Transient Receptor Potential Melastatin 4 (TRPM4) is associated with an immuno-stimulatory TME and better prognosis in BLCA

Abstract

Objective

Bladder cancer (BLCA) is one of the most invasive and malignant tumors in the world, thus leading to the urgent need of an accurate and recognizable biomarker. Several previous research has shown that Transient Receptor Potential Melastatin 4 (TRPM4) plays an important role in measuring vascular fragility and endothelial function. However, the role of TRPM4 in tumor hasn’t been explored.

Methods

We obtained the clinical data and gene expression profiles of The Cancer Genome Atlas (TCGA) datasets via UCSC Xena. The tumor microenvironment (TME) was also evaluated by the amount of immunomodulators. The role of TRPM4 was demonstrated using the dataset GSE130001 via the website (http://tisch.comp-genomics.org/). We also validated the results of the pathway analysis through the Polymerase Chain Reaction (PCR).

Results

Patients with higher TRPM4 expression showed better prognosis. We found out that high TRPM4 tended to shape an immuno-stimulatory TME with less T cell exhaustion and lower level of immune overdrive. We demonstrated that TRPM4 could predict the molecular subtype and different therapy options in BLCA. We also found the important influence of TRPM4 in P53 pathway and TNF-a pathway. Last but not least, we developed a prognostic model centered on TRPM4 using ten machine learning algorithms, which showed excellent predictive ability.

Conclusion

High TRPM4 shapes a more immuno-stimulatory tumor microenvironment and can not only predict the immune phenotypes but also predict the clinical phenotypes in BLCA. High TRPM4 predicts better prognosis. TRPM4 is a potential biomarker in the future for identifying the immunogenicity and immune-therapy probability of BLCA.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-04080-z.

Introduction

BLCA is one of the most common urogenital malignancies in the world, with high aggressiveness and malignancy potential [1, 2]. In 2022, approximately 92,000 new cases of BLCA were reported in China [3]. As a result, an effective and precise biomarker is urgently needed. TRPM4 is a nonselective sodium ion channel, which can be regulated and controlled by the NC1,thus leading to sodium ions overload [4]. TRPM4 is associated with secondary bleeding due to vascular fragility. Because genes related to the fragility of blood vessels are involved in the process of angiogenesis, and once the regulatory mechanisms of angiogenesis are disrupted, it can lead to excessive or abnormal formation of blood vessels. This, in turn, makes tumor blood vessels prone to rupture, potentially facilitating tumor invasion and metastasis. Therefore, TRPM4 may be related to the prognosis of tumors [5]. In addition, there is also previous research showing that TRPM4 is highly expressed in human colorectal tumor buds and contributes to proliferation, cell cycle, and invasion of colorectal cancer cells [6]. However, the role of TRPM4 in bladder cancer hasn’t been explored and demonstrated. As a result, we highlighted TRPM4 to explore whether it plays an important role in BLCA.

Immuno-stimulatory tumor microenvironment (TME) is a local ecosystem within and around a tumor that is conducive to the initiation, activation, and effector function of the anti-tumor immune response. Unlike an immuno-suppressive TME, which inhibits immune cells, this favorable milieu is characterized by a high influx and infiltration of cytotoxic T cells. This environment promotes immune cell survival and cytotoxicity, ultimately leading to enhanced tumor cell killing and is often associated with a positive response to cancer immunotherapies [7]. T cell exhaustion is a unique adaptive cellular state that develops in response to sustained antigen stimulation, particularly in the context of chronic infections and tumors. Its hallmark is the progressive loss of effector function, accompanied by the sustained expression of various inhibitory receptors [8]. This progressive dysfunction impairs their effector mechanisms, ultimately diminishing cytotoxic efficacy against neoplastic cells and compromising immune surveillance of pathological tissues. An immune overdrive TME with high immune cell infiltration, low tumor purity, and high levels of ICs. Immune overdrive is often associated with the expression levels of the TGF-β family [9]. However, the relationship between TRPM4 and TME in bladder cancer hasn’t been explored, and the extent of T cell exhaustion and immune overdrive within the TME shaped by high expression of TRPM4 has also not been investigated.

In this study, we firstly described the role of TRPM4 as an important tumor-suppressing gene, thus shaping an immuno-stimulatory TME and was associated with better prognosis in BLCA patients by controlling the immune overdrive and decreasing the T cell exhaustion. The degree of immune infiltration was measured by 133 immunomodulators, including chemokines, immunoinhibitors, immunostimulators, MHC and receptors. We found out that high expression of TRPM4 decreased the expression of immunomodulators. However, the degree of T cell exhaustion is lower in the high-TRPM4 group. The immune overdrive is also controlled in the high TRPM4 group, thus leading to better prognosis in the BLCA patients. The bioinformatics analysis also showed that TRPM4 can not only predict the immune phenotypes but also predict the clinical phenotypes in BLCA. To further explore the role of TRPM4 in BLCA, we finished the single cell analysis using the dataset GSE130001 [10]. The result showed that TRPM4 is highly related to the epithelial C0 and C5 cluster and enriched in the P53 pathway and TNF-a pathway, which means it may play an irreplaceable role in P53 pathway and TNF-a pathway, thus explaining its powerful anticancer effects. The subsequent PCR experiments also corroborated our findings from the single-cell analysis. Since TRPM4 shaped such an anticancer TME, we hypothesize that similar results could be obtained in other types of cancer, such as breast cancer, lung cancer and so on [11–14]. The pan cancer analysis demonstrated our thinkings, such as the relationship between TRPM4 and CSscore [15], immune infiltration [16], immune inflammation [17] and ICBs [18]. Finally, the immunohistochemistry of TRPM4 in several cancers from the HPA database was also finished, thus demonstrating our previous conclusion. In summary, our findings demonstrate TRPM4 is highly correlated with BLCA TME and clinical prognosis, highlighting the potentiality of TRPM4 as an efficient biomarker in BLCA.

