Abstract
Background:
Clear cell renal cell carcinoma (ccRCC) is an angiogenic tumor originating from proximal tubule epithelial cells. Ammonia induced cell death is closely associated with carcinogenesis, but its potential mechanism in ccRCC remains unclear and requires further investigation.
Methods:
The transcriptomic data of ccRCC and the genes related to ammonia induced cell death were retrieved from public resources. Candidate genes were ascertained by taking the common part of the Differentially Expressed Genes (DEGs) and the Ammonia death-related genes (ADRGs). Prognostic genes were filtered using machine learning and a prognostic model was established. ccRCC patients were segmented into a high-risk group (HRG) and a low-risk group (LRG) in accordance with the risk score values. Functional enrichment, immune infiltration analysis, somatic mutation and Reverse transcription quantitative PCR (RT-qPCR) were also executed. Key cells were identified at the single-cell level, and analyses including cell communication and pseudo-temporal analysis were conducted.
Results:
The risk model was constructed from 6 prognostic genes (RGS20, ADA, AICDA, SLC12A5, RUFY4 and CDK5RAP3). RGS20, ADA, AICDA, SLC12A5, RUFY4 and CDK5RAP3 were significantly upregulated in ccRCC group, and the RT - qPCR analysis results were consistent (p < 0.05). Enrichment analysis suggested that HRG and LRG were linked to pathways, such as “olfactory transduction” signaling pathway (p < 0.05). Sixteen immune cells with differential abundance between HRG and LRG were identified, such as activated B cells. The most frequent type of mutation between HRG and LRG was missense mutation. Malignant cells (MCs) were identified as the key cells. Moreover, frequent interactions between MCs and T cells were observed in ccRCC, and 6 prognostic genes were involved in the regulation of MCs’ expression.
Conclusion:
This study identified 6 prognostic genes for ccRCC, MCs were identified as the key cells. We explored potential mechanisms and prognostic associations in ccRCC, providing new insights for potential therapeutic strategies.
Introduction
Clear cell renal cell carcinoma (ccRCC) originates from the epithelial cells of the proximal renal tubules and is the most common subtype of renal cancer, accounting for approximately 70%-80% of all renal cancer case (). This disease exhibits not only high tumor heterogeneity, a high metastasis rate, and poor prognosis in advanced stages (), but also distinct metabolic abnormalities, specifically manifested as altered lipid and glycogen storage patterns (). Notably, the disease often remains asymptomatic until advanced stages (III or IV), leading to delayed diagnosis and dismal survival outcomes (). These challenges underscore the urgent need to identify novel therapeutic targets and optimize treatment strategies for ccRCC.
Ammonia-induced cell death (ammonitosis) is a recently described form of regulated cell death triggered by excessive intracellular ammonia accumulation. This process disrupts lysosomal and mitochondrial function, primarily through dysregulated glutamine metabolism (). Under physiological conditions, ammonitosis plays a critical role in immune homeostasis by eliminating effector CD8+ T cells post-infection to prevent autoimmune responses (). Emerging evidence links aberrant ammonia metabolism to various pathologies, including cancer, liver diseases, and neurodegenerative disorders (, ). However, the mechanistic role of ammonitosis in ccRCC pathogenesis remains poorly understood. Given the metabolic reprogramming inherent to ccRCC, particularly its reliance on glutamine metabolism, investigating ammonia-induced cell death could unveil novel pathways driving tumor progression and therapy resistance.
Single cell RNA sequencing (scRNA seq) technology has become a powerful tool for studying cell biology and is widely used in basic and clinical research. It can characterize cellular heterogeneity, identify rare but important cell types, and also explore communication networks and interactions between cells (). In ccRCC, scRNA-seq has identified malignant cell subpopulations, immune microenvironment interactions, and lineage-specific transcriptional programs that conventional bulk sequencing cannot resolve (). Integrating scRNA-seq with functional studies offers a powerful approach to elucidate how ammonia metabolism influences key cellular players in ccRCC progression.
This study utilized transcriptomic data from ccRCC and control samples in public databases to perform differential expression analysis. Through bioinformatics methods, we further identified prognostic genes playing crucial roles in ccRCC and associated with ammonia metabolism. We then conducted an in-depth analysis of the potential mechanisms by which these prognostic genes function in ccRCC. Subsequently, single-cell analysis was performed to identify key cells, and pseudo-time analysis was conducted to explore the dynamic expression of prognostic genes during the differentiation processes of these key cells. Finally, RT-qPCR validation was performed to elucidate the prognostic significance of ammonia metabolism in ccRCC, aiming to provide a theoretical basis for optimizing treatment strategies and advancing personalized therapy for ccRCC.
2 Materials and methods
2.1 Data collection
The transcriptomic, mutation, clinical, and survival data for ccRCC were retrieved from TCGA database (https://www.cancer.gov/ccg/research/genome-sequencing/tcga, downloaded on April 17th, 2025). Specifically, the TCGA-KIRC dataset included 538 ccRCC kidney tissue samples along with 72 paracancerous tissue samples, and 534 tumor kidney tissue samples with survival data. The `createFolds` function from the caret package was used to split the samples with survival information in the TCGA-KIRC dataset into 10 folds. A random seed of 512000 was set to ensure reproducibility of the results. The first 7 folds (70%) were used as the training set 1 (374 samples), and the remaining 3 folds (30%) were designated as validation set 1 (160 samples). The E-MTAB-1980 dataset containing transcriptome data of ccRCC and survival information was Retrieved from the ICGC database (https://icgc.org/content/icgc-home-0) as validation set 2. It included 101 ccRCC renal tissue samples with survival information and was downloaded on April 17, 2025. scRNA-seq dataset related to ccRCC was obtained from the GEO database. Specifically, the GSE159115 dataset (platform: GPL16791) included 7 ccRCC patient kidney tissue samples and 6 paracancerous samples. A total of 467 ADRGs were obtained from the literature () (Supplementary Table 1).
2.2 Differential expression analysis
The DEGs between ccRCC (Tumor) and normal samples in TCGA-KIRC were discovered using the R package “DESeq2” (v 3.54.0) (). The filtering conditions were adjusted p - value < 0.01 and |log2 fold change (FC)| > 2. The top 10 up- and down-regulated DEGs ranked by log2FC value from high to low were marked on the volcano plot using R package “ggplot2” (v 3.5.1) (). The expression trends DEGs were visualized in the heatmap using R package “ComplexHeatmap” (v 2.18.0) ().
