Graphical AbstractHighlights d Ligand-receptor interactions in tumors were investigated using single-cell RNA-seq d Identified interactions were regressed against phenotypic measurements of tumors d The approach provides a tool for studying cell-cell interactions and their variability
In BriefTumors are composed of cancer cells and many non-malignant cell types, such as immune and stromal cells. To better understand how all cell types in a tumor cooperate to facilitate malignant growth, Kumar et al. studied communication between cells via ligand and receptor interactions using single-cell data and computational modeling.
SUMMARYTumor ecosystems are composed of multiple cell types that communicate by ligand-receptor interactions. Targeting ligand-receptor interactions (for instance, with immune checkpoint inhibitors) can provide significant benefits for patients. However, our knowledge of which interactions occur in a tumor and how these interactions affect outcome is still limited. We present an approach to characterize communication by ligand-receptor interactions across all cell types in a microenvironment using single-cell RNA sequencing. We apply this approach to identify and compare the ligand-receptor interactions present in six syngeneic mouse tumor models. To identify interactions potentially associated with outcome, we regress interactions against phenotypic measurements of tumor growth rate. In addition, we quantify ligand-receptor interactions between T cell subsets and their relation to immune infiltration using a publicly available human melanoma dataset. Overall, this approach provides a tool for studying cell-cell interactions, their variability across tumors, and their relationship to outcome.We thank Merrimack Pharmaceuticals, Inc. for sponsoring the research and supporting publication of the results and the internal review team for reviewing the article prior to submission. We also acknowledge the NIGMS Interdisciplinary Biotechnology Training Program (T32-GM008334) and NCI (U01-CA215798) for funding (to M.P.K. and D.A.L.). K. performed all computational analyses. J.D. and Y.J. processed the mouse syngeneic tumor samples, performed the flow cytometry analysis, and obtained the scRNA-seq data. A.S. provided the tumor growth data. M.P.K., A.R., and G.L. helped design the computational analysis. D.C.