Single-cell transcriptomics in ovarian cancer identify a metastasis-associated cell cluster overexpressed RAB13

Abstract

Background: Metastasis, the leading cause of cancer-related death in patients diagnosed with ovarian cancer (OC), is a complex process that involves multiple biological effects. With the continuous development of sequencing technology, single-cell sequence has emerged as a promising strategy to understand the pathogenesis of ovarian cancer.

Methods: Through integrating 10 × single-cell data from 12 samples, we developed a single-cell map of primary and metastatic OC. By copy-number variations analysis, pseudotime analysis, enrichment analysis, and cell-cell communication analysis, we explored the heterogeneity among OC cells. We performed differential expression analysis and high dimensional weighted gene co-expression network analysis to identify the hub genes of C4. The effects of RAB13 on OC cell lines were validated in vitro.

Results: We discovered a cell subcluster, referred to as C4, that is closely associated with metastasis and poor prognosis in OC. This subcluster correlated with an epithelial-mesenchymal transition (EMT) and angiogenesis signature and RAB13 was identified as the key marker of it. Downregulation of RAB13 resulted in a reduction of OC cells migration and invasion. Additionally, we predicted several potential drugs that might inhibit RAB13.

Conclusions: Our study has identified a cell subcluster that is closely linked to metastasis in OC, and we have also identified RAB13 as its hub gene that has great potential to become a new therapeutic target for OC.

Keywords: Heterogeneity; Metastasis; Ovarian cancer; RAB13; Single-cell transcriptomics.

Fig. 1

Single-cell atlas of 12 patients. A, B UMAP of the all 155,173 cells. Colored by cell type or patient. NM: normal epithelium; PT: primary tumor; MT: metastatic tumor. C Violin plot showed the markers of each cell type. D Bar plot showed the cell proportion among patients. E Chromosomal landscape of inferred CNVs among cancer cell subclusters

Fig. 2

Heterogeneity of transcriptomes among OC cells. A Dimplot showed the UMAP of epithelial cell subclusters. B Violin plot demonstrated the difference in CNV scores among benign and malignant cell subclusters. C Malignant cell proportion among patients. D Hallmark and Reactome pathways of malignant cell subclusters determined by GSVA. E Kaplan–Meier analysis for patients from TCGA cohort with high and low GSVA score based on the top 20 markers of C4 (top) and C7 (bottom). F, G Pseudotime trajectory analysis highlights subcluster C4 that mainly derived from metastatic samples with the largest pseudotime

Fig. 3

Cell–cell Communications between OC and TME cells. A, B Cell communications among cancer cell subclusters and TME. The thickness of the line indicates numbers of signaling targeting malignant cells or TME cells. C–F Circle plots demonstrating the interactions of CXCL, Collagen, Laminin and VEGF pathways. G, H Bubble plots showing the ligands and receptors of VEGF and Laminin pathways

Fig. 4

Identification of gene co-expression modules among OC cells. A, B Weighed gene co-expression network analysis was constructed among malignant cells. (See “Materials and methods” section) C The first 20 eigengenes of each module, ranked by eigengene-based connectivity (kME). D Heatmap showed the relationships between modules and clinical phenotypes. E UMAP of the expression of Epi7 among all epithelial cells. F Protein–protein interaction network demonstrated the interactions within Module Epi7. G, H Dot plot of the GO (G) and KEGG (H) functional enrich analysis of the module Epi7. (*p < 0.05, **p < 0.01, ***p < 0.001 in a spearman test.)

Fig. 5

RAB13 is overexpressed in OC and promotes OC cells migration and invasion. A Volcano plot revealed the DEGs between primary and metastatic OC cells (left) and the five genes with the greatest difference (right). B Intersection of the upregulated genes in metastatic OC cells and the hub genes of Epi7. C RAB13 and HES1 in the UMAP plot colored by expression level with pseudotime trajectory. D KM curve of overall survival time and RAB13 (top) or HES1 (bottom), n = 1656. E, F The mRNA expression levels of RAB13 were higher in OC cell lines than normal ovary cell line (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 in t-test. N = 3), as well as the tissue protein levels based on the Human Protein Atlas (HPA) database. G, H Western blot analysis and qRT-PCR analysis of the expression of RAB13. The expression level of protein was quantified by grey analysis. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 in t-test. N = 3). I, J The wound healing assay and Transwell migration assay showed decreased migration ability of SK-OV-3 and A2780 cell lines treated with si-RAB13. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 in t-test. N = 3). K The Transwell invasion assay showed decreased invasion ability of SK-OV-3 and A2780 cell lines treated with si-RAB13. (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 in t-test. N = 3)

Fig. 6

Functions and Drug Response Prediction of RAB13 based on TCGA cohort. A Correlations between RAB13 and the enrichment scores of GOBP (left bottom) as well as KEGG (right top) pathways. B–F Boxplots displaying the sensitivity of the conventional chemotherapy drugs (B, C) or potential cytoskeleton inhibitors (D–F) among OC patients

Fig. 7

RAB13 promotes cell migration via regulating the tight junction and actin cytoskeleton. RAB13 inhibits the phosphorylation of VASP and suppress the binding of TJP1 with claudin and occludin by suppressing the ability of PKA, leading to decreased cell adhesion; Meanwhile, RAB13 forms a complex with JRAB and inhibits the remodeling of cytoskeleton by affecting filamin, thereby promoting the migration ability of cancer cells