Identifying PLAU as a shared pathogenic gene in type 2 diabetes and bladder urothelial carcinoma through integrated transcriptomic analysis and machine learning for diagnostic and therapeutic value Authors: Shiyong Xin, Guanyu Li, Xingqiang Yang, Junjie Su, Ruixin Li, Wang Qin, Zheng Zhang, Chu Wang, Yingao Zhu, Liming Feng, Sichang Ren, Zhongwei Gao Published: 2026 ABSTRACT Type 2 diabetes mellitus (T2DM) and bladder urothelial carcinoma (BLCA) are two kinds of diseases that seriously threaten human health. Their pathogenesis is complex and involves the interaction of multiple genes and multiple pathways. Recent epidemiological studies have shown that the risk of BLCA in patients with T2DM is significantly higher than that in non-diabetic people, suggesting that there may be a potential biological correlation between the two. Genomic studies have opened up new ways to reveal the common genetic characteristics of T2DM and BLCA. However, most of the current studies only focus on a single disease, and the comorbidity mechanism of these two diseases still needs to be further explored. Firstly, the datasets of BLCA and T2DM were downloaded from the The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases respectively. Differential expression genes (DEGs) were identified using the Limma package. Weighted gene co-expression network analysis (WGCNA) was employed to determine the co-expression modules related to BLCA and T2DM, and the common potential target genes were obtained. Correlation analysis and enrichment analysis were conducted on these target genes. Then, the best diagnostic biomarker - plasminogen activator (PLAU) was selected using machine learning algorithms. Additionally, the role of PLAU in the progression of T2DM and BLCA was confirmed through immunohistochemistry, Western Blot, and Edu experiments. Finally, small molecule compounds targeting PLAU were discovered through molecular docking and virtual screening, and the inhibitory effect of these small molecules on the progression of bladder urothelial carcinoma was verified through experiments. This study conducted a combined limma and WGCNA analysis on the T2DM and BLCA datasets to identify 42 common potential target genes, which were enriched in pathways such as innate immunity. Using machine learning algorithms such as LASSO and SVM, PLAU was identified as the best diagnostic marker for T2DM combined with BLCA. It was significantly highly expressed in both T2DM and BLCA samples, and high expression of PLAU predicted a shorter overall survival period for BLCA patients. Experimental results confirmed that PLAU was highly expressed in BLCA tissues and increased with the severity of malignancy. Knockdown (sh-PLAU) of PLAU could inhibit cancer cell proliferation and migration in a high-glucose environment, while overexpression (oe-PLAU) still promoted cancer cell progression in a low-glucose environment. Finally, molecular docking virtual screening revealed that the small molecule compound epigallocatechin gallate (EGCG) could target and inhibit PLAU, and effectively inhibited the proliferation and invasion of BLCA cells in experiments. The results of this study reveal the role of PLAU, a common characteristic gene of T2DM and BLCA, whose high expression drives tumor progression and poor prognosis. Moreover, small molecule drugs targeting PLAU, such as EGCG, have therapeutic potential. This study provides a new direction for accurate diagnosis and treatment of BLCA patients with T2DM.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10238-026-02211-6. Introduction Bladder cancer (BCa) is one of the common malignant tumors of the genitourinary system, ranking the sixth in the incidence of male malignant tumors in the world. Bladder urothelial carcinoma (BLCA) is the most common type, accounting for 90% of all BCa cases [1]. It is projected that by 2050, the global population aged 65 and above will reach 1.6 billion. Due to the higher prevalence of urinary system tumors such as BCa, kidney cancer (KCa), and prostate cancer (PCa) among the elderly population, the burden on the healthcare system is increasing day by day. Moreover, according to the Norpred prediction model, the number of BCa cases worldwide is expected to increase significantly from 2022 to 2046 [2]. Recent studies have shown that metabolic abnormalities (such as T2DM and obesity) are closely related to the occurrence and development of BLCA [3]. Hyperglycemic microenvironment can induce chronic inflammation, oxidative stress and extracellular matrix remodeling, which promote tumor cell proliferation, invasion and immune escape. Epidemiological evidence shows that the risk of BLCA in patients with T2DM is higher than that in non-diabetic patients, and the survival time of BLCA patients with T2DM is shortened [4]. The diagnostic methods for BLCA have achieved significant breakthroughs: at the diagnostic level, precise, non-invasive, and convenient technologies such as liquid biopsy, multimodal imaging, endoscopic fusion, and artificial intelligence pathology are gradually becoming widespread [5]. However, the molecular mechanisms by which metabolic disorders regulate the malignant progression of BLCA are still unclear, especially the cross-disease core targets and precise treatment strategies for these patients. Therefore, it is of great scientific significance and clinical application value to reveal the comorbidity molecular network of T2DM and BLCA, screen the key factors driving the metabolism-tumor interaction, and develop targeted interventions to improve the clinical prognosis of BLCA. A number of studies have shown that the risk of cancer in patients with T2DM is significantly higher than that in the general population, with