Prognosis and immune landscape of bladder cancer can be predicted using a novel miRNA signature associated with cuproptosis

Data collection

RNA sequencing data and clinical characterization for BLCA were retrieved from the TCGA database as of November 28th, 2022. The dataset comprised 418 tumor samples and 19 normal tissue samples, and RNA-Seq data of miRNAs were also obtained according to the gene annotations from TCGA. Additionally, one Gene Expression Omnibus (GEO) dataset, GSE13507, consisting of 256 samples, was obtained from GEO.

Cell lines

The human bladder urothelial carcinoma cell line T24 and UMUC3 and human normal urine epithelial cell line SV-HUC-1 were supplied by the Cell Bank of the Chinese Academy of Sciences. The cell lines were cultivated in standard DMEM medium (MedChem Express, Monmouth Junction, NJ, USA) supplemented with 10% foetal calf serum (FCS) and 1% penicillin‒streptomycin (P/S) purchased from Gibco (Grand Island, NY, USA).

Tissue samples

Paired human samples of BLCA tissues and neighbouring ordinary mucosa were obtained from patients treated at the First Affiliated Hospital of Shandong Second Medical University (n = 3) for the following FISH assay. All cases were diagnosed as primary BLCA without any treatment before surgery. Informed consent was obtained from all participants before sample collection. This research received permission (No. KYLL20240613-1) from the ethical committee of the First Affiliated Hospital of Shandong Second Medical University and was conducted according to the Helsinki.

Screening and differential expression analysis of cuproptosis-associated miRNAs

Firstly, we conducted a co-expression correlation analysis of miRNA and cuproptosis-related gene expression profiles using the “limma” “edgeR” and “pheatmap” packages. We obtained 19 cuproptosis related genes from previous studies (Tsvetkov et al., 2022; Xu et al., 2022; Zhang, Sun & Zhang, 2022). miRNAs differentially expressed between normal and bladder cancer tissues were identified using a Log2FC cutoff of > 1.5 and p < 0.01.

Construction of the prognostic cuproptosis related miRNA signature

The BLAC samples were divided into training and testing groups randomly. Univariate Cox regression was conducted in the training group, and cuproptosis related miRNAs were identified by Lasso. These miRNAs were further analyzed by multivariate Cox regression analysis, and only those with a p value < 0.01 were selected as the basis for the best model for signature construction. Then, risk scores were investigated using the equation: Risk score = Exp miRNA1 × β miRNA1 + Exp miRNA2 × β miRNA2 + Exp miRNA3 × β miRNA3 + • + Exp miRNAn × β miRNAn.

Nomogram and calibration

To develop a comprehensive prognostic model for bladder cancer patients, the risk scores were combined with various clinical pathological factors, and a nomogram was constructed. The nomogram was designed to predict overall survival (OS) rates of patients. The Hosmer-Lemeshow test was applied to draw calibration curves, which assessed the predictive performance of the nomogram.

PCA, GO, and KEGG analysis

miRNAs associated with cuproptosis in bladder cancer samples were analyzed using PCA analysis to visualize distribution of samples. The intersected genes from different risk groups and different immune-score groups were then subjected to Gene Ontology (GO) analysis, which comprised three components: biological process (BP), cellular component (CC), and molecular function (MF). Additionally, differentially expressed KEGG pathways were evaluated using the Hs. “eg.db”, “clusterProfiler”, and “enrichplot” packages.

Immune profile analysis

Immune cells and mediators in oncology are directly involved in the process of cell death in the tumor microenvironment (TME) (Hinshaw & Shevde, 2019). To analyze the relationship of immunity and stromal in TME, the immune and stromal score of each sample was calculated by the “estimate” package. We also utilized the CIBERSORT algorithm to evaluate infiltration of 22 immune cells in TME, which including Macrophages M0, Macrophages M1, Macrophages M2, T cells CD4 naive, T cells CD4 memory resting, T cells CD4 memory activated, T cells follicular helper, T cells regulatory (Tregs), NK cells resting, NK cells activated, T cells gamma delta, neutrophils, monocyters, CD8+ T cells, plasma cells, mast cells resting, mast cells activated, dendritic cells resting, dendritic cells activated, B cells naive, B cells memory.

