Mapping Organism-wide Single Cell mRNA Expression Linked to Extracellular Vesicle Biogenesis, Secretion, and Cargo

Abstract

Extracellular vesicles (EVs) are functional lipid-bound nanoparticles trafficked between cells and found in every biofluid. It is widely claimed that EVs can be secreted by every cell, but the quantity and composition of these EVs can differ greatly among cell types and tissues. Defining this heterogeneity has broad implications for EV-based communication in health and disease. Recent discoveries have linked single-cell EV secretion to the expression of genes encoding EV machinery and cargo. To gain insight at single-cell resolution across an entire organism, we compared the abundance, variance, and co-expression of 67 genes involved in EV biogenesis and secretion, or carried as cargo, across >44 000 cells obtained from 117 cell populations in the Tabula Muris. Our analysis provides both novel holistic and cell population-specific insight into EV biology. The highest overall expression of EV genes occurs in secretory cells of the pancreas and perhaps more surprisingly, multiple non-neuronal cell populations of the brain. We find that the most abundant EV genes encode the most abundant EV cargo proteins (tetraspanins and syndecans), but these genes are highly differentially expressed across functionally distinct cell populations. Expression variance identifies dynamic and constitutively expressed EV genes while co-expression analysis reveals novel insights into cell population-specific coordination of expression. Results of our analysis illustrate the diverse transcriptional regulation of EV genes which could be useful for predicting how individual cell populations might communicate via EVs to influence health and disease.

Keywords: cell atlas; endocrine; exosome; extracellular vesicle; secretion; single cell; tetraspanin.

Graphical Abstract

Figure 1.

Overview of genes selected based on links to endosome biogenesis, endosome protein sorting, secretion of exosomes and microvesicles and common extracellular vesicle (EV) protein cargo. Made using Biorender.

Figure 2.

Identification of “core” EV genes in Mus musculus. (A) Mean and frequency of EV gene expression in all cells irrespective of cell population. (B) Mean and frequency of EV gene expression as a function of cell population. (C) Abundant (>10% of all cell populations) EV gene co-expression pairs illustrated via Circos plot. (D) Mean-adjusted variance in EV gene expression.

Figure 3.

Identification of cell populations with high and low overall EV gene expression. (A) Z-scores were calculated for each gene across all cell populations then summed. Z-scores were then calculated from summed z-scores, and then plotted against frequency of EV gene expression. Text boxes and colored dots indicate cell populations with high (Z-score > 1) or low (Z-score < −1). -(B) Relative expression (reported as z-score) in high EV gene expressing cells. (C) Relative gene expression (reported as z-score) in low EV gene expression cells. (D) Cells with high (Z > 1) co-expression of EV genes. (E) Identification of cell populations with both high EV gene expression, and high co-expression of EV genes. N = 117 cell populations.

Figure 4.

Cell population-specific expression of abundant EV cargo genes. Cell populations with extremely high transcript abundance (Z-score > 2) of Cd9 (A), Cd63 (B), Cd81 (C), Sdc4 (D), and Sdcbp (E) relative to all other cell populations (n = 117).