were: (I) failed tumor take, indicated by the absence of a palpable tumor by day 7–10 post-implantation; (II) post-surgical complications, including severe surgical site infections, neurological deficits (e.g., persistent seizures, paresis), or greater than 15% loss of initial body weight within the first week; and (III) tumor-related morbidity necessitating the intervention of predefined humane endpoints during the experimental period. These endpoints included tumor volume exceeding 1,500 mm³, tumor ulceration, body weight loss surpassing 20%, or severe clinical distress (e.g., lethargy, hunched posture, inability to access food or water). All outcome assessments were performed by investigators blinded to the group assignments. To minimize environmental variability, cage positions on the animal rack were rotated weekly across the different experimental groups. Furthermore, all treatments and tumor measurements were carried out at a consistent time each day.

Statistical analysis

Data analysis was performed utilizing GraphPad Prism (version 10.1.3). The findings from three independent biological experiments are presented as the mean ± standard deviation (SD). Distribution normality was evaluated using Shapiro-Wilk normality test. The homogeneity of variance was tested by Brown-Forsythe test. Differences with statistical significance were evaluated using two-tailed Student’s t-tests or one-way analysis of variance (ANOVA), and P value <0.05 was deemed to be statistically significant.

Results

Identification of DEGs between BC tissues and normal bladder tissues

The study analyzed DEGs between BC tissues and normal bladder tissues through the TCGA database and GSE13507 dataset. The screening criteria were P<0.05 and |log₂FC| >1.2. Ultimately, 6,041 and 712 DEGs were identified, respectively, and the results are presented as volcano plots in Figure 1A,1B. The DEGs identified above were intersected with ferroptosis-related genes obtained from the GeneCards database, resulting in 84 overlapping genes (Figure 1C). Subsequently, a PPI network of these 84 genes was constructed using the STRING database, and the result is shown in Figure 1D. The study further performed GO and KEGG pathway analysis of these genes. For GO-biological process (BP), the genes were primarily enriched in mitotic nuclear division (Figure 1E). In terms of GO-cellular component (CC), the genes were mainly distributed in the ECM, external encapsulating structure, contractile muscle fiber, spindle, and actin filament bundle (Figure 1E). Regarding GO-molecular function (MF), the primary enriched terms included actin binding, transmembrane receptor protein kinase activity, ECM structural constituent, growth factor binding, and transmembrane receptor protein tyrosine kinase activity (Figure 1E). Finally, KEGG pathway analysis showed significant enrichment in Motor proteins, cytoskeleton in muscle cells, immunoglobulin superfamily (IgSF) cell adhesion molecule (CAM) signaling, focal adhesion, and epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor resistance (Figure 1F).

Figure 1 Identification of DEGs between BC tissues and normal bladder tissues. (A) The volcano map showing DEGs between BC tissues and normal bladder tissues through the TCGA database. (B) The volcano map showing DEGs between BC tissues and normal bladder tissues through the GSE13507 dataset. (C) A Venn diagram showing the intersection of DEGs identified from the TCGA database, DEGs identified from the GSE13507 dataset, and ferroptosis-related genes. (D) The PPI network of the 84 genes. (E) Bubble plot of GO enrichment analysis for the 84 genes. The figure displays the top 5 significantly enriched terms in BP, CC, and MF. (F) Bubble plot of KEGG pathway analysis for the 84 genes. The figure displays the top 6 significantly enriched pathways. BC, bladder cancer; BP, biological process; CAM, cell adhesion molecule; CC, cellular component; DEG, differentially expressed gene; EGFR, epidermal growth factor receptor; GEO, Gene Expression Omnibus; GO, Gene Ontology; IgSF, immunoglobulin superfamily; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function; PPI, protein-protein interaction; TCGA, The Cancer Genome Atlas.