Methods and materials

Data collection

Crucial signature-related information pertinent to immunotherapy and other treatments were extracted from the link at the end of a 2021 publication (http://www.thno.org/v11p3089s4.zip/) [19]. We also obtained clinical data and gene expression profiles from the TCGA-BLCA dataset via the UCSC Xena platform (https://xenabrowser.net/datapages/) [20]. Transcripts Per Million (TPM) is chosen for data normalization to effectively eliminate technical biases and provide reliable data support for subsequent analysis. In the filtering step, we removed meaningless genes, such as those with an expression level of zero or close to zero. We employed the ComBat method, which is based on the “sva” package, for batch correction. Additionally, we sourced the gene list of 133 immunomodulators from the referenced article [21]. What’s more, we also downloaded the dataset of BLCA from the GEO website (https://www.ncbi.nlm.nih.gov), which is called GSE32894 and GSE31684 [22, 23].

Survival analysis

The prognostic significance of the TRPM4 concerning overall survival (OS), disease free survival (DSS), progression free interval (PFI) analysis and disease free interval (DFI) plot was assessed by a bioinformatics website (http://www.sxdyc.com/) [24]. The outcome of Kaplan–Meier (KM) curve was illustrated alongside clinical data from TCGA-BLCA, GSE32894 and GSE31684 cohort.

Estimation of the immunological characteristics of TME

It is known to all that the stromalscore, tumor purity, immune score and ESTIMATE score can be calculated when employing ESTIMATE R package [25]. As a result, we partitioned the samples medially into high-TRPM4 group and low-TRPM4 group on the basis of the TRPM4 value. What’s more, we estimated the immunological characteristics of the TME of each patient [26]. We gathered information of 133 immunomodulators such as chemokines, immunoinhibitors, immunostimulators, MHC, and receptors from prior studies. Last but not least, in an effort to mitigate the potential miscalculations stemming from diverse algorithms when estimating the levels of tumor-infiltrating immune cells (TIICs), we performed a thorough analysis of their relative abundance incorporating the following algorithms: CIBERSORT-ABS [27], TIMER [28], EPIC [29], MCP-counter [30] and quanTIseq [31].

Enrichment scores of different gene signatures

To visually delineate oncogenic signaling cascades within the inflammatory tumor microenvironment, alongside the responses to molecularly targeted drug therapy and immune checkpoint modulation, pathway enrichment analysis was conducted using established computational frameworks from our prior investigations [32]. All the gene sets are available in the website. (https://www.gsea-msigdb.org/) Hypoxia score [33], Tumor Stemness score [34] and EMT score [35] were all derived using the R package “GSVA”. The enrichment scores of these signatures were computed by the “GSVA” algorithms [36].

Bioinformatics analysis

We used the CNSknowall platform (https://cnsknowall.com/) and generated the analysis results of heatmap plots. R version 4.3.3 was applied to create several plots such as violin plots and triangle plots. What’s more, some analysis were conducted by using tools in Hiplot Pro (https://hiplot.com.cn/), a comprehensive platform for bioinformatics analysis and visualization [37].

Statistical analysis

Quantitative analyses were executed through GraphPad Prism 8.0, ensuring stringent statistical validation. Intergroup variations were determined through parametric one-way ANOVA complemented by Bonferroni-corrected post hoc examinations to address type I error inflation in multigroup comparative analyses. For comparisons between two groups, an unpaired two-tailed Student’s t-test was employed. Correlation analyses were evaluated using Pearson correlation coefficient. Kaplan–Meier (KM) analyses were conducted with the Log-Rank test. For all analyses, a two-sided P ≤ 0.05 was considered statistically significant, unless otherwise specified. Statistical significance was denoted as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001 [38].

Cell culture

Human BLCA cell lines RT4(Cat. No.VGC-0331, Vigen, Zhenjiang, China,), T24 (Cat. No.VGC-0168, Vigen, Zhenjiang, China,) and normal bladder cells lines SVHUC1(Cat. No.VGC-0352, Vigen, Zhenjiang, China, ) were purchased from the Vigen (https://www.vigenbio.com/). All cells were cultured in DMEM (Cat. No.C11995500BT, Gibco, USA) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin in a humidified environment of 37 °C, 95% air, and 5% CO₂.

Quantitative PCR (qPCR)

Details of cell culture are detailed in Supplementary Methods 1.2. Our previous study described specific methods for total RNA extraction and qRT-PCR. Total RNA was extracted from three cell lines, sv-huc-1, T24 and RT4, using the total RNA extraction reagent (Cat. No. R401-01, Vazyme, Nanjing, Jiangsu, China). The RNA concentration and purity were detected using a NanoDrop micro spectrophotometer, ensuring that the A260/A280 ratio was between 1.8 and 2.0. Subsequently, 1 µg of total RNA was transcribed into cDNA using the reverse transcription kit. The qPCR reaction was carried out using ChamQ SYBR qPCR Master Mix (Cat. No. Q341-02, Vazyme) on the QuantStudio 5 real-time PCR system. Three technical replicates were set for each experimental group, and the biological experiments were independently repeated three times (n = 3) to ensure the reliability of the results. The reaction procedure was as follows: 30 s of pre-denaturation at 95 °C; then 40 cycles of 10 s of denaturation at 95 °C and 30 s of annealing/extension at 60 °C. The quantitative analysis of gene expression was performed using the 2^(–ΔΔCt) method. The Ct values of the target genes were standardized (ΔCt) using the housekeeping gene GAPDH as the internal reference control. Subsequently, the sv-huc-1 cell group was used as the calibration sample to calculate the relative expression levels of each gene in T24 and RT4 cells (2^(–ΔΔCt)). The primer sequences used in qPCR are detailed in supplement excel 1.

