Single-cell RNA sequencing of mouse brain and lung vascular and vessel-associated cell types

Abstract

Vascular diseases are major causes of death, yet our understanding of the cellular constituents of blood vessels, including how differences in their gene expression profiles create diversity in vascular structure and function, is limited. In this paper, we describe a single-cell RNA sequencing (scRNA-seq) dataset that defines vascular and vessel-associated cell types and subtypes in mouse brain and lung. The dataset contains 3,436 single cell transcriptomes from mouse brain, which formed 15 distinct clusters corresponding to cell (sub)types, and another 1,504 single cell transcriptomes from mouse lung, which formed 17 cell clusters. In order to allow user-friendly access to our data, we constructed a searchable database (http://betsholtzlab.org/VascularSingleCells/database.html). Our dataset constitutes a comprehensive molecular atlas of vascular and vessel-associated cell types in the mouse brain and lung, and as such provides a strong foundation for future studies of vascular development and diseases.

Figure 1. Work flow.

(a) Flowchart of the procedure of generating the single cell dataset. (b) Whole adult mouse brains from the indicated mouse reporter lines were mechanically and enzymatically digested, and single cells were isolated by FACS, cDNA libraries prepared and sequenced. (c) Single cell transcriptomes were clustered by BackSPIN. The black bar-plot shows Actb expression (sequence counts) in the 38 clusters (0–37) generated at split-level 6: these clusters were given a preliminary cell class assignment (black-colored bars) using canonical cell type-specific markers. (d) After cluster consolidation, a final annotation was provided for individual cells. (e) The average expression (+/− standard error) of each cluster is summarized. Gene-by-gene expression figures are available at http://betsholtzlab.org/VascularSingleCells/database.html. Figure c–d overlap with Extended Data Figure 1b–c,i in our related publication (ref. 4).

Figure 2. Overview of the single cell data in the adult mouse brain and lung.

(a) The 3,418 brain single cells were analyzed by the T-Distributed Stochastic Neighbor Embedding (t-SNE) method to visualize their similarities, and the first two dimensions were used to plot the cells. Each cell is color-coded and also shape-coded according to its classified cell types from BackSPIN result annotation. (b) The same analysis of the 1,504 lung single cells as in panel a.

Figure 3. A screenshot of the database search outputs.

An example search of ribosomal gene Rpl13a in the online database http://betsholtzlab.org/VascularSingleCells/database.html. Four figures are displayed. (a) The detailed expression in each cell in the brain dataset. (b) The average expression level in each of the 15 clusters in the brain. (c) The detailed expression in each cell in the lung dataset. (d) The average expression level in each of the 17 clusters in the lung.

Figure 4. Quality control of the single cell sequencing library preparation.

(a) Representative cDNA graphs from a validation plate, analyzed with a High Sensitivity D5000 ScreenTape on a TapeStation 4200. The size distribution of the cDNA was established by running a ladder (top left). A1 and A2 represent positive controls (20 cells), while P1 and P2 are empty negative controls. The other graphs display cDNA size distribution of randomly picked single cells of a validation plate after amplification. Graph E1 has colored boxes to assist in clarifying peak significance: Red boxes indicate upper and lower markers for size selection, the green box shows the cDNA size distribution and the blue box highlights the primer dimers. (b) Quality control of the pooled sequencing library after tagmentation with homemade Tn5 and indexing with the Nextera Indexing XT kit. The sequencing pools were analyzed on a BioAnalyzer with a High Sensitivity DNA Chip. Colored boxes describe peak significance as described above.

Figure 5. The distribution of replications of endothelial cells and mural cells.

(a) The mouse origin and the technical plate replicates for the four plates of mural cells were color-coded in t-SNE and bar-plot display. The strong endothelial cell marker gene Cldn5 is illustrated. (b) The same analysis of the four plates of mural cells as in a. The strong mural cell marker Pdgfrb is illustrated.

Figure 6. Analysis of the gene expression level of five pericyte-specific genes.

The gene names are labeled on the top and right side of the figure. The five density plots in the diagonal line shows expression summary of each individual gene from all pericyte cells. The scatter plot on the lower left panels shows the pair-wise comparison of genes in each cell and the correlation coefficients are indicated on the upper right panels.

Figure 7. Analysis of the gene expression level of five broadly expressed genes.

Description refers to Figure 6.