Quantitative single-cell interactomes in normal and virus-infected mouse lungs.

Margo P Cain,Belinda J Hernandez,Jichao Chen

doi:10.1242/dmm.044404

Margo P Cain, Belinda J Hernandez + Show 1 more

Open Access

https://doi.org/10.1242/dmm.044404

Copy DOI

Abstract

ABSTRACTMammalian organs consist of diverse, intermixed cell types that signal to each other via ligand-receptor interactions – an interactome – to ensure development, homeostasis and injury-repair. Dissecting such intercellular interactions is facilitated by rapidly growing single-cell RNA sequencing (scRNA-seq) data; however, existing computational methods are often not readily adaptable by bench scientists without advanced programming skills. Here, we describe a quantitative intuitive algorithm, coupled with an optimized experimental protocol, to construct and compare interactomes in control and Sendai virus-infected mouse lungs. A minimum of 90 cells per cell type compensates for the known gene dropout issue in scRNA-seq and achieves comparable sensitivity to bulk RNA sequencing. Cell lineage normalization after cell sorting allows cost-efficient representation of cell types of interest. A numeric representation of ligand-receptor interactions identifies, as outliers, known and potentially new interactions as well as changes upon viral infection. Our experimental and computational approaches can be generalized to other organs and human samples.

Highlights

In multicellular mammalian organs, specialized cell types, such as beta cells in the pancreas and cardiomyocytes in the heart, perform organ-specific functions in coordination with generic cell types, such as the omnipresent endothelial and immune cells
RESULTS scRNA-seq has cell type resolution with sensitivity comparable to that of bulk RNA-seq One challenge to construct an interactome using scRNA-seq is the so-called gene dropout issue, where only a few thousand genes are detected in a given cell due to technical inefficiency, compared to the 20,000-30,000 genes expected and obtained by bulk RNA-seq of typical mammalian cells (Hicks et al, 2018; Kharchenko et al, 2014)
We used as a standard our published bulk RNA-seq data of fluorescence-activated cell sorting (FACS)-purified alveolar type 1 (AT1) and alveolar type 2 (AT2) cells (Little et al, 2019), and evaluated scRNA-seq gene dropouts as a function of expression level (Fig. 1A; Fig. S1A)

Summary

Introduction

In multicellular mammalian organs, specialized cell types, such as beta cells in the pancreas and cardiomyocytes in the heart, perform organ-specific functions in coordination with generic cell types, such as the omnipresent endothelial and immune cells. Delineating genome-wide ligand-receptor interactions between all pairwise combinations of constituent cell types in a given organ, hereinafter named interactomes, becomes feasible with the advancement of single-cell RNA sequencing (scRNA-seq) technology, using which individual cell types can be profiled upon computational, instead of physical, purification (Han et al, 2018; Tabula Muris Consortium et al, 2018) Several such singlecell interactomes have been constructed to characterize organs and cell culture systems at baseline and upon perturbation (Camp et al, 2017; Cohen et al, 2018; Kumar et al, 2018; Raredon et al, 2019; Skelly et al, 2018; Vento-Tormo et al, 2018; Vieira Braga et al, 2019). Current algorithms and outputs are not readily adopted by bench scientists without advanced computational skills

Methods

Results

Conclusion