2.3 Identification and analysis of candidate genes
To identify candidate genes, the R package “ggvenn” (v 0.1.10) () was used to find the intersection of DEGs and ADRGs to obtain candidate genes. KEGG and GO analyses were executed for candidate genes using the R package “clusterProfiler” (v 4.10.1) () (adjusted p < 0.05, corrections were carried out using the Benjamini-Hochberg (BH) method) to investigate the functions of candidate genes. Additionally, the candidate genes were evaluated by protein - protein interaction (PPI) network using the Search Tool for the Retrieval of Interacting Genes (STRING) database (https://string-db.org/) (confidence > 0.4). Subsequently, the candidate genes were visualized by means of Cytoscape software (v 3.10.2) ().
2.4 Identification and analysis of prognostic genes
To further screen for prognostic genes, based on ccRCC samples with survival information in the training set 1, a univariate Cox regression analysis (p < 0.001, Hazard Ratio (HR) ≠ 1) was executed on the candidate genes using the R package “survival” (v 3.7-0) (). Additionally, the PH assumption test (p > 0.05) was executed by means of the R package “survival” (v 3.7-0) () to obtain genes with prognostic value. Subsequently, based on the genes with prognostic value, LASSO regression analysis was executed by means of the R package “glmnet” (v 4.1.8) () (family =“cox”, nfolds = 5). After 10 - fold cross - validation, the model corresponding to the lambda.min value was selected as the optimal model to screen out the prognostic genes.
2.5 Construction of risk model
The prognostic genes were used for constructing the risk model with the following formula: Riskscore = ∑iCoefficient(i)*Expression of gene(i). In the training set 1, ccRCC samples with survival information were divided into a HRG and a LRG in accordance with the median risk score. The Kaplan-Meier (K-M) curves and risk curve of the HRG and LRG were plotted, and the prognostic ability of the risk model was evaluated using the R package “survminer” (v 0.4.9) (). Furthermore, the model’s predictive performance of 3, 5, 7 years survival rates was evaluated using ROC. 1 > AUC values > 0.6, created with the R package “survivalROC” (v 1.0.3.1) (). Then, to validate the risk model, the same analytical approach served to assess the model in ccRCC samples from validation set 1 and validation set 2.
2.6 Correlation analysis of clinical characteristics, human protein atlas, and expression analysis of prognostic genes
To explore the differences in risk scores among different clinical characteristic subgroups (including Age, Stage, Gender, T stage, M stage, N stage), based on the ccRCC samples from the training set 1, the disparities in risk scores among clinical characteristic subgroups were evaluated by means of the Wilcoxon test (p < 0.05). To explore the protein manifestation of prognostic genes in ccRCC tissue samples and paracancerous samples, the protein manifestation of prognostic genes was analyzed utilizing the HPA database. Subsequently, to profile the expression manifestation of prognostic genes in tumor and normal samples, the manifestation levels of prognostic genes were compared through the Wilcoxon test (p < 0.05).
2.7 GSEA and GSVA
To further elucidate the signaling pathways associated with the HRG and LRG in the training set 1, the DEGs between the HRG and LRG were detected through the R package “DESeq2” (v 3.54.0) (), and calculated the log2FC. The “c2.KEGG.v2022.1.Hs.symbols” was retrieved as the background set from the MSigDB, and GSEA of the DEGs was performed using the R package “clusterProfiler” (v 4.10.1) () with adjust p < 0.05 (corrections were carried out using the BH method), and |(NES)| > 1. Then, to analyze the disparities in the biological functions and signaling pathways of genes between the HRG (n = 141) and LRG (n = 233), based on the ccRCC samples with survival information in the training set 1, the “c2.cp.kegg.v2023.1.Hs.symbols.gmt” was downloaded from MSigDB to serve as the background. GSVA scores for each sample were calculated using the R package “GSVA” (v 1.50.0) (). The R package “limma” (v 3.56.2) () was then applied to compare functional enrichment pathway differences between HRG and LRG, with the thresholds set at |t| > 2 and p < 0.05.
2.8 Immune checkpoints and immune factors analysis
In order to understand the associations between immune checkpoints, immune factors and the occurrence of ccRCC, 65 immune checkpoints were obtained from the literature (). Then, based on the ccRCC samples in training set 1, a correlation analysis between the prognostic genes and immune checkpoints was conducted using the R package “psych” (v 2.4.3) () (|(cor)| > 0.30, p < 0.05). After that, in order to explore the correlation between the prognostic genes and immune factors, different immune factor sets (including immune inhibitors, immune stimulants, chemokines, receptors, major histocompatibility complex (MHC)) were collected from TISIDB. Based on the ccRCC samples in the training set 1, a correlation analysis between the prognostic genes and these immune factors was conducted using the R package “psych” (v 2.4.3) (|cor| > 0.30, p < 0.05).
2.9 Immunotherapy analysis
To evaluate the infiltration status of stromal cells and immune cells in ccRCC, the stromal score, immune score, and ESTIMATE score, tumor purity of ccRCC samples were calculated using the ESTIMATE algorithm, and the differences in each score between the HRG and LRG were compared through the Wilcoxon test (p < 0.05).
2.10 Immune infiltration analysis
To investigate immune cell infiltration in HRG and LRG from the training set 1, the ssGSEA algorithm from the R package “GSVA” (v 1.50.0) () served to calculate infiltration scores for 28 immune cells (MMC3.GMT immune gene set) () across both HRG and LRG. The Wilcoxon test served to evaluate discrepancies in infiltration scores between HRG and LRG. Immune cell types with significant differential infiltration (adjust p < 0.05, corrections were carried out using the BH method) were selected. Then, correlation analysis was executed using the R package “psych” (v 2.4.3) to explore relationships among differentially infiltrated immune cell types and between these immune cells and prognostic genes, with thresholds set at |cor| > 0.30 and p < 0.05.
2.11 Analysis of tumor mutation and drug sensitivity
To explore the somatic mutation status between the HRG and LRG in ccRCC patients in the training set 1, we retrieved somatic mutation data (in MAF format) of ccRCC directly from TCGA database using the tcga_load(study = "KIRC") function in the TCGAmutations package. TCGAmutations packages are typically based on the standardized processing version of the TCGA Firehose, containing somatic mutations identified through a unified workflow to ensure data comparability and reliability. Based on the HRG and LRG in the ccRCC samples of the training set 1, the gene mutation status of each ccRCC sample was characterized by means of R package “Maftools” (v 2.18.0) (), and the top 20 common mutant genes in the HRG and LRG were presented. Generally, a higher half-maximal inhibitory concentration (IC50) value indicates lower drug sensitivity. To evaluate the drug sensitivity profiles of ccRCC patients in training set 1, IC50 values for 138 conventional drugs from the GDSC database were calculated using the R package “pRRophetic” (v 0.5) (). Wilcoxon test was then utilized for identify notable discrepancies in drug sensitivity between HRG and LRG from ccRCC samples in the training set 1 (p < 0.05).