an increased risk of 34% in men and 62% in women [6]. Studies have also found that T2DM is associated with an increased risk of more than 10 types of cancer, such as colorectal cancer, liver cancer, and pancreatic cancer [7], with a 107% increase in the risk of pancreatic cancer and a 239% increase in the risk of liver cancer [8]. In addition, the mortality rate of cancer in patients with diabetes is 30%-50% higher than that in non-diabetic people, which may be due to various reasons: high blood glucose can supply energy to tumor cells and promote their proliferation, and also increase the risk of gene mutations [9]; Patients with T2DM are often accompanied by hyperinsulinemia. Excessive insulin stimulates tumor growth and enhances the biological activity of IGF-1 [10]. T2DM is a chronic low-grade inflammatory state, and inflammatory factors can promote angiogenesis and tumor metastasis. Obesity-related hormones may also increase the risk of specific cancers [11]. The imbalance of intestinal flora is closely related to gastrointestinal tumors, and the diversity of intestinal flora is significantly reduced in patients with T2DM [12]. At the same time, the treatment effect of cancer patients with diabetes is poor in clinical practice. On the one hand, the prognosis of cancer patients with diabetes is worse, and hyperglycemia will increase the risk of postoperative infection and shorten the survival time. For example, the 5-year mortality of breast cancer patients with diabetes is significantly increased [13]. On the other hand, the two treatments interfere with each other. Cancer treatment may induce hyperglycemia, and improper management of diabetes will affect the efficacy of cancer [14]. Tumor cell metastasis needs to break through the extracellular matrix, and it secretes a variety of proteases including serine proteases to decompress the matrix [15]. Urokinase-type plasminogen activator (uPA, PLAU), a serine protease, exists in the blood and extracellular matrix, and its main function is to convert plasminogen into plasmin. In addition, previous studies have shown that PLAU plays a crucial role in tumor occurrence and tumor immunity, making it a potential therapeutic target and prognostic marker for various malignant tumors. Moreover, the expression level of PLAU is closely related to immune cells and key immune-related genes, suggesting that it has great potential as a therapeutic target in cancer immunotherapy [16]. For example, PLAU can induce cell proliferation in tumors such as breast cancer and head and neck squamous cell carcinoma, and it is also abnormally expressed in other tumors, such as gastric cancer and pancreatic cancer [17, 18]. However, the biological role of PLAU in T2DM and BLCA remains unclear. In this study, by analyzing multiple datasets, PLAU expression was found to be increased in BLCA tumor samples and T2DM pancreatic tissues compared with adjacent tissues, confirming that it is a key common characteristic gene between T2DM and BLCA patients. Materials and methods Data sources The data in this study were collected from TCGA and GEO databases. Which from TCGA database (https://www.cancer.gov/ccg/research/genome-sequencing/tcga) to download the bladder urothelial carcinoma clinical and RNA sequencing (RNA-seq) data as the training of the present study the queue, Nineteen normal bladder tissues and 412 bladder urothelial carcinoma (BLCA) tissues were included. In addition, from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) to download the transcriptome data of patients with T2DM, There are GSE41762 (including human islet tissue from 57 nondiabetic individuals and 20 patients with T2DM) and GSE38642 (including human islet tissue from 54 nondiabetic individuals and 9 patients with T2DM). WGCNA screened common module genes To identify gene modules associated with BLCA and T2DM, WGCNA was used to analyze TCGA-BLCA and GSE38642 datasets. Firstly, according to the scale-free topological criterion, a “soft” threshold power (β) is established, which is helpful for the construction of biologically meaningful scale-free networks. Then, the topological overlap matrix (TOM) was derived from the adjacency matrix, and the dynamic pruning tree algorithm was used to identify the gene modules. Finally, the p-value of Pearson correlation coefficient, characteristic genes and disease traits were used to identify the key modules associated with BLCA and T2DM, with a P < 0.05 criterion. Finally, the important module genes in WGCNA of BLCA and T2DM were combined, and the module intersection genes were identified by Venn diagram. GO and KEGG enrichment analysis GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment analyses of common potential target genes were performed using the clusterProfiler package (an R package for comparing biological themes among gene clusters). GO is used to annotate the biological processes, molecular functions, and cellular components of genes. Gene pathways were annotated using the KEGG. Enrichment was statistically significant when P < 0.05. Construction of PPI networks The 42 common potential target genes were entered into the String platform, independent genes were removed, and the key genes were screened using Cytoscape software to establish a PPI network. Subsequently, the TOP10 core genes in the PPI network were obtained using the Stress algorithm and were used for subsequent analyses. Machine learning algorithms were used to screen key genes Two machine learning algorithms, LASSO and support vector machine recursive feature elimination (SVM-RFE), were used to screen the feature genes associated with bladder urothelial carcinoma in patients with T2DM. LASSO is a regression method used for variable selection to