Abstract 2451: Physical, genomic, and proteomic characterization of a cancer cell line panel in an integrated dataset

Abstract

Abstract Introduction: Recent work has linked cancer phenotype (e.g., metastatic ability) to physical properties (e.g., nuclear deformability), as well as genetic alterations (e.g., mutations in lamin genes affecting nuclear stiffness). However, while the genomics of immortalized human cell lines - important models for studying cancer disease pathogenesis and progression - have been relatively well studied, an integrative analysis of physical cell phenotypes with genomic, transcriptomic, and proteomics assays, spanning tissues of origins and culture environments (CE), has been lacking. The Physical Sciences in Oncology Network (PS-ON), a trans-disciplinary project established by the NCI, has generated a curated open dataset that is accessible to researchers via various database interfaces to address this need. Methods: A panel of 39 cancerous and non-malignant cell lines frequently used in cancer research were selected to cover a set of nine tissues: breast, lung, pancreas, brain, prostate, colon, ovary, blood, and skin. Thirty of these cell lines (from all tissues except blood) were assayed by traction and atomic force microscopy and imaging to measure their cell morphology, proliferation, motility, and nuclear volume. Measurements were performed in seven CE of varying physical properties (e.g., stiffness) to recapitulate a range of realistic tissue conditions and frequently used reagents. All 39 cell lines were assayed via whole exome sequencing, mRNA-seq, and miRNA-seq (in a single CE), while nine cell lines were subjected to proteomic analysis (across seven CEs). Results: All raw measurements and summary metrics from the study were incorporated into a data model compatible with established data stores (e.g., GDC, miRBase) and integrated in an open-access relational database (RDB). We performed several integrative analyses using the RDB. A broad set of interactions emerged between cell line CE characteristics (e.g., integrin ligands), the phenotypes of cells (e.g., motility) and gene expression, with effect-size varying by cell-line tissue of origin and cancer diagnosis. For instance, motility was significantly (FDR < 5%) correlated with expression of a small set of genes across tissues (e.g., BNP1 and PFKP in both skin and prostate cancers). Conclusion: This PS-ON cell line characterization allows correlative analyses across CE, cells’ morphological and proliferation properties, and omics data. The integrated RDB facilitates these analyses via a unified interface and data model for the large collection of files comprising the resource. We demonstrated the utility of the RDB by using it to infer the relationship between differential gene expression, CE and resulting phenotypes, with potentially important downstream effects on hallmarks of cancer malignancy. Citation Format: Milen Nikolov, Brian White, Adrian Pegoraro, Debra Hope, Mariam Eljanne, James Eddy, Paul Janmey, Parag Mallick, Kristen Dang. Physical, genomic, and proteomic characterization of a cancer cell line panel in an integrated dataset [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2451.