Modeling the Diversity of Epithelial Ovarian Cancer through Ten Novel Well Characterized Cell Lines Covering Multiple Subtypes of the Disease

Abstract

Cancer cell lines are amongst the most important pre-clinical models. In the context of epithelial ovarian cancer, a highly heterogeneous disease with diverse subtypes, it is paramount to study a wide panel of models in order to draw a representative picture of the disease. As this lethal gynaecological malignancy has seen little improvement in overall survival in the last decade, it is all the more pressing to support future research with robust and diverse study models. Here, we describe ten novel spontaneously immortalized patient-derived ovarian cancer cell lines, detailing their respective mutational profiles and gene/biomarker expression patterns, as well as their in vitro and in vivo growth characteristics. Eight of the cell lines were classified as high-grade serous, while two were determined to be of the rarer mucinous and clear cell subtypes, respectively. Each of the ten cell lines presents a panel of characteristics reflective of diverse clinically relevant phenomena, including chemotherapeutic resistance, metastatic potential, and subtype-associated mutations and gene/protein expression profiles. Importantly, four cell lines formed subcutaneous tumors in mice, a key characteristic for pre-clinical drug testing. Our work thus contributes significantly to the available models for the study of ovarian cancer, supplying additional tools to better understand this complex disease.

Keywords: biomarkers; carboplatin; cell lines; clear cell; epithelial ovarian cancer; gene expression; high-grade serous; mucinous; mutation profile; xenograft.

Figure 1

Morphology of 10 new patient-derived epithelial ovarian cancer (EOC) cell lines. Shown are brightfield microscopy pictures of each cell line, between passages 60 and 75, except in the case of OV3291, represented at passage 33. At these passages, cells exhibited uniform morphology, and cell lines were devoid of fibroblast-shaped cells. All pictures were taken at a magnification of 100×.

Figure 2

Gene expression analysis of our EOC cell lines, and comparison to tumor samples from the TCGA high-grade serous carcinoma (HGSC) dataset. (a,b) The complete normalized gene expression data for each of the 12 analyzed cell lines (10 from this work plus the two matched tumor cell lines from patients 2978 and 3291) was subjected to (a) unsupervised hierarchical clustering and (b) principal component analysis (PCA) using the TM4 MeV software. (c) Expression of the 1000 most up- or downregulated genes in HGSC tumors from the TCGA cohort (left) was verified in 10 of our HGSC cell lines (8 from this work plus the two matched TOV2978G and TOV3291G cell lines) (right) and plotted as bar graphs (mean ± SEM). (d) Expression of the 1000 most variably expressed genes in the TCGA HGSC cohort was verified in the 10 HGSC cell lines mentioned in (c), and Pearson correlation analysis was performed.

Figure 3

Immunohistochemistry staining for ovarian cancer markers. Shown are immunohistochemistry (IHC) staining of the tumor of origin from which each cell line was derived, separated by subtype. Each tumor was tested for relevant biomarkers for its respective subtype.

Figure 4

Protein expression of EOC subtype-specific markers in tumor cell lines. Detection of characteristic subtype-specific EOC markers (p53; cytokeratins 7, 8, 18, and 19; WT1; PAX8; and ER) of whole cell lysates of each cell line. β-Actin was used as control (n = 3).

Figure 5

Confluence-based proliferation curves by live cell imaging. Cell proliferation of each cell line was determined by measuring confluence every 2 h. Initial values of confluence were between 5 and 10%, and cells were left to proliferate until confluence reached approximately 100%. Grey zones represent SEM (n = 3).

Figure 6

In vitro culture phenotypes. Spheroid formation of cell lines after 5–8 days using the hanging droplet technique with 2000 cells seeded, and migration evaluated by wound-healing scratch assay. Photos for migration were taken 0, 12, and 24 h after the plate was scratched. All photos are representative of three independent experiments.

Figure 7

In vivo growth characteristics. (a) Evolution of tumor volume after SC injection in NRG mice, for cell lines that induced observable tumor growth (n = 5). Points represent average ± SEM, and curves were plotted when end points were attained per group, when the first animal was sacrificed. (b) Kaplan–Meyer survival curves of NRG mice after IP injection with each of the cell lines (n = 5). For clarity, cell lines were separated and grouped by HGSC that formed (top) or did not form (middle) tumors, ascites, and/or metastases in at least 3/5 mice, and non-HGSC cell lines (bottom). Censored data points represent mice that had reached end points. (c) Summarizing table of in vivo growth characteristics.