Single cell sequencing reveals heterogeneity within ovarian cancer epithelium and cancer associated stromal cells

Abstract

Objectives: The purpose of this study was to determine the level of heterogeneity in high grade serous ovarian cancer (HGSOC) by analyzing RNA expression in single epithelial and cancer associated stromal cells. In addition, we explored the possibility of identifying subgroups based on pathway activation and pre-defined signatures from cancer stem cells and chemo-resistant cells.

Methods: A fresh, HGSOC tumor specimen derived from ovary was enzymatically digested and depleted of immune infiltrating cells. RNA sequencing was performed on 92 single cells and 66 of these single cell datasets passed quality control checks. Sequences were analyzed using multiple bioinformatics tools, including clustering, principle components analysis, and geneset enrichment analysis to identify subgroups and activated pathways. Immunohistochemistry for ovarian cancer, stem cell and stromal markers was performed on adjacent tumor sections.

Results: Analysis of the gene expression patterns identified two major subsets of cells characterized by epithelial and stromal gene expression patterns. The epithelial group was characterized by proliferative genes including genes associated with oxidative phosphorylation and MYC activity, while the stromal group was characterized by increased expression of extracellular matrix (ECM) genes and genes associated with epithelial-to-mesenchymal transition (EMT). Neither group expressed a signature correlating with published chemo-resistant gene signatures, but many cells, predominantly in the stromal subgroup, expressed markers associated with cancer stem cells.

Conclusions: Single cell sequencing provides a means of identifying subpopulations of cancer cells within a single patient. Single cell sequence analysis may prove to be critical for understanding the etiology, progression and drug resistance in ovarian cancer.

Keywords: Molecular subtypes; Ovarian cancer; Single cell sequencing and cancer stem cells.

Figure 1

Three clustering methods reveal two major groups of cells. A) Unsupervised hierarchical clustering based on 412 variably expressed genes. B) K-means clustering using the same 412 genes. C & D) Principle component analysis (PCA) 3D plot of 66 cells based on first three principle components analysis using 4,673 highly expressed genes. Cells are colored in C) based on color bar underneath hierarchical clustering heat map. Cells in D) are colored based on K-means color bar. The two major groups defined by all three methods (Group 1 = blue/lightblue vs Group 2 = red/lightred) are identical except for a single cell.

Figure 2

High expression levels of 8 ECM-related-genes define cell subpopulations. PCA plots using 4,673 genes × 66 samples with cells colored based on average expression levels of A) 8 highly variably expressed ECM genes, and B) EMT transcription factors (TWIST, SNAIL, ZEB). Red = high, yellow = medium, blue = low. C) PAX8 high in green. D) CA125 (MUC16) high in black.

Figure 3

PCA plots of all 66 cells colored based on expression of the indicated stem cell gene. A) CD44 in green. B) CD24 in red. C) EPCAM in blue. D) PROM1 in yellow. E) KIT in pink. F) ALDH1A1, ALDH1A2 and ALDH1A3 in brown. G) MYD88 in cyan. H) ABCG2 in red. In A-H, cells are colored if expression of the indicated gene(s) is > 2 (LogTPM), otherwise cells are in white. I) CK18 expression based on expression tertiles: low/none = blue, middle = yellow, high = red. Red and blue circles delineate stroma and epithelial subgroups identified in Figure 1, respectively.

Figure 4

Pie charts depicting distribution of single cells based on molecular subtype. A) Epithelial group cells (n=45) and B) Stroma group cells (n=21). Differentiated = green, proliferative = blue, mesenchymal = red, immunoreactive = yellow.

Figure 5

PCA plot with single cells colored based on presence of functional markers: Cancer epithelial cells (dark blue), cancer EMP cells (blue), cancer EMT cells (yellow), non-cancer EMP cells (red), fibroblasts (activated = black, not activated = grey), and myofibroblasts (activated = dark green, not activated = light green).