Immunohistochemical analysis

Immunohistochemical images for the gene were sourced from The Human Protein Atlas, comparing both cancerous and non-cancerous bladder tissues [39].

Western blot

Lysate the tissues or cells using RIPA lysis buffer containing protein phosphatase inhibitor. Centrifuge at 12,000 rpm for 15 min at 4 °C, then collect the supernatant. Determine the protein concentration using BCA method. Add sample buffer and heat for 10 min at 95 °C. Run the protein samples through 10–15% SDS-PAGE gel electrophoresis, then transfer to PVDF membrane. After incubating the PVDF membrane with 5% skimmed milk at room temperature for 2 h, incubate with the following primary antibodies overnight at 4 °C: anti-TNF-α (Cat. No. SC52746, 1:500, Santa Cruz), anti-P53 (Cat.No. ab179477, 1:800, Abcam), and anti-GAPDH (Cat. No. 60004-1-Ig, 1:5000, Proteintech). Incubate with HRP-labeled secondary antibody at room temperature for 1.5 h: goat anti-mouse (Cat. No.SA00001–1, 1:5000, Proteintech) and goat anti-rabbit (Cat. No. SA00001–2, 1:5000, Proteintech). Finally, develop the bands using ECL detection kit and Tanon 5200 automatic imaging system for chemiluminescence imaging analysis.

Ten machine learning algorithms

Ten machine learning algorithms generally refers to the 101 different combinations of 10 machine learning algorithms that can be obtained through a specific study, including Random Survival Forest (RSF), Elastic Net (ENET), Lasso regression, Ridge, Step Cox, Coxboost, plsRcox, SuperPC, Generalized Boosted Regression Model (GBM) and Survival Support Vector Machine (Survival SVM) [40].

Results

High expression of TRPM4 predicts better prognosis

TRPM4 is one of the most important nonselective sodium ion channels on the surface of several cells, such as epithelial cells, cardiomyocytes, endothelial cells and T cells [41]. First of all, we compared the expression of TRPM4 in different tumors and normal tissue. (Fig. 1A) The analysis of TRPM4 in different tumors and normal tissue from tumors and normal tissue also confirmed our result. (Figure S1) We found out that TRPM4 was expressed higher in most of tumors than normal tissue, such as BLCA, BRCA, CESC and so on. This phenomenon has drawn our attention and we choose BLCA as the type of cancer being studied. Using the median expression of TRPM4 as a stratification threshold, TCGA-BLCA patients were categorized into high-TRPM4 and low-TRPM4 groups. The Kaplan-Meier overall survival analysis based on the optimal cut-off value showed that the high-TRPM4 group showed better overall survival (OS) compared with low-TRPM4 group, which represented better prognosis outcomes. (Fig. 1B) Considering that expression of TRPM4 was higher in the tumor tissue than that in the normal tissue, intriguingly, the Kaplan-Meier overall survival analysis based on the optimal cut-off value indicated TRPM4 may play a compensatory role in the pathophysiological process of BLCA. We also made the Kaplan-Meier Progression Free Interval (PFI) (Fig. 1C) analysis, Disease Free Interval (DFI) (Fig. 1D) analysis and Disease Free Survival (DSS) (Fig. 1E) analysis based on the optimal cut-off value, and most of them showed the same result. What’s more, as we all know, EMT score [35], Tumor Stemness score [34] and Hypoxia score [33] are all associated with poor prognosis, thus predicting the worse outcome of patients. (Fig. 1F-H) The correlation plot showed that the expression of TRPM4 was negatively correlated with the three score, indicating that high expression of TRPM4 predicted the better prognosis.

Fig. 1.

High expression of TRPM4 predicts better prognosis. A The expression of TRPM4 in normal and tumor tissue among different types of cancers. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. B–E Overall survival (OS), Progression Free Interval (PFI), Progression Free Interval (PFI) and Disease Free Survival (DSS) Kaplan-Meier curve in different TRPM4 groups in BLCA. F–H The correlation plot between TRPM4 and EMT score, Tumor Stemness score and Hypoxia score

High expression of TRPM4 shapes an immuno-stimulatory tumor microenvironment by inhibiting T cell exhaustion and immune overdrive