Identification of hub genes through WGCNA

To further precisely mine hub genes associated with the BC phenotype, the study constructed gene co-expression networks using the WGCNA algorithm based on the TCGA database and the GSE13507 dataset. For the TCGA database analysis, the soft threshold was set to 4 to satisfy the scale-free topology of the network, with a corresponding R² of 0.86 and high mean connectivity (Figure 2A). A hierarchical clustering tree was constructed based on gene correlations, and a total of 16 similar gene modules were identified (Figure 2B,2C). For the GSE13507 dataset analysis, the soft threshold was set to 7 to satisfy the scale-free topology of the network, with a corresponding R² of 0.87 and high mean connectivity (Figure 2D). A hierarchical clustering tree was constructed based on gene correlations, and a total of 18 similar gene modules were identified (Figure 2E,2F). Ultimately, based on the correlation between module eigengenes and the BC phenotype, the “red” module (TCGA data, r=0.57, P=1e−37) and the “greenyellow” module (GSE13507 dataset, r=0.49, P=4e−12) were identified as the most clinically significant modules in BC, as shown in Figure 2G,2H. The intersection of the genes from these two modules and ferroptosis-related DEGs was taken, resulting in four genes including JAM3, COL14A1, FBLN5, and FXYD6 (Figure 2I).

Figure 2 Identification of hub genes through the WGCNA. (A,D) Analysis of the optimal soft threshold in WGCNA network construction. The left panel displays the relationship between the scale-free topology model fit (R²) and different soft thresholds, while the right panel shows the trend of the mean connectivity of the network as the soft threshold changes. (B,E) Gene clustering dendrogram. This figure illustrates the hierarchical clustering of genes based on expression similarity. Branches in the dendrogram represent genes, and the color bands below the branches partition genes with highly coordinated expression changes into different modules. (C,F) Module-trait heatmap showing the correlation between each gene module and tumor/normal status. This figure displays the correlation between each gene module and sample grouping. Each cell contains two numbers: the number outside the parentheses is the correlation coefficient, and the number inside the parentheses is the P value. Red indicates a positive correlation, while blue indicates a negative correlation; the depth of the color represents the strength of the correlation. (G) TCGA database analysis: scatter plot of GS versus MM for genes within the most correlated “red” module. (H) GSE13507 dataset analysis: scatter plot of GS versus MM for genes within the most correlated “greenyellow” module. (I) Venn diagram showing the intersecting genes between the above WGCNA analysis results and ferroptosis-related DEGs. DEG, differentially expressed gene; GEO, Gene Expression Omnibus; GS, gene significance; MM, module membership; TCGA, The Cancer Genome Atlas; WGCNA, weighted gene co-expression network analysis.

COL14A1 was identified as a hub gene through machine learning algorithm

The four genes identified by WGCNA were subsequently analyzed using machine learning algorithms. For the WGCNA results derived from the TCGA database, Lasso regression analysis was performed using the R package ‘glmnet’, which identified 4 feature genes. Additionally, the SVM-RFE algorithm identified 3 feature genes. The intersection of these results, visualized via a Venn diagram, yielded 3 feature genes (Figure 3A). For the WGCNA results derived from the GSE13507 dataset, Lasso regression analysis using the R package ‘glmnet’ screened 2 feature genes, while the SVM-RFE algorithm identified 3 feature genes. The intersection of these results, shown in a Venn diagram, yielded 2 feature genes (Figure 3B). Finally, the intersection of the results from both datasets was taken, identifying the common gene COL14A1 and FBLN5 (Figure 3C). Given that COL14A1 was the only ubiquitination substrate identified through the UbiBrowser_v2 database, it was selected as the focus of this study. Subsequent GSEA of TCGA and GSE13507 data showed that COL14A1 was associated with BOQUEST_STEM_CELL_UP, LIM_MAMMARY_STEM_CELL_UP, LINDGREN_BLADDER_CANCER_CLUSTER_2B, and SMID_BREAST_CANCER_NORMAL_LIKE_UP (Figure 3D,3E).

