Unveiling distinctions between mesenchymal stromal cells and stem cells by single-cell transcriptomic analysis

Abstract

Mesenchymal stromal cells (MSCs) and stem cells are distinct types of cells, but they are practically undistinguishable by currently commonly-used identification markers. A single-cell transcriptomic analysis was used to solve this problem. There are eight critical genes involved in self-renewal and differentiation, SOX2, NANOG, POU5F1, SFRP2, DPPA4, SALL4, ZFP42 and MYCN expressed in ESCs, iPSCs and adult stem cells (ASCs), but not in MSCs. There are five functional genes of MSCs, TMEM119, FBLN5, KCNK2, CLDN11 and DKK1, which are not expressed in stem cells. Trajectory analysis displayed clear developmental cliffs from ESCs/iPSCs to ASCs and to MSCs. Adipose-derived MSCs, relative to other types of MSCs, exhibit a more consistent and broader spectrum of gene expression for regulatory and excrete function. This study identifies distinction markers between MSCs and stem cells, providing an alternative approach for quality control of MSCs in their propagation and further mechanistic insights into their action.

Highlights

• •Single-cell RNA-sequencing identifies distinctions between MSCs and stem cells.
• •MSCs do not express any of 8 self-renewal and differentiation genes of stem cells.
• •Stem cells do not express any of herein identified 5 functional genes of MSCs.
• •AD-MSCs are more functionally capable relative to other types of MSCs.

1. Introduction

Mesenchymal stromal cells (MSCs) have become the mainstay of cell therapy research and clinical application [[1], [2], [3], [4]], but their identity and mechanism of action are confounded with that of stem cells [5,6], although MSCs and stem cells are two distinguishable types of cells [5,[7], [8], [9]]. The minimum identification criteria for MSCs were first proposed by the International Society for Cellular Therapy (ISCT) in 2006 [10], and updated in 2019 [6]. According to these criteria, MSCs must (1) be plastic-adherent in cultures; (2) express CD73, CD90, CD105, and lack expression of CD11b or CD14, CD19 or CD79α, CD34, CD45, and HLA-DR; and (3) differentiate to osteoblasts, adipocytes and chondroblast in vitro. However, these criteria do not distinguish MSCs from stem cells. With regard to this, the ISCT has called for: 1) clarification of tissue origin for MSCs; 2) provide rigorous evidence of stemness with in vitro and in vivo validation; 3) using functional matrix assays to assess therapeutic properties like immunomodulation and paracrine functions [6].

However, the current controversy remains in the interchangeable use of MSCs and stem cells [[7], [8], [9],11,12]. The identification of distinct properties of MSCs is a Critical Quality Attribute (CQA) for cell-based products [[13], [14], [15]]. Practically, the currently used confounding criteria for the identification of these cells can lead to an inadvertent mixture of stem cells and MSCs in clinical application, impacting the exploration of therapeutic mechanism, the control of safety, and the effectiveness of MSCs [9].

Poor-quality control, including issues such as immune compatibility, stability, heterogeneity, tumorigenicity and migration ability would affect the outcome of these cells in clinical application [4,16]. An undefined mixture of MSCs and stem cells in clinical application would present a significant concern of all the issues above.

It is thus an urgent and practical need to specify the identification criteria for the distinction between MSCs and stem cells. We thus made an effort to fulfill this purpose by developing a single-cell transcriptome and trajectory analysis procedure and identified fundamental differences between MSCs and stem cells in signal transduction pathways involved in self-renewal and differentiation. This analysis unambiguously discriminated against MSCs from stem cells, or vice versa. In addition, the same principle was extended to identify functional heterogeneity among MSCs obtained from different tissues, including adipose, bone marrow, umbilical cord, and blood, possibly prescribing therapeutic potentials of different MSCs.

2. Materials and Methods

2.1. Human MSCs culture and resuscitation

Human adipose tissues were collected as approved by the Ethics Committee of the Tianjin First Central Hospital, Tianjin, China (2021N020KY). Written informed consent was secured from all donors prior to their participation in this study. There was a total of 42 AD-MSC samples collected from female donors with signed consent.

