Cytokine priming alters the heterogeneity of hUC-MSCs

hUC-MSCs were isolated from UC Wharton’s jelly and cultured according to previously described methods [19]. These primary cells can be MSCs, including those that undergo tri-lineage differentiation into adipogenic, chondrogenic, and osteogenic cells (Additional file 1: Fig S1A). The hUC-MSCs were positive for typical mesenchymal cell surface markers (CD105, CD90, and CD73), while hematopoietic cell markers (CD45, CD34, and CD19) were almost entirely absent (Additional file 1: Fig S1B). To analyse the biological characteristics of various cytokine-primed MSCs, hUC-MSCs were primed with IFN-γ, TNF-α, IL-4, IL-6, IL-15, and IL-17 in vitro. These samples were subsequently used to generate scRNA-seq data following the 10× Genomics protocol (Fig. 1A). After stringent cell filtration, we removed genes based on the following criteria: unique genes < 200 or > 9,000, number of unique molecular identifiers (UMI) < 2,000 or > 100,000, mitochondrial counts < 10% and ribosomal counts < 30%. A total of 23,250 cells were ultimately obtained for downstream analysis (Additional file 1: Fig S1C). We isolated four cell clusters using a graph-based method alongside visualization via tSNE. The analysis demonstrated that hUC-MSCs had four distinct subpopulations: cluster 0, cluster 1, cluster 2 and cluster 3 (Fig. 1B). All the subpopulations in the tSNE plots were positive for the expression of ENG (CD105), THY1 (CD90), and NT5E (CD73) and negative for the expression of PTPRC (CD45), CD34, and CD19 (Fig. 1C). Multiple cytokines can potentially influence the function of MSCs in vivo following transplantation through ligand–receptor binding interactions. Subsequently, the expression of receptors for these six cytokines in MSC subpopulations was assessed through scRNA-seq. Our analysis revealed the expression of IFN-γ, TNF-α, IL-4, and IL-17 receptors in hUC-MSCs, while the receptors for IL-6 and IL-15 (IL6R and IL15RA) were expressed at low levels. Among these receptors, the IFN-γ receptor (IFNGR2) and TNF-α receptor (TNFRSF1A) exhibited greater expression than did the other receptors. Additionally, IFNGR1, TNFRSF1B, IL4R, IL17RA, and IL17RC exhibited moderate expression levels (Fig. 1D). Protein expression levels of surface receptors were assessed via flow cytometry analysis (Additional file 1: Fig S1D). The results indicated that IFNGR1 and IL-17RA exhibited increased protein levels, whereas TNFR1, IL4R, IL6R, and IL15R demonstrated moderate expression. On the other hand, IFNGR2 and TNFR2 were found to be expressed at low levels (Additional file 1: Fig S1E). Thus, it is likely that hUC-MSCs can respond to these cytokines. scRNA-seq further demonstrated that the priming of different cytokines could induce the differential distribution of MSC subpopulations. The distribution of some cytokine-primed hUC-MSCs was predominantly reduced, demonstrating that cytokines, especially IFN-γ and TNF-α, at the single-cell level could significantly lessen the transcriptomic heterogeneity of MSCs (Fig. 1E). In unprimed MSCs, three large subpopulations were identified (clusters 0, 1, and 3), and one small subpopulation was identified (cluster 2). Only two large clusters (clusters 0 and 1) persisted following IFN-γ and TNF-α priming, which demonstrated that cytokine priming could lower the heterogeneity of hUC-MSCs, especially when IFN-γ and TNF-α were used (Fig. 1F). While different cytokine-primed hUC-MSCs expressed MSC-related surface markers, the expression levels of these markers differed among the groups. For example, the expression of ENG and NT5E in IFN-γ-primed hUC-MSCs was slightly elevated compared to that in other groups, while their THY1 expression was the lowest (Fig. 1G). The heterogeneity and surface markers of hUC-MSCs can be altered by cytokine priming, but their biological function requires further analysis via scRNA-seq.