Figure 3 COL14A1 was identified as a hub gene through machine learning algorithm. (A) The four intersecting genes were subsequently analyzed using LASSO and SVM-RFE algorithms through the TCGA database. The intersection of the genes identified by both algorithms was visualized using a Venn diagram. (B) The four intersecting genes were subsequently analyzed using LASSO and SVM-RFE algorithms through the GSE13507 dataset. The intersection of the genes identified by both algorithms was visualized using a Venn diagram. (C) A Venn diagram was used to visualize the intersection of the results from the above algorithms. (D) GSEA of COL14A1 through the TCGA database. (E) GSEA of COL14A1 through the GSE13507 dataset. COL14A1, collagen type XIV alpha 1 chain; GSEA, gene set enrichment analysis; LASSO, least absolute shrinkage and selection operator; SVM-RFE, support vector machine recursive feature elimination; TCGA, The Cancer Genome Atlas.

USP5 stabilized COL14A1 protein expression through deubiquitination activities in BC cells

The results showed that COL14A1 protein expression was upregulated in BC cells (5637 and HT1376) in comparison with immortalized cells of human bladder epithelium (SV-HUC-1) (Figure 4A). Analysis using the UbiBrowser online platform predicted the deubiquitinases for COL14A1, showing that COL14A1 was a substrate of the deubiquitinases USP13 and USP5 (Figure 4B). Given the inhibitory effect of USP13 on BC progression (27), the study selected USP5 as a deubiquitinase of COL14A1. Subsequent data showed that USP5 expression was increased in 5637 and HT1376 cells when compared with SV-HUC-1 cells (Figure 4C). Co-IP assays demonstrated that the USP5 antibody successfully immunoprecipitated the COL14A1 protein, and conversely, the COL14A1 antibody pulled down the USP5 protein in 5637 and HT1376 cells (Figure 4D). Furthermore, USP5 knockdown accelerated the degradation of the COL14A1 protein (Figure 4E). As shown in Figure 4F, the downregulation of USP5 led to increased ubiquitination levels of COL14A1 in 5637 and HT1376 cells. Thus, USP5 overexpression stabilized COL14A1 protein expression in 5637 and HT1376 cells.

Figure 4 USP5 stabilized COL14A1 protein expression in BC cells through deubiquitination activities. (A) Western blotting was used to detect COL14A1 protein expression in SV-HUC-1, 5637, and HT1376 cells. (B) The UbiBrowser online platform was used to predict the deubiquitinases for COL14A1. (C) Western blotting was used to detect USP5 protein expression in SV-HUC-1, 5637, and HT1376 cells. (D-F) Co-IP, CHX, and ubiquitination assays were used to analyze the association of USP5 and COL14A1 in 5637 and HT1376 cells. (G) The effects of USP5 knockdown and COL14A1 overexpression on COL14A1 protein expression were analyzed by Western blotting assay in 5637 and HT1376 cells. ***, P<0.001. BC, bladder cancer; CHX, cycloheximide; Co-IP, co-immunoprecipitation; COL14A1, collagen type XIV alpha 1 chain; IB, immunoblotting; IgG, immunoglobulin G; IP, immunoprecipitation; sh-NC, shRNA targeting negative control; sh-USP5, shRNA targeting USP5; shRNA, small hairpin RNA; USP5, ubiquitin specific peptidase 5; USP13, ubiquitin specific peptidase 13.