Since TRPM4 was so important to the prognosis, previous research showed that the increase in intracellular calcium caused by ErSO leads to the opening of TRPM4 channels, causing sodium ions and water to flow into the cell, which in turn leads to cell swelling, rupture, and leakage, activating immune cells to rush to the location of dead cells [42]. we began the research on the function of TRPM4 in shaping BLCA immune microenvironment. Given the high potential correlation between TRPM4 and several immune factors in BLCA [43], we investigated the relationship between TRPM4 and 133 immunomodulators, including chemokines, immunoinhibitors, immunostimulators, MHC and receptors. The heatmap plot between 133 immunomodulators and the expression of TRPM4 showed that high-TRPM4 group tend to behave lower expression of immunomodulators, thus representing lower exist of immune molecule. (Fig. 2A) In addition to this, we calculated the average of immunomodulators, thus described by the violin plot (Fig. 2B). The violin plot showed that no matter chemokines, immunoinhibitors, immunostimulators, MHC or receptors are all lower in the high-TRPM4 group, thus meaning the lower immune molecule in the high-TRPM4 group. In addition, we calculated the immunescore, stromalscore, estimatescore and tumor purity of all samples using the R package “ESTIMATE”, thus showing that high-TRPM4 group’s immunescore, stromalscore and estimatescore are all lower than that in the low-TRPM4 group. (Fig. 2C) We made the box plot of different immune cells between different TRPM4 groups using CIBERSORT-ABS algorithms. (Fig. 2D) A triangle heatmap plot between TRPM4 and different immune cells has been made, showing that most of immune cells are negative correlated with TRPM4 expression.(Fig. 2E) Tumor purity is the content of tumor parenchyma, which is higher in the high-TRPM4 group [44]. A heatmap was generated to visualize the relationship between TRPM4 and gene markers of immune cells. The result indicated that tumors with higher expression of TRPM4 tend to be infiltrated by less immune cells. (Figure S2A) We also use different immune algorithm to describe the immune infiltration of the tumor, including TIMER, EPIC, MCPCOUNTER and QUANTISEQ algorithm. (Figure S2B) The heatmap plot result also indicated that the extend of immune infiltration was lower in the high-TRPM4 group.Meanwhile, the triangle heatmap plot was also made to calculate the relationship between TRPM4 and several ICBs, such as CD274, CD80, CD86 and CD276. We found out that TRPM4 was mostly negatively correlated with most ICBs except BTLA and CEACAM1. (Fig. 2F) According to the norms we are familiar with, tumors with lower immune infiltration tend to exhibit worse prognosis [45]. However, patients with high expression of TRPM4 tend to exhibit better prognosis. In order to explore the reason why patients with high-TRPM4 exhibit better prognosis, we have thoroughly investigated the status of T cells and the immune state in the immunological microenvironment [46]. Comprehensive phenotypic profiling of eight functionally distinct T cell subsets (quiescent, regulatory, proliferating, helper, cytotoxic, progenitor-exhausted, terminally-exhausted, and senescent populations) was performed across TRPM4 stratified groups. The violin plot showed that progenitor-exhausted (Fig. 3A), terminally-exhausted (Fig. 3B), and senescent populations T cells (Fig. 3C) were lower in the high-TRPM4 group, meaning that high-TRPM4 group’s T cells were younger than those in the low-TRPM4 group, thus making up for the shortage in the number of T cells. In addition, we also made the violin plot of T cell exhaustion related genes, such as CD3E, PDCD1, CTLA4, HAVCR2, LAG3 and TIGIT, which were all lower in the high-TRPM4 group, indicating that the exhaustion of T cells was lower in the high-TRPM4 group, thus explaining why patients with high-TRPM4 group tend to exhibit better prognosis [47]. (Fig. 3D) We also made the correlation plot between TRPM4 and T cell exhaustion related genes, which also confirmed our previous conjectures and judgments. (Fig. 3E-J) The violin plot of immune overdrive related genes such as CTLA4, CD274, PDCD1, LAG3, HAVCR2, TNFRSF9 and TIGIT were also calculated between high and low TRPM4 groups, showing that the degree of immune overdrive was lower in the high-TRPM4 group. (Fig. 3K) The expression of TGF-β family was also calculated between high and low TRPM4 group, also indicating that the degree of immune overdrive was lower in the high-TRPM4 group, thus being the possible explanation for this phenomenon [48]. (Fig. 3L) In a word, high expression of TRPM4 shapes an immuno-stimulatory tumor microenvironment.

Fig. 2.

High expression of TRPM4 shapes an immuno-stimulatory tumor microenvironment. A Heatmap plot of 133 immunomodulators in different TRPM4 groups. B Box plot of different immunomodulators such as chemokines, immunoinhibitors, immunostimulators, MHC or receptors in different TRPM4 groups. C Immunescore, stromalscore, estimatescore and tumor purity in different TRPM4 groups using the “ESTIMATE” R package. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. D Box plot of different gene markers using CIBERSORT-ABS algorithm in different TRPM4 groups. E Triangle heatmap plot between TRPM4 and different immune cells. F Triangle heatmap plot between TRPM4 and different immune check points

Fig. 3.

High expression of TRPM4 shapes an immuno-stimulatory tumor microenvironment. A–C Progenitor-exhausted, terminally-exhausted, and senescent populations T cell evaluations between different TRPM4 groups. D T cell exhaustion genes expression in different groups. E–J Correlation plot between TRPM4 and T cell exhaustion genes. K Immune overdrive related genes expression in different TRPM4 groups. L TGF-β expression in different TRPM4 groups. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

