Single-cell analysis of immune cells on gingiva-derived mesenchymal stem cells in experimental autoimmune uveitis

Abstract

Gingiva-derived mesenchymal stem cells (GMSCs) have shown astonishing efficacy in the treatment of various autoimmune diseases. However, the mechanisms underlying these immunosuppressive properties remain poorly understood. Here, we generated a lymph node single-cell transcriptomic atlas of GMSC-treated experimental autoimmune uveitis mice. GMSC exerted profound rescue effects on T cells, B cells, dendritic cells, and monocytes. GMSCs rescued the proportion of T helper 17 (Th17) cells and increased the proportion of regulatory T cells. In addition to globally altered transcriptional factors (Fosb and Jund), we observed cell type-dependent gene regulation (e.g., Il17a and Rac1 in Th17 cells), highlighting the GMSCs' cell type-dependent immunomodulatory capacity. GMSCs strongly influenced the phenotypes of Th17 cells, suppressing the formation of the highly inflammatory CCR6-CCR2+ phenotype and enhancing the production of interleukin (IL) -10 in the CCR6+CCR2+ phenotype. Integration of the glucocorticoid-treated transcriptome suggests a more specific immunosuppressive effect of GMSCs on lymphocytes.

Keywords: Immunology; Stem cells research; Transcriptomics.

Graphical abstract

Figure 1

Study design and clustering of CDLN immune cells by scRNA-seq (A) Study design. CDLNs of EAU mice, GMSCs-treated EAU mice (therapeutic group), and normal controls were utilized. GMSC, gingiva-derived mesenchymal stem cell. (B) Left: fundus photography of three groups of mice. Middle: Bar plots of the clinical scores of three groups of mice. Mean +SD with individual values is presented. Data are sampled from six mice per group. Right: Histological assessment of the retina in three groups of mice. Retinal sections were stained with hematoxylin and eosin, and evaluated for histological damage on day 14. Representative H&E staining images of three groups of mice (six mice per group) were presented. p-values are generated by one-way ANOVA. ∗∗∗∗p< 0.0001. Normal, normal subject; EAU, experimental autoimmune uveitis mouse; GMSC, GMSC-treated (therapeutic) mouse group. (C) Design of the scRNA-seq and data processing. CDLNs were mixed samples of three groups of mice. A total of six mixed samples (three samples from normal controls, two samples from EAU mice, one sample from GMSC-treated mice) were sequenced without pre-selection of flow cytometry. (D) UMAP embedding of CDLN immune cells colored by cell types. B, B cell; cDC, conventional dendritic cell; macro, macrophage; mono, monocytes; neu, neutrophil; pDC, plasmacytoid dendritic cell; T, T cell. (E) UMAP embeddings of lymph node immune cells colored by origins. (F) Dot plots depict marker gene expression of each cell type. Abbreviations were the same as (D). (G) Barplots of cell type proportions of seven types of immune cells. (H) Flow cytometry and quantification of T and B cells (gate on CD45⁺ immune cells). CD3 staining represents T cells, and CD19 staining represents B cells. (I) Left: Barplots of proportion of T cells of three groups of mice in CDLNs; Right: Barplots of proportion of B cells of three groups of mice in CDLNs. Mean ± SD with individual values is presented. Data are sampled from six mice per group. p-values are exact two-sided generated by one-way ANOVA. N, not significant, ∗p< 0.05, ∗∗p< 0.01, ∗∗∗p< 0.001, ∗∗∗∗p< 0.0001. (J) GSEA analysis of upregulated/downregulated EAU DEGs across all CDLN cells. (K) GSEA analysis of upregulated/downregulated GMSC DEGs across all CDLN cells.