USP5 knockdown inhibited the malignant phenotypes of BC cells by regulating COL14A1 expression

The study then transfected USP5 shRNA, COL14A1 overexpression plasmid, and the matched controls (sh-NC and pcDNA) into 5637 and HT1376 cells to determine the effects on the key malignant phenotypes of tumor cells. As shown in Figure 4G, USP5 silencing inhibited COL14A1 protein expression, whereas the effect was relieved after COL14A1 overexpression. Subsequently, USP5 knockdown reduced the number of positive colonies, but the effect was counteracted after COL14A1 overexpression (Figure 5A). The results also revealed that USP5 silencing induced cell apoptosis, whereas the effect was relieved by ectopic COL14A1 expression (Figure 5B). Moreover, the downregulation of USP5 expression increased the levels of ROS, MDA, and Fe²⁺ and decreased GSH levels, however, these effects were relieved after COL14A1 overexpression (Figure 5C-5F). The study further injected 5637 cells stably expressing sh-USP5 or sh-NC into nude mice to validate the in vitro data regarding the effects of USP5 silencing on the key malignant phenotypes of BC cells. The results showed that USP5 silencing inhibited tumor weight and volume (Figure S1A,S1B). In addition, USP5 and COL14A1 protein expression was lower in the tumors resulting from 5637 cells transfected with sh-USP5 when compared with those resulting from 5637 cells transfected with sh-NC (Figure S1C). Thus, USP5 knockdown inhibited the malignant progression of BC cells through the regulation of COL14A1 expression.

Figure 5 USP5 knockdown inhibited BC cell proliferation and induced cell apoptosis and ferroptosis by regulating COL14A1 expression. 5637 and HT1376 cells were divided into the sh-NC group, the sh-USP5 group, the sh-USP5 + pcDNA group, and the sh-USP5 + COL14A1 group. (A) Cell proliferation was analyzed by cell colony formation assay (crystal violet staining). (B) Cell apoptosis was analyzed by flow cytometry. (C) ROS levels were quantified by microscopic method (DCFH-DA staining). (D-F) Colorimetric assays were used to detect the levels of MDA, GSH, and Fe²⁺. **, P<0.01; ***, P<0.001. BC, bladder cancer; COL14A1, collagen type XIV alpha 1 chain; DCFH-DA, 2',7'-dichlorodi-hydrofluorescein diacetate; GSH, glutathione; MDA, malondialdehyde; pcDNA, protruding clustered DNA; ROS, reactive oxygen species; sh-NC, shRNA targeting negative control; sh-USP5, shRNA targeting USP5; shRNA, small hairpin RNA; USP5, ubiquitin specific peptidase 5.

Discussion

The pathogenesis of BC is complex and involves multiple signaling pathways; among these, ferroptosis, a novel iron-dependent form of cell death, plays an indispensable and key role in the progression of the disease. In the pathological context of BC, the inhibition of ferroptosis is closely associated with enhanced resistance to immunotherapy and platinum-based chemotherapy, as well as the significant promotion of malignant progression (9,28,29). Multiple lines of evidence indicate that specifically inducing ferroptosis in cancer cells may represent a promising therapeutic strategy (30). Therefore, elucidating the specific molecular mechanisms of ferroptosis in BC progression is expected to provide potential targets for the development of innovative therapeutic approaches. The study identified ferroptosis-related DEGs between BC tissues and normal bladder tissues through WGCNA and machine learning algorithm. The results showed that COL14A1 was a hub gene, and subsequent analyses revealed that USP5 knockdown inhibited the proliferation of BC cells and induced cell apoptosis and ferroptosis by destabilizing COL14A1 expression.

