Single-Cell RNA Sequencing Reveals the Tumor Heterogeneity and Immunosuppressive Microenvironment in Urothelial Carcinoma

Abstract

Urothelial carcinoma (UC) can arise from either the lower urinary tract or the upper tract; they represent different disease entities and require different clinical treatment strategies. A full understanding of the cellular characteristics in UC may guide the development of novel therapies. Here, we performed single-cell transcriptome analysis from four patients with UC of the bladder (UCB), five patients with UC of the ureter (UCU), and four patients with UC of the renal pelvis (UCRP) to develop a comprehensive cell atlas of UC. We found the rare epithelial cell subtype EP9 with epithelial-to-mesenchymal transition (EMT) and cancer stem cell (CSC) features, and specifically expressed SOX6, which was associated with poor prognosis. We also found that ACKR1+ endothelial cells and inflammatory cancer-associated fibroblasts (iCAFs) were more enriched in UCU, which may promote pathogenesis. While ESM1+ endothelial cells may more actively participate in UCB and UCRP tumorigenesis by promoting angiogenesis. Additionally, CD8 + effector T cells were more enriched in UCU and UCRP patients, while Tregs were mainly enriched in UCB tumors. C1QC+ macrophages and LAMP3+ dendritic cells were more enriched in UCB, which is closely related to the formation of the heterogeneous immunosuppressive microenvironment. Furthermore, we found strong interactions between iCAFs, EP9, and Endo_ESM1, and different degrees of activation of the FGF-FGFR3 axis and immune checkpoint pathway were observed in different UC subtypes. Our study elucidated the cellular heterogeneity and the components of the microenvironment in UC arising from the upper and lower urinary tracts and provided novel therapeutic targets.

Keywords: lower and upper urinary tract; single‐cell RNA sequencing; tumor heterogeneity; tumor microenvironment; urothelial carcinoma.

FIGURE 1

Single‐cell RNA‐seq of UC. (A) The design and workflow of this study. (B) tSNE of the 95,225 cells profiled here, with each cell color‐coded for its sample origin, the corresponding sample type, the corresponding patient, the associated cell type, the cell phase, and the cell clusters. (C) Dot plot of known marker genes of each major cell type. (D) Heatmap of top 20 marker genes of each major cell type. (E) The proportion of each cell type in nonmalignant and UC samples. (F) The proportion of each cell type in UCU, UCRP, and UCB.

FIGURE 2

Reclustering of epithelial cells in UC. (A) tSNE plot of subgroups of epithelial cells. (B) Uroepithelial differentiation‐related marker gene expression. (C) Heatmap showing marker gene expression between the luminal, basal, and EMT subtypes in epithelial cells. Representative genes are indicated (below). (D) Representative HALLMARK enriched by DEGs of EP6, EP7, EP8, and EP9 epithelial cell subsets. (E) Violin plot showing cancer stem cell marker genes’ expression in major cell types and epithelial cell subsets. (F) Immunofluorescence (IF) staining of SOX6 and ZEB1. Scale bar represents 50 μm. (G) RNA velocity of epithelial cells in UC. (H) Epithelial population distribution across latent time for UC samples.

FIGURE 3

Reclustering of endothelial cells in UC. (A) UMAP plot of subgroups of endothelial cells. (B) Heatmap showing marker genes’ expression in endothelial cell subtypes. (C) IF staining of ACKR1 and CD31. Scale bar represents 50 μm. (D) Representative HALLMARK enriched by DEGs of endothelial cell subsets. (E) Bar chart showing the relative proportion of ACKR1_Endo and ESM1_Endo in UCB, UCRP, and UCU patients. (F) High population of ACKR1_Endo predicted poor prognosis in the TCGA BLCA cohort. Log‐rank p value < 0.05 was considered as statistically significant. (G) Violin plot showing representative genes expression in major cell types and endothelial cell subsets.

FIGURE 4

Reclustering of T cells in UC. (A) UMAP plot of subgroups of T cells. (B) Heatmap showing marker genes’ expression in T cell subtypes. (C) Bar chart showing the relative proportion of T cell subsets in UCB, UCRP, and UCU patients, respectively. (D) Representative HALLMARK and KEGG pathways enriched by DEGs of T cell subtypes. (E) Violin plot showing representative immunosuppression‐related genes in major cell types and T cell subsets. (F) Receptor‐ligand interactions significantly enriched between T cell subtypes and epithelial cells’ population in UC. Displayed in UCB, UCRP, and UCU, respectively.

FIGURE 5

Reclustering of myeloid cells in UC. (A) UMAP plot of subgroups of myeloid cells. (B) Heatmap showing marker genes’ expression in myeloid cell subtypes. (C) Bar chart showing the relative proportion of myeloid cell subsets in UCB, UCRP, and UCU patients, respectively. (D) Violin plot showing representative immunosuppression‐related genes in major cell types and macrophage clusters. (E) Representative HALLMARK and KEGG pathways enriched by DEGs of myeloid cell subtypes. (F) Violin plot showing representative immunosuppression related genes in major cell types and DC subtypes. (G) High population of TAM_C1QC predicted poor prognosis in the TCGA BLCA cohort. Log‐rank p value < 0.05 was considered as statistically significant. (H) Correlation between DC_LAMP3 and Tregs in UCB, UCRP, and UCU patients, respectively.

FIGURE 6

Reclustering of fibroblasts in UC. (A) UMAP plot of subgroups of fibroblasts. (B) Heatmap showing marker genes’ expression in fibroblast subtypes. (C) Bar chart showing the relative proportion of fibroblast subsets in UCB, UCRP, and UCU patients, respectively. (D) Violin plot showing representative immunosuppression‐related genes in major cell types and fibroblast clusters. (E) Representative HALLMARK and KEGG pathways enriched by DEGs of myeloid cell subtypes.

FIGURE 7

Cell and gene interaction networks. (A) Heatmap showing the interaction intensity among major cell subsets from UCs. (B) Selected ligand–receptor pairs between iCAFs, endothelial cells, and EP9. (C) The interaction of basal/luminal with other major cell subsets in UC via the FGFR3/FGF interaction pair.