Integrated Genomic, Epigenomic, and Expression Analyses of Ovarian Cancer Cell Lines

Abstract

To improve our understanding of ovarian cancer, we performed genome-wide analyses of 45 ovarian cancer cell lines. Given the challenges of genomic analyses of tumors without matched normal samples, we developed approaches for detection of somatic sequence and structural changes and integrated these with epigenetic and expression alterations. Alterations not previously implicated in ovarian cancer included amplification or overexpression of ASXL1 and H3F3B, deletion or underexpression of CDC73 and TGF-beta receptor pathway members, and rearrangements of YAP1-MAML2 and IKZF2-ERBB4. Dose-response analyses to targeted therapies revealed unique molecular dependencies, including increased sensitivity of tumors with PIK3CA and PPP2R1A alterations to PI3K inhibitor GNE-493, MYC amplifications to PARP inhibitor BMN673, and SMAD3/4 alterations to MEK inhibitor MEK162. Genome-wide rearrangements provided an improved measure of sensitivity to PARP inhibition. This study provides a comprehensive and broadly accessible resource of molecular information for the development of therapeutic avenues in ovarian cancer.

Keywords: cancer cell lines; cancer genomics; drug response; gene fusions; ovarian cancer; structural variants.

Figure 1.. Overview of Genomic, Epigenomic, Expression, and Therapeutic Analyses of Ovarian Cancer Cell Lines

Figure 2.. Number of False-Positive Somatic Structural Variant Identifications in Lympho- blastoid Cell Lines

(A and B) Estimated number of false-positive somatic deletions and duplications (A) and somatic intrachromosomal and inter-chromosomal rearrangements (B).

Figure 3.. Trellis Approach for Characterization of Genomic Structural Alterations

(A) Circos plot displaying focal deletions (green), amplifications (orange), and intra- and inter-chromosomal rearrangements (blue) for cell line FU-OV-1. (B) Improperly paired reads established connections (edges) between distant amplicons (nodes) visualized as a graph. The size of the plotting symbols is proportional to the number of sites in which the amplicon was inserted, and the triangle shape indicates an amplicon involving a known driver. (C) The average maximum copy number (top) and mean number of amplicon links (bottom) for amplicon groups with and without drivers. (D) Top: Segmented normalized coverage identified a homozygous deletion (shaded). Bottom: Rearranged read pairs improved the precision of the deletion breakpoints. Lines connecting the read pairs indicate whether the positive or negative strand was sequenced (blue, positive; green, negative).

Figure 4.. Methylation of CpG Sites in Ovarian Cancers and Normal Fallopian Tissue

(A) The proportion of methylated CpG sites (mean β > 0.3) in the lymphoblastoid cell lines, ovarian cell lines, TCGA ovarian cancers, and TCGA normal fallopian tissues. (B) Plotted are 96 probes that were differentially methylated between normal TCGA fallopian tissue and 100 randomly selected TCGA ovarian tumors (blue points, A).Among these probes, the lymphoblastoid cell lines were most correlated with normal fallopian tissue and the ovarian cell lines were most correlated with TCGA ovarian tumors, suggesting that the cell line effect does not dominate among probes that were differentially methylated in these tissues. Among probes that were methylated in TCGA ovarian and unmethylated in TCGA fallopian, the ovarian cell lines were predominantly methylated and have quantitatively higher β values. While copy number analyses suggested that the purity in the ovarian cell lines was ≈00%, the median tumor purity of TCGA ovarian tumors was 85% (interquartile range, 78%−88%). (C) Genes CDKN2A and ESR1 exhibit bimodal gene expression explained by homozygous copy number deletions (blue points in x-axis margin) or methylation levels above 0.2.

Figure 5.. Sequence, Structural, Epigenomic, and Expression Alterations in Ovarian Cancer Cell Lines

Cell lines were grouped by tumor subtype (E, endometrioid; Und, undifferentiated; M, mixed). For many of the pathways, mutual exclusivity of genomic alterations within the pathway is evident (e.g., cell cycle, TK receptors, TGFBR, BRCA, and WNT). The group indicated as Other contains genes that are clinically relevant for ovarian cancer but cannot be easily categorized by a single molecular process. Methylation and expression were not evaluated for the Large Gene group.

Figure 6.. Sensitivity and Resistance to Pathway Inhibitors

(A) Bayesian model averaging was used to identify features associated with response to drug. Features selected in fewer than half of the multi-variate models have a posterior probability of being non-zero ≤ 0.5 (vertical dashed line, left) and a posterior median of zero (right). (B) Boxplots of inhibitor concentrations for features selected by this approach, as well as HRD, PARP1, and BRCA1/2 (left). The two cell lines with BRCA1/2 mutations are indicated by triangles inthe PARP pathway. Right: The difference in mean logIC50 concentrations by alteration status and the90% highest posterior density (HPD) interval for the difference.