Existing literature has established that USP5 is upregulated in BC, with specific studies indicating that its overexpression stabilizes the c-Jun protein to facilitate tumor progression (20). Further research has expanded on this oncogenic role, demonstrating that USP5 enhances proliferation and invasion by stabilizing snail family transcriptional repressor 2 (SLUG), a key regulator of epithelial-mesenchymal transition (31). While these findings underscore the importance of USP5 in tumor cell survival and migration, they overlook the potential metabolic mechanisms, specifically ferroptosis, that may be regulated by this deubiquitinase. A recent study did observe that USP5 ablation impacts ferroptosis markers by stabilizing GPX4 (32); however, the precise substrate linking USP5 to the ferroptosis pathway remained insufficiently defined. The present results addressed this gap by identifying COL14A1 as a novel, critical substrate of USP5. The study showed that COL14A1 expression was upregulated in BC cells. In contrast to the c-Jun and SLUG pathways, this study elucidated a distinct mechanism wherein USP5 interacted with COL14A1 to prevent its ubiquitination and degradation. This interaction was proven essential for cell fate, as USP5 knockdown destabilized COL14A1 and triggered a robust ferroptotic response characterized by increased Fe²⁺, ROS, and MDA levels and decreased GSH levels, coupled with the induction of apoptosis and the suppression of proliferation. Moreover, in vivo studies further confirmed that USP5 knockdown inhibited tumor growth and reduced COL14A1 expression. Crucially, the rescue of these phenotypes by COL14A1 overexpression in vitro confirmed the specificity of this axis, offering a novel and specific molecular explanation for USP5-mediated ferroptosis inhibition that extended beyond the previously described GPX4 stabilization.

The mechanistic analysis revealed that COL14A1 functioned as a critical oncogenic driver in BC progression through its potent inhibition of ferroptosis. The current data demonstrated that COL14A1 overexpression promoted cell proliferation while simultaneously suppressing apoptosis. More importantly, COL14A1 exerted a protective effect against oxidative stress by maintaining redox homeostasis. This was evidenced by the decreased accumulation of ferroptosis markers, including ROS, MDA, and Fe²⁺, coupled with elevated GSH levels following COL14A1 upregulation. These observations are consistent with the emerging consensus that evasion of ferroptosis is a hallmark of aggressive malignancies, allowing cancer cells to survive under high metabolic stress (33,34). These findings suggest that COL14A1 creates an antioxidant microenvironment that shields BC cells from iron-dependent cell death. The GSEA analysis provided additional mechanistic context, revealing that COL14A1 was positively associated with stem cell-like signatures (BOQUEST_STEM_CELL_UP, LIM_MAMMARY_STEM_CELL_UP) and specific BC molecular subtypes (LINDGREN_BLADDER_CANCER_CLUSTER_2B). Such an association with stemness signatures is particularly noteworthy, as cancer stem cells are known to exhibit heightened resistance to conventional therapies (35,36). This stem-like phenotype may contribute to therapy resistance and tumor aggressiveness. Collectively, these results indicate that COL14A1 facilitates BC progression by establishing a ferroptosis-resistant state, wherein elevated GSH and suppressed lipid peroxidation enable sustained tumor cell survival and proliferation.

Despite the compelling evidence presented, several limitations warrant consideration. First, while the interaction between USP5 and COL14A1 was confirmed, the precise structural domains mediating this deubiquitination process remain to be elucidated through detailed mapping experiments. Furthermore, although ferroptosis markers were assessed, the specific downstream signaling pathways linking COL14A1 to the regulation of iron metabolism and redox homeostasis require deeper mechanistic exploration. Finally, the in vivo experiments were restricted to subcutaneous xenograft models; thus, future studies employing orthotopic BC models or patient-derived xenografts (PDX) are essential to better simulate the complex tumor microenvironment and validate the therapeutic efficacy of targeting this pathway.