TMB analysis and pharmaceutical screening

The mutation data was retrieved from the TCGA website, and then integrated and analyzed using the “maftools” package. Differences in TMB and survival rates were assessed between high-risk and low-risk groups, with a significance threshold of <0.05. Furthermore, the IC50 values of drugs commonly employed in bladder cancer were predicted for the high- and low-risk groups using “pRRophetic” package.

Validation the prognostic of cuproptosis‑related miRNAs in OncomiR datasets

Bladder cancer patients were stratified into high and low expression groups based on the levels of cuproptosis-related miRNA expression in the OncomiR online database. Additionally, a signature composed of eight cuproptosis-related miRNAs was established and applied to assess the survival analysis of these miRNAs in the OncomiR dataset.

Quantitative real-time polymerase chain reaction

We first extracted total RNA and reverse transcribed into cDNA using a reverse transcription kit (Accurate Biotechnology (Human) Co., Ltd., Guangzhou, China), according to the manufacturer’s instructions. All steps were performed according to the manufacturer’s instructions and our previous study (Yu et al., 2024). Primers for amplification were synthesized by Huada Gene (Beijing, China) and were listed in Table S1. U6 was used as the internal control of miRNA.

Fluorescence in situ hybridization

The subcellular localization of four kinds of miRNA were measured using the fluorescence in situ hybridization (FISH) kit (F11101/50 and F21101/50, Genepharma, shanghai, China). The cell and tissue slides were treated with miRNA probe hybridization solution labeled by SA-Cy3 according to the manufacturer’s instructions. Brifely, the slide was hybridized at 42 °C for 16 h and immersed in 2 × SSC (saline sodium citrate buffer), followed by immersion in 70% ethanol for 3 min and stained with DAPI for 10 min. The slide was imaged using the Olympus fluorescence microscope and AxioVision image analysis software. Probes for FISH were listed in Table S2.

Transwell migration and invasion assays

All steps were performed as our previous study (Yu et al., 2024). Brifely, the BLCA cells (5 × 10⁴ cells/well) were placed on the upper Transwell chamber (8 μm, Corning, NY, USA) with complete culture medium incubated for 24 h, then stained with 0.1% crystal violet after fixed with 10% formaldehyde. Images were taken under a light microscope. The sequence of interfered inhibitors and mimics for cells were listed in Tables S3 and S4.

Statistics

The quantitative variables were analyzed by the independent-samples t-test. Univariate and multivariate Cox regression analysis were performed to identfy the independent prognostic factors of OS. The nomogram was constructed using the survival and rms package in R based on the coefficients of the multivariable model. The Kaplan-Meier method was utilized to estimate the survival rate of OS between groups, and the log-rank test was employed to compare significant differences. The discriminatory performance of the nomogram was assessed by calculating AUC. Calibration curves were also plotted to compare the predicted nomogram results with the actual Kaplan–Meier estimates of OS possibility in the cohort. The significant level in the present study was two tailed p < 0.05.

The expression of cuproptosis related genes in bladder cancer and prognosis-related miRNAs with coexpression of cuproptosis

As depicted in Fig. 1A, approximately nine out of nineteen cuproptosis-related genes, including NFE2L2, NLRP3, ATP7A, SLC31A1, LIPT2, MTF1, CDKN2A, GCSH and DLST, displayed significant differences between normal and bladder cancer tissues in the TCGA database. In addition, these cuproptosis related genes were all significantly associated with the prognosis of bladder cancer patients in both TCGA and GEO dataset (Figs. 1B–1F, Figs. S1A–S1G). Subsequently, the TargetScan database was utilized to screen 25,702 miRNAs as cuproptosis related miRNAs. In TCGA database, 135 up-regulated miRNAs and 96 down-regulated miRNAs were identified as differentially expressed DE-miRNAs (Figs. 1G, 1H).

