Single-cell sequencing reveals the heterogeneity and intratumoral crosstalk in human endometrial cancer

Abstract

Background: Endometrial cancer (EC) is one of the most common gynecologic malignancies with increasing morbidity. Cell-cell and cell-matrix interactions within the tumour microenvironment (TME) exert a powerful influence over the progression of EC. Therefore, a comprehensive exploration of heterogeneity and intratumoral crosstalk is essential to elucidate the mechanisms driving EC progression and develop novel therapeutic approaches.

Methods: 4 EC and 2 normal endometrium samples were applied for single-cell RNA sequencing (scRNA-seq) analysis. In addition, we also included the public database to explore the clinical benefits of the single cell analysis.

Results: 9 types of cells were identified with specific expression of maker genes. Both the malignant epithelial cells and cells comprising the immune microenvironment displayed a high degree of intertumoral heterogeneity. Notably, the proliferation T cells also showed an exhausted feature. Moreover, the malignant cells may induce an immunosuppressive microenvironment through TNF-ICOS pair. Cancer-associated fibroblasts (CAFs) were divided into four subsets with distinct characteristics and they maintained frequent communications with malignant cells which facilitating the progression of EC. We also found that the existence of vascular CAF (vCAF) may indicate a worse prognosis for EC patients through integrating TCGA database.

Conclusion: The TME of human EC remains highly heterogeneous. Out finding that malignant cells interact closely with immune cells and vCAFs identifies potential therapeutic targets.

FIGURE 1

Comprehensive overview of human endometrial cancer. (A) Schematic diagram of scRNA‐seq analysis workflow; (B) UMAP plotting of the 41,358 cells showing 27 cell clusters; (C) The sample origin of the cells; (D) The distinct cell types identified by marker genes; (E) The number of cells in each cell type; (F) Bubble plots showing marker genes for 9 distinct cell types; (G) Bar plots showing the proportion of cell types in each sample

FIGURE 2

Transcriptomic heterogeneity of malignant cells in EC. (A) The cell cycling status of distinct cell types; (B) The heatmap of the relative expression density of genes on each chromosome by comparing the tumour cell genome with a series of normal cell reference genomes; (C) The CNV scores of each k‐means class; (D) The expression level of marker gene in TCGA dataset, (*p < 0.05); (E) Heatmap of DEGs in each k‐means class; (F) Differences in pathway activity (scored per cell by GSVA) in 6 epithelial cell sub‐clusters

FIGURE 3

Profiling of immune microenvironment in EC and intratumoral crosstalk with malignant cells. (A) t‐SNE plotting of the T cells showing 11 cell clusters; (B) The sample origin of the cells; (C) t‐SNE plots of marker genes for each cell type as indicated; (D) Violin plots of selected cytotoxicity, proliferation, and suppressive genes in distinct T cell subclusters; (E) Bar plots showing the proportion of cell types in each sample; (F) Interaction analysis showing enriched receptor‐ligand pairs in subsets of T cells and malignant cells; (G) Trajectory of differentiation from CD8+ Tcyto into Tex predicted by monocle 2; (H) Significantly up‐regulated genes in the differentiation process coloured by cell clusters

FIGURE 4

Distinct cancer‐associated fibroblasts subpopulations detected in human EC. (A) H&E, picrosirius red staining in EC and normal tissues; (B) t‐SNE plotting of the cancer‐associated fibroblasts (CAFs) showing 4 cell clusters; (C) The sample origin of the cells; (D) Heatmap showing the top 10 DEGs (Wilcoxon test) for each cluster; (E) GO analysis of DEGs in distinct CAF subclusters; (F) GSVA analysis revealing the hallmark pathways in distinct CAF subclusters; (G) Dot plot showing receptor‐ligand pair analysis of the interactions between malignant cells and distinct cell types

FIGURE 5

Prognostic significance of vCAF. (A) Heatmap showing the clustering result for the value of consensus clustering based on the vCAF markers; (B) Kaplan‐Meier survival analysis of tumour samples grouped in A; (C) Violin plots showing the estimated scores of TME in each cluster, (*p < 0.05, **p < 0.01, ***p < 0.001); (D) The expression level of classic stroma markers in each cluster; (E) Kaplan‐Meier survival curve of the prognostic model for TCGA EC patients; (F) Time‐dependent ROC curves of the prognostic model for 1‐,3‐ and 5‐year overall survival in EC; (G) The infiltrating immune cells in different cluster