Conclusions

In summary, this study identified a novel USP5/COL14A1 regulatory axis that functioned as a critical inhibitor of ferroptosis in BC. The findings demonstrate that USP5 stabilizes COL14A1 by inhibiting its ubiquitination, thereby promoting tumor progression and protecting malignant cells from iron-dependent cell death. These results position the USP5/COL14A1 axis as a potential therapeutic target, offering a strong theoretical foundation for the development of novel deubiquitinase-targeted therapies designed to induce ferroptosis in BC.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the MDAR and ARRIVE reporting checklists. Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0311/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0311/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0311/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-2026-0311/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All animal experiments were performed under a project license (No. 2018035) granted by the Animal Care Committee of Shantou Longhu People’s Hospital, in compliance with institutional guidelines for the care and use of animals.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Hasan N, Yang D, Gibson S, et al. Advancements in bladder cancer treatment: The synergy of radiation and immunotherapy. Oncotarget 2025;16:337-46. [Crossref] [PubMed]
3. van Hoogstraten LMC, Vrieling A, van der Heijden AG, et al. Global trends in the epidemiology of bladder cancer: challenges for public health and clinical practice. Nat Rev Clin Oncol 2023;20:287-304. [Crossref] [PubMed]
4. Dixon SJ, Olzmann JA. The cell biology of ferroptosis. Nat Rev Mol Cell Biol 2024;25:424-42. [Crossref] [PubMed]
5. Aschner M, Skalny AV, Martins AC, et al. Ferroptosis as a mechanism of non-ferrous metal toxicity. Arch Toxicol 2022;96:2391-417. [Crossref] [PubMed]
6. Ozkan E, Bakar-Ates F. Ferroptosis: A Trusted Ally in Combating Drug Resistance in Cancer. Curr Med Chem 2022;29:41-55. [Crossref] [PubMed]
7. Zhou Q, Meng Y, Li D, et al. Ferroptosis in cancer: From molecular mechanisms to therapeutic strategies. Signal Transduct Target Ther 2024;9:55. [Crossref] [PubMed]
8. Ma Y, Wang M, Zhang C, et al. Resveratrol regulates ferroptosis and endometrial cancer progression by inhibiting CEBPA-mediated FAM83B expression. Current Proteomics 2025;22:100061.
9. Shen L, Zhang J, Zheng Z, et al. PHGDH Inhibits Ferroptosis and Promotes Malignant Progression by Upregulating SLC7A11 in Bladder Cancer. Int J Biol Sci 2022;18:5459-74. [Crossref] [PubMed]
10. Jiang P, Ning J, Yu W, et al. FLRT2 suppresses bladder cancer progression through inducing ferroptosis. J Cell Mol Med 2024;28:e17855. [Crossref] [PubMed]
11. Jin S, Kudo Y, Horiguchi T. The Role of Deubiquitinating Enzyme in Head and Neck Squamous Cell Carcinoma. Int J Mol Sci 2022;24:552. [Crossref] [PubMed]
12. Dewson G, Eichhorn PJA, Komander D. Deubiquitinases in cancer. Nat Rev Cancer 2023;23:842-62. [Crossref] [PubMed]
13. Wu Y, Deng Y, Ai Q, et al. Deubiquitinating enzymes in parkinson’s disease: molecular mechanisms and therapeutic potential. Mol Med 2025;31:329. [Crossref] [PubMed]
14. Liu J, Tan X, Fang X, et al. Deubiquitinating enzymes as therapeutic targets in diabetic nephropathy. Trends Pharmacol Sci 2026;47:198-215. [Crossref] [PubMed]
15. Fu Z, Wei B, Huang Y, et al. Research on the mechanism by which quercetin affects ferroptosis in breast cancer cells through regulating USP9X/PIAS4. Letters in Drug Design & Discovery 2025;22:100169.