Construction of cuproptosis-related miRNA signature related to prognosis of bladder cancer patients in TCGA cohort

A total of 61 candidate miRNA with a p-value less than 0.01 were identified by univariate cox regression analysis. Subsequently, these miRNAs were further screened using LASSO-Cox regression analysis (Figs. 1I, 1J). Ultimately, a miRNA signature consisting of eight-miRNA was developed through multivariate cox regression analysis. The corresponding risk scores of bladder cancer sample were calculated using multivariate cox regression analysis in the TCGA datasets, where the risk score was computed as follows: miR-125b-2-3p * 0.43647 − let-7c-5p * 0.45759 + miR-145-3p * 0.32586 + miR-409-3p * 0.22515 + miR-548a-5p * 1.69183 − miR-5682 * 0.49332 − miR-625-3p * 0.27289 + miR-634 * 30.88948. All of the eight miRNAs were significantly associated with patients’ prognosis (Fig. 2A). The overall survival rate of patients in the high-risk group was significantly shorter than that of those in the low-risk group in overall, training and testing groups (Fig. 2B).

Survival analysis of the signature and independent analysis of prognostic factors

Risk curve suggested that the mortality rate of high-risk patients was higher than that of low-risk patients (Fig. 3A). The heatmap demonstrated that the eight miRNAs could be used to discriminate between high- and low risk levels, where let-7c-5p, miR-125b-2-3p and miR-548a-5p were the high-risk miRNAs, whereas miR-625-3p was the low-risk miRNAs (Fig. 3B). Additionally, the predictive accuracy of the risk score was evaluated using ROC curves for the all, training and testing group, with the AUC of the three groups for the risk score being 0.711, 0.736 and 0.687 respectively (Fig. 3C).

Furthermore, univariate and multivariate cox regression analysis were employed to determine whether the risk score might be independent of other clinical features as a prognostic factor. The results of multivariate cox regression analysis indicated that age (HR = 1.021, 1.004–1.04, P < 0.05), gender (HR = 0.654, 0.460–0.929; P < 0.05), T (HR = 1.844, 1.223–2.779; P < 0.05) and risk score (HR = 3.424, 2.342–5.004; P < 0.05) were significantly associated with overall survival, indicating that the risk signature is an independent prognostic factor for patients with bladder cancer (Fig. 3D). A nomogram including independent prognostic factors was constructed to evaluate the prognosis of bladder cancer patients at 1, 3, and 5 years (Fig. 3E). Moreover, the calibration curve of nomogram model for the prognosis of survival probability displayed a high consistency between the prognostic and measured values (Fig. 3F).

Stratified prognostic analyses of patient clinical features

Patients with bladder cancer were stratified based on various clinicopathological characteristics and molecular subtype to investigate the association between survival probability and risk score. Notably, the overall survival rate was markedly higher in the low-risk group than in the high-risk group across most subgroups, except for the low-grade group (Figs. 4A–4H). These findings imply that the risk score model has promising prognostic value for predicting the survival outcome of bladder cancer patients with diverse clinicopathological features.

Immune characteristics

To investigate the significance of miRNA risk status in tumor microenvironment, we quantified the level of immune cell infiltration between high-risk and low-risk groups of bladder cancer using a heatmap (Fig. 5A). The results suggested that the CD8+ T cells and CD4+ T cells were negatively correlated with the risk score, while Treg cells, CD4+ memory T cells, mast cells and M0 macrophages were positively associated with the risk score (Figs. 5B–5G). The correlations of infiltrating immune cells and related functions were presented in (Figs. 5H, 5I). Specifically, CD8+ T cells, DCs cells and Th1 cells were positively correlated with tumor-infiltrating lymphocytes (TIL) (Fig. 5H). In addition, a positive association was found between T-cell co-inhibition and check point in the immune function (Fig. 5I). We also used the TIDE score to evaluate whether the BLCA patients would respond well to ICB therapy. Our findings indicated that the high-risk group had a higher TIDE score, although there was no significant difference (Fig. S1H). Furthermore, with regard to TME scores, stromal scores and ESTIMATE scores were higher in high-risk patients, but there was no difference in immune scores between the two risk groups (Fig. 5J). Overall, these results suggest that miRNA risk status could be used as biomarker for predicting the infiltration of immune cells and TME status in bladder cancer.