enhance model interpretability and prediction accuracy. SVM-RFE performs feature selection based on the maximum interval principle of SVM, and iteratively optimizes training and feature importance score to screen features. The intersection genes of the results of the two methods were considered as key common feature genes. Expression of key common feature genes and receiver operating characteristic (ROC) assessment The expression levels of the key common characteristic genes were evaluated in the TCGA-BLCA, GSE38642, and GSE41762 cohorts respectively. Subgroups were compared with the use of the Wilcoxon rank-sum test, and P < 0.05 were considered to be significant. ROC curve was used to evaluate the value of genes in the diagnosis of BLCA and T2DM, and AUC > 0.7 indicated a significant diagnostic difference. In TCGA bladder urothelial carcinoma dataset, survival differences between groups with different expression levels of key common feature genes were studied, and survival curves were drawn. Clinical samples and immunohistochemistry In this study, 112 bladder cancer tissue samples (including 28 lymph node metastasis tissue samples of bladder cancer patients and 15 bladder cancer tissue samples of patients with T2DM) were collected from the First Affiliated Hospital of Henan University of Science and Technology. This study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Henan University of Science and Technology. Informed consent was obtained from all patients before the study. Paraffin-embedded samples were first sequentially immersed in xylene I/II to achieve complete deparaffinization, hydrated in graded alcohol, and subsequently impregnated in 3% H2O2 -methanol solution to block endogenous catalase activity. After antigen repair, the cells were blocked for 30 to 60 min at room temperature and incubated overnight at 4 ° C with appropriate dilutions of the primary antibody solution. After washing, the secondary antibody working solution matching the primary antibody was added and incubated for 1 h at room temperature. Finally, the tissue specimens were stained with DAB and observed under a microscope. Cell culture The human bladder urothelial carcinoma cell lines T24 and RT112 used in this study were purchased from HyCyteTM (Starfish TCH-C352, TCH-C448) and cultured in 1640 medium containing 10% fetal bovine serum. All cells were cultured under standard culture conditions. To mimic the high glucose microenvironment in diabetic pathological state, human bladder urothelial carcinoma cell lines were cultured in high D-glucose (25 mM) medium 24 h before the experiment, while the control group was cultured in normal D-glucose (5.5 mM) medium. shRNA for PLAU was designed and synthesized by Ribo-Bio, and transfection was performed according to its standard procedures. Western Blot After washing the cells twice, RIPA lysate was added and the supernatant collected by centrifugation. Protein concentration was determined using the BCA kit, and proteins were separated by SDS-PAGE electrophoresis and transferred to PVDF membranes. This was followed by blocking with skim milk powder for 1 h, adding primary antibody to incubate overnight, washing membranes three times with TBST the next day, incubating with secondary antibody for another 2 h at room temperature, and finally washing membranes three times with TBST. Transwell 80 µL of matrix glue was added to the chamber and placed in a 37 °C incubator until it set completely. Subsequently, the cells were digested and centrifuged, and 5 × 105 cells were seeded into chambers containing matrigel. 600 µL of medium was added to the lower chamber and incubated for 48 h. At the end of the incubation period, the cells were fixed with 4% paraformaldehyde for 30 min, further stained with 0.1% crystal violet, and finally observed and imaged under a microscope. Edu assay Cells were seeded in 6-well plates at a density of 20 × 105/ well and cultured in an incubator at 37 ° C. After the cells were stable, the cells were processed according to the standard operating procedure of BeyoClickTM EDU-555 cell proliferation assay kit (Beyotime). Finally, confocal microscopy was used to photograph and unfold the analysis. Screening of small molecule compounds or drugs by molecular docking In this study, the PLAU protein structure was first obtained from the PDB database, imported into Autodock software and saved in pdbqt format. Then, PLAU gene was searched in CTDbase database to obtain related drug and disease information. Having identified small molecule compounds that reduced PLAU expression, these small molecules were processed using Autodock, including hydrogenation, setting as ligands, selection of torsion bonds, and saved in pdbqt format. After setting the docking parameters, the files of the optimal conformations were studied and visualized by pymol software. Statistic and software Data processing and analysis were accomplished by R 4.4.3. We conducted statistical analysis employing Student’s t-test. Overall survival was analyzed using the Kaplan‒Meier method and compared using the log-rank test. Differences with P < 0.05 were considered statistically significant. Results Identification of DEGs in type 2 diabetes and BLCA Volcano plot analysis (Fig. 1A) identified differentially expressed genes (DEGs) in BLCA relative to adjacent normal tissues, with thresholds of |log2(FC)| > 1 and P < 0.05, revealing distinct upregulated and downregulated gene subsets, which were further validated by hierarchical clustering (Fig. 1B) that effectively segregated BLCA tumor samples from normal controls. Parallel analysis of diabetic samples (Fig. 1C) using identical thresholds detected DEGs that also formed distinct expression signatures, as confirmed