2.12 Data quality control and cell annotation
The GSE159115 dataset was evaluated with the aid of the R package “Seurat” (v 5.1.0) (). The following screening guidelines were adopted: Cells containing fewer than 200 genes were excluded from the analysis. Moreover, cells had to meet the following criteria: 200 < nFeature_RNA (genes per cell) < 6000, 200< nCount_RNA (total RNA count per cell) < 40000, and percentage.mt (proportion of mitochondrial gene expression) < 25%. After QC, low - quality cells were removed. Subsequently, LogNormalize normalization was performed using the NormalizeData function. The FindVariableFeatures function in the R package “Seurat” (v 5.1.0) () was utilized to identify highly variable genes (HVGs). The data was normalized utilizing the ScaleData function in the R package “Seurat” (v 5.1.0) (), and then the batch effect among samples was observed. The HVGs were subjected to Principal Component Analysis (PCA) utilizing the R package “Seurat” (v 5.1.0) (). In the JackStraw function, the principal components with genes having smaller enrichment p were retained (p < 0.05). The cells in the GSE159115 dataset were classified with the use of the R package “Seurat” (v 5.1.0) (resolution = 0.2). Subsequently, the classified cell clusters were annotated on the basis of the marker genes in the literature (, ) and CellMarker2 Website. The R package “Seurat” (v 5.1.0) () was used to draw a Uniform Manifold Approximation and Projection (UMAP) plot to display the annotated cell clusters. To understand the distribution differences of the annotated cell clusters between different groups, the R package “ggplot2” (v 3.5.1) () was used to plot the distribution proportions of the annotated cell clusters.
2.13 Identification of key cells and function analyses, cell-to-cell communication, pseudo-temporal analyses
To understand the expression differences of the annotated cell clusters between the normal and ccRCC groups, the R package “ggplot2” (v 3.5.1) () was used to plot the distribution proportions of the annotated cell clusters in the normal and ccRCC groups, the Wilcoxon test was employed to contrast the prognostic genes expression disparity of the annotated cell clusters between the normal and tumor (p < 0.05). The cells that were differentially expressed and had high expression of most prognostic genes were selected as key cells. To delve into the biological pathways and functions related to the key cells and annotated cell clusters, ReactomeGSA (https://github.com/reactome/ReactomeGSA) was utilized to perform functional enrichment analysis on the annotated cell clusters in the GSE159115 dataset (p < 0.05). Additionally, cell-to-cell communication analysis was executed using the R package “CellChat” (v 1.6.1) () to investigate interactions between key cell types and other annotated cell types in both the disease and control samples from the GSE159115 dataset. This analysis generated cell-cell communication networks and determined the expression levels of receptors and ligands, highlighting the number and strength of interactions and the intensity of receptors and ligands. Subsequently, the reduceDimension functions of the R package “Monocle” (v 2.30.1) () and R package “DDRTree” (v 0.1.5) () were used to conduct secondary dimensionality reduction on the key cells (max_components=2, resolution = 0.2) in the GSE159115 dataset. Then, the R package “Monocle” (v 2.30.1) was used to analyze the cell differentiation trajectories and to construct the cell trajectory plot of key cell types. The expression changes of prognostic genes throughout the pseudo-temporal progression were also analyzed.
2.14 RT-qPCR
To preliminarily verify the mRNA expression trend of prognostic genes in clinical samples, RT-qPCR experiments were conducted. Specifically, 10 tissue samples (5 ccRCC tumors and 5 normal controls) were derived from the clinic in the department of Urology, Air Force 986th Hospital of Air Force Medical University. All participants were given informed consent. This study has been approved by the Ethics Committee of the Air Force 986th Hospital (approval number: KYLL2025-986-177). All primers were listed in Supplementary Table 2. We used the CFX96 Connect Real - time Quantitative Fluorescence PCR Instrument (Bio - Rad, USA) to perform the qPCR assay. The conditions for thermal cycling were as follows: First, there was a pre - denaturation step. During this step, the temperature was set at 95°C, and it was maintained for 1 minute. After that, a total of 40 cycles were performed. The 2-ΔΔCT method was employed to calculate the relative quantification of mRNAs. Once the RT - qPCR results were generated, they were first exported to Excel and subsequently imported into Graphpad Prism 10 (https://www.graphpad.com/) for statistical analysis and visualization (p < 0.05).
2.15 Statistical analysis
Bioinformatics were implemented with the R (v 4.3.3). Wilcoxon test was utilized to evaluate disparities. For qRT-PCR analysis, the T-test was employed for statistical comparisons, and p < 0.05.
3 Results
3.1 Ascertainment and exploration of functions of 96 candidate genes
By comprehensively analyzing the DEGs in the TCGA-KIRC, we ascertained a total of 2,422 DEGs, and it was found that there were 1,602 up - regulated genes and 820 down - regulated genes in ccRCC samples (Figures 1a, b). The intersection of 2,422 DEGs and 467 ADRGs yielded 96 candidate genes (Figure 1c). Subsequently, a total of 517 GO functional annotation results were obtained, including 414 enrichments in BPs like “regulation of T cell activation”, 30 in CCs like “cytolytic granule” and 73 in MFs like “CXCR3 chemokine receptor binding” (adj.p < 0.05) (Figure 1d) (Supplementary Table 3). Next, 102 KEGG pathways such as “Endocrine resistance” were enriched (p < 0.05) (Figure 1e) (Supplementary Table 4). Finally, the outcomes of the PPI network disclosed that 15 candidate genes were isolated, and among the remaining 81 candidate genes had interactions (Figure 1f).
Figure 1
3.2 Construction and of validation a well-performing risk model in 6 prognostic genes
Overall, 20 candidate prognostic genes were detected (Figure 2a) and the PH assumption test was performed (Supplementary Table 5) (p > 0.05). Overall, 6 prognostic genes (RGS20, ADA, AICDA, SLC12A5, RUFY4 and CDK5RAP3) were then selected using LASSO, with a minimum lambda of 0.050488458597895 as the threshold (Figures 2b, c). Then, a risk model was developed based on the 6 prognostic genes, successfully stratifying 374 ccRCC patients from the training set 1 into 2 groups: HRG (141 samples) and LRG (233 samples), using a median risk score of 0.7354565. Risk curves and survival status distribution in the training set 1 clearly distinguished the two groups, with HRG patients showing higher mortality rates and shorter survival times (Figure 2d). K-M survival further demonstrated notably lower survival probabilities for HRG patients (p < 0.0001) (Figure 2e). ROC curves indicated strong predictive performance, with all AUC values exceeding 0.6 in the training set 1, and the AUC values of the 3-, 5-, and 7-year survival rates were 0.656, 0.728, and 0.747, respectively (Figure 2f). Similarly, 160 ccRCC patients from the validation set 1 were classified into HRG (88 samples) and LRG (72 samples) using the same median risk score of 0.6410694. In the risk curves and survival status, the HRG had a high mortality rate (Figure 2g). K-M survival curves further demonstrated notably lower survival probabilities for HRG patients (p = 0.00065) (Figure 2h), and the AUC values of the 3-, 5-, and 7-year survival rates were 0.668, 0.689, and 0.7, respectively (Figure 2i). Finally, these results were successfully validated in the validation set 2 (Figures 2j–l). These findings suggest that the developed risk model has the potential to stratify ccRCC patients based on their prognostic profiles, though its clinical utility requires further validation against established prognostic factors.