For the preparation of AD-MSCs, adipose tissue was collected via abdominal liposuction surgery. The collected tissue was washed thoroughly with saline solution, and the floating layer was digested with 0.3 U/mL collagenase (Nordmark, Germany) at 37 °C for 1 h. After digestion, the cell suspension was centrifuged, filtered, and seeded onto T75 culture flasks containing MEM Alpha medium (Gibco, USA) supplemented with 5 % UltraGRO-Advanced (HELIOS, USA) for 5–10 days (P0 cells). Once the monolayer cultures reached 50–70 % confluence, the cells were passaged using 1x TrypLE Express (Gibco, USA) and reseeded at a density of 5000 cells/cm², passaging under the same condition for several passages as indicated. Cells were collected at P2 for stocking and at P5 for experimental usage. For the experimentation, cells were washed thoroughly with saline solution and suspended in saline supplemented with 1 % human albumin preparing cell density of 1 million cells per milliliter.

2.2. Single-cell RNA sequencing

Single-cell RNA-sequencing (scRNA-seq) was performed using the BD Rhapsody platform (BD, USA) in accordance with the manufacturer's protocol. Briefly, cells were randomly distributed in microwells for capture, and oligonucleotide barcode-equipped beads were loaded on a cartridge, allowing for the pairing of one bead with one cell. After the application of a cell-lysis buffer, RNA molecules were hybridized with the beads. The beads were then pooled in a single tube for complementary DNA (cDNA) synthesis via reverse transcription. Each cDNA molecule was tagged at its 5′ end with a unique molecular identifier (UMI) and cell label information. Subsequently, second-strand complementary DNA was generated and ligated with a universal amplification adaptor. A random priming polymerase chain reaction (PCR) was then performed to enrich the 3′ end of the transcripts. Finally, the prepared libraries were sequenced on an Illumina NextSeq platform (Illumina, San Diego, CA, USA) with a 100-bp paired-end run.

2.3. scRNA-seq data collection

The public sample data of AD-MSCs, UC-MSCs, PM-MSCs, and BM-MSCs, three samples each, were collected from the GSE200161 dataset available in the GEO database (https://www.ncbi.nlm.nih.gov/geo) [17]. The public sample data of stem cells were also collected from the GEO database with accession number: ESCs from GSE136447 [18], iPSCs from GSE208625 [19], HSCs from GSE201333 [20], NSCs from GSE136719 [21] and GSE165388 [22], amniotic fluid cells from GSE220994 [23].

The sample data of AD-MSCs collected and subjected in this study was deposited in the Genome Sequence Archive (GSA) database (https://ngdc.cncb.ac.cn/gsa-human/) with accession number HRA006953, HRA006954, HRA006955, HRA003811. Additional details can be found in Table 1.

Table 1.

Details of scRNA-seq data.

Cell type	Total detected cells (n)	No. of samples	Sample source	Data source
ESC	549	1	Human embryos cultured in vitro for 6–14 days (after fertilization)	GSE136447
iPSC	6124	3	iPSC lines from the American Type Culture Collection (cat. no. ACS-1011, healthy newborn male)	GSE208625
HSC	499	2	Blood and bone marrow samples in the Tabula Sapiens benchmark dataset	GSE201333
NSC	30717	2	Subpallium of fetal brain at 9 and 10 weeks after pregnancy	GSE136719
				GSE165388
aAD-MSC	309630	42	The healthy human adipose tissue	HRA006953
				HRA006954
				HRA006955
				HRA003811
AD-MSC	11742	3	The healthy human adipose tissue	GSE200161
UC-MSC	13390	3	The umbilical cord of neonatal tissue after delivery	GSE200161
PM-MSC	10610	3	The placentas of neonatal tissue after the delivery	GSE200161
BM-MSC	15410	3	The iliac crest bone marrow aspiration.	GSE200161
Amniotic fluid cells	34568	11	Amniotic fluid cells from 11 samples	GSE220994

scRNA-seq: single-cell RNA sequencing, ESC: Embryonic Stem Cell, iPSC: Induced Pluripotent Stem Cell, HSC: Hematopoietic Stem Cell, NSC: Neural Stem Cell, AD-MSC: Adipose-Derived MSC, UC-MSC: Umbilical Cord-Derived MSC, PM-MSCs: Placenta-Derived MSCs, BM-MSC: Bone Marrow-Derived MSCs.