Fig. 1

Cellular heterogeneity of hUC-MSCs before and after cytokine priming (A) Schematic overview of the study design. (B) t-SNE plots of the scRNA-seq clusters of hUC-MSCs. (C) t-SNE plots showing the expression levels of marker genes of hUC-MSCs. (D) Violin plots showing the expression levels of the receptor genes IFN-γ (IFNGR1, IFNGR2), TNF-α (TNFRSF1A, TNFRSF1B), IL-4 (IL4R), IL-6 (IL6R), IL-15 (IL15RA), and IL-17 (IL17RA, IL17RC, IL17RD). (E) t-SNE plots of the scRNA-seq data of hUC-MSCs after cytokine priming. (F) Sector graph illustrating the percentage change in the four clusters in unprimed, IFN-γ-primed, or TNF-α-primed hUC-MSCs; this change reflects the change in heterogeneity of the cell subpopulations. (G) Violin plots showing the expression levels of the marker genes of hUC-MSCs after cytokine priming

Analysis of differentially expressed genes (DEGs) in hUC-MSCs primed with various cytokines

Compared to those in the unprimed hUC-MSCs, we found DEGs across IFN-γ, TNF-α, IL-4, IL-6, IL-15, and IL-17-primed hUC-MSCs (Additional file 11–16: Table S2-7). GBP1, IDO1, WARS, HLA-B, and NME1-NME2 were highly expressed in IFN-γ-primed hUC-MSCs. CXCL5, CXCL1, CCL2, and IL8 were highly expressed in TNF-α-primed hUC-MSCs. Similarly, BCYRN1, MT-ND3, MT-ATP6, MT-ND5 and MT-ND1 were highly expressed in IL-4-primed hUC-MSCs. NME1-NME2, RPL17, NBEAL1, RPS10, and MIF were highly expressed in IL-6-primed hUC-MSCs. BCYRN1, MT-ND3, MT-ATP6, MT-ND5, and MT-ND2 were highly expressed in IL-15-primed hUC-MSCs. RPL17, and NME1-NME2, RPS10, NBEAL1, and MIF were highly expressed in IL-17-primed hUC-MSCs (Fig. 2A). To clarify the number of upregulated and downregulated DEGs across cytokine-primed hUC-MSCs and unprimed hUC-MSCs, we used the Seurat FindAllMarkers function for every sample. We performed a Wilcoxon rank sum test, with the DEGs of samples selected based on a p < 0.05 and a fold change > 0.8. The DEG data indicated that 101 DEGs (74 upregulated genes and 27 downregulated genes) were upregulated in IFN-γ-primed hUC-MSCs compared to unprimed hUC-MSCs; moreover, 61 DEGs (42 and 19), 48 DEGs (36 and 12), 28 DEGs (13 and 15), 27 DEGs (25 and 2), and 21 DEGs (14 and 7) were upregulated in the TNF-α, IL-4, IL-6, IL-15, and IL-17-primed hUC-MSC groups, respectively (Fig. 2B). There were fewer DEGs in the IL-6, IL-15, and IL-17-primed groups, and these three cytokines had minimal effects on cellular functions, as shown by GO and KEGG analyses. For example, ribosome function was enriched in IL-6- and IL-17-primed hUC-MSCs, and extracellular matrix (ECM) function was slightly altered in IL-15-primed hUC-MSCs (Additional file 2: Fig. S2A, B). Considering the low expression of IL-6 and IL-15 receptors (IL6R and IL15RA) and moderate expression of IL-17 receptors (IL17RA and IL17RC) (Fig. 1D), we investigated these two groups of cells. Therefore, we investigated only IFN-γ-, TNF-α-, and IL-4-primed hUC-MSCs in the following analysis. Furthermore, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis demonstrated that antigen processing and presentation, cell adhesion, and T helper cell differentiation were enriched in IFN-γ-primed hUC-MSCs, whereas the tumor necrosis factor (TNF)/IL-17/NF-κB signalling pathway, chemokine signalling pathway, and cytokine‒cytokine receptor interaction were associated with TNF-α-primed hUC-MSCs, and focal adhesion and leukocyte transendothelial migration were enriched in IL-4-primed hUC-MSCs (Fig. 2C). Upon GO enrichment analysis, negative regulation of immune system processes (associated genes IDO1, GBP1, HLA-E, and CD74), response to interferon-gamma (associated genes GBP1-4, IFITM1, and STAT1), and cytokine-mediated signalling pathway (associated genes IFITM1, STAT1, IRF1, and PARP14), were associated with the IFN-γ-primed hUC-MSCs (Fig. 2D, E); leukocyte migration, leukocyte chemotaxis, and cell chemotaxis were enriched in TNF-α-primed hUC-MSCs (associated genes CXCL1, CXCL5, CXCL6, CCL2, and MIF) (Fig. 2D, F); adhesion pathways, such as focal adhesion and cell-substrate junction (associated genes RPS8, RPS29, RPL31, RPL38, ALCAM, MME, HSP90B1, and HSPA5), collagen-containing extracellular matrix (associated genes COL6A1, COL6A2, COL1A1, COL3A1, FN1, DCN, FBN1, and FBLN1), were associated with the IL-4-primed hUC-MSCs (Fig. 2D, G). Overall, we found that IFN-γ-primed hUC-MSCs can regulate the interferon-mediated immune system response, TNF-α-primed hUC-MSCs can promote leukocyte migration and chemotaxis, and IL-4-primed hUC-MSCs play key roles in focal adhesion, cell-substrate junctions, and the collagen-containing extracellular matrix.