Figure 2
3.3 Clinical features, transcript abundance of prognostic genes
Through a comprehensive series of analyses, results illustrated that differences in risk score manifestation were found among the subgroups of T stage and Stage (p < 0.05), and the expression levels of risk scores were higher in the Stage IV and T4 Stage subgroup (Figure 3a). The HPA results denoted that the manifestation levels of the corresponding proteins of ADA, CDK5RAP3, and RUFY4 were detected to be drastically higher in ccRCC tissues (Figure 3b). In addition, the expression of prognostic genes showed significant differences between ccRCC and control samples. Among them, RGS20, ADA, AICDA, SLC12A5, RUFY4 and CDK5RAP3 were significantly upregulated in ccRCC group (p < 0.05) (Figure 3c). Similarly, in the RT-qPCR analysis, ADA, AICDA, SLC12A5, RUFY4 and CDK5RAP3 were notably upregulated in the ccRCC group, RGS20 expression trended upward, but no biologically relevant (p < 0.05) (Figure 3d).
Figure 3
3.4 Signaling pathways enrichment analysis in HRG and LRG patients of ccRCC
A total of 27 pathways were obtained between the HRG and LRG, such as olfactory transduction, oxidative phosphorylation (Figure 4a) (Supplementary Table 6) (adjust p < 0.05, |NES| >1). These outcomes demonstrated that these pathways participated in the pathogenesis of ccRCC. In the GSVA analysis, 141 pathways with differential expression were detected between HRG and LRG patients (Figure 4b) (Supplementary Table 7) (|t| > 2, p < 0.05). Notable up-regulated pathways in HRG included “citrate cycle TCA cycle”, etc. and homologous recombination in HRG included “propanoate metabolism”. These outcomes demonstrated that these pathways participated in the pathogenesis of ccRCC.
Figure 4
3.5 The immune functions of the 6 prognostic genes
In the immune factor analysis, RUFY4 showed the highest direct association with the chemokine CXCL13 (cor = 0.57, p < 0.001), and RGS20 showed the highest negative association with the chemokine CX3CL1 (cor = - 0.38, p < 0.001) (Figure 5a). CDK5RAP3 showed the highest direct association with the immune stimulant TNFRSF25 (cor = 0.83, p < 0.001), and ADA showed the highest negative association with the immune stimulants TNFSF13 (cor = - 0.40, p < 0.001) (Figure 5b). RUFY4 showed the highest direct association with the immune inhibitors CTLA4 (cor = 0.74, p < 0.001), and RGS20 showed the highest negative correlation with the immune inhibitors factor KDR (cor = - 0.46, p < 0.001) (Figure 5c). RUFY4 showed the highest positive correlation with the MHC HLA-DOB (cor = 0.64, p < 0.001) (Figure 5d). ADA showed the highest positive association with the chemokine receptor CXCR4 (cor = 0.60, p < 0.001) (Figure 5e). In addition, the immune checkpoint CTLA4 demonstrated the highest direct association with RUFY4 (cor = 0.74, p < 0.001), and RGS20 demonstrated the highest negative correlation with the immune checkpoint BTNL9 (cor = - 0.41, p < 0.001) (Figure 5f). Meanwhile, the immune checkpoint LAG3 showed the highest direct association with RUFY4 (cor = 0.68, p < 0.001) (Figure 5g).
Figure 5
3.6 Immunotherapy and immune microenvironment in ccRCC patients
Immunity is crucial for the initiation and progression of ccRCC (), immune infiltration analysis identified that such as activated B cells accounted for the highest proportion in HRG patients (Figure 6a) and 15 immune cells with differential abundance between HRG and LRG, such as eosinophils. For instance, eosinophils were higher in HRG (Figure 6b) (adjust p < 0.001). Correlation analysis demonstrated that activated B cells were marked positive association with immature B cells (cor = 0.85, p < 0.05), and RUFY4 showed the highest direct association with the effector memory CD8 T cells (cor = 0.52, p < 0.001), CDK5RAP3 illustrated the highest negative correlation with the Monocyte (cor = - 0.50, p < 0.001) (Figure 6c). The Immune Score, ESTIMATE Score were all found to be higher in the HRG, and tumor purity were all found to be higher in the LRG (Figure 6d). It was indicated that a stronger immune response was present in the samples of the HRG.
Figure 6
3.7 Tumor mutation and drug sensitivity in the HRG and LRG of ccRCC patients
In the results of the mutation analysis, the top 20 most frequently mutated genes in the HRG and in the LRG were presented. It was found that genes such as VHL and PBRM1 were significantly mutated in both the HRG (Figure 7a) and LRG (Figure 7b). In addition, the most frequent type of mutation among these genes was missense mutation. After that, in the analysis of chemotherapeutic drug sensitivity, 19 chemotherapeutic drugs showed significant differences in sensitivity between the HRG and LRG patients (p<0.05). The top 5 drugs such as CHIR.99021, Bexarotene with significant differences IC50 values were sorted in ascending order according to the p values (p < 0.05) (Figure 7c). The results showed that these drugs might have potential therapeutic effects on ccRCC.
Figure 7
3.8 Identification of MCs as key cells in ccRCC
A total of 28,537 cells and 23,541 genes were obtained in the GSE159115 dataset (Figure 8a). After data quality control, the count of cells was 26,378, and the count of genes remained 23,541 (Figure 8b). Then, 2,000 HVGs were identified, with the top 10 HVGs highlighted (Figure 8c). Later, PCA was conducted, and it was found that the distribution between the samples and the groups was relatively uniform, and there was no batch effect (Figure 8d). And the standard deviation of the first 30 PCs changed significantly, while the change trend became smaller after the 30th PC, so the top 30 PCs were selected (p < 0.05) (Figure 8e). The UMAP analysis illustrated that the cells were clustered into 11 distinct clusters (Figure 8f). These clusters were then annotated as belonging to 10 cell clusters (Figure 8g). Further stacked bar chart analysis demonstrated that the proportions of MCs were different between the ccRCC and control samples, and accounted for the highest proportion in tumor samples (Figure 8h). Notably, RGS20, ADA, AICDA, and SLC12A5 were differentially expressed in MCs, and their expression levels were relatively high (Supplementary Table 8) (p < 0.05). Based on these findings, MCs were identified as key cells associated with ccRCC.