TRPM4 predicts the blca’s molecular subtypes and response to therapeutic choices

BLCA’s molecular subtype was usually divided into several subtypes, such as UNC subtype, Baylor subtype [49], TCGA subtype [50], MDA subtype, Lund subtype, CIT subtype and Consensus subtype [51]. First of all, we made the clinical heatmap between different TRPM4 groups, indicating that patients in high-TRPM4 group were younger (age < = 65), and M, N,T stage were all lower in this group, thus aligning with our previous research on prognosis. (Fig. 4A) Meanwhile, we also finished the heatmap plot on different molecular subtypes and TRPM4 expression, thus demonstrating that TRPM4 could predict the BLCA’s molecular subtypes. (Fig. 4B) In addition, we also made the heatmap correlation between TRPM4 and the response to other therapies. The results from the Drug Bank database (https://go.drugbank.com/) demonstrated that patients with high expression of TRPM4 tend to be sensitive to Cetuximab, Pazopanib, Sorafenib and Atezolizumab, which were all common targeted drugs. However, patients with low TRPM4 expression were sensitive to Trastuzumab and Afatinib. (Fig. 4C) The box plot of different targeted drugs also confirmed our previous research, thus predicting different groups’ response to therapeutic choices. (Fig. 4D) The Oncoplot, which is a combination of heatmap and conventional stacked bar chart, shows the mutation status and composition of genes in different samples [52]. We made the oncoplot of high-TRPM4 group, (Fig. 4E) low-TRPM4 group (Fig. 4F)and all samples (Fig. 4G). The results indicated that several common gene mutation sites such as TP53 and TTN were obvious in both high-TRPM4 group and low-TRPM4 group. What’s more, some common gene mutation sites such as RB1 were also obvious in the low-TRPM4 group, which can be a potential treatment strategy in the future targeting low expression of TRPM4 BLCA tumors. In one word, the expression of TRPM4 could predict the BLCA’s molecular subtypes and response to therapeutic choices, thus holding great potential in future targeted therapies for bladder cancer.

Fig. 4.

TRPM4 predicts the BLCA’s molecular subtypes and response to therapeutic choices. A Clinical heatmap plot between different TRPM4 groups. B The heatmap plot between TRPM4 and molecular subtypes. C The heatmap plot between TRPM4 and several drug-target genes. D Box plot of different targeted drugs between different TRPM4 groups. E, F The Oncoplots of high and low-TRPM4 groups. G The Oncoplot of all samples. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

Single cell analysis and PCR experiments of TRPM4’s potential role in P53 pathway and TNF-a pathway

Since our previous research were all based on a comprehensive study of the tumor as a whole, but lacked in-depth research on bladder cancer cell clusters. We chose GSE130001, a sample of bladder urothelial carcinoma from the website (http://tisch.compbio.cn/). [53] Based on cell clustering, all the cells (totally 4129 cells) were clustered into 13 clusters. (from 0 to 12) (Fig. 5A) What’s more, all the cells were divided into different cell populations based on markers, such as endothelial cells, epithelial cells, fibroblasts and myofibroblasts. (Fig. 5B) Among all the 4129 cells, 3772 cells were epithelial cells, 143 cells were endothelial cells, 70 cells were myofibroblasts and 144 cells were fibroblasts. (Fig. 5C) Based on the distribution plot, TRPM4 was mostly expressed on epithelial cells, thus clarifying the direction for our future research endeavors. (Fig. 5D) In addition, we also described the violin plot of the expression of TRPM4 in different cells and clusters, indicating that TRPM4 was mostly expressed on epithelial cells and the 0 and 5 clusters. (Fig. 5E) Since TRPM4 was mostly expressed in 0 and 5 clusters, we focused our research efforts on these two clusters mostly. With the motivation of understanding how different cell types communicate through molecular mechanisms such as ligand-receptor interactions. We finished the cell chat analysis, which was showed in the CCI outcome [54]. C0 cluster was associated with C5 and C12 cluster closely. (Fig. 5F) While the C5 cluster was closely associated with C12, C0 and C7 cluster. (Fig. 5G) As we all know, most genes exert its influence on the entire individual by affecting certain pathways [55]. However, previous research about the role of TRPM4 in different pathways was very little. Based on the above situation, we made the heatmap plot of TRPM4 in different hallmark pathways, such as DNA, cellular, development, immune, metabolic, several pathways like hypoxia, proliferation and signaling. We found out that C0 cluster was mostly positive in hypoxia pathway while C5 cluster was mostly positive in P53 pathway and TNF-a pathway. (Fig. 5H) The P53 pathway was a potent tumor suppressor, and its functions in DNA damage repair, cell cycle regulation, apoptosis, senescence, and metabolism collectively contribute to preventing uncontrolled cell growth and the development of cancer [56]. As a central node that connects inflammation, immunity, metabolism, and cell fate regulation, the TNF signaling pathway had always been a focus of research in life sciences [57]. Meanwhile, we made the heatmap plot of TRPM4 in different KEGG pathways. The heatmap plot showed that C5 cluster was positive in the ribosome pathway. (Fig. 5I)

Fig. 5.

Single cell analysis a of TRPM4’s potential role in P53 pathway and TNF-a pathway. A Different clusters of 4129 single cells. B Cell annotation plot of GSE130001. C The pie plot of GSE130001. D The distribution of TRPM4 in GSE130001. E The violin plot of TRPM4 expression in different clusters and cells. F, G Cell chat analysis of C0 and C5 cluster. H Heatmap plot of Hallmark gene analysis in different clusters and cells. I Heatmap plot of KEGG gene analysis in different clusters and cells