by hierarchical clustering (Fig. 1D) showing clear separation between diabetic and normal control samples. Venn diagram analysis (Fig. 1E-F) revealed the overlap between BLCA- and diabetes-related DEGs: 8 genes were commonly upregulated, while 14 genes were commonly downregulated, highlighting potential molecular candidates linking BLCA and diabetes pathogenesis. Fig. 1Identification of differentially expressed genes (DEGs). (A) The volcano plot shows the distribution of DEGs in the TCGA-BLCA dataset. (B) The heatmap shows the top 20 genes with the most significant differential expression in the TCGA-BLCA dataset. (C-D) Volcano plot and heatmap of DEGs in the GSE38642 dataset of diabetes. (E-F) The overlapping Venn diagram of the upregulated and downregulated genes for BLCA and diabetes Identification of differentially expressed genes (DEGs). (A) The volcano plot shows the distribution of DEGs in the TCGA-BLCA dataset. (B) The heatmap shows the top 20 genes with the most significant differential expression in the TCGA-BLCA dataset. (C-D) Volcano plot and heatmap of DEGs in the GSE38642 dataset of diabetes. (E-F) The overlapping Venn diagram of the upregulated and downregulated genes for BLCA and diabetes Module genes were screened by weighted gene co-expression network analysis (WGCNA) Firstly, WGCNA analysis was conducted on the TCGA bladder urothelial carcinoma dataset, which included 19 normal tissues and 412 BLCA tissues. During the construction of the co-expression network, a “soft” threshold power (β) of 8 was set based on the scale-free topological standard, thereby identifying 9 gene modules related to bladder urothelial carcinoma (Fig. 2A-B). Similarly, a WGCNA analysis was conducted on the GSE38642 diabetes dataset, which included human islet tissues from 54 non-diabetic individuals and 9 diabetic patients. According to the scale-free topology standard, the “soft” threshold power (β) was set to 7, and thereby 9 gene modules related to diabetes were identified (Fig. 2D-E). Through heat map and Spearman correlation coefficient evaluation, it was found that the MEblue (r = 0.39, P = 1e-08) and MEgreenyellow (r = 0.16, P = 0.02) modules were positively correlated with bladder urothelial carcinoma, and thus were confirmed as key modules for further analysis (Fig. 2C). They contain 3808 and 341 genes respectively. Meanwhile, the MEblack (r = 0.28, P = 0.03) and MEpink (r = 0.39, P = 0.002) modules were positively correlated with diabetes and were selected as the key modules for subsequent analysis (Fig. 2F), containing 60 and 60 genes, respectively. Fig. 2Weighted gene co-expression network analysis (WGCNA) was performed. (A-B) The related parameters of WGCNA analysis in TCGA-BLCA dataset. (C) Heat maps of gene module-trait relationships in patients with BLCA, with numbers indicating the corresponding correlations and p-values. (D-E) Relevant parameters for WGCNA analysis of the GSE38642 dataset. (F) Heat maps of gene module-trait relationships in patients with T2DM, with numbers indicating the corresponding correlations and p-values Weighted gene co-expression network analysis (WGCNA) was performed. (A-B) The related parameters of WGCNA analysis in TCGA-BLCA dataset. (C) Heat maps of gene module-trait relationships in patients with BLCA, with numbers indicating the corresponding correlations and p-values. (D-E) Relevant parameters for WGCNA analysis of the GSE38642 dataset. (F) Heat maps of gene module-trait relationships in patients with T2DM, with numbers indicating the corresponding correlations and p-values Identification and enrichment analysis of common characteristic genes Venn diagram analysis identified 21 overlapping genes across four datasets (GEO black, TCGA blue, TCGA greenyellow, GEO pink), which were subjected to subsequent functional and interaction analyses (Fig. 3A). By adding the WGCNA genes to the DEGs, a total of 42 genes were obtained. A protein-protein interaction (PPI) network (Fig. 3B) was constructed, revealing functional enrichment in processes including response to type I interferon, cellular response to type I interferon, response to virus, regulation of viral genome replication, and negative regulation of viral process. GO enrichment analysis (Fig. 3C) highlights the top 10 biological processes (BP), molecular functions (MF), and cellular components (CC). Among them, the most prominent BP term is the innate immune response, the CC term is cytosol, and the MF term is protein binding. Disease and pathway enrichment analysis (Fig. 3D) demonstrated significant associations with coronavirus disease (COVID-19), hepatitis C, influenza A, measles, and pathways such as cytokine-cytokine receptor interaction, estrogen signaling pathway, and relaxin signaling pathway. Further modular analysis of the PPI network (Fig. 3E) and a key subnetwork (Fig. 3F) centered was performed, and a correlation heatmap (Fig. 3G) confirmed strong positive co-expression among these core genes, underscoring their coordinated role in antiviral immune responses. Fig. 3Construction and enrichment analysis of PPI network. (A) Venn diagram to identify common pathogenic genes. (B) PPI networks created using the GeneMANIA database. (C) Conducted GO analysis on 42 candidate common feature genes. (D) Conducted KEGG analysis on 42 candidate common feature genes. (E) Perform PPI network analysis on the candidate common feature genes using Cytoscape. (F) The top ten common feature genes in PPI network analysis. (G) Matrix diagram shows the correlations among the top ten common feature genes Construction and enrichment analysis of PPI network. (A) Venn diagram to identify common pathogenic genes. (B) PPI networks created using the GeneMANIA database. (C) Conducted