^aRepresents samples collected and subjected to scRNA-seq in our study.

2.4. Single-cell RNA-sequencing data analysis

The scRNA-seq data were analyzed using the Seurat (v4.3.0) package [24] in R. The expression matrices were processed using the BD Rhapsody™ cwlref-runner (v3.1.2) pipeline, being specifically designed for BD platform scRNA-seq data (https://www.bdbiosciences.com/content/dam/bdb/marketing-documents/BD_Single_Cell_Genomics_Analysis_Setup_User_Guide_v2.pdf). Subsequent data processing, including quality control, normalization, data scaling, sample integration, dimensionality reduction, and clustering, was carried out using the Seurat package. Genes expressed in more than three cells were retained for analysis. Cells expressing less than 200 genes, or more than 25 % mitochondria content were considered poor-quality data and were excluded from further analysis.

Counts for genes expressed in each cell were normalized using the “NormalizedData” function with a scale factor of 10,000. To identify the variable feature, the “FindVariableFeatures” function was used to extract the top 2000 variable features. Count matrices for cells from each sample were integrated into one large count matrix using the merge function from the Seurat R package. Batch effect correction was performed using the RunHarmony function from the Harmony R package (v0.1.1). Data scaling was performed using the “ScaleData” function, which was also regressed out UMIs. Dimensionality reduction was achieved using principal component analysis (PCA) and uniform manifold approximation and projection (‌UMAP) methods, being implemented through the “RunPCA” and “RunUMAP” functions, respectively. Thirty principal components were selected for downstream analysis. Clustering results were visualized by Uniform Manifold Approximation and Projection (UMAP). Differential expression features were determined using the Wilcoxon test, as implemented in the “FindAllMarkers”.

2.5. Identification of marker genes distinguishing between MSCs and stem cells

The percentage of cells expressing each self-renewal or differentiation gene was calculated for each MSC and stem cell sample. A gene was defined as a negative marker for MSCs if its expression was observed in less than 1 % of cells across all MSC samples and in more than 90 % of cells in ESC or iPSC samples. The expression levels and percentage of cells expressing these marker genes in different samples were visualized using bar charts.

2.6. Pseudo-time trajectory analysis

To investigate the developmental progression of stem cells and MSCs, Pseudo-time analysis was conducted using the R package monocle (v2.21.1) [25,26]. This method arranges individual cells along a continuum of developmental states, simulating dynamic changes during development based on the expression patterns of critical genes. For the pseudo-time analysis of samples of MSCs, adult stem cells and pluripotent stem cells, approximately 80 feature genes related to self-renewal and differentiation were selected as ordering genes. For the pseudo-time analysis of samples from different sources of MSCs, the differentially expressed genes (DEGs) with a q-value of less than 0.01 were used as ordering genes. Dimensionality reduction was performed using “reduceDimension” function with the DDRTree method, and cells were ordered in pseudotime using the “orderCells” function with default parameters.

2.7. Functional enrichment analysis

To explore the functional implications of the differential expression genes, reactome pathway enrichment analysis of differentially expressed genes was performed using the enrichPathway function from the ReactomePA R package (v1.46.0) [27] and the REACTOME pathway database from the R package org.Hs.eg.db (v3.16.0). The top pathways identified through this analysis were selected for further investigation.

2.8. Analysis of MSCs secretome properties

To assess and compare the secretome properties of MSCs from different sources, the Exosome Gene Database from the Human Protein Atlas (HPA) (https://www.proteinatlas.org/) was utilized. The “FindMarker” function of Seurat was employed to identify genes uniquely expressed in MSCs from different tissues, using a cutoff of log₂FC > 0.25 and p-value <0.05. Subsequently, corresponding receptor genes were then identified based on the HPA database and categorized to various functional groups such as immune regulation, promotion of vascularization, anti-apoptosis, and extracellular matrix generation. The frequencies of gene expression associated with each functional category were calculated and visualized using a Circos plot for a comprehensive illustration.

2.9. Statistical analysis

All statistical analyses were conducted using R (v4.2.3). The differences in gene expression between the two cell groups were evaluated using a two-group t-test, with the false discovery rate (FDR) method applied to correct for multiple testing. A p-value of <0.05 was considered indicative of statistical significance.

