Identifying Pulmonary Endothelial Cell Zonation From Single-Cell Transcriptomics
Stefanie N. Sveiven, Carsten Knutsen, Fabio Zanini, David N. Cornfield, Cristina M. AlviraBACKGROUND:
The lung vasculature is comprised of a series of branching vessels extending from the main pulmonary artery to the alveolar capillaries, then back to the pulmonary veins. Lung endothelial cells (EC) exist along this continuum, exposed to gradients of shear stress, oxygen tension, and pressure. Single-cell RNA sequencing has identified lung EC subsets, but many aspects of the vascular continuum, including vessel size and capillary polarity, remain undefined from transcriptomic data.
METHODS:
We created an EC-enriched single-cell RNA sequencing data set from the P3 mouse lung. Using diffusion pseudotime, we developed an analytical framework to delineate transcriptomic gradients and assign vessel-size scores to categorize individual EC along the vascular continuum. We validated size-related gene expression patterns with fluorescence in situ hybridization and tested the application of this framework to transcriptomic data sets derived from diverse species and developmental stages.
RESULTS:
We categorized capillary 1, arterial, and venous EC along 2 gradients: arterio-venous zonation and vessel size, distinguishing large arteries from arterioles, large veins from venules, and revealing arterio-venous polarity within the capillaries. Our data recapitulated previously established zonally defined cell signaling axes, identified unique cellular communication patterns in large versus small vessels, and localized injury-induced venous EC proliferation to vessels of specific size. This analytical framework was successfully applied to categorize lung EC by size in several published mouse and human data sets across different stages of lung development.
CONCLUSIONS:
Our findings provide a new approach to analyze transcriptional data to map EC along the pulmonary vascular tree, enabling the assignment of individual EC to vessels of a relative size. This framework allows the inference of spatial information from gene expression alone, thus providing novel mechanistic insights into pulmonary vascular diseases affecting specific vascular segments.