GO analysis on 42 candidate common feature genes. (D) Conducted KEGG analysis on 42 candidate common feature genes. (E) Perform PPI network analysis on the candidate common feature genes using Cytoscape. (F) The top ten common feature genes in PPI network analysis. (G) Matrix diagram shows the correlations among the top ten common feature genes Screen key diagnostic biomarkers using machine learning algorithms This study further adopted a multi-stage machine learning strategy to systematically screen for comorbidity diagnostic markers related to BLCA and T2DM. Firstly, in the TCGA-BLCA dataset, 10 core genes were subjected to feature dimension reduction through LASSO regression analysis, and the optimal penalty coefficient (λ) was determined through ten-fold cross-validation, thereby selecting 7 key candidate genes (Fig. 4A-B). Subsequently, the Support Vector Machine Recursive Feature Elimination algorithm (SVM-RFE) was applied to identify 6 most representative common feature genes from the original feature set based on the ranking of feature weights (Fig. 4C-D). The intersection of these two feature selection algorithms was analyzed through a Venn diagram, and 5 common feature genes were identified (Fig. 4E). Using the same method in the GSE38642 dataset, 1 common characteristic gene was identified (Fig. 4F-J). Ultimately, the PLAU gene was identified as a key common characteristic gene for both BLCA and T2DM. Subsequently, the optimal cut-off point for grouping in the survival analysis of the PLAU gene was determined using the maximum selection test statistics from the R package survminer, and survival analysis was conducted to compare the survival differences between groups and the impact of individual genes on prognosis. The results showed that in the TCGA-BLCA dataset, the survival time of the PLAU high-expression group was shorter compared with the PLAU low-expression group (Fig. 4K). This gene not only demonstrated significant importance in the feature screening of both algorithms, but also the survival differences between groups further verified its key role in the prognosis of patients with bladder urothelial carcinoma. Fig. 4Machine-learning algorithms were used to screen for key diagnostic biomarkers. (A-B) Through LASSO analysis, 7 candidate genes were selected in the TCGA-BLCA dataset. (C-D) Through SVM-RFE analysis, 6 candidate genes were selected in the TCGA-BLCA dataset. (E) Venn diagram. (F-G) Through LASSO analysis, 5 candidate genes were selected in the GSE38642 dataset. (H-I) Through SVM-RFE analysis, 1 candidate gene was selected in the GSE38642 dataset. (J) Venn diagram. (K) Compare the survival differences between the group with low expression of PLAU and the group with high expression of PLAU Machine-learning algorithms were used to screen for key diagnostic biomarkers. (A-B) Through LASSO analysis, 7 candidate genes were selected in the TCGA-BLCA dataset. (C-D) Through SVM-RFE analysis, 6 candidate genes were selected in the TCGA-BLCA dataset. (E) Venn diagram. (F-G) Through LASSO analysis, 5 candidate genes were selected in the GSE38642 dataset. (H-I) Through SVM-RFE analysis, 1 candidate gene was selected in the GSE38642 dataset. (J) Venn diagram. (K) Compare the survival differences between the group with low expression of PLAU and the group with high expression of PLAU Evaluation of the expression characteristics and diagnostic capabilities of the key common feature gene PLAU In this study, the expression of PLAU in the patient cohort was analyzed in the TCGA-BLCA dataset and compared with that in the control group. The results showed that the expression of PLAU increased in bladder urothelial carcinoma tissues (Figure S1A), and its diagnostic ability for bladder urothelial carcinoma was demonstrated through the ROC curve, with an AUC value of 0.787 (Figure S1B). In the GSE38642 dataset for T2DM, the expression level of PLAU in diabetic pancreatic tissue is also relatively high (Figure S1C), and its AUC value for the diagnostic ability of T2DM is 0.818 (Figure S1D). Subsequently, in the GSE41762 dataset for T2DM, further analysis confirmed these findings (Figure S1E-F). Data validation indicates that PLAU has diagnostic value in diabetes-related bladder urothelial carcinoma, supporting its potential as a diagnostic biomarker. The relationship between PLAU expression and the pathological characteristics of bladder urothelial carcinoma as well as diabetes This study, through bioinformatics and literature analysis, found that high expression of PLAU is closely related to poor prognosis in patients with bladder urothelial carcinoma and is positively correlated with the malignancy of the tumor. Although PLAU has shown carcinogenic effects in various cancers, its specific role and mechanism in bladder cancer remain unclear. To verify the expression of PLAU protein and its relationship with the pathological characteristics of bladder cancer, immunohistochemical staining was performed on bladder cancer samples and normal tissues. The results showed that the protein expression of PLAU in bladder urothelial carcinoma tissues was significantly higher than that in normal tissues. In addition, the protein expression of PLAU in bladder urothelial carcinoma tissues with diabetes mellitus was significantly increased compared with that in tissues without diabetes mellitus (Fig. 5A). Meanwhile, with the increase of malignancy degree of bladder urothelial carcinoma, the protein expression of PLAU also significantly increased (Fig. 5B-C). The elevated expression of PLAU suggests its involvement in the malignant progression of bladder cancer and may serve as a potential biomarker for prognosis assessment and targeted therapy. To further explore the