2.10. Flow cytometry

The FBLN5 expression of AD-MSCs was assessed using flow cytometry. Fixation and permeabilization of AD-MSCs were carried out for 20 min at 4 °C using the Cytofix/Cytoperm™ Fixation/Permeabilization Kit (BD Biosciences, Cat# 554714). The cells were then washed and incubated with mouse anti-human monoclonal primary antibodies against FBLN5 (Invitrogen, Cat# MA5-42544; 1:100 dilution) for 1 h at 4 °C, followed by incubation with FITC-conjugated goat anti-mouse secondary antibodies (Abcam, Cat# ab150113; 1:2000 dilution) for 30 min at 22 °C in the dark. Mouse IgG1 isotype control antibodies (Abcam, Cat# ab81216) were used as negative controls. The cells were assessed using a NovoCyte flow cytometer (Agilent, USA) with NovoExpress software (Agilent, USA).

3. Results

3.1. Distinction between MSCs and stem cells revealed by single-cell transcriptome

We integrated the scRNA-seq data from all datasets and samples included in the study and found that pluripotent stem cells (PSCs), adult stem cells (ASCs), and MSCs exhibited distinct clustering through UMAP analysis (Additional file 1: Fig. S1A). We then examined potential differences in the profiles of single-cell transcriptomes among PSCs, ASCs, and MSCs (Fig. 1A). The functional enrichment analysis of gene expression revealed that the genes highly expressed in PSCs are mainly involved in self-renewal and proliferation, whereas the highly expressed genes in ASCs are shifted to differentiation and metabolic processes. The genes highly expressed in MSCs are involved in cellular metabolism and interactions, lacking those involved in self-renewal or differentiation (Fig. 1B).

Fig. 1.

Single-cell transcriptomes of MSCs and various stem cells.

(A) A heatmap of single cell gene expression of pluripotent stem cells (PSCs), adult stem cells (ASCs), and MSCs. Each row represents a gene, and each column represents a single cell. Different types of cell samples are separated by black borders, while samples of the same cell type are separated by gray lines. PSCs include embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs). ASCs include hematopoietic stem cells (HSCs) and neural stem cells (NSCs). MSCs include adipose-derived (AD-MSCs), umbilical cord (UC-MSCs), placenta (PM-MSCs) and bone marrow (BM-MSCs). These cells displayed distinct spectra of gene expression, as further analyzed in Fig. 1B. (B) Enriched functional pathways based on the genes highly expressed in PSCs, ASCs and MSCs. The length of horizontal bars represents the number of highly expressed genes in the pathway, and the color of the bars represents the significance of enrichment, ranging from blue to red indicating increased significance. The FDR method is used to adjust p-values. (C) Cell fate trajectory analysis for the lineage development of PSCs, ASCs and MSCs based on the expression profiles of self-renewal and differentiation genes. The dotted arrow indicates the pseudotime developmental trajectory from PSCs to MSCs.

In order to define the developmental states and interlinking between different types of cells, we conducted a pseudo-temporal developmental trajectory analysis of the cells based on the expression of self-renewal/differentiation genes (Additional file 1: Table S1). The scRNA-seq analysis revealed that the cells are distinctly clustered according to their development stages (Fig. 1C, Additional file 1: Fig. S1b). The expression of self-renewal/differentiation genes exhibited early, middle, and late phases of temporal trajectories respectively relevant to PSCs, ASCs, and MSCs. MSCs represent a cell type in the late stage of development, suggesting that the functional enrichment of self-renewal and differentiation genes would be a critical feature distinguishing between stem cells and MSCs.

3.2. Null expression of critical self-renewal and differentiation genes in MSCs

To identify critical genes involved in self-renewal and differentiation, we constructed a core interaction map of self-renewal and differentiation based on public data and identified critical proteins involved in interaction. The genes encoding these proteins are considered critical genes for self-renewal and differentiation. There were eight critical genes identified, including SOX2, NANOG, and POU5F1 that are core genes maintaining the self-renewal capacity [28], SFRP2, an important signaling molecule in self-renewal [29,30], and DPPA4, SALL4 and ZFP42 involved in downstream transcriptional regulation of self-renewal [[31], [32], [33], [34], [35]], and MYCN, a critical driver gene for ectodermal differentiation (Fig. 2A) [36].