Fig. 2

Predicted functions and pathways associated with DEGs after cytokine priming in hUC-MSCs. (A) Differential expression gene analysis showing upregulated and downregulated genes across cytokine-primed hUC-MSCs. An adjusted p value < 0.01 is indicated in red, while an adjusted p value > = 0.01 is indicated in gray. (B) Statistical table of genes significantly differentially expressed between cytokine-primed hUC-MSCs and unprimed hUC-MSCs. (C-D) KEGG (C) and GO (D) enrichment analyses of IFN-γ-, TNF-α-, or IL-4-primed hUC-MSCs; dot plot showing the most significant terms. The size of each dot indicates the gene ratio (the total number of DEG-enriched genes). The color indicates the adjusted p value for enrichment analysis. (E) GO enrichment network of IFN-γ-primed hUC-MSCs. (F) GO enrichment network of TNF-α-primed hUC-MSCs. (G) GO enrichment network of IL-4-primed hUC-MSCs

Gene set enrichment analysis of hUC-MSCs primed with various cytokines

We performed GSEA on our scRNA-seq data using gene sets from the GO and KEGG databases to better understand the underlying mechanism of IFN-γ, TNF-α, or IL-4-primed hUC-MSCs and unprimed hUC-MSCs. As expected, compared with unprimed hUC-MSCs, IFN-γ-primed hUC-MSCs had increased interferon-gamma-mediated signalling pathway activity (NES = 1.89), increased negative regulation of the innate immune response (NES = 2.03), and increased negative regulation of immune system processes (NES = 1.94) (Fig. 3A). We also found that TNF-α-primed hUC-MSCs activated cytokine activity (NES = 2.36), the TNF signalling pathway (NES = 2.18), and cytokine‒cytokine receptor interactions (NES = 2.31) (Fig. 3B). Similarly, our GSEA showed that IL-4-primed hUC-MSCs had an activated extracellular structure organization (NES = 2.36), an enhanced ECM-receptor interaction (NES = 2.31), and a cytokine‒cytokine receptor interaction (NES = 1.74) (Fig. 3C). These results suggest that diverse cytokines stimulate various signalling pathways in IFN-γ-, TNF-α-, and IL-4-primed hUC-MSCs, possibly altering the biological functions of hUC-MSCs in vitro and in vivo.