relationship between PLAU and bladder urothelial carcinoma as well as diabetes, Western blot and qRT-PCR experiments were conducted for verification. The results showed that the Western blot, qRT-PCR and immunohistochemical staining results were consistent (Fig. 5D-I), once again confirming that the expression of protein and mRNA of PLAU was significantly upregulated in bladder urothelial carcinoma tissues with diabetes and those with a higher degree of malignancies. Fig. 5Relationship between PLAU expression and pathological features of bladder urothelial carcinoma and diabetes mellitus. (A) Immunohistochemical results of PLAU in normal bladder tissue, bladder urothelial carcinoma tissue without diabetes, and bladder urothelial carcinoma tissue with diabetes. (B) Immunohistochemical results of PLAU in normal bladder tissue, low-grade bladder urothelial carcinoma tissue, high-grade bladder urothelial carcinoma tissue, and lymph. (C) Immunohistochemical results of PLAU in stage T1, T2, T3, and T4 bladder urothelial carcinoma tissues. (D-E) Western blot and qRT-PCR results of PLAU in normal bladder tissues, bladder urothelial carcinoma tissues without diabetes mellitus and bladder urothelial carcinoma tissues with diabetes mellitus. (F-G)Western blot and qRT-PCR results of PLAU in normal bladder tissues, low-grade bladder urothelial carcinoma tissues, high-grade bladder urothelial carcinoma tissues and lymph node metastasis tissues. (H-I)Western blot and qRT-PCR results of PLAU in T1, T2, T3 and T4 bladder urothelial carcinoma tissues Relationship between PLAU expression and pathological features of bladder urothelial carcinoma and diabetes mellitus. (A) Immunohistochemical results of PLAU in normal bladder tissue, bladder urothelial carcinoma tissue without diabetes, and bladder urothelial carcinoma tissue with diabetes. (B) Immunohistochemical results of PLAU in normal bladder tissue, low-grade bladder urothelial carcinoma tissue, high-grade bladder urothelial carcinoma tissue, and lymph. (C) Immunohistochemical results of PLAU in stage T1, T2, T3, and T4 bladder urothelial carcinoma tissues. (D-E) Western blot and qRT-PCR results of PLAU in normal bladder tissues, bladder urothelial carcinoma tissues without diabetes mellitus and bladder urothelial carcinoma tissues with diabetes mellitus. (F-G)Western blot and qRT-PCR results of PLAU in normal bladder tissues, low-grade bladder urothelial carcinoma tissues, high-grade bladder urothelial carcinoma tissues and lymph node metastasis tissues. (H-I)Western blot and qRT-PCR results of PLAU in T1, T2, T3 and T4 bladder urothelial carcinoma tissues sh-PLAU can reverse the promoting effect on the proliferation and migration ability of bladder urothelial cancer cells caused by a high glucose environment In order to investigate the biological role of PLAU in bladder urothelial carcinoma, this study further constructed a cell line with PLAU deficiency. The knockdown effect of shRNA on the PLAU gene in BLCA cells was verified by Western blot. The results showed that the expression level of PLAU in the sh-PLAU group was significantly lower than that in the NC control group (Fig. 6A-B), indicating that the expression of PLAU in BLCA cells was effectively inhibited after plasmid transfection. Subsequently, through EdU experiments, this study found that T24 cells in a high-glucose environment had a stronger proliferation ability compared to those in a low-glucose environment, while T24 cells treated with sh-PLAU reversed the promoting effect of the high-glucose environment on their proliferation ability. Furthermore, the BLCA cells that have undergone oe-PLAU treatment still maintain a strong proliferative capacity in a low-glucose environment (Fig. 6C-D). The effect of sh-PLAU-treated T24 cells on migration ability was further explored through the Transwell assay. The results showed that the PLAU gene could promote the migration of BLCA cells, and the BLCA cells treated with oe-PLAU exhibited significantly enhanced migration ability in a low-glucose environment. It is considered that the low-glucose stress response of the cells led to this phenomenon (Fig. 6E-F). Fig. 6sh-PLAU can reverse the promoting effect of high glucose environment on the proliferation and migration of bladder urothelial carcinoma cells. (A-B) Western blot demonstrated the difference in PLAU protein expression between the knockdown group and the control group. (C-D) The proliferation ability of bladder urothelial cancer cells in NC + low glucose, NC+ high glucose, oe-PLAU + low glucose, and sh-PLAU+ high glucose environments was determined by Edu assay. (E-F) The migration ability of bladder urothelial cancer cells in NC + low glucose, NC+ high glucose, oe-PLAU + low glucose, and sh-PLAU+ high glucose environments was determined by Transwell assay sh-PLAU can reverse the promoting effect of high glucose environment on the proliferation and migration of bladder urothelial carcinoma cells. (A-B) Western blot demonstrated the difference in PLAU protein expression between the knockdown group and the control group. (C-D) The proliferation ability of bladder urothelial cancer cells in NC + low glucose, NC+ high glucose, oe-PLAU + low glucose, and sh-PLAU+ high glucose environments was determined by Edu assay. (E-F) The migration ability of bladder urothelial cancer cells in NC + low glucose, NC+ high glucose, oe-PLAU + low glucose, and sh-PLAU+ high glucose environments was determined by Transwell assay Molecular docking screening for targeting PLAU small molecule compounds or drugs In order to find drugs that can target the PLAU protein and treat patients with bladder urothelial carcinoma who have overexpression of PLAU, this study further conducted molecular