Figure 2

GMSCs exerted immunosuppressive effect on T cell subsets of EAU mice (A) UMAP embeddings of T cell subtypes. CD8 eff: CD8⁺ effector T cells; Th17: Th17 cells; CD4n: CD4⁺ naive T cells; Th1: Th1 cells; Treg: regulatory T cells; Pro-T: proliferating T cells; NK: NK cells; CD8n: CD8⁺ naive T cells; Tfh: follicular T cells. (B) Heatmap of marker genes of T cell subtypes in three groups of mice. (C) Barplots of cell proportions of T cell subtypes. Annotation colors were the same as (A). (D) Barplots of cell proportions of Treg (gate on CD4⁺T cells), Th1, Tfh (gate on CD4⁺B220^-T cells) and Th17 cells. Mean ± SD with individual values is presented. Data are sampled from six mice per group. p-values are exact two-sided generated by one-way ANOVA. N, not significant, ∗p< 0.05, ∗∗p< 0.01, ∗∗∗p< 0.001, ∗∗∗∗p< 0.0001. (E) GO analysis of upregulated/downregulated GMSC DEGs of all T cells. (F) Schematic diagram of upregulated/downregulated rescue DEGs. (G) GO analysis of upregulated/downregulated DEGs in CD4⁺/CD8⁺T cells. (H) Venn diagram of commonly/specifically downregulated rescue DEGs among five types of CD4⁺T cells. The black dot at the bottom of the diagram indicates that genes were downregulated in certain CD4⁺T cell subset(s). The number over each bar indicates the overall counts of commonly/specifically downregulated genes in certain T cell subset(s).

Figure 3

GMSCs partially reversed the effect of EAU in Th17 cells (A) Pie charts indicate the Th17/Treg ratios in three groups of mice based on scRNA-seq. Ratios are shown under the pie charts. Normal, normal subject; EAU, experimental autoimmune uveitis mouse; GMSC, GMSC-treated mouse group. (B) Barplots of Th17/Treg ratios in three groups of mice based on flow cytometry. Mean ± SD with individual values is presented. Data are sampled from six mice per group. ∗∗∗p< 0.001. p-values are generated by one-way ANOVA. (C) Dot plots of rescue DEGs of GMSCs in Th17 cells. (D) UMAP embeddings show expression of Il17a, Rora and Cd48 of Th17 cells in three groups of mice. (E) GO analysis of upregulated/downregulated EAU/GMSC DEGs of Th17 cells. EAU up: upregulated EAU DEGs, GMSC down: downregulated GMSC DEGs, EAU down: downregulated EAU DEGs, GMSC up: upregulated GMSC DEGs. (F) GO analysis of upregulated/downregulated rescue DEGs in Th17 cells.

Figure 4

GMSCs influenced Th17 phenotype formation (A–D) Left: Flow cytometry and quantification of GM-CSF or IL-10 staining in Th17 cells in lymph nodes (A and C) or spleen (B and D). The percentage of cells was labeled over each gate. Right: Barplots of GM-CSF or IL-10 in Th17 cells across three groups of mice in lymph nodes or spleens. Mean ± SD with individual values is presented. Data are sampled from six mice per group. p-values are exact two-sided generated by one-way ANOVA. N, not significant, ∗p< 0.05, ∗∗p< 0.01, ∗∗∗p< 0.001, ∗∗∗∗p< 0.0001. Normal, normalsubject; EAU, experimental autoimmune uveitis mouse; GMSC, GMSC-treated mouse group. (E) Violin plots of expression of Ccr2, Ccr6 and Il17a of three groups of mice in Th17 cells. (F) UMAP embeddings of Ccr2 and Ccr6 of Th17 cells in three groups of mice. (G) Heatmap of phenotype-associated genes of Th17 cells in three groups of mice. (H and I) Left: Flow cytometry of staining of CCR6/CCR2 on Th17 cells in lymph nodes (H) or spleens (I) in three groups of mice. Data are sampled from six mice per group. Right: Bar plots summarize the proportion of CCR6⁺CCR2^-, CCR6⁻CCR2⁺ and CCR6⁺CCR2⁺ Th17 phenotypes in three groups of mice. Mean ± SD with individual values is presented. Data are sampled from six mice per group. p-values are exact two-sided generated by one-way ANOVA. N, not significant, ∗p< 0.05, ∗∗p< 0.01, ∗∗∗p< 0.001. (J) Bar plots of the mean fluorescent index (MFI) of IL-10 staining in CCR6⁺CCR2⁺ Th17 cells of lymph nodes (left) or spleens (right) in three groups of mice. Mean ± SD with individual values is presented. Data are sampled from six mice per group. p-values are exact two-sided generated by one-way ANOVA. N, not significant, ∗∗∗p< 0.001, ∗∗∗∗p< 0.0001.