Fig. 2.

Characteristics of self-renewal and differentiation in MSCs and stem cells.

(A) Interaction map of critical self-renewal and differentiation (SD) genes. Three critical SD genes (SOX2, NANOG, and POU5F1) maintaining self-renewal are highlighted in blue, while four signaling and effector genes maintaining self-renewal, and one differentiation-driving gene (MYCN) are highlighted in purple. The functional locations include the cell membrane, cytoplasm, and nucleus. (B) Percentage of positive cells for important self-renewal and differentiation (SD) genes in PSC, ASC, and MSC samples. Important SD genes are defined as SD genes expressed in over 90 % of ESC or iPSC cells. The white dashed line separates important SD genes with negative expression in MSCs (top, positive cell rate <1 %) from those with positive expression in MSCs (bottom, positive cell rate ≥1 %). (C) Percentage of cells expressing eight critical SD genes in MSCs and stem cells. The height of the bars represents the proportion of positive cells in a population, with the 95 % confidence intervals marked on the bar charts. (D) Percentage of cells expressing one or more critical SD genes from panel (B) in different types of MSC samples. There are three samples of each type, and the numbers on the histogram represent the mean percentage of the three samples.

We then focused on differentiating MSCs from stem cells based on the expression of these eight critical genes. MSCs obtained from four different tissues were included: AD-MSCs, UC-MSCs, PM-MSCs and BM-MSCs. Stem cells include four cell types: ESCs, iPSCs, HSCs and NSCs. It was found that all of the MSCs lack the expression of any of the eight genes (Fig. 2B). In contrast, all of the stem cells express these genes but with different intensities. ESCs and iPSCs express all eight genes, and other stem cells express two or three of these genes (Fig. 2B).

We made a further distinction among these cells in their expression of the eight critical genes by comparing the expression intensity at a single gene level (Fig. 2C). This analysis showed that most of ESCs/iPSCs extensively express all of the eight genes. There was a high percentage of cells expressing SOX2 and very small percentage of cells expressing MYCN and SFRP2 in NSCs. There was a moderate percentage of HSCs that express MYCN and DPPA4 (Fig. 2C). There was very small percentage of MSCs that express any of the eight genes (Fig. 2C). To further precisely quantify the percentage of different sources of MSCs that express the eight genes, we analyzed the samples of different MSCs and found that there were only 1.55 %, 0.16 %, 0.48 % and 0.23 % of, respectively, BM-MSCs, UC-MSCs, PM-MSCs and AD-MSCs that express one or more of these genes (Fig. 2D).

3.3. Functional genes of MSCs not expressed in stem cells

The single-cell transcriptome and trajectory analysis above indicated that in addition to the loss of expression of the eight critical genes involved in self-renewal and differentiation, MSCs gain an enhanced expression of unique functional genes. We then made a comparation of the profiles of gene expressions among different sources of MSCs and found there were different patterns of gene expressions in different types of MSCs (Fig. 3A). We identified, through a comparison between MSCs and stem cells, five common critical functional genes of MSCs. As shown in Fig. 3B, these functional genes include TMEM119, FBLN5, KCNK2, CLDN11 and DKK1 that were expressed in all types of MSCs, but not expressed in stem cells (Fig. 3B, Additional file 1: Table S2). We also found that there were a large set of genes that were expressed in all of the MSCs (Fig. 3C), and a small set of genes that were expressed only in unique type of MSCs, especially in AD-MSCs (Fig. 3C), suggesting functional differences among different types of MSCs.

Fig. 3.

Functional gene analysis in MSCs at single-cell level.