We explored the single cell signature of P53 pathway, (Fig. 6A) hypoxia pathway (Fig. 6B) and TNF-a pathway (Fig. 6C), whose range were almost similar to the range of TRPM4, thus validating our previous findings and demonstrating that TRPM4 was significantly important in these three pathways. Cell-cell communication, particularly through the ligand-receptor pathway, plays a crucial role in regulating cell function and distribution in the tumor microenvironment. The bubble plots of C0 and C5 cluster were also finished to describe the cell chat from epithelial C0 and C5 cluster. (Fig. 6D-E) We found that FN1-SDC4 ligand-receptor pathway was particularly remarkable when both C0 and C5 cluster was source cluster, revealing that it might play a potentially significant role in cellular communication. Meanwhile, a heatmap plot of TF enrichment was also finished, indicating that FOXP1’s p-value was highest in the C5 cluster and BRD4’s p-value was highest in the C0 cluster, thus indicating that FOXP1 and BRD4 could be a transcription factor that regulated gene expression in each cell population. (Fig. 6F-H) To further validate our single cell analysis outcome, we finished the PCR experiments using SV-HUC-1, T24 and RT4 cell lines. SV-HUC-1 cell line was normal bladder immortal cells, which was used to be as the control group [58]. T24 and RT4 were BLCA cell lines, which was usually used to finish PCR experiments [59]. We found that the expression of P53 was higher in RT4 cell line than that in T24 cell line (Fig. 6I). The outcome also indicated that expression of TNF-a was lower in the RT4, while in T24 and SVHUC1 its expression was high. (Fig. 6J) Last but not least, we found that the expression of TRPM4 was higher in RT4 and lower in T24, which was absolutely consistent with P53 and TNF-a outcomes, thus validating our single cell analysis outcome. (Fig. 6K) Same results were validated through western blot experiments. (Fig. 6L-N) As a result, TRPM4 may be an important role in P53 and TNF-a pathway.

Fig. 6.

Single cell analysis and PCR experiments of TRPM4’s potential role in P53 pathway and TNF-a pathway. A–C The single cell signature of P53 pathway, hypoxia pathway and TNF-a pathway. D, E The bubble plots of C0 and C5 cluster. F–H The heatmap plot of TF enrichment and the TF enrichment analysis of C0 and C5 cluster. I–K PCR experiments outcomes of TRPM4 in TNF-a pathway and P53 pathway. L–N Western blot experiments outcomes of TRPM4 in TNF-a pathway and P53 pathway

Pan cancer analysis of TRPM4

Since TRPM4 had been validated that it was a tumor suppressor gene in BLCA according to our previous bioinformatics analysis, single cell analysis and experiments, we also completed the pan cancer analysis of TRPM4 in other type of cancers to verify whether our conclusions are universally applicable. First, we made the overall survival and disease free survival Kaplan–Meier plots of different kinds of cancers, such as breast cancer (BRCA), glioblastoma multiforme + brain lower grade glioma (GBMLGG), kidney renal clear cell carcinoma (KIRC), brain lower grade glioma (LGG), pancreatic adenocarcinoma (PAAD) and uveal melanoma (UVM). (Fig. 7A-B) The Kaplan–Meier plots showed that in BRCA, patients with high expression of TRPM4 tend to show better prognosis, while in other cancers, such as KIRC and PAAD it displayed the opposite results. In addition, we also completed the prognosis in different cancers about OS (Fig. 7C), DSS (Fig. 7D), DFI (Fig. 7E) and PFI (Fig. 7F). The results indicated that TRPM4 was associated with better prognosis in BLCA, BRCA, KIRP, KICH and OV, while associated with poorer prognosis in LGG, KIRC, PAAD and UVM. Tumor Mutation Burden (TMB) was regarded as an important predictive marker for the effectiveness of immunotherapy [60]. High TMB is an approved positive predictive biomarker for immune checkpoint inhibitor-mediated survival, and this information can be easily calculated from large-scale, exome, or whole-genome sequencing in routine clinical practice. We completed the correlation plot between TRPM4 and TMB of several types of cancer, such as BLCA (Fig. 7G), THYM (Fig. 7H), KIRP (Fig. 7I), UCEC (Fig. 7J), DLBC (Fig. 7K) and ESCA (Fig. 7L). All the expression of TRPM4 was positively correlated with TMB, thus indicating that high-TRPM4 may be associated with better response to immunotherapy. Meanwhile, we also described the relationship between TRPM4 and CSscore (Fig. 8A), immune infiltration (Fig. 8B), immune inflammation (Fig. 8C) and ICBs (Fig. 8D). Immunohistochemistry was a common technology to described the real expression of protein. Last but not least, we use the immunohistochemistry results from the HPA website to validate our previous research conclusion in Fig. 1A. We found out that the expression of TRPM4 was higher in the tumor group than that in the normal group, such as BLCA, COAD, LIHC, READ, BRCA, ECSC, KICH and PRAD, thus validating our bioinformatics analysis result. (Fig. 8E)

Fig. 7.

Pan cancer analysis of TRPM4. A, B OS Kaplan-Meier curve and DSS Kaplan-Meier curve of different TRPM4 groups in pan cancers. C–F OS, DSS, DFI and PFI prognostic analysis of pan cancers. G–L Correlation plot between TMB and TRPM4 expression in pan cancers. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

Fig. 8.

Pan cancer analysis of TRPM4. A The relationship between TRPM4 and CSscore. B The relationship between TRPM4 and immune infiltration. C The relationship between TRPM4 and immune inflammation. D The relationship between TRPM4 and ICBs. E The immunohistochemistry outcomes in normal ant tumor tissue in pan cancers. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