Figure 5

GMSCs influenced B and myeloid cells (A) UMAP embeddings colored by B cell subtypes (left) or origins (right). Germinal: germinal B cells; naive: naive B cells; plasma: plasma B cells. Normal, normal subject; EAU, experimental autoimmune uveitis mouse; GMSC, GMSC-treated (therapeutic) mouse group. (B) Heatmap of marker genes of B cell subtypes in three groups of mice. (C) Bar plots of rescue/EAU DEGs ratios. Left: downregulated rescue DEGs/upregulated EAU DEGs; right: upregulated rescue DEGs/downregulated EAU DEGs. (D) Heatmap of rescue DEGs of naive B and plasma B cells in three groups of mice. (E) GO analysis of upregulated/downregulated rescue DEGs of naive B cells. (F) Heatmap of rescue DEGs in monocytes and cDCs in three groups of mice. (G) GO analysis of upregulated/downregulated rescue DEGs of monocytes.

Figure 6

GMSCs exhibited more specific rescue effect on EAU mice than prednisone (A) UMAP embeddings of lymph node immune cells colored by cell subtypes. B, B cell; cDC, conventional dendritic cell; macro, macrophage; mono, monocytes; neu, neutrophil; pDC, plasmacytoid dendritic cell; T, T cell. Prednisone: prednisone-treated mice. (B) UMAP embeddings of lymph node immune cells colored by origins. Normal, normalsubject; EAU, experimental autoimmune uveitis mouse; GMSC, GMSC-treated (therapeutic) mouse group; Prednisone: prednisone-treated mouse group. (C) Left: schematic Venn diagram of rescue DEGs of GMSCs or prednisone. Right: GO analysis of common/specific rescue DEGs of T cells of GMSCs or prednisone. GMSC down: specifically downregulated GMSC rescue DEGs, common: common downregulated rescue DEGs, pre up: specifically upregulated prednisone rescue DEGs, pre down: specifically downregulated prednisone rescue DEGs. The number of commonly rescued upregulated DEGs and specifically upregulated GMSC rescue DEGs are too small for GO analysis. (D) Pie charts of Th17/Treg ratios of GMSC-treated and prednisone-treated mice based on scRNA-seq. (E and F) Violin plots of commonly rescued DEGs (E) or specific rescued DEGs by GMSCs (F) in Th17 cells. (G) Left: schematic Venn diagram of rescue DEGs of GMSCs or prednisone in Th17 cells. Right: GO analysis of common/specific rescue DEGs of Th17 cells of GMSCs or prednisone. GMSC up: specifically upregulated GMSC rescue DEGs, GMSC down: specifically downregulated GMSC rescue DEGs, common: commonly downregulated rescue DEGs, pre up: specifically upregulated prednisone rescue DEGs. The number of commonly rescued upregulated DEGs and specifically downregulated GMSC rescue DEGs are too small for GO analysis. (H) Network plots showed the cell-cell communications change between different types of immune cells based on different groups of DEGs. Lines represent cell-cell communication pairs, the thickness of the lines represents the number of intercellular ligand-receptor pairs, and the color of the lines represents that the ligand-receptor pairs are strengthened (red) or weakened (blue). The circles indicate the total number of pairs identified in each type of cells. (I) Same as A, network plots of the changes of the cell-cell communication between 17 immune cell subtypes.