Fig 1.
Single-cell RNA-Seq analysis of ccRCC.
(A) Quality control metrics of single cells, including the number of detected genes per cell (nFeature_RNA), total UMI counts (nCount_RNA), percentage of mitochondrial transcripts (percent.mt), and hemoglobin gene expression ratio (percent.HB).(B) Principal component analysis (PCA) of single cells, colored by tissue type (normal vs. tumor), with an elbow plot showing the variance explained by the top principal components.(C) Uniform Manifold Approximation and Projection (UMAP) visualization of major cell clusters, annotated with distinct colors and corresponding cell-type proportions.(D) UMAP representation showing distribution of annotated cell types between tumor (T) and normal (N) samples.(E) Heatmap of average gene expression correlation across cell types, indicating the transcriptional similarity between clusters.(F) Bar plot showing the relative proportion of each cell type in tumor and normal tissues. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 2.
Identification of key cell populations by augur.
(A) Heatmap of representative gene modules (C1–C8) across major cell types, with module eigengene expression patterns (left), top representative genes (middle), and enriched biological processes (right). (B) UMAP visualization of single cells colored by module activity scores, demonstrating cell type–specific enrichment of functional gene sets. (C) Ranking of cell types based on average module activity scores, with NK cells exhibiting the highest specificity (0.79), followed by Mac/Mono (0.78), T cells (0.76), and other cell lineages. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 3.
Dissection of critical NK cell subclusters.
(A) UMAP visualization of tumor (T) and normal (N) tissues showing distinct NK cell subclusters (0–9). (B) Bar plot of the proportional distribution of NK cell subclusters between tumor and normal tissues, indicating tumor-specific enrichment of certain clusters. (C) Ranking of NK cell subclusters by functional activity scores, with subcluster 3 exhibiting the highest enrichment (0.76). (D) Network plots depicting intercellular communication among NK cell subclusters and other immune/stromal populations. The left panel shows the number of ligand–receptor interactions, while the right panel represents interaction strength. (E) Dot plot of significant ligand–receptor pairs mediating communication between NK subclusters and other cell types, with dot size indicating statistical significance and color representing communication probability. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 4.
Signaling network analysis and pseudotime trajectory of NK cell subpopulations.
(A) Heatmaps showing outgoing (left) and incoming (right) signaling patterns across different cell populations. The relative strength of each signaling pathway is depicted, highlighting major pathways involved in NK cell–mediated intercellular communication. (B) Detailed pathway-specific signaling networks, including ANNEXIN, CX3C, BAG, and IFN-II pathways. For each pathway, cell populations are annotated as senders, receivers, mediators, or influencers, demonstrating their roles in modulating NK cell signaling. (C) Pseudotime trajectory analysis of NK cell subpopulations. Left: trajectory plot colored by pseudotime progression. Middle: trajectory branches annotated by distinct cellular states. Right: trajectory colored by NK cell subtypes (IS_NK and Other_NK), suggesting differentiation dynamics within NK lineages. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 5.
Identification of key gene modules of NK cell by hd-WGCNA.
(A) Determination of the optimal soft-thresholding power for WGCNA. Plots show the scale-free topology model fit (top left), mean connectivity (top right), median connectivity (bottom left), and maximum connectivity (bottom right) across a range of soft-threshold powers. A soft-threshold power of 6 was selected to achieve a scale-free topology with balanced network connectivity. (B) Module membership (kME) distribution of key co-expression modules (yellow, blue, turquoise, red, green, and brown), with representative hub genes labeled for each module. (C) Correlation analysis among identified modules, presented as a module–module relationship matrix. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 6.
Module characterization of NK cell subpopulations.
(A) The UMAP plot shows cell populations marked by different modules. (B) The bubble chart displays the average expression levels and expression percentages of modules in different cell populations. (C) The network diagram shows genes associated with specific cell populations. (D) GO biological process enrichment analysis. The bar chart presents significantly enriched biological processes related to different cell populations. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 7.
Identification of core prognostic genes.