Figure 8
3.9 Exploring the cell enrichment pathways, interactions, and differentiation of MCs
Cell enrichment analysis demonstrated that MCs were primarily associated with the “Electron transport from NADPH to Ferredoxin” (p < 0.05) (Figure 9a) (Supplementary Table 9). Cell-to-cell communication analysis emphasized frequent interactions between MCs and Pericytes in ccRCC (Figure 9b), the number of interactions between macrophages and MCs was increased (Figure 9c). At the same time, the number of communications between MCs and T cells increased (Figure 9d). The frequent interactions between MCs and Endothelials in control samples (Figures 9e–g), highlighting their potential roles in immune response and cellular signaling networks of the tumor microenvironment. Notably, compared with the control samples (Figure 9h), the receptor-ligand pair with a relatively high frequency of communication was CCL2 − ACKR1 of MCs and Endothelials in the ccRCC samples (Figure 9i), indicating potential differences in functions such as Inflammation and immunity, cell adhesion and migration. These findings demonstrated different signaling mechanisms of angiogenesis and cell migration in the context of the disease. Further analysis of macrophages through secondary dimensionality reduction and clustering revealed 9 distinct subclusters (Figure 10a). Pseudo-temporal analysis of macrophages contained 5 differentiation stages (Figure 10b), and differentiation uncovered a progressive trajectory from an early (light red) to a more mature (dark purple) state (Figure 10c). RGS20, SLC12A5, and ADA exhibited markedly higher expression in the middle-to-late differentiation stages; AICDA and RUFY4 demonstrated peak expression during the middle stage; CDK5RAP3 showed robust early-stage expression, followed by a gradual decline (Figure 10d). This indicated that 6 prognostic genes were involved in the expression of key cells.
Figure 9
Figure 10
4 Discussion
ccRCC remains a major clinical challenge due to its late presentation, high metastasis rate, and resistance to conventional therapies (, ). Ammonia-induced cell death (ammonitosis), a newly characterized regulated cell death pathway, has emerged as a critical player in cancer progression and immune modulation (), but its role in ccRCC pathogenesis remains underexplored. In this study, we integrated bulk transcriptomics, scRNA seq, and experimental validation. Using differential analysis and machine learning algorithms, six differentially expressed genes (RGS20, ADA, AICDA, SLC12A5, RUFY4, CDK5RAP3) related to ammonia death were identified and obtained. MCs were identified as key cells, and the functional enrichment and immune infiltration of high-risk patients were explored. Meanwhile, the pseudo temporal analysis explored the dynamic expression of prognostic genes during key cell differentiation processes. This study systematically analyzed the role of ADRGs in ccRCC for the first time and found that six prognostic genes were significantly upregulated in ccRCC and associated with reduced patient survival. This discovery suggests a potential link between ammonia metabolism-related genes and disease progression in ccRCC, warranting further investigation into its clinical relevance.
Regulator of G protein signaling 20 (RGS20) is a negative regulator of GPCR signaling, but its role in cancer is context-dependent (). In ccRCC, we found that RGS20 is highly expressed in tumor tissues and correlates with advanced stage and poor survival. This is consistent with a recent study showing that RGS20 promotes ccRCC cell migration and invasion by activating the PI3K/AKT pathway (). GPCRs are major drivers of ccRCC progression, for example, the von Hippel-Lindau (VHL) tumor suppressor gene, frequently mutated in ccRCC, regulates GPCR-mediated signaling (). Our data suggest that RGS20 may enhance GPCR signaling in ccRCC by modulating downstream effectors, thereby promoting tumor growth.
Adenosine deaminase (ADA) and activation-induced cytidine deaminase (AICDA) are key regulators of immune function. ADA catalyzes the conversion of adenosine to inosine, a process that suppresses T cell activation by reducing extracellular adenosine levels (). Research has shown that the decrease in ADA activity is closely related to disease progression in cancers such as head and neck cancer, prostate cancer, and laryngeal squamous cell carcinoma (). In ccRCC, we found that ADA expression is positively correlated with immune checkpoint molecules (e.g., PD-L1) and regulatory T cell (Treg) infiltration. This suggests that ADA may promote immune evasion by creating an immunosuppressive TME.
AICDA (Activation-Induced Cytokine Deaminase) is a core regulatory enzyme of adaptive immunity, mainly expressed in activated B cells (germinal center B cells) and present in small amounts in T cells and other immune cells, mediating class switching recombination (). However, recent work has identified AICDA in tumor cells, where it promotes genetic instability and therapy resistance (). In addition, studies have found that AICDA is an oncogene that may induce tumorigenesis through DNA demethylation, and targeting this gene may represent a novel therapeutic approach for metastatic ccRCC (). In our study, AICDA was highly expressed in ccRCC and correlated with B cell infiltration. Given that B cells in the TME can either promote (e.g., regulatory B cells) or inhibit (e.g., effector B cells) tumor growth, future studies should clarify whether AICDA modulates B cell function in ccRCC.
Solute carrier family 12 member 5 (SLC12A5) is a potassium chloride co transporter protein primarily involved in maintaining intracellular ion homeostasis. Research has shown that SLC12A5 has carcinogenic effects in various cancers, and its overexpression is associated with tumor progression, invasion, metastasis, and poor prognosis. For example, SLC12A5 inhibits apoptosis by suppressing apoptosis inducing factor (AIF) and Endonuclease G(EndoG)-dependent apoptosis signaling pathways, thereby promoting tumor cell survival. Additionally, it can promote tumor cell invasion and metastasis by regulating matrix components such as Matrix Metalloproteinases (MMPs) and Fibronectin (FN) (). In prostate cancer, SLC12A5 may promote tumor progression and drug resistance by interacting with m6A reader YTHDC1 and transcription factor HOXB13 (). The above evidence is consistent with our research conclusion that high SLC12A5 expression is associated with poor survival rate.
RUFY4 (RUN and FYVE domain-containing 4) regulates cytoskeleton dynamics by interacting with Rab GTPases (), while CDK5RAP3 (CDK5 regulatory subunit-associated protein 3) modulates cell cycle progression via CDK5 inhibition (). In addition, RUFY4 positively correlates with the expression of programmed death-ligand 1 (PD-L1), and the absence of RUFY4 leads to a decrease in PD-L1 expression, indicating that RUFY4 plays a significant role in immune regulation and immunotherapy (). In ccRCC, we found that both genes are upregulated in tumor tissues and correlate with increased cell proliferation (e.g., Ki-67 expression). These results indicate that RUFY4 and CDK5RAP3 are associated with poor survival in patients.
