Single-cell transcriptome analysis of uncultured human umbilical cord mesenchymal stem cells

Umbilical cord mesenchymal stem cells (UC-MSCs) have certain advantages over other MSCs and about 300 clinical trials have been registered using UC-MSCs to treat diseases such as osteoarthritis, autoimmune diseases, and degenerative disorders, yet, only limited success has been achieved. One reason is that in vitro expanded UC-MSCs show tremendous heterogeneity and their relationship to in vivo UC-MSCs remains unknown. Here, we investigated freshly isolated, uncultured UC-MSCs by single-cell RNA sequencing (scRNA-seq) and found two populations of UC-MSCs. Although UC-MSCs share many expressed genes and may have the same origin, they can be clearly separated based on differentially expressed genes including CD73 and other markers. Moreover, group 1 MSCs are enriched in expression of genes in immune response/regulatory activities, muscle cell proliferation and differentiation, stemness, and oxidative stress while group 2 MSCs are enriched with gene expression in extracellular matrix production, osteoblast and chondrocytes differentiation, and bone and cartilage growth. These findings suggest that UC-MSCs should be separated right after isolation and individually expanded in vitro to treat different diseases.

cycle-regulating proteins. In this study, we analyzed MSCs freshly isolated from Warton's jelly of human umbilical cord and compared them to cultured UC-MSCs. We used a widely-used protocol to isolate human UC-MSCs (after removing blood vessels). We found that UC cells isolated with this protocol could adhere to plastic petri dishes and proliferate for at least 4 passages. Flow cytometry analysis showed that the cultured cells were negative for CD34, CD45, CD11b, and HLA-DR but positive for CD73, CD90, CD105, and CD44 (Supplementary Figure S1a). They also expressed very low levels of CD146 and CD200 (Supplementary Figure S1a). They could differentiate into osteoblasts, chondrocytes, or adipocytes in vitro (Supplementary Figure S1b). Thus, these UC-MSCs meet the criteria of MSCs. We then carried out scRNA-seq on freshly isolated UC-MSCs. A total of 5330 cells were sequenced at the depth of 3800 genes per cell (Fig. 1a). t-SNE analysis revealed two populations of epithelial cells and two populations of MSCs with the former expressing epithelial cell signature genes, e.g., EPCAM, KRT13, KRT14, and KRT17, and the latter expressing mesenchymal signature genes, e.g., PDGFRA, COL1A1, COL1A2, and COL3A1 ( Fig. 1b-d).
The two epithelial subpopulations have different features with one group expressing ECM genes and the other expressing development-related genes (Supplementary Figure S2A and B), suggesting that group 1 may represent epithelial progenitors while group 2 may represent the amniotic or cord-lining epithelia. Interestingly, both groups express CD29 and CD44, which are believed to be stem/progenitor cell markers, consistent with the primitive nature of cells of embryonic tissues. We then focused on the MSCs. The gene expression profiles of the two MSC subpopulations were similar ( Fig. 1b and d), and both groups expressed PDGFRA, VIM, COL1A2, and ACTA2 ( Fig. 1d and supplementary Figure S3), suggesting that the two MSC groups might have the same origin. However, 176 genes were differentially expressed ( Fig. 1e and supplementary Table S2), suggesting that they may have distinct functions.
Examination of cell surface marker gene expression revealed that neither group expressed CD31, CD34, CD45, or CD11b ( Fig. 1d and supplementary Figure S4). They expressed low levels of CD73, CD90, and CD105, the common MSC markers, although CD73 was mainly expressed in group 1 MSCs ( Fig. 1d and supplementary Figure S4). Moreover, they expressed low levels of CD200 but not CD106 or CD146 ( Fig. 1d and supplementary Figure S4). Thus, the surface marker expression pattern of uncultured UC-MSCs is different from cultured UC-MSCs.
MSCs in the bone marrow are believed to be skeletal stem cells, which can be marked by PRRX1, TWIST2, LEPR, GREMLIN1, GLI1, PTHRP, and/or CTSK [8]. We analyzed the expression of these markers and found that both UC-MSC groups expressed TWIST2 (and TWIST1) but not PRRX1, GLI1, GREMLIN 1, or LEPR, and a portion of group 2 cells expressed PRRX1, PTHRP, or CTSK ( Fig. 1d and supplementary Figure S5a). In addition, there is evidence that MSCs are pericytes although later studies produced conflicting results [9,10]. Our scRNAseq data showed that UC-MSCs expressed some of the pericyte markers including DESMIN, CD13, and CD248 but not NG2, ANG1/2, or RGS5 (Supplementary Figure  S5b), suggesting that UC-MSCs are not typical pericytes. The lack of pericytes can be explained by removal of the blood vessels and associated cells during MSC isolation.
KEGG pathway analysis revealed that group 1 UC-MSCs were enriched with TNFα, IL17, TLR, TGFβ, infection, NOD, NF-κB, and PGE pathways, many of which are immune-related ( Fig. 1f and Supplementary Figure S6a-c). These pathways drive the expression of chemokines and immunomodulatory including PGE2, suggesting that group 1 MSCs may play a role in immune response and/or regulation. In addition, group 1 cells were enriched in the expression of genes in controlling pluripotency (Fig. 1f), whereas group 2 cells were enriched in the expression of genes in protein metabolism, extracellular matrix, and glucose and amino acid metabolism pathways (Fig. 1f). These results suggest that the two UC-MSC groups may have different functions.
Gene Ontology enrichment analysis revealed that group 1 MSCs expressed genes in biological activities including inflammation, muscle proliferation, cell differentiation, and oxidative stress response while group 2 cells expressed genes enriched in ECM synthesis, bone and cartilage growth, and glucose metabolism (Fig. 1g), confirming that the two UC-MSC subpopulations may have different functions.
This study shows that the standard UC-MSC isolation protocol also yields epithelial cells, which can be removed by FACS sorting based on their cell surface markers (CD24 + CD44 + PTHRP + ) (Supplementary Figures  S4-S5). More importantly, we find that UC-MSCs can be divided into two subpopulations based on differentially expressed genes especially cell surface markers such as CD73. Group 1 MSCs have features that are reminiscent of the therapeutic activities of MSCs including immunomodulation, pro-survival, and differentiation potentials, whereas group 2 MSCs have features suggesting that they are at a more differentiated state than group 1 MSCs and may be useful for repairing degenerated cartilage.
While this study identifies two subpopulations in freshly isolated UC-MSCs, a recent study shows that cultured human UC-MSCs can be divided into 11 groups. KEGG and GO analyses show that none of these 11 subpopulations expresses immunomodulatory genes [7].  These findings, taken together, suggest that freshly isolated UC-MSCs give rise to more subpopulations during in vitro expansion, and moreover, these cells may lose their original gene expression patterns and activities. It will be interesting to compare the therapeutic effects of the two uncultured UC-MSC subpopulations against cultured cells on osteoarthritis and autoimmune disorders. Moreover, right culture conditions are needed to maintain pluripotency and the major features of the UC-MSC subpopulations for future cell-based therapy.