Single-Cell Transcriptome Integration Analysis Reveals the Correlation Between Mesenchymal Stromal Cells and Fibroblasts

Abstract

Background: Mesenchymal stromal cells (MSCs) and fibroblasts show similar morphology, surface marker expression, and proliferation, differentiation, and immunomodulatory capacities. These similarities not only blur their cell identities but also limit their application. Methods: We performed single-cell transcriptome sequencing of the human umbilical cord and foreskin MSCs (HuMSCs and FSMSCs) and extracted the single-cell transcriptome data of the bone marrow and adipose MSCs (BMSCs and ADMSCs) from the Gene Expression Omnibus (GEO) database. Then, we performed quality control, batch effect correction, integration, and clustering analysis of the integrated single-cell transcriptome data from the HuMSCs, FMSCs, BMSCs, and ADMSCs. The cell subsets were annotated based on the surface marker phenotypes for the MSCs (CD105 ⁺ , CD90 ⁺, CD73 ⁺, CD45 ^-, CD34 ^-, CD19 ^-, HLA-DRA ^-, and CD11b ^-), fibroblasts (VIM ⁺, PECAM1 ^-, CD34 ^-, CD45 ^-, EPCAM ^-, and MYH11 ^-), and pericytes (CD146 ⁺, PDGFRB ⁺, PECAM1 ^-, CD34 ^-, and CD45 ^-). The expression levels of common fibroblast markers (ACTA2, FAP, PDGFRA, PDGFRB, S100A4, FN1, COL1A1, POSTN, DCN, COL1A2, FBLN2, COL1A2, DES, and CDH11) were also analyzed in all cell subsets. Finally, the gene expression profiles, differentiation status, and the enrichment status of various gene sets and regulons were compared between the cell subsets. Results: We demonstrated 15 distinct cell subsets in the integrated single-cell transcriptome sequencing data. Surface marker annotation demonstrated the MSC phenotype in 12 of the 15 cell subsets. C10 and C14 subsets demonstrated both the MSC and pericyte phenotypes. All 15 cell subsets demonstrated the fibroblast phenotype. C8, C12, and C13 subsets exclusively demonstrated the fibroblast phenotype. We identified 3,275 differentially expressed genes, 305 enriched gene sets, and 34 enriched regulons between the 15 cell subsets. The cell subsets that exclusively demonstrated the fibroblast phenotype represented less primitive and more differentiated cell types. Conclusion: Cell subsets with the MSC phenotype also demonstrated the fibroblast phenotype, but cell subsets with the fibroblast phenotype did not necessarily demonstrate the MSC phenotype, suggesting that MSCs represented a subclass of fibroblasts. We also demonstrated that the MSCs and fibroblasts represented highly heterogeneous populations with distinct cell subsets, which could be identified based on the differentially enriched gene sets and regulons that specify proliferating, differentiating, metabolic, and/or immunomodulatory functions.

Keywords: fibroblast; integration analysis; mesenchymal stromal cells; pericytes; single-cell transcriptome sequencing.

FIGURE 1

Characterization of the 15 cell subsets derived from ADMSCs, BMSCs, FSMSCs, and HuMSCs using various bioinformatics analysis methods. (A) The PCA plots show the distribution of ADMSCs (blue), BMSCs (red), FSMSCs (green), and HuMSCs (yellow), which are derived from different tissues before and after correcting the batch effects. The TSNE plot shows distribution of the 15 cell subsets (C0–C14). The clustering tree shows similarity between the 15 cell subsets. (B–D) The violin plot shows the expression status of ( B ) MSC-related markers, ( C ) fibroblast-related markers, and ( D ) pericyte-related markers in the 15 different cell subsets. ( E ) The violin plot shows the expression of classical fibroblast markers in the 15 different cell subsets. ( F ) The bubble plot shows the differential expression of the top three highly expressed genes in each cellular subset among all the 15 cellular subsets. The size of the bubble is inversely proportional to the p-value; that is, the larger the bubble, the smaller the p-value. The expression level is color coded, with dark red representing high expression and blue representing low expression.

FIGURE 2

Hallmark gene set enrichment analysis of the 15 cell subsets. The heatmap plot shows the enrichment scores and p-values of different hallmark gene sets among the 15 different cellular subsets. The color of the grids denotes the enrichment scores of the gene set, with red representing high enrichment and blue representing low enrichment. The bar graphs (top) represent different molecular phenotypes (fibroblasts, MSCs, and/or pericytes) represented by the 15 different cell subsets and is shown with different colors. The asterisk indicates that the p-value was less than 0.05. The bar charts (below) represent the cell cycle phases of different proportions of cells in the 15 cell subsets.

FIGURE 3

Evaluation of the differentiation status of cells in the 15 cellular subsets. ( A ) TSNE_plot shows the distribution of the 15 different cell subsets with different color codes in the low-dimensional space. ( B ) TSNE plot shows the distribution of the 15 different cell subsets based on the degree of differentiation. The degree of differentiation is denoted by the color code, with red representing low differentiation status and blue denoting high differentiation status. ( C ) The box plot shows the differentiation status of all cells in the 15 different cell subsets. The cells are ordered from low to high degrees of differentiation.

FIGURE 4

Identification of key gene regulatory networks in the 15 cell subsets. ( A ) The scatter plots show expression of the top five highly expressed regulons in each of the 15 cell subsets. ( B ) The heatmap plot shows expression levels of the top five highly expressed regulons in the 15 different cell subsets. SMC3, E2F2, and IRF2 regulons are labeled. ( C ) The plots show the putative binding motifs of SMC3, E2F2, and IRF2, as well as their corresponding expression levels in different cells among the 15 cell subsets.