The construction of prognostic model relating TRPM4

With the motivation of predicting the prognosis better with clinical data better, as a result, we decided to construct a prognostic model, thus predicting prognosis more precisely and efficiently. Considering that if samples were only from TCGA-BLCA, it might be not universally applicable. Therefore, we included two additional BLCA cohorts, GSE32894 and GSE31684 [22, 23]. First of all, we made the correlation analysis between TRPM4 and other genes, thus filtering all the genes whose correlation with TRPM4 was more than 0.6 or less than − 0.6. All the genes are: ARHGAP27, TNK2, ACOT11, TJP3, ARRDC1, CAPS, TMC4, SH3GLB2, PPP1R13L, TMC6, EFCAB4A, CYP4F12, IQSEC2, CBLC, PROM2, PLXNB1, RASSF7, SHROOM1, TNFRSF14, SLC9A1, SPATA20, NR2F6, KCNN4, SLC44A2, NUDT11, ZBTB7A, MYH14, MUC20, CNKSR1, DOCK6, EVPL, TRPM4. Meanwhile, we used the ten machine learning algorithms to construct the prognostic model relating TRPM4.The result showed that “CoxBoost + RSF” could predicted the prognosis most effectively. The C-index of the model could be 0.879 in TCGA-BLCA. (Fig. 9A) What’s more, according to the riskscore, we divided all the samples into high-riskscore group and low-riskscore group. The Kaplan–Meier plots showed that patients in low-riskscore group showed better prognosis than that in high-riskscore group among BLCA cohorts and GSE32894 cohort. (Fig. 9B-C) The clinical data also showed that patients in low-riskscore group tend to behave lower grade and lower age, thus predicting its better prognosis in BLCA cohorts and GSE32894 cohort. (Fig. 9D-E) The ROC curve of the model in TCGA-BLCA was 0.91 and in GSE32894 was 0.77, indicating that the model could predict the dataset excellently. (Fig. 9F-G) The Calibration curve of the model, the DCA curve, the Nomogram of the model was also described. (Fig. 9H-J). In one word, the prognostic model relating TRPM4 could perfectly predict the prognosis. Last but not least, with the intention of letting readers understand the role of TRPM4, we made a figure summarizing the proposed TRPM4-related mechanisms. (Fig. 10)

Fig. 9.

The construction of prognostic model relating TRPM4. A The outcome of ten machine learning algorithms in TCGA-BLCA, GSE32894 and GSE31684. B, C The Kaplan-Meier curve of different riskscore in TCGA-BLCA and GSE32894 cohort. D, E Clinical data analysis of TCGA-BLCA and GSE32894 cohort. H Calibration curve of the model. I DCA curve of the model. J The Nomogram of the model. (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001The construction of prognostic model relating TRPM4 (A) The outcome of ten machine learning algorithms in TCGA-BLCA, GSE32894 and GSE31684. (B-C) The Kaplan-Meier curve of different riskscore in TCGA-BLCA and GSE32894 cohort. (D-E) Clinical data analysis of TCGA-BLCA and GSE32894 cohort. (F-G) The ROC curve of the model in TCGA-BLCA and GSE32894 cohort. (H) Calibration curve of the model. (I) DCA curve of the model. (J) The Nomogram of the model. (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001

Fig. 10.

Figure summarizing the proposed TRPM4-related mechanisms

Discussion

Existing research highlights that TRPM4 and TRPM5 are channels with permeability to monovalent cations, with TRPM4 having the following ion permeability order: Na + >K+ >Cs + >Li+ >>Ca2+, Cl-. TRPM4 is expressed in various tissues, primarily the intestines and prostate, and it is also present in the innate and adaptive immune responses [61]. In the article called “Persistent Activation of TRPM4 Triggers Necrotic Cell Death Characterized by Sodium Overload ” ,the researcher systematically reveals the molecular mechanism by which NC1, a small molecule, directly and continuously activates TRPM4 channels [4]. However, the role of TRPM4 in tumors hasn’t been explored, especially BLCA. In this article, we firstly viewed the expression of TRPM4 in different tumors and normal tissue. According to the expression of TRPM4, we divided all the TCGA-BLCA patients into two groups: high-TRPM4 group and low-TRPM4 group. The Kaplan-Meier curve of TRPM4 groups in BLCA was made, thus indicating that TRPM4 was associated with better prognosis in BLCA, the correlation between TRPM4 and EMT score, Tumor Stemness score and Hypoxia score was finished to validate our conclusion. We also found that TRPM4 shaped an immuno-stimulatory tumor microenvironment by inhibiting the T cell exhaustion and immune overdrive. The heatmap plot of several immunomodulators and gene markers confirmed our conclusion. TRPM4 can be a biomarker to predict the molecular subtypes and response to therapeutic choices. Single cell analysis, PCR experiments and Western blot experiments confirmed TRPM4’s potential important effect in P53 and TNF-a pathway using T24, SVHUC1 and RT4 cell lines. This makes our conclusions even more severe, convincing and scientific. We suspect that the possible mechanism is that the downregulation of p53 results in decreased apoptosis of tumor cells, and the upregulation of TNF-α may be related to the immuno-stimulatory microenvironment of the tumor. In addition, we also view the effect of TRPM4 in pan cancers, finding that in other cancers TRPM4 may perform the opposite function. We speculate that the effect of TRPM4 activation is likely integrated with other dominant signaling pathways unique to each tissue. For example, in BLCA, TRPM4 function might boosts pathways such as P53 and TNF-α pathways, making its activity appear tumor-suppressive. This has been confirmed in our article. In CRC, according to previous research, it may contributes to proliferation, cell cycle, and invasion of colorectal cancer cells and synergizes with oncogenic drivers like Wnt/β-catenin signaling, thereby exerting an oncogenic effect [5]. The immunohistochemistry outcomes in normal ant tumor tissue in pan cancers validated our previous outcomes too. Last but not least, we constructed the prognostic model relating TRPM4 by ten machine learning algorithms, which could predict the prognosis well. Through our construction of prognostic model, we found that “CoxBoost + RSF” could predict the prognosis most effectively. The C-index of the model could be 0.879 in TCGA-BLCA, thus demonstrating the accuracy and effectiveness of our model.

