Abstract 214: The Breast Single-Cell Atlas: Single-cell transcriptomics for personalised medicine

• Published in: Cancer research (Chicago, Ill.) Ill.), 2021-07, Vol.81 (13_Supplement), p.214-214
• Main Authors: Viscido, Gaetano, Gambardella, Gennaro, di Bernardo, Diego
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2021-214

Breast Cancer (BC) patient stratification is driven by receptor status and histological grading and subtyping, with about 20% of patients for which absence of any actionable biomarkers results in no clear therapeutic intervention. Here, we evaluated the potentiality of single cell RNA-sequencing (scRNA-seq) for automated diagnosis and drug treatment of BC. We transcriptionally profiled 35,276 individual cells from 33 BC cell-lines covering all BC subtypes plus primary cancer cells from three patients to obtain a BC cell atlas. scRNA-seq successfully measured the expression of clinically relevant receptors and was used to automatically group cancer cell lines according to their tumor subtype. In most cell lines, as well as patients' cells, we observed a high degree of heterogeneity in the expression of BC receptors. We thus asked whether such heterogeneity impacts a cell line overall drug sensitivity. By correlating the percentage of cells expressing a given drug target (e.g. HER2, etc.) to the known IC50 of the relevant drug across the 33 cell lines, we observed a significant negative correlation (the higher the % of cells, the lower the IC50). This means that even within a genomically stable isogenic cell line, cells with differential drug sensitivity can co-exist. We then focused on the MDAMB361 cell-line of the luminal B subtype with a gain in genomic copy number of the locus containing the ERRB2 gene coding for HER2. Despite HER2 amplification, scRNA-seq showed that only 64% of cells express its mRNA. We used fluorescence-activating cell sorting to isolate HER2 expressing cells (HER2+) from non-expressing cells (HER2-). After a week, both subpopulations re-established the original heterogeneity, thus showing that heterogeneity in HER2 expression in these cells is dynamic. As expected, HER2 targeting drugs such as Afatinib and Erlotinib, have a higher IC50 in this cell line as compared to cell lines uniformly expressing HER2. We developed a bioinformatics approach named DEEP (Drug Estimation from Expression Profiles) to automatically predict responses to more than 450 anticancer agents starting from scRNA-seq and confirmed the validity of the approach using published large-scale studies on drug sensitivity. We then applied DEEP to the MDAMB361 cell-line to identify drugs able to selectively inhibit growth of the HER2- subpopulation. Etoposide was predicted to selectively inhibit growth of the HER2- cells but not HER2+ cells. We experimentally confirmed