(A) A heatmap showing the percentage expression of MSC functional genes in MSC and stem cell samples Functional genes are defined as those with an expression percentage exceeding 90 % in at least one type of MSC cell. Each row represents a functional gene (1329 in total), and each column represents a sample. The color of the grid indicates the percentage of that gene expressed in the cells of that sample, with the color gradient ranging from blue to red corresponding to expression percentages from 0 % to 100 %. (B) The single-cell gene expression computation displays percentage of positive cells (circle size) and expression level (intensity of red) of 5 MSC-specific functional genes in different types of MSCs and stem cells. (C) Distribution of percentage of cells expressing common and unique functional genes across different types of MSC cells. Each point on the graph represents the percentage of a functional gene expressed in a particular MSC sample. The color of the dots indicates the sample group. Functional genes are categorized on the X-axis according to their Venn diagram groups and arranged in descending order based on the number of genes expressed. The numbers between the arrows indicate the number of functional genes within each category. The “Others” category includes the following groups and the number of genes in each: AD (43), PM (42), AD ∩ PM (14), AD ∩ BM (13), AD ∩ UC (11), BM (10), AD ∩ BM ∩ PM (9), BM ∩ PM ∩ UC (7), AD ∩ BM ∩ UC (2), BM ∩ PM (1), and BM ∩ UC (1).

To validate the gene markers identified by the scRNA-seq analysis for identification of MSCs, we performed a flow cytometry analysis for the expression of FBLN5 protein in AD-MSC samples. FBLN5 is an important secreted protein that plays a crucial role in vascular remodeling and neointima formation [37]. The results showed that 90.32 % of the AD-MSCs expressed FBLN5, consisting with the result observed in the scRNA-seq analysis (Fig. 3B, Additional file 1: Fig. S2).

3.4. Distinct single-cell transcriptomes in MSCs from different tissues

We performed a cluster analysis of MSCs from four different tissues and found that MSCs from each different tissue exhibited a distinct clustering (Additional file 1: Fig. S3A), indicating that MSCs from different tissues would perform distinguishable function. Then, we conducted a single-cell transcriptomic analysis of different MSCs, revealing clear differences in the expression of functional genes in MSCs from different tissues (Fig. 4A). Genes highly expressed in AD-MSCs are related to “chemical signal reception” and “ECM formation”, while those in UC-MSCs and PM-MSCs are enriched in “chemokine perception” and “ROBO-mediated cell adhesion”. Gene expression in BM-MSCs tends to be associated with “PDGF-induced cell movement” and “IGF transport and uptake regulated by IGFBP” (Fig. 4B). To further explore the distinction among different MSCs, we conducted a cell fate trajectory analysis for the lineage development of different MSCs against ESCs. It was found that PM-MSCs and UC-MSCs are located in the same cell fate cluster. BM-MSCs are located in a different cell fate cluster but with a close proximity to PM-/UC-MSCs. However, AD-MSCs are distantly clustered from the other three types of MSCs. Regardless of the intrinsic differences among MSCs, all of the four types of MSCs exhibit a clear developmental cliff from ESCs along the cell fate trajectory (Fig. 4C).

Fig. 4.

Single-cell transcriptome analyses of different MSCs.

(A) The heatmap of gene expression of MSCs from different tissues (AD-, UC-, PM-, and BM-) displaying a single-cell transcriptomic heterogeneity. (B) Enriched functional pathways based on the genes expressed in different MSCs. The length of horizontal bars represents the number of highly expressed genes in the pathway, and the color of the bars represents the significance of enrichment, ranging from blue to red indicating increased significance. The FDR method is used to adjust p-values. (C) Relative positioning diagram of 4 different MSCs along the lineage development against ESCs. Each point represents a cell. Black circles with numbers represent the turning points of cell fate decision. The dotted arrow indicates the pseudotime developmental trajectory of ESCs and various MSCs. (D) Enriched pathways based on the expression of common exocrine genes in different MSCs. The length of horizontal bars represents the number of highly expressed genes in the pathway, and the color of the bars represents the significance of enrichment, ranging from blue to red indicating increased significance. The FDR method is used to adjust p-values. (E) A heatmap of significantly highly expressed exocrine genes in different tissues of MSCs. Each row represents a gene, and each column represents a single cell. The color of the grid indicates the expression of that gene in the cell, with the color gradient ranging from pink to red indicates the intensity of gene expression from low to high. (F) A Circos plot showing the number and composition of functional pathways activated by highly expressed exocrine proteins in different MSCs. The outer circle shows 4 types of MSCs, and the inner circle shows four therapeutic effects (each containing multiple functional pathways). The right half of the outer circle shows the composition of exocrine proteins activated relative to the therapeutic effects of MSCs. The left half of the inner circle shows the composition of different tissues of MSCs involved in the therapeutic effect.