Fig. 3

Gene set enrichment analysis of cytokine-primed hUC-MSCs. (A) Three representative significantly enriched gene sets from IFN-γ-primed hUC-MSCs; the normalized enrichment score and adjusted p value were calculated via permutation tests. (B) Three representative significantly enriched gene sets from TNF-α-primed hUC-MSCs; the normalized enrichment score and adjusted p value were calculated via permutation tests. (C) Three representative significantly enriched gene sets from IL-4-primed hUC-MSCs; the normalized enrichment score and adjusted p value were calculated via permutation tests

Analysis of chemotaxis, immunomodulation, and collagen synthesis

Tri-lineage differentiation gene score analysis suggested that there was no significant difference between unprimed and cytokine-primed hUC-MSCs, demonstrating that IFN-γ, TNF-α, or IL-4 priming had little effect on MSC differentiation, including adipogenic, chondrogenic, and osteogenic ability (Additional file 2: Fig S2C-E). Moreover, the chemotaxis gene score indicated that 99.4% of the TNF-α-primed hUC-MSCs had high chemotactic ability, while the other three groups had chemotactic ability, with 8.8% (unprimed), 1.8% (IFN-γ), and 12.4% (IL-4) (Fig. 4A). The immunomodulatory gene score indicated that 93.6% of the IFN-γ-primed hUC-MSCs and 58.8% of the TNF-α-primed hUC-MSCs had high immunomodulatory ability, while 3.2% of the unprimed hUC-MSCs and 10.1% of the IL-4-primed hUC-MSCs (Fig. 4B). The collagenic gene score showed that 94.0% of the IL-4-primed hUC-MSCs and 79.6% of the TNF-α-primed hUC-MSCs had high collagenic ability, whereas 45.3% of the unprimed hUC-MSCs and 19.8% of the IFN-γ-primed hUC-MSCs (Fig. 4C). We also assessed the consistency of the AddModuleScore results mentioned above by utilizing methods based on the gene expression ranking of a single sample, such as AUCell and Ucell. The results obtained from the AUCell and Ucell methods were consistent with the results obtained from the AddModuleScore (Additional file 3: Fig S3A-C). To confirm the accuracy of these scores, we performed functional experiments to enhance the applicability of our bioinformatics data. Specifically, we found that TNF-α-primed hUC-MSCs exhibited greater chemotactic migration than did the other groups (Additional file 4: Fig S4A, B). Additionally, compared with those in the other MSC groups, the immunosuppressive ability of IFN-γ-primed hUC-MSCs was greater (Additional file 4: Fig S4C). Moreover, the collagen secretion assay results indicated significant increases in the collagen I and collagen V levels in comparison to those in the other three groups (Additional file 4: Fig S4D-F). These functional findings provide further confirmation and validation of the scRNA-seq results described in Fig. 4A-C. scRNA-seq data also revealed that IFN-γ-primed hUC-MSCs expressed high levels of immunomodulatory genes, including IDO1, HLA-G, CD274, and FAS (p ≤ 0.0001; Fig. 4D). The expression of chemotactic-associated genes, including CCL2, CXCL1, CXCL2, CXCL5, and IL8 (CXCL8), was significantly upregulated in TNF-α-primed hUC-MSCs compared to unprimed hUC-MSCs (p ≤ 0.0001; Fig. 4E). The expression of the collagen-associated genes COL1A1, COL3A1, COL6A1, COL6A2, and COL5A1 was greater in the IL-4-primed hUC-MSCs than in the other three groups (p ≤ 0.0001; Fig. 4F). The duration of cytokine priming may influence the expression of functional genes in hUC-MSCs. We selected five time points for cytokine priming with IFN-γ, TNF-α, and IL-4: 6 h, 12 h, 24 h, 36 h, and 48 h. The results of our qPCR analysis indicated that the expression of immunomodulatory genes, chemotactic genes, and collagen genes in hUC-MSCs was significantly influenced by the duration of cytokine priming. Specifically, the optimal priming times were identified as 24 or 36 h, as these time points had the most pronounced effects on gene expression (Additional file 5: Fig S5A-C). In the present study, only single donor-derived hUC-MSCs were primed with various cytokines, and single-cell bioinformatics data were analysed. Considering the cellular heterogeneity among different individuals, we included qPCR data from 3 different donors (Additional file 6: Fig S6A-C). The results indicated that the expression levels of immunomodulatory genes (IDO1 and PDL1), chemotaxis genes (IL-8 and CXCL1), and collagen genes (COL3A1 and COL8A1) were different among these three donors. However, many genes exhibit increased expression in these three donors following IFN-γ, TNF-α and IL-4 priming. This trend was similar to the findings from the scRNA-seq data shown in Fig. 4D-F. Together, the scRNA-seq data demonstrated that IFN-γ-primed hUC-MSCs possess a strong immunomodulatory ability, TNF-α-primed hUC-MSCs exhibit high chemotaxis, and IL-4-primed hUC-MSCs express elevated levels of collagens.