Collectively, these data suggest that the 6 ADRGs promote ccRCC through a network of metabolic, immune, and proliferative pathways, highlighting the complexity of ammonia toxicity mediated tumor progression. The scRNA-seq revealed that MCs are the dominant cell type in ccRCC and the primary expressers of the 6 prognostic genes. This is consistent with previous scRNA-seq studies showing that MCs drive tumor progression by secreting growth factors (e.g., VEGF) and modulating the TME (). MCs refer to the basic building blocks of cancer that break through normal growth regulation mechanisms due to genetic mutations or epigenetic changes. MCs can experience immune escape, leading to uncontrolled proliferation and enhanced invasion and metastasis capabilities (). Research has shown that the MCs of ccRCC are prone to invade the renal vein and form cancer thrombi, which can later metastasize to organs such as the lungs, bones, and liver (, ). Our intercellular communication analysis further confirmed that MCs interact with T cells and macrophages via ligand-receptor pairs such as CXCL12-CXCR4 and IL-6-IL6R. These interactions are known to suppress anti-tumor immunity: CXCL12-CXCR4 recruits Tregs and myeloid-derived suppressor cells (MDSCs), while IL-6-IL6R promotes M2 macrophage polarization ().
Through pseudo-time series analysis, it was found that the expression of the 6 prognostic genes increases as MCs differentiate from a “quiescent” to a “proliferative” state, suggesting that these genes are involved in tumor progression and therapy resistance. For example, RGS20 and CDK5RAP3 expression was highest in late-stage MCs, correlating with increased cell cycle activity (, ). This aligns with studies showing that tumor cell differentiation status is a key predictor of ccRCC survival.
Notably, MCs from high-risk patients had higher expression of immune checkpoint molecules (e.g., PD-L1) and chemokines (e.g., CCL2) compared to those from low-risk patients. This suggests that the prognostic model not only predicts survival but also reflects the immunogenicity of the tumor—an important consideration for immunotherapy selection.
While our study provides valuable insights into ccRCC pathogenesis, several limitations warrant consideration. First, the bulk transcriptomic data were derived from TCGA and ICGC, which may have geographic biases and limited clinical information (e.g., treatment history). Second, the scRNA-seq cohort was small (7 ccRCC patients, 6 normal controls), limiting the generalizability of our cell-type-specific findings. Third, we only validated gene expression using RT-qPCR; functional studies (e.g., knockdown/overexpression in cell lines, animal models) are needed to confirm the causal roles of the 6 ADRGs in ccRCC. Fourth, the mechanism by which these genes regulate ammonitosis (e.g., ammonia production, lysosomal stability) remains unclear and requires further investigation.
Future studies should address these limitations by: (1) Validating the prognostic model in multi-center cohorts with detailed clinical data; (2) Expanding scRNA-seq analysis to larger cohorts to characterize MC heterogeneity; (3) Performing functional experiments to elucidate the mechanistic links between ADRGs, ammonitosis, and ccRCC progression; (4) Exploring combination therapies targeting the 6 ADRGs and immune checkpoints (e.g., anti-PD-L1 + anti-ADA); (5) Developing circulating biomarkers (e.g., plasma ADRG mRNA) for non-invasive prognosis.
5 Conclusion
In summary, our study screened out six ADRGs in ccRCCC and constructed a risk model based on these six genes. The risk model shows promise as a potential adjunct tool for risk stratification in ccRCC, and the mechanistic insights into ammonia metabolism suggest novel avenues for therapeutic exploration. However, further validation in independent cohorts and comparison with established prognostic models are needed before clinical translation. These findings bridge the gap between basic science and clinical oncology, paving the way for more personalized and effective treatments for ccRCC patients.
Statements
Data availability statement
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.
Ethics statement
The studies involving humans were approved by the Ethics Committee of the Air Force 986th Hospital (approval number: KYLL2025-986-177). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements. Ethical approval was not required for the studies on animals in accordance with the local legislation and institutional requirements because only commercially available established cell lines were used.
Author contributions
YJ: Conceptualization, Writing – original draft. TZ: Validation, Formal analysis, Visualization, Software, Methodology, Writing – review & editing. JF: Validation, Formal analysis, Writing – review & editing. YS: Data curation, Writing – original draft. SQ: Visualization, Investigation, Writing – review & editing. JJ: Supervision, Writing – review & editing. XH: Software, Writing – review & editing. SZ: Software, Writing – review & editing. YW: Investigation, Writing – review & editing. LD: Validation, Writing – review & editing. BY: Methodology, Writing – review & editing, Funding acquisition. WZ: Methodology, Writing – review & editing, Funding acquisition.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This project is supported by the "New Talent Program" of the 986th Hospital of the Air Force (986-2024-SXRC-XXZT-001).
Acknowledgments
We would like to express our sincere gratitude to the Scientific Research Office of Air Force 986 Hospital for their generous financial support through the "New Talent Program". Their funding was instrumental in enabling us to conduct this research and achieve our objectives.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu.2026.1765098/full#supplementary-material
Supplementary Table 1ADRGs information.
Supplementary Table 2Primer sequence table.
Supplementary Table 3GO enrichment analysis pathway results.
Supplementary Table 4KEGG enrichment analysis pathway results.
Supplementary Table 5Schoenfeld test scatter plot of candidate prospective genes.
Supplementary Table 6Enrichment analysis of GSEA pathway in HRG and LRG patients.
Supplementary Table 7Enrichment analysis of GSVA pathway in HRG and LRG patients.
Supplementary Table 8Box plot of differential analysis of prognostic genes in cells, where ns represents insignificant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
Supplementary Table 9Cell enrichment analysis.
Abbreviations
Clear cell renal cell carcinoma, ccRCC; ammonia-induced cell death, ADRGs; Differentially Expressed Genes, DEGs; Ammonia death-related genes, ADRGs; high-risk group, HRG; low-risk group, LRG; Reverse transcription quantitative PCR, RT-qPCR; Malignant cells, MCs; Single cell RNA sequencing, scRNA seq; highly variable genes, HVGs; Principal Component Analysis, PCA; Uniform Manifold Approximation and Projection, UMAP; Kaplan-Meier, K-M; Ammonia-induced cell death, ammonitosis; von Hippel-Lindau, VHL; Adenosine deaminase, ADA; activation-induced cytidine deaminase, AICDA; apoptosis inducing factor, AIF; Endonuclease G, EndoG; Matrix Metalloproteinases, MMPs; Fibronectin, FN; myeloid-derived suppressor cells, MDSCs.