16. Gao ST, Xin X, Wang ZY, et al. USP5: Comprehensive insights into structure, function, biological and disease-related implications, and emerging therapeutic opportunities. Mol Cell Probes 2024;73:101944. [Crossref] [PubMed]
17. Cao L, Liu H, Huang C, et al. USP5 knockdown alleviates lung cancer progression via activating PARP1-mediated mTOR signaling pathway. Biol Direct 2023;18:16. [Crossref] [PubMed]
18. Peng ZM, Han XJ, Wang T, et al. PFKP deubiquitination and stabilization by USP5 activate aerobic glycolysis to promote triple-negative breast cancer progression. Breast Cancer Res 2024;26:10. [Crossref] [PubMed]
19. Meng J, Ai X, Lei Y, et al. USP5 promotes epithelial-mesenchymal transition by stabilizing SLUG in hepatocellular carcinoma. Theranostics 2019;9:573-87. [Crossref] [PubMed]
20. Zhang HH, Zhang AQ, Peng P, et al. USP5 facilitates bladder cancer progression by stabilizing the c-Jun protein. Cancer Cell Int 2024;24:32. [Crossref] [PubMed]
21. Lo Buglio G, Lo Cicero A, Campora S, et al. The Multifaced Role of Collagen in Cancer Development and Progression. Int J Mol Sci 2024;25:13523. [Crossref] [PubMed]
22. Calò CM, Onali F, Robledo R, et al. Worldwide variation of the COL14A1 gene is shaped by genetic drift rather than selective pressure. Mol Genet Genomic Med 2021;9:e1629. [Crossref] [PubMed]
23. Chitty JL, Cox TR. The extracellular matrix in cancer: from understanding to targeting. Trends Cancer 2025;11:839-49. [Crossref] [PubMed]
24. Baghy K, Tátrai P, Regős E, et al. Proteoglycans in liver cancer. World J Gastroenterol 2016;22:379-93. [Crossref] [PubMed]
25. Edwards IJ. Proteoglycans in prostate cancer. Nat Rev Urol 2012;9:196-206. [Crossref] [PubMed]
26. Huang Y, Fang S, Xie W, et al. Collagen gene signature in the tumor microenvironment predicts survival and guides prognosis in bladder cancer. Discov Oncol 2025;16:1479. [Crossref] [PubMed]
27. Man X, Piao C, Lin X, et al. USP13 functions as a tumor suppressor by blocking the NF-kB-mediated PTEN downregulation in human bladder cancer. J Exp Clin Cancer Res 2019;38:259. [Crossref] [PubMed]
28. Ding Q, Tang W, Li X, et al. Mitochondrial-targeted brequinar liposome boosted mitochondrial-related ferroptosis for promoting checkpoint blockade immunotherapy in bladder cancer. J Control Release 2023;363:221-34. [Crossref] [PubMed]
29. Xiang L, Zeng Q, Liu J, et al. MAFG-AS1/MAFG positive feedback loop contributes to cisplatin resistance in bladder urothelial carcinoma through antagonistic ferroptosis. Sci Bull (Beijing) 2021;66:1773-88. [Crossref] [PubMed]
30. Mou Y, Wang J, Wu J, et al. Ferroptosis, a new form of cell death: opportunities and challenges in cancer. J Hematol Oncol 2019;12:34. [Crossref] [PubMed]
31. Wan QK, Li TT, Liu BB, et al. USP5 promotes tumor progression by stabilizing SLUG in bladder cancer. Oncol Lett 2024;28:572. [Crossref] [PubMed]
32. Liu C, Deng Y, Huang L, et al. USP5 Suppresses Ferroptosis in Bladder Cancer Through Stabilization of GPX4. Curr Issues Mol Biol 2025;47:211. [Crossref] [PubMed]
33. Singh M, Arora HL, Naik R, et al. Ferroptosis in Cancer: Mechanism and Therapeutic Potential. Int J Mol Sci 2025;26:3852. [Crossref] [PubMed]
34. Dos Santos AF, Fazeli G, Xavier da Silva TN, et al. Ferroptosis: mechanisms and implications for cancer development and therapy response. Trends Cell Biol 2023;33:1062-76. [Crossref] [PubMed]
35. Huang T, Song X, Xu D, et al. Stem cell programs in cancer initiation, progression, and therapy resistance. Theranostics 2020;10:8721-43. [Crossref] [PubMed]
36. Lee H, Kim B, Park J, et al. Cancer stem cells: landscape, challenges and emerging therapeutic innovations. Signal Transduct Target Ther 2025;10:248. [Crossref] [PubMed]