docking research between the PLAU protein and small molecule compounds. We screened out 81 small molecule compounds that interact with PLAU proteins from the CTDbase database and used Autodock software to perform molecular docking one by one to obtain the corresponding binding energies (Table S1). Based on the lowest binding energy, we selected three compounds (EGCG, 6-chloro-HBICA, Cra_8696) and visualized them using PyMOL (Fig. 7A). The results showed that the small molecule compound epigallocatechin gallate (EGCG) exhibited a good effect in targeting the PLAU protein. Fig. 7Molecular docking. (A) Mutual binding sites between PLAU protein and the small molecule compounds EGCG, 6-chloro-HBICA and Cra_8696. (B) The proliferative capacity of bladder urothelial cancer cells in NC, oe-PLAU, and oe-PLAU+EGCG as determined by Edu assay. (C) The invasive ability of bladder urothelial cancer cells in NC, oe-PLAU, and oe-PLAU+EGCG was determined by Transwell assay Molecular docking. (A) Mutual binding sites between PLAU protein and the small molecule compounds EGCG, 6-chloro-HBICA and Cra_8696. (B) The proliferative capacity of bladder urothelial cancer cells in NC, oe-PLAU, and oe-PLAU+EGCG as determined by Edu assay. (C) The invasive ability of bladder urothelial cancer cells in NC, oe-PLAU, and oe-PLAU+EGCG was determined by Transwell assay Based on the above research results, in order to further explore the role of small molecule compounds in targeting PLAU proteins in BLCA, in this study, appropriate amounts of EGCG were added and not added respectively to the oe-PLAU bladder urothelial carcinoma cell line, and the same amount of EGCG was added to the NC group bladder urothelial carcinoma cell line at the same time. After a period of culture and through Edu and Transwell experiments, the results showed that compared with the oe-PLAU without EGCG group, the proliferation and invasive ability of bladder urothelial cancer cells in the oe-PLAU plus EGCG group decreased significantly. The small molecule compound EGCG can reverse the enhanced proliferation and invasive ability of bladder urothelial cancer cells caused by overexpression of PLAU (Fig. 7B-C). Disscusion Bladder cancer is a common malignant tumor in the urinary system, originating from abnormal proliferation of epidermal cells in the bladder. In recent years, the main treatment methods for bladder cancer patients have been surgical resection, radiotherapy and chemotherapy, etc. However, they are prone to metastasis, have a poor prognosis, and have a high recurrence rate and mortality rate [1]. This study, through WGCNA and differential gene intersection analysis, found that PLAU is a common pathogenic gene for T2DM and BLCA. Previous studies have reported that PLAU promotes tumor invasion by regulating the fibrinolytic system [19]. In pan-cancer analysis, the expression level of PLAU has been confirmed to be an independent prognostic factor for various cancers, and high expression is associated with a shortened overall survival period for patients [16]. Meanwhile, PLAU is highly expressed in various cancers including glioma and lung cancer, and is significantly associated with poor prognosis in patients. In addition, Fang et al. found that PLAU can promote the proliferation and migration of esophageal squamous cell carcinoma through the MAPK signaling pathway [20]. Ning et al. found that in lung squamous cell carcinoma (LUSC), overexpression of PLAU significantly enhances cell migration ability and is a key driver of metastasis [21]. In cervical cancer, PLAU is regulated by the transcription factor YY1 and can promote cell migration and invasion [22]. Machine learning screening confirmed that PLAU is the best biomarker for differentiating BLCA from normal tissues, and its expression level is significantly positively correlated with tumor stage and lymph node metastasis, suggesting that PLAU may become a specific indicator for cancer risk stratification in diabetic patients. This is consistent with previous studies. It is the first to reveal the characteristics related to hypoxia in bladder cancer, which can more accurately predict the survival prognosis of patients compared to traditional pathological parameters [23]. Relevant epidemiological investigations have found that patients with diabetes have an increased risk of cancer and a higher mortality rate [24, 25]. Therefore, understanding the molecular mechanism of the association between T2DM and cancer has received extensive attention. Metabolic disorders such as hyperinsulinemia and dyslipidemia existing in T2DM are considered to be the causal relationship between diabetes and cancer [26, 27]. Studies have found that in patients with hyperinsulinemia, high levels of insulin occur and insulin /IGF signaling is activated. Subsequently, the activation of the PI3K/AKT/mTOR and MAPK signaling pathways promotes cancer cell proliferation and drug resistance [28–31]. Multiple cohort studies have shown that patients with T2DM have a 1.3 to 1.5 times higher risk of developing BLCA than those without diabetes, and the overall survival of BLCA patients with T2DM is significantly shortened [32, 33]. Previous studies have shown that in the TCGA-BLCA cohort, three samples of PLAU have undergone mutations, and these are very likely to be missense mutations [16]. In the Type 2 Diabetes Knowledge Portal (T2DKP) database, the HuGE score between PLAU and T2DM is 45 (including 45 common variant BF and 1 rare variant BF), which falls within the very strong evidence range [34]. This also provides a certain degree of theoretical basis for the comorbidity analysis in this study. It is worth noting that PLAU has been confirmed to be regulated by AGEs/RAGE signaling during the