Fig. 4

Cytokine-primed and unprimed hUC-MSCs exhibited different predictive efficacies. (A-C) Ridge plot showing the chemotaxis score (A), immunomodulation score (B), and collagenic score (C) of hUC-MSCs with or without cytokine priming. The zero line was established as a threshold for discriminating cell potential, and the percentage of high-score cells is also shown. (D) Violin plots showing immunomodulatory-related gene expression after cytokine priming. (E) Violin plots showing chemotaxis-related gene expression after cytokine priming. (F) Violin plots showing collagen-related gene expression after cytokine priming. The Wilcoxon rank sum test was performed for significance; ****P < 0.0001

Changes in the subpopulations of hUC-MSCs primed by various cytokines

Using a graph-based method and visualization through t-distributed stochastic neighbor embedding (tSNE), we found four cell clusters in unprimed hUC-MSCs. Many cells were distributed in cluster 0 and cluster 1, demonstrating that these two cell groups are the main subpopulations of hUC-MSCs, while cluster 2 and cluster 3 were relatively smaller. We also found that, compared with no priming, priming with different cytokines could alter the distribution of MSC subpopulations; for example, cluster 2 in IL-4-primed and IL-15-primed hUC-MSCs (Fig. 5A). Due to the change in cell distribution, different MSC subpopulations were upregulated or downregulated after cytokine priming. Compared to those of unprimed hUC-MSCs, the percentage of cells in cluster 0 was greater after IFN-γ, TNF-α, IL-4, IL-6, IL-15, and IL-17 priming, especially for IL-4-primed and IL-15-primed hUC-MSCs. Cluster 1 included IFN-γ-, TNF-α-, IL-6-, and IL-17-primed hUC-MSCs. The expression of cluster 2 genes was increased largely in IL-4- and IL-15-primed hUC-MSCs. The percentage of cells in cluster 3 was reduced mainly in the cytokine-primed hUC-MSCs compared to the unprimed hUC-MSCs (Fig. 5B). Subpopulation markers were identified, and the top ten DEGs are listed. Notably, cluster 2 expressed high levels of collagen fibril organization genes (COL1A1, COL1A2, and LUM) and wound healing genes (FN1 and SERPINE2) (Fig. 5C). GO enrichment analysis revealed cellular functions associated with clusters 0, 1, 2, and 3 (Fig. 5D-G). RNA/mRNA splicing, regulation of chromosome organization, and regulation of cell cycle processes were enriched in cluster 0 (Fig. 5D). Regulation of the cell cycle process, mitotic nuclear division, and DNA replication were enriched in cluster 1 (Fig. 5E). Extracellular structure organization, extracellular matrix organization, external encapsulating structure organization, and wound healing were enriched in cluster 2 (Fig. 5F). Cytoplasmic translation, ribonucleoprotein complex biogenesis/assembly, and oxidative phosphorylation/stress were enriched in cluster 3 (Fig. 5G).

Fig. 5

Single-cell RNA sequencing analysis revealed functional heterogeneity among the different clusters. (A) Cell type identification in the t-SNE plot of cytokine-primed and unprimed hUC-MSCs. (B) The relative contribution of each cluster was weighed using the number of cells per sample and scaled to 100%. (C) Heatmap of the top 10 DEGs in each cluster. (D) GO enrichment analyses of cluster 0. (E) GO enrichment analyses of cluster (1) (F) GO enrichment analyses of cluster (2) (G) GO enrichment analyses of cluster 3

Changes in various biological functions of the hUC-MSC subpopulations after cytokine priming