(A) The network diagram shows the interactions of differentially expressed genes in the turquoise and yellow cell populations. Nodes represent genes, and edges indicate interactions between genes. Genes in the turquoise and yellow cell populations are represented by nodes of different colors. (B) The forest plot illustrates the hazard ratios of significantly differentially expressed genes in univariate regression along with their 95% confidence intervals. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 8.
Construction and evaluation of prognostic models.
(A) The heatmap shows the prediction performance of different models in four cohorts, using AUC (Area Under the Curve) as the metric. The models include CoxBoost, Elastic_net (with different parameter settings), Lasso, Ridge, Stepcox (forward and backward selection), and Stepcox (combination of both). The depth of color indicates the level of AUC values, with red representing higher AUC values and blue representing lower AUC values. The legend on the right shows the range of AUC values for different cohorts. (B) The survival curve plot displays the prediction performance of different models across six cohorts. Each subplot shows the survival rates over time for the high-risk group (red) and low-risk group (cyan). All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 9.
Dissecting the molecular characterization of model genes.
(A) The box plot shows the expression level differences of multiple genes between normal tissue and tumor tissue. (B) The pie chart displays the results of gene enrichment analysis, categorized by biological process. (C) The bubble chart presents the results of gene set enrichment analysis (GSEA), categorized by signaling pathway. (D) The heatmap illustrates the coefficients of genes across different algorithm models. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 10.
Prognostic and therapeutic implications of UBE2S.
(A) The violin plot shows the expression distribution of the UBE2S gene in normal tissues and tumor tissues. (B) The survival curve illustrates the overall survival rate of the high UBE2S expression group (red, n = 21) and the low UBE2S expression group (cyan, n = 35) in the GSE167573 cohort. (C) The bar chart displays the distribution of patient survival status across different clinical stages (Q1 to Q4) in the KIRC-UBE2S dataset. (D) The stacked bar chart shows the distribution of patient survival status across different clinical stages (Stage I to Stage IV). (E) The violin plot illustrates the distribution of UBE2S gene expression across different clinical stages (G1 to G4) in the KIRC dataset. (F) Drug sensitivity analysis shows the potential therapeutic drugs for the gene. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 11.
Immune regulatory correlates of UBE2S.
(A) The heatmap shows the distribution of different immune response and genome status indicators across four different groups (Q1 to Q4).(B) The heatmap displays the expression of different immune checkpoints, antigen presentation, cell adhesion, receptors, and ligands in the four different groups.(C) The scatter plot illustrates the correlation between UBE2S expression and the expression of related genes in different biological processes (such as angiogenesis, apoptosis, cell cycle, differentiation, DNA damage, DNA repair, and epithelial-mesenchymal transition).(D) The scatter plot shows the correlation between different transcription factors from the Cistrome data browser and UBE2S gene expression. Each point represents a sample, with the color indicating the size and direction of the correlation coefficient; green represents a positive correlation, and purple represents a negative correlation. All plots were generated using R software (version 4.2). Individual panels were assembled in Adobe Illustrator.
Fig 12.
Knockdown validation of siUBE2S in ccRCC.
(A) The bar chart shows the relative mRNA expression levels of the UBE2S gene in ACHN and 786-O cell lines. The black bars represent the control group (NC), while the gray bars represent the UBE2S gene silencing group (siUBE2S). (B) Upper panel: Western blot analysis shows the expression of UBE2S protein in 786-O and ACHN cell lines. Lower panel: The bar chart displays the relative expression levels of UBE2S protein.
Fig 13.
Experimental validation of UBE2S in ccRCC.
(A) The cell viability assay results show the impact of UBE2S gene silencing (siUBE2S) on cell viability in the 786-O and ACHN cell lines. (B) The cell proliferation assay results illustrate the cell proliferation in the 786-O and ACHN cell lines for the control group (NC) and the UBE2S gene silencing group (siUBE2S). (C) The bar graph shows the number of cells in the control group (NC) and the UBE2S gene silencing group (siUBE2S) in the ACHN and 786-O cell lines.