fibrosis process of diabetic nephropathy [35–37], and this study further expands its function, suggesting that there may be a “PLAU-dependent” molecular bridge between metabolic disorders and tumor progression. In recent years, the association between diabetes and bladder cancer has gradually become a hot topic in tumor metabolism research. Several large-scale cohort studies have shown that the risk of bladder cancer in patients with diabetes is significantly higher than that in non-diabetic patients. A meta-analysis involving 16 studies revealed that diabetes was associated with an increased risk of bladder cancer compared to no diabetes, and there was strong evidence of heterogeneity among these studies (P < 0.0001) [38]. It is worth noting that bladder urothelial carcinoma, as the main subtype of bladder cancer, may have a unique pathophysiological basis for its association with diabetes. For instance, a retrospective study of the population in Taiwan found that the risk of high-grade urothelial carcinoma in diabetic patients was 1.7 times that of non-diabetic patients, and there was a synergistic effect with smoking status [39]. These data suggest that diabetes is not only an independent risk factor for bladder cancer but may also promote the malignant transformation of urothelial cells through specific molecular pathways. Although most studies support a positive correlation between T2DM and BLCA, some cohort studies have not found a significant association. For instance, a multi-center European study revealed that diabetes was only associated with the risk of bladder cancer in women, while no statistical difference was observed in male patients. This heterogeneity may be related to differences in the genetic background of the population, types of diabetes, or confounding factors such as obesity and smoking [40]. In addition, apart from the aforementioned abnormalities in apoptosis and metabolic pathways, inducing parapoptosis has become a promising strategy for reversing drug resistance in urinary system tumors. Parapoptosis leads to cell death through mechanisms such as endoplasmic reticulum-mitochondrial calcium overload and accumulation of reactive oxygen species, and does not rely on caspase activation. This provides a new entry point for the treatment of bladder cancer related to oxidative stress induced by high blood sugar in the diabetic microenvironment [41]. Zhang et al. recently demonstrated that curcumol, a bioactive sesquiterpenoid derived from Curcumae Rhizoma, effectively overcomes cisplatin resistance and suppresses bladder cancer progression [42]. Similarly, the PLAU inhibitor with the lowest binding energy obtained through virtual screening in this study was epigallocatechin gallate (EGCG). Epidemiological studies on T2DM have shown that foods rich in flavane-3-alcohol, especially EGCG in green tea, may help reduce the risk of diabetes [43]. EGCG is the main polyphenol in green tea, accounting for more than 50% of its total polyphenol components. A large number of scientific studies have shown that EGCG is the cause of many potential health benefits of green tea [44]. It is reported that EGCG has a wide range of potential health-promoting effects, such as reducing inflammation, preventing cancer, combating obesity, inhibiting microbial growth, resisting infection, protecting the nervous system and regulating immune responses [45–48]. According to the experimental research results, EGCG can still effectively inhibit the proliferation of bladder urothelial cancer cells in a high-glucose microenvironment and reverse the invasion phenotype induced by high glucose. It is worth noting that EGCG has relatively low toxicity to normal bladder epithelial cells. This selectivity may stem from the abnormal activation state of the PLAU/uPAR pathway in bladder urothelial cancer cells with T2DM. Compared with the indirect effects of traditional hypoglycemic drugs on cancer, directly targeting PLAU provides a new approach for simultaneously intervening in metabolic abnormalities and tumor progression. In the future, the combined therapy of EGCG and immune checkpoint inhibitors (such as anti-PD-1) can be explored to synergistically improve the immune microenvironment of diabetes-related tumors. Furthermore, As we continue to unravel the complexities of the human proteome, noncanonical proteins are likely to play an increasingly important role in our understanding of human biology and in the development of innovative diagnostic and therapeutic strategies across a wide spectrum of diseases. It is expected to reshape our approaches to disease treatment and personalized medicine in the coming years [49]. In conclusion, this study utilized multiple algorithms to screen out the common characteristic gene PLAU for BLCA and T2DM, and this gene was highly expressed in samples of both diseases. In addition, through molecular docking, it was found that the PLAU small molecule inhibitor EGCG can significantly inhibit the proliferation and migration of bladder cancer cells, providing a new approach for the treatment of bladder cancer complicated with T2DM. However, this study has certain limitations. Firstly, there is a lack of further research on the impact of different culture conditions on the therapeutic effect of the candidate drugs. Secondly, the experimental work mainly focuses on phenotypic observation and lacks validation through in vivo animal experiments. Therefore, in the future, it is necessary to conduct in-depth research on the therapeutic effect of the drugs under different glucose concentration culture conditions, and collect clinical data and conduct in vivo animal experiments for verification. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3