The therapeutic effects of MSCs are believed to be highly associated with secrete bioactive factors [38]. We thus analyzed the differences in the expression of exocrine genes (Additional file 2: Table S1) to explore differential potentials of therapeutic application among different types of MSCs. We found that MSCs from different tissues collectively expressed 455 types of secrete factors involved in the interaction with 217 cell membrane receptors, playing roles in ECM formation and regulation, and reshaping the cellular microenvironment, which are a source-free function of MSCs (Fig. 4D).

There was a consistence in the expression of these secrete factors among samples obtained from AD-MSCs and some slight variation in those from UC-MSCs and PM-MSCs. However, the expression of these genes in different samples of BM-MSCs varied distinctly (Fig. 4E). Overall, there were different patterns in the expression of these genes among different types of MSCs (Fig. 4E, Additional file 2: Table S1). These differences were further analyzed by a Circos plot [39]. The exosome proteins that can interact with cell surface receptors were identified, and the Circos plot was constructed displaying the number of exocrine genes associated with a specific function in different MSCs. It was found (Fig. 4F, Additional file 1: Fig. S2b) that AD-MSCs are most active in the expression of genes involved in vascularization (208), immune regulation (211) and ECM remodeling (171), and less active in anti-apoptosis (29). BM-MSCs are moderate active in the expression of genes involved in vascularization (60), immune regulation (65) and ECM remodeling (56), but do not express genes involved in anti-apoptosis. PM-MSCs express all of the above-mentioned genes with much less intensity relative to AD-MSCs. UC-MSCs weakly express genes involved in vascularization (22), immune regulation (42) and anti-apoptosis (6), but do not express genes involved in ECM remodeling.

3.5. Utilization of the developed criteria for analysis of mixed cell samples

We made an effort to apply the above developed criteria for analysis of mixed cell samples. In this context, we analyzed MSCs and stem cells in amniotic fluid samples (n = 11). We found high levels of expression for critical stem cell marker genes and low levels of expression for MSC functional genes in amniotic fluid (Fig. 5A). Further analysis revealed that the proportion of stem cells was 13.66 % (range: 10.45–21.82 %), while the proportion of MSCs was only 1.52 % (range: 0.06–4.49 %) in amniotic fluid samples (Fig. 5B).

Fig. 5.

Content of MSC and stem cells and marker gene expression in amniotic fluid samples (n = 11).

(A) Heat map showing the proportion of cells expressing key stem cell genes and MSC functional genes in amniotic fluid samples. The color in each grid represents the proportion of cells expressing a certain gene (vertical axis) in a certain sample (horizontal axis). From blue to red, the cell proportion ranges from 0 to 100 %. (B) Percentage of cells expressing one or more functional MSC genes (red histogram) and critical stem cell genes (blue histogram) from panel (A) in 11 amniotic fluid samples (GSE220994).

4. Discussion

MSCs have been the focus of extensively fundamental and clinical research for cell-based therapy, however, the clinical application of these cells has progressed much slower than its anticipation [40,41]. There were only 12 MSC products that have been approved for clinical application globally [41]. and the US-FDA only very recently approved one MSC product for clinical use [42]. The FDA has disclosed adverse safety events associated with stem cell-based therapy, including unexpected cell changes and the growth of tumors [43].

MSCs have been considered as stem cells due to their multidirectional differentiation potential in vitro [10,44,45]. Recent studies from a large amount of therapeutic data indicated that the fundamental mechanism of action of MSCs is not implicated by their differentiation rather is indicated by their secretory and regulatory actions [8,9], clearly distinguishing from the mechanism of action for stem cells in therapeutic application [6]. However, what are unique criteria for the identification of MSCs, or how to unambiguously distinguish MSCs from stem cells? This is a critical question needed to be addressed in order to promote the clinical translation of MSC's therapeutics.

Specifically, single-cell transcriptomic and functional gene enrichment analysis helps identify fundamental biologic signatures of different cells. This analysis used in the present study thus revealed fundamental differences in the expression of genes involved in self-renewal and differentiation between MSCs and stem cells. We identified eight critical genes involved in self-renewal and differentiation that are expressed in stem cells, lacking the expression in MSCs. From developmental trajectory analysis, we found that there is a clear developmental cliff between MSCs and stem cells, indicating that the lineage commitment of MSCs silences their expression of these genes. Therefore, the lack of expression of the collective eight genes involved in self-renewal and differentiation would help distinguish MSCs from stem cells.

