Description

Flexible Single Cell Isolation Technologies Combined With High-Throughput Targeted Sequencing Enable the Characterization of Intratumor Heterogeneity

W. Hamou^1*, N. Bozinovic^2*, T. Silvers^1*, K. Beaumont², B. Bruhn², H. Shah¹, B. Evans¹, K. Allette¹, M. Strahl¹, H. Arib¹, A. Antoine¹, P. Dottino¹, M. Smith¹, J. Martignetti¹, M. White², E. Schadt¹, R. Sebra¹

¹Icahn School of Medicine at Mount Sinai & Icahn Institute for Genomics and Multi-scale Biology, New York, NY 10029; ²Berkeley Lights Inc., Emeryville, CA 94608

*Equal contribution

Bulk tumor tissue is known to be heterogeneous and genetically complex. Detailing this complexity is limited by current methods that are often destructive to the genomic material. The need to observe real-time cellular function in combination with single cell genetic information has driven the development of less destructive, integrated single cell technologies.

Here we present methods for single- and multi-cellular selection, manipulation and genomic characterization using the Berkley Lights Inc., Optoselect technology. This nanofluidics-based workflow facilitates cell selection based on morphology and/or fluorescence from a broad range of inputs. Selected cells are manipulated using light-induced dielectrophoresis into holding areas for isolation, culture, or real-time functional surveillance. Platform sensitivity and specificity was determined by identifing cells based on marker expression or secretion profile, allowing for the detection of one cell out of 100,000. To examine tumor heterogeneity, we selected single cells (n=51) as well as cell pools (i.e. 10-, 100- or 1000-cell pools and bulk material) from primary ovarian cancer tumors, extracted genomic DNA and characterized known tumor variants in each sample with the Ion Torrent HotSpot panel. These data identified multiple subclones in this tumor, demonstrating heterogeneity at the subclonal level compared to variant detection in bulk tissue. Together, the Optoselect technology and molecular methods developed here have the potential to enable studies that integrate functional interrogation and genomic profiling, generating genotype-to-phenotype models for cancer biology.