Single-Cell RNA Sequencing Reveals Distinct Cardiac-Derived Stromal Cell Subpopulations

Abstract

Human cardiac-derived c-kit+ stromal cells (CSCs) have demonstrated efficacy in preclinical trials for the treatment of heart failure and myocardial dysfunction. Unfortunately, large variability in patient outcomes and cell populations remains a problem. Previous research has demonstrated that the reparative capacity of CSCs may be linked to the age of the cells: CSCs derived from neonate patients increase cardiac function and reduce fibrosis. However, age-dependent differences between CSC populations have primarily been explored with bulk sequencing methods. In this work, we hypothesized that differences in CSC populations and subsequent cell therapy outcomes may arise from differing cell subtypes within donor CSC samples. We performed single-cell RNA sequencing on four neonatal CSC (nCSC) and five child CSC (cCSC) samples. Subcluster analysis revealed cCSC-enriched clusters upregulated in several fibrosis- and immune response-related genes. Module-based analysis identified upregulation of chemotaxis and ribosomal activity-related genes in nCSCs and upregulation of immune response and fiber synthesis genes in cCSCs. Further, we identified versican and integrin alpha 2 as potential markers for a fibrotic cell subtype. By investigating differences in patient-derived CSC populations at the single-cell level, this research aims to identify and characterize CSC subtypes to better optimize CSC-based therapy and improve patient outcomes.

Keywords: c-kit+ cardiac stromal cells; heart failure; single cell RNA sequencing.

Figure 1

Clustering and cluster compositions of nCSCs and cCSCs. (A) Top sequence: CSC isolation from CHD patients and culture. Bottom sequence: analysis pipeline and computational tool summary. Figure generated in BioRender. UMAP projections of all patient-derived CSCs colored by (B) cell cluster and (C) age group. Cluster composition as grouped by (D) age group and (E) patient sample. (F) Enrichment of selected pathways from positive cluster marker gene sets, as listed in Table S6.

Figure 2

Trajectory analysis and gene clustering. UMAP projections with trajectories determined by Monocle colored by (A) monocle clusters and (B) pseudotime with root nodes set to cluster 2 (top) and cluster 6 (bottom) from the Seurat analysis. (C) gene module expression heatmap by Seurat cluster. Modules were determined through a Leiden clustering of highly co-expressed genes along trajectories. Module expression levels are computed by age group and are shown in the two rightmost columns in the heatmap. Cluster proportions by age group are illustrated by bar charts on top of the heatmap. Pathway analysis for the highest expressing modules among (D) neonates and (E) children highlights important biological differences between the age groups.

Figure 3

Characterization of inflammatory and fibrotic cell clusters 4 and 6. (A) Top 25 differentially expressed genes ordered by log fold change between cluster 4 cells and non-cluster 4 cells. Inflammatory cytokines are highlighted in purple. (B) Pathway analysis barplot of upregulated differentially expressed genes for clusters 4 and 6 cells. (C) Top 25 differentially expressed genes ordered by log fold change between cluster 6 cells and non-cluster 6 cells. Long non-coding RNAs (lncRNAs) are highlighted in green and fibrosis and extracellular matrix-related RNAs are highlighted in blue. (D) Dot plot of selected genes relating to fibrosis, angiogenesis, and proliferation. (E) Transcriptional expression of conserved differentially expressed surface proteins in cluster 6 cells. (F) Identification of a cluster 6-like population of interest in pooled child CSCs: ITGA2+, VCAN+.