The principal component analysis and biological pathways analyses

Principal component analysis (PCA) was employed to examine the distinctions between two groups in terms of four expression profiles (cuproptosis associated miRNAs, total gene expression profiles, cuproptosis genes, and risk model) (Figs. 6A–6D). Risk model exhibited the greatest discriminatory capability in distinguishing the low and high-risk populations. The Venn diagram in Fig. 6E depicts the cross-genes among different risk scores and different immune scores (Fig. 6E). KEGG analysis of these genes indicated a predominance of focal adhesion and PI3K-AKT signaling pathway (Fig. 6F), while Gene Ontology analysis revealed a strong association between cuproptosis associated genes and extracellular adhesion (Fig. 6G). Furthermore, a correlation analysis between the intersection genes and immune cells following univariate cox regression indicated a positive correlation between these genes and Treg cells and M0 macrophage, and a negative correlation with CD8+ T cells (Fig. 6H).

TMB analysis and drug sensitivity analysis

We examined mutations in cohorts of both high-risk and low-risk individuals. The results indicated that for the majority of genes, the mutation frequency in the low-risk group was higher than that in the high-risk group (TP53: low risk, 49%; high risk, 48%. TTN: low risk, 49%; high risk, 38%; MUC16: low risk, 26%; high risk, 25%) (Figs. 7A, 7B). Additionally, a statistically significant difference in TMB was observed between the high-risk and low-risk cohorts (Fig. 7C). Moreover, survival analysis suggested a probable distinction between patients with high and low TMB, with those in the high TMB group exhibiting significantly better overall survival than those in the low TMB group (Figs. 7D, 7E). The present study utilized the “pRRophetic” package to identify potentially effective antitumor drugs for bladder cancer treatment, including gemcitabine, 5-fluorouracil, methotrexate, mitomycin, doxorubicin, and axitinib. Subsequently, the sensitivity of these drugs was evaluated, revealing that the high-risk group displayed a higher IC50 value (concentration that inhibits cell growth by 50%) compared to the low-risk group for all drugs except axitinib, suggesting that high-risk patients may have lower sensitivity to these antitumor drugs (Figs. 8A–8F).

Prognostic analysis of the external database and validation of cuproptosis-related miRNAs in vitro experimental

The current study examined a set of eight cuproptosis related miRNAs (hsa-let-7c-5p, hsa-miR-125b-2-3p, hsa-miR-145-3p, hsa-miR-409-3p, hsa-miR-548a-5p, hsa-miR-625-3p, hsa-miR-634 and hsa-miR-5682) listed in OncomiR database. With the exception of hsa-miR-5682, all eight miRNAs were detected in the observed sample (Figs. 9A–9D and Figs. S2A–S2C). Further, survival analysis were conducted to evaluate the association between single cuproptosis-related miRNAs and prognosis of patients with bladder cancer, with results indicating that four of the seven miRNAs analyzed were significantly associated with patient prognosis (Figs. 9A–9D). In addition, a miRNAs signature constructed to evaluate the prognosis of bladder cancer patients suggested that patients with high risk score demonstrated a poor prognosis than the patients with low risk score (Fig. 9E). These findings were consistent with the results obtained from the TCGA dataset. Moreover, to further evaluate the prognostic value of this cuproptosis associated miRNAs, in vitro cell experiments were performed to assess the expression level of the four identified miRNAs. As the RT-qPCR results suggested, these four miRNAs had an overall trend of increased expression levels in bladder cancer cells compared to bladder normal cells (Figs. 9F–9I). In addition, the FISH experiment confirmed the expression levels of four miRNAs in bladder cancer cell lines T24 and UMUC-3. It was found that miR-125-b-2-3p, miR-145-3p, Mir-4093p, and miR-625-3p were expressed in both cell lines (Figs. 10A, 10B). Additionally, these four miRNAs showed significantly higher expression in three pairs of bladder cancer tissues compared to adjacent tissues (Figs. 10C, 10D). Functional verification revealed that all four miRNAs promoted bladder cancer migration, invasion (Figs. 10E, 10F), and proliferation (Figs. 10G–10J) to varying degrees.