References
1
MotzerRJJonaschEAgarwalNAlvaABaineM. Beckermann K et al: Kidney Cancer, Version 3.2022, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. (2022) 20:71–90. doi: 10.6004/jnccn.2022.0001
2
LvZWangMHouHTangGXuH. Wang X et al: FOXM1-regulated ZIC2 promotes the Malignant phenotype of renal clear cell carcinoma by activating UBE2C/mTOR signaling pathway. Int J Biol Sci. (2023) 19:3293–306. doi: 10.7150/ijbs.84067
3
AlbiñanaVRecio-PovedaLGonzález-PeramatoPMartinez-PiñeiroLBotellaLMCuestaAM. Blockade of β2-adrenergic receptor reduces inflammation and oxidative stress in clear cell renal cell carcinoma. Int J Mol Sci. (2022) 23:1325 doi: 10.3390/ijms23031325
4
FengDGaoPHenleyNDubuissezMChenN. Laurin LP et al: SMOC2 promotes an epithelial-mesenchymal transition and a pro-metastatic phenotype in epithelial cells of renal cell carcinoma origin. Cell Death Dis. (2022) 13:639. doi: 10.1038/s41419-022-05059-2
5
LiZLinJYinP. Ammonia death: a novel potential strategy to augment immunotherapy in cancer. Cancer Gene Ther. (2024) 31:1751–3. doi: 10.1038/s41417-024-00851-y
6
ZhangHLiuJYuanWZhangQLuoX. Li Y et al: Ammonia-induced lysosomal and mitochondrial damage causes cell death of effector CD8(+) T cells. Nat Cell Biol. (2024) 26:1892–902. doi: 10.1038/s41556-024-01503-x
7
RehmanUGuptaGSahebkarAKesharwaniP. Ammonia-induced cell death: A novel frontier to enhance cancer immunotherapy. Immunology. (2025) 175:16–20. doi: 10.1111/imm.13918
8
AngelovaPRKerbertAJCHabtesionAHallAAbramovAYJalanR. Hyperammonaemia induces mitochondrial dysfunction and neuronal cell death. JHEP Rep. (2022) 4:100510. doi: 10.1016/j.jhepr.2022.100510
9
ChengCChenWJinHChenX. A review of single-cell RNA-seq annotation, integration, and cell-cell communication. Cells. (2023) 12:1970 doi: 10.3390/cells12151970
10
JikuyaRMurakamiKNishiyamaAKatoIFuruyaM. Nakabayashi J et al: Single-cell transcriptomes underscore genetically distinct tumor characteristics and microenvironment for hereditary kidney cancers. iScience. (2022) 25:104463. doi: 10.1016/j.isci.2022.104463
11
LoveMIHuberWAndersS. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. (2014) 15:550. doi: 10.1186/s13059-014-0550-8
12
GustavssonEKZhangDReynoldsRHGarcia-RuizSRytenM. ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinformatics. (2022) 38:3844–6. doi: 10.1093/bioinformatics/btac409
13
GuZ. Complex heatmap visualization. Imeta. (2022) 1:e43. doi: 10.1002/imt2.43
14
XiaWTQiuWRYuWKXuZCZhangSH. Identifying TME signatures for cervical cancer prognosis based on GEO and TCGA databases. Heliyon. (2023) 9:e15096. doi: 10.1016/j.heliyon.2023.e15096
15
YuGWangLGHanYHeQY. clusterProfiler: an R package for comparing biological themes among gene clusters. Omics. (2012) 16:284–7. doi: 10.1089/omi.2011.0118
16
SmootMEOnoKRuscheinskiJWangPLIdekerT. Cytoscape 2.8: new features for data integration and network visualization. Bioinformatics. (2011) 27:431–2. doi: 10.1093/bioinformatics/btq675
17
JiaWShiWYaoQMaoZChenC. Fan A et al: Identifying immune infiltration by deep learning to assess the prognosis of patients with hepatocellular carcinoma. J Cancer Res Clin Oncol. (2023) 149:12621–35. doi: 10.1007/s00432-023-05097-z
18
ShiYWangYDongHNiuKZhangW. Feng K et al: Crosstalk of ferroptosis regulators and tumor immunity in pancreatic adenocarcinoma: novel perspective to mRNA vaccines and personalized immunotherapy. Apoptosis. (2023) 28:1423–35. doi: 10.1007/s10495-023-01868-8
19
LinZHeYWuZYuanYLiXLuoW. Comprehensive analysis of copper-metabolism-related genes about prognosis and immune microenvironment in osteosarcoma. Sci Rep. (2023) 13:15059. doi: 10.1038/s41598-023-42053-w
20
HänzelmannSCasteloRGuinneyJ. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinf. (2013) 14:7. doi: 10.1186/1471-2105-14-7
21
RitchieMEPhipsonBWuDHuYLawCW. Shi W et al: limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. (2015) 43:e47. doi: 10.1093/nar/gkv007
22
HuFFLiuCJLiuLLZhangQGuoAY. Expression profile of immune checkpoint genes and their roles in predicting immunotherapy response. Brief Bioinform. (2021) 22:bbaa176 doi: 10.1093/bib/bbaa176
23
Nazari-GhadikolaeiAFikseFGelinder ViklundÅErikssonS. Factor analysis of evaluated and linearly scored traits in Swedish Warmblood horses. J Anim Breed Genet. (2023) 140:366–75. doi: 10.1111/jbg.12764
24
BlanchePDartiguesJFJacqmin-GaddaH. Estimating and comparing time-dependent areas under receiver operating characteristic curves for censored event times with competing risks. Stat Med. (2013) 32:5381–97. doi: 10.1002/sim.5958
25
WangWLuZWangMLiuZWuB. Yang C et al: The cuproptosis-related signature associated with the tumor environment and prognosis of patients with glioma. Front Immunol. (2022) 13:998236. doi: 10.3389/fimmu.2022.998236
26
SaoutJRLecuyerGLéonardSEvrardBKammerer-JacquetSFNoëlLet al. Single-cell deconvolution of a specific Malignant cell population as a poor prognostic biomarker in low-risk clear cell renal cell carcinoma patients. Eur Urol. (2023) 83:441–51. doi: 10.1016/j.eururo.2023.02.008
27
ZhangYNarayananSPMannanRRaskindGWangXVatsPet al. Single-cell analyses of renal cell cancers reveal insights into tumor microenvironment, cell of origin, and therapy response. Proc Natl Acad Sci U.S.A. (2021) 118:e2103240118 doi: 10.1073/pnas.2103240118
28
CaoTLiQHuangYLiA. plotnineSeqSuite: a Python package for visualizing sequence data using ggplot2 style. BMC Genomics. (2023) 24:585. doi: 10.1186/s12864-023-09677-8
29
LuoJDengMZhangXSunX. ESICCC as a systematic computational framework for evaluation, selection, and integration of cell-cell communication inference methods. Genome Res. (2023) 33:1788–805. doi: 10.1101/gr.278001.123
30
QiuXMaoQTangYWangLChawlaRPlinerHAet al. Reversed graph embedding resolves complex single-cell trajectories. Nat Methods. (2017) 14:979–82. doi: 10.1038/nmeth.4402