In this article, we firstly use TRPM4 to predict the BLCA prognosis as a biomarker, thus highlighting the value of TRPM4 in predicting the migration and invasion ability of tumor cells in BLCA. Past research mainly focused on the role of TRPM4 in angiogenesis, while our study innovatively focuses on the role of TRPM4 within tumors, and has discovered that TRPM4 may play a potential role in inhibiting bladder cancer. In addition, we firstly expand our research on the role of TRPM4 in various aspects to the field of pan-cancer, not just bladder cancer. We describe the relationship between TRPM4 and CSscore, immune infiltration, immune inflammation What’s more, we firstly build the prognostic model relating TRPM4, which indicates that TRPM4 can be a potential biomarker to identify the outcome of BLCA patients.

Previous research about TME showed that TRPM4 is highly expressed in human colorectal tumor buds and contributes to proliferation, cell cycle, and invasion of colorectal cancer cells [6]. Furthermore, the team in Shanghai has discovered that continuous activation of TRPM4 can lead to a novel necrotic cell death mechanism (NECSO) [4]. However, there has been no research conducted on the role of TRPM4 in BLCA. Previous research has shown that through the mitochondria pathway and the NDUFB10 gene, high MitoPS scores can be used to identify the “cold” immune phenotype (low number of CD8 + T cells), providing evidence for the metabolic-immune axis and supporting the existence of an immunosuppressive tumor microenvironment (TME) in this article [62, 63]. What’s more, some articles suggest that dynamic network biomarkers and high expression of the ASPH gene can facilitate immune evasion and poor prognosis, thereby providing insights into the mechanisms by which key genes regulate immune phenotypes. This supports the rationale for considering TRPM4 as an immunomodulatory factor. The three research papers together reveal the synergistic effect of the p53 pathway and the TNF-α pathway in shaping the tumor immune microenvironment. The innovation of this study lies in directly linking the ion channel mechanism to the immune phenotype. The previous two research papers provided the mechanism-based extension and validation from the metabolic and transcriptional regulation levels.

Compared to previous studies, our study expands on their conclusions by demonstrating that the high expression of TRPM4 bladder cancers may predict better prognosis, which is opposite to their conclusion. However, there are still several limitations of our study. Due to time limitation, we only validate our conclusion in BLCA through bioinformatics analysis, single cell analysis, Western blot and PCR. However, our experimental validation is limited to PCR, without functional assays, such as knockdown, overexpression and immune cell assays. What’s more, our conclusion may be not suitable in other cancers, thus needing more in-depth and extensive research to explain. We found that Disease Free Survival of different TRPM4 groups showed the opposite outcomes, which may be due to the small sample size. We do believe that if the number of samples could be expanded, we may get the same correct outcome. In order to overcome this limitation, we plan to expand the sample size, collect samples from multiple different hospitals, and attempt to expand to other types of cancer to verify whether our conclusions can be validated across all types of cancer. Meanwhile, we plan to combine TRPM4 with several BLCA’s common prognosis biomarkers such as TMN stage together to construct a powerful and sensitive prognostic model, thus helping in better predicting the prognosis of the patients. To approach our model’s interpretive power and predictive capabilities, we plan to use machine learning to build our prognostic model, which will be the focus of our future work.

The findings of this study theoretically provide new insights or evidence for the function of TRPM4 in shaping an immuno-stimulatory TME and is associated with better prognosis in BLCA, thus contributing to several drugs’ development targeting TRPM4 and the application of agonists targeting TRPM4 such as NC1 in BLCA. Based on our current research results, we hypothesize that NC1 may promote TRPM4, induce sodium death, thereby inhibiting BLCA, and subsequently cause the death of tumor cells due to the influx of sodium ions. Additionally, these results have significant practical applications, such as the application of NC1 in BLCA and several other drugs targeting TRPM4. In the future, these drugs, such as NC1, may lead to potential advances in the treatment of bladder cancer. We do believe that TRPM4 is bound to become an important target for future targeted therapy of bladder cancer.

Conclusion

The current study reveals that high TRPM4 shapes a more immuno-stimulatory tumor microenvironment in BLCA. What’s more, high expression of TRPM4 is often associated with better prognosis. Moreover, the study suggests that TRPM4 is an identifying biomarker to predict different molecular subtypes and therapy options in BLCA. In addition, the single cell analysis and PCR experiment reveal that TRPM4 may play an important role in P53 and TNF-a pathway. Last but not least, a prognostic model relating TRPM4 is built to predict the prognosis of BLCA. Overall, we identified TRPM4 as a novel and potential BLCA target for identifying tumor immunogenicity.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

YD, YZ and YG finished the study. YD downloaded and handled the public data and finished bioinformatics analysis. LY, YG and XC performed the cell experiment and the figures. ZZ, YL, LY, and YZ reviewed the manuscript.

Funding

None.

Data availability

All data is available from the authors upon reasonable request. The clinical data and gene expression profiles of the TCGA-BLCA dataset are obtained from the UCSC Xena platform (https://xenabrowser.net/datapages/). The dataset of GSE32894 and GSE31684 are got from the GEO website (https://www.ncbi.nlm.nih.gov). Crucial signature-related information pertinent to immunotherapy and other treatments were extracted from the link at the end of a 2021 publication (http://www.thno.org/v11p3089s4.zip/).

Declarations

Ethics approval and consent to participate

This study utilized publicly available, de-identified datasets. The original collection of these data was conducted in accordance with relevant ethical guidelines and received appropriate ethical approval (including informed consent from participants) by the original investigators. As this secondary analysis involves no new data collection or interaction with human subjects, and uses only anonymized data, additional ethical approval for this specific analysis was not required. All analyses were performed in compliance with the terms of use of the respective datasets. As a result, the ethics, consent to participate, and consent to publish declarations is not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.