In consideration of regulation of cell cycle processes, DNA replication, and epithelial cell proliferation were enriched in the MSC clusters (Fig. 5D-G), we further analysed the proliferation, DNA repair, and cellular senescence scores in the different clusters. The results revealed that cluster 1 displayed an increased score for proliferation and DNA repair (Fig. 6A, B). Correspondingly, cluster 2 exhibited lower proliferation and DNA repair scores but a relatively greater cellular senescence score (Fig. 6C). Tri-lineage differentiation gene score analysis indicated that the hUC-MSCs in cluster 2 had greater chondrogenic potential than did those in cluster 0, cluster 1, and cluster 3 (54.3% vs. 29.2% vs. 31.4% vs. 5.3%). This difference may be related to the greater expression of extracellular matrix-associated genes in cluster 2 (Fig. 6D). In contrast, the adipogenic and osteogenic potentials were similar across Cluster 0, Cluster 1, and Cluster 2 (Fig. 6E, F). We found that, according to the chemotaxis gene score, 48.9% of the cells in cluster 2 exhibited high chemotactic ability, while 28.7% (cluster 0), 32.3% (cluster 1), and 25.7% (cluster 3) of the cells in the other subpopulations exhibited chemotactic ability (Fig. 6G). Cluster 2 had a higher immunomodulatory gene score than did the other subsets, with scores of 28.9% (cluster 0), 37.8% (cluster 1), 48.4% (cluster 2), and 12.1% (cluster 3) (Fig. 6H). Cluster 2 also demonstrated a significantly greater collagenic score than did the other subsets, with scores of 64.0% (cluster 0), 54.0% (cluster 1), 96.1% (cluster 2), and 33.6% (cluster 3) (Fig. 6I). Additionally, various cellular biological functions, including proliferation and senescence, tri-lineage differentiation, chemotaxis, immunomodulation, and collagen synthesis, were further evaluated in different clusters following priming with the six cytokines. (Additional file 7–9: Fig S7-9). First, IL-4- and IL-15-primed hUC-MSCs had significantly lower proliferation and DNA repair scores than other hUC-MSCs did, especially in cluster 0, cluster 2, and cluster 3 (p ≤ 0.0001; Additional file 7: Fig S7A, B), while they had relatively higher cellular senescence scores in cluster 0, cluster 1, and cluster 2 (p ≤ 0.001; Additional file 7: Fig S7C). Moreover, compared to those of other cytokine-primed hUC-MSCs, the chondrogenic score of IL-4- and IL-15-primed hUC-MSCs was greater in cluster 2 (p ≤ 0.0001; Additional file 8: Fig S8A), while in cluster 3, IL-4-hUC-MSCs exhibited strong chondrogenic, adipogenic, and osteogenic potential (p ≤ 0.001; Additional file 8: Fig S8A-C). More importantly, compared to those of other cytokine-primed hUC-MSCs, our findings demonstrated that cluster 0, cluster 1, cluster 2, and cluster 3 all exhibited greater chemotaxis in TNF-α-primed hUC-MSCs (p ≤ 0.0001); moreover, compared with unprimed hUC-MSCs, IL-17-primed hUC-MSCs exhibited an increased chemotaxis score (Additional file 9: Fig S9A). The results demonstrated that the IFN-γ-primed hUC-MSCs exhibited greater immunomodulatory potential than the other groups in all four clusters (p ≤ 0.0001); additionally, we observed an increased immunomodulatory score in cluster 0, cluster 1, and cluster 2 of the TNF-α-primed hUC-MSCs compared to the unprimed hUC-MSCs (Additional file 9: Fig S9B). Furthermore, IL-4-primed hUC-MSCs exhibited a relatively greater degree of collagen synthesis than did the other groups in cluster 0, cluster 1, and cluster 2 (p ≤ 0.001; Additional file 9: Fig S9C). Together, the different subpopulations of hUC-MSCs had different biological functions, and compared with those of cluster 0, cluster 1 and cluster 3, cluster 2 exhibited strong potential for chondrogenic ability, chemotaxis capacity, immunomodulatory potential, and collagen secretion; however, cluster 2 demonstrated less potential for cell proliferation than the other clusters.

Fig. 6

Different clusters had different predicted biological potencies. (A-F) Violin plots showing the proliferation score (A), DNA repair score (B), cellular senescence score (C), chondrogenic score (D), adipogenic score (E), osteogenic score (F), chemotaxis score (G), immunomodulatory score (H), and collagenic score (I) for the four candidate clusters