31
WangJGuanXShangNWuDLiuZGuanZet al. Dysfunction of CCT3-associated network signals for the critical state during progression of hepatocellular carcinoma. Biochim Biophys Acta Mol Basis Dis. (2024) 1870:167054. doi: 10.1016/j.bbadis.2024.167054
32
KrishnaCDiNataleRGKuoFSrivastavaRMVuongLChowellDet al. Single-cell sequencing links multiregional immune landscapes and tissue-resident T cells in ccRCC to tumor topology and therapy efficacy. Cancer Cell. (2021) 39:662–677.e666. doi: 10.1016/j.ccell.2021.03.007
33
SchiavoniVCampagnaRPozziVCecatiMMilaneseGSartiniDet al. Recent advances in the management of clear cell renal cell carcinoma: novel biomarkers and targeted therapies. Cancers (Basel). (2023) 15:3207 doi: 10.3390/cancers15123207
34
GuanBLiMCuiDShenCHaoZLiX. Single-cell transcriptomic analysis in clear cell renal cell carcinoma: Deciphering the role of APP within the tumour microenvironment. J Cell Mol Med. (2024) 28:e18186. doi: 10.1111/jcmm.18186
35
LiLXuQTangC. RGS proteins and their roles in cancer: friend or foe? Cancer Cell Int. (2023) 23:81. doi: 10.1186/s12935-023-02932-8
36
ShiDTongSHanHHuX. RGS20 promotes tumor progression through modulating PI3K/AKT signaling activation in penile cancer. J Oncol. (2022) 2022:1293622. doi: 10.1155/2022/1293622
37
TianZHYuanCYangKGaoXL. Systematic identification of key genes and pathways in clear cell renal cell carcinoma on bioinformatics analysis. Ann Transl Med. (2019) 7:89. doi: 10.21037/atm.2019.01.18
38
ZhuWWuFQiaoZZhaoMHuH. Nanomaterials-mediated adenosine pathway inhibition for strengthening cancer immunotherapy. Theranostics. (2025) 15:5007–28. doi: 10.7150/thno.108931
39
SunCChenYCMartinez ZuritaABaptistaMJPittalugaS. Liu D et al: The immune microenvironment shapes transcriptional and genetic heterogeneity in chronic lymphocytic leukemia. Blood Adv. (2023) 7:145–58. doi: 10.1182/bloodadvances.2021006941
40
KremMMPressOWHorwitzMSTidwellT. Mechanisms and clinical applications of chromosomal instability in lymphoid Malignancy. Br J Haematol. (2015) 171:13–28. doi: 10.1111/bjh.13507
41
CheZFanJZhouZLiQMaZ. Hu Z et al: Activation-Induced Cytidine Deaminase Expression Facilitates the Malignant Phenotype and Epithelial-to-Mesenchymal Transition in Clear Cell Renal Cell Carcinoma. DNA Cell Biol. (2020) 39:1299–312. doi: 10.1089/dna.2019.5119
42
XuLLiXCaiMChenJLiX. Wu WK et al: Increased expression of Solute carrier family 12 member 5 via gene amplification contributes to tumour progression and metastasis and associates with poor survival in colorectal cancer. Gut. (2016) 65:635–46. doi: 10.1136/gutjnl-2014-308257
43
YuanSHeSHLiLYXiSWengH. Zhang JH et al: A potassium-chloride co-transporter promotes tumor progression and castration resistance of prostate cancer through m(6)A reader YTHDC1. Cell Death Dis. (2023) 14:7. doi: 10.1038/s41419-022-05544-8
44
CharRPierreP. The RUFYs, a family of effector proteins involved in intracellular trafficking and cytoskeleton dynamics. Front Cell Dev Biol. (2020) 8:779. doi: 10.3389/fcell.2020.00779
45
ZhangCWangDShenYZhangYLiuJ. Nuclear translocation of CDK5RAP3 regulated by NXF3 promotes the progression of gastric cancer. Cell Mol Life Sci. (2025) 82:100. doi: 10.1007/s00018-025-05630-y
46
MiaoDShiJXiongZXiaoWMengX. Lv Q et al: As a prognostic biomarker of clear cell renal cell carcinoma RUFY4 predicts immunotherapy responsiveness in a PDL1-related manner. Cancer Cell Int. (2022) 22:66. doi: 10.1186/s12935-022-02480-7
47
RastegarSKallenMSusterDI. Updates in the role of the tumor microenvironment cellular crosstalk and genetic signatures in diffuse large B-cell lymphoma: a narrative review. Chin Clin Oncol. (2024) 13:38. doi: 10.21037/cco-23-124
48
ShiYZhangQBiHLuMTanY. Zou D et al: Decoding the multicellular ecosystem of vena caval tumor thrombus in clear cell renal cell carcinoma by single-cell RNA sequencing. Genome Biol. (2022) 23:87. doi: 10.1186/s13059-022-02651-9
49
BeiderKBitnerHLeibaMGutweinOKoren-MichowitzM. Ostrovsky O et al: Multiple myeloma cells recruit tumor-supportive macrophages through the CXCR4/CXCL12 axis and promote their polarization toward the M2 phenotype. Oncotarget. (2014) 5:11283–96. doi: 10.18632/oncotarget.2207
50
JiangLShenJZhangNHeYWanZ. Association of RGS20 expression with the progression and prognosis of renal cell carcinoma. Oncol Lett. (2021) 22:643. doi: 10.3892/ol.2021.12904
51
CharoentongPFinotelloFAngelovaMRiederDHackHTrajanoskiZ. et al: pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep. (2017) 18:248–62. doi: 10.1016/j.celrep.2016.12.019
Summary
Keywords
ammonia induced cell death, clear cell renal cell carcinoma, metabolic reprogramming, prognostic genes, single-cell sequencing analysis
Citation
Jiang Y, Zhang T, Fan J, Su Y, Qiao S, Ji J, Hu X, Zhou S, Wei Y, Du L, Yang B and Zhang W (2026) Identification of a prognostic signature based on ammonia metabolism-related genes in clear cell renal cell carcinoma: an integrated analysis of bulk and single-cell transcriptomics. Front. Immunol. 17:1765098. doi: 10.3389/fimmu.2026.1765098
Received
10 December 2025
Revised
14 March 2026
Accepted
18 March 2026
Published
23 April 2026
Volume
17 - 2026
Edited by
Xiulin Jiang, Beckman Research Institute, City of Hope, United States
Reviewed by
Tanxin Liu, University of Southern California, United States
Jingfen Su, Mayo Clinic Florida, United States
Updates
Copyright
© 2026 Jiang, Zhang, Fan, Su, Qiao, Ji, Hu, Zhou, Wei, Du, Yang and Zhang.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Bo Yang, Drboyung@aliyun.com; Wuhe Zhang, zwh74129@126.com
†These authors have contributed equally to this work and share first authorship
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.