The lineage commitment of MSCs makes these cells lose the capability of self-renewal and differentiation, but gain other biological functions. From the pathway enrichment analysis, we found that MSCs express a unique set of genes associated with cellular metabolism and secrete regulation. Among these genes, we identified five MSC-specific functional genes that are not expressed in stem cells. These unique functional genes would then serve as “positive” markers for MSCs, although not all of MSCs express the genes with the same intensity. To validate these MSC identification markers, we performed a flow cytometry analysis for one critical gene marker, FBLN5, an important secreted protein that plays a crucial role in vascular remodeling and neointima formation [37]. There were 90.32 % of AD-MSCs expressing FBLN5 protein, partially validating the gene markers for identification of MSCs. We also found public data that show immunostaining of TMEM119, FBLN5, and CLDN11 proteins in adipose tissue in Human Protein Atlas Database (https://www.proteinatlas.org), consisting with the gene markers identified for MSCs.

MSCs that are currently available for experimental and clinical studies include AD-MSCs [[46], [47], [48], [49]], BM-MSCs [44,50,51], UC-MSCs [52], and PM-MSCs [53]. These different types of MSCs share the same feature of lacking expression of eight critical genes involved in self-renewal genes and differentiation, but differentially express the MSC-specific functional genes.

MSCs obtained from different tissues would have varied therapeutic functions. As revealed by the analysis of the enriched functional pathways, AD-MSCs would have more versatile therapeutic functions relative to other MSCs, as indicated by the expression of more genes involved in multiple functional pathways. In addition, trajectory analysis indicates that AD-MSCs have a farther developmental distance from ESCs relative to MSCs from other tissues. This would suggest that AD-MSCs would have less concerns for tumorigenesis risk derived from proliferation and differentiation. It was also found that the profiles of functional gene expression were consistent among samples obtained from AD-MSCs, but those from other types of MSCs were less consistent. With regards to this aspect, the pattern of gene expression displayed more heterogenic among the samples obtained from BM-MSCs.

The differences revealed here by scRNA-seq and enriched functional pathway analyses would not only help distinguish MSCs from stem cells, but also provide further insights into mechanisms of action for these cells in therapeutic application. The self-renewal and differentiation capacity of stem cells, such as iPSCs, would help these cells repair the injured tissues by regeneration through replacement of the damaged cells. In contrast, MSCs would help repair the injured tissues by promoting rejuvenation of the existing cells, as reported in most recent studies [[54], [55], [56]].

In the present study, we made an effort to apply the developed criteria for analysis of mixed cell samples. In a recent study, De Coppi et al. performed an outstanding single-cell transcriptomic analysis of amniotic fluid, identifying epithelial stem/progenitor cells and culturing them into organoids, highlighting the potential of amniotic fluid-derived organoids for modeling congenital diseases and exploring prenatal therapies [23]. In this context, we analyzed the cell composition of amniotic fluid samples, focusing on the distinction between MSCs and stem cells. As shown in the results, the analysis revealed a higher proportion of stem cells and a lower proportion of MSCs in the amniotic fluid samples. This application underscores the significance of the criteria developed in this study for the practical analysis of cell products in clinical applications.

Limitations of the study:There are some limitations in the present study. First, there are very small percentage (less than 1.76 % of the overall population) of MSCs that express one or more of the eight critical genes involved in self-renewal or differentiation. What is the nature of these small percentage of cells? Are they adult stem cells that are mixed with MSCs and simultaneously isolated during the propagation process? Or, these are originally MSCs, but activated to express these genes during propagation in cultures. It is a rather difficult undertaking in the future. Second, the single cell transcriptomic analysis revealed that AD-MSCs may have a better functional capability in some therapeutic applications than MSCs obtained from other tissues, but it is unknown whether there is a difference in homing ability between AD-MSCs and other types of MSCs. It is important to note that the therapeutic efficacy of MSCs is closely related to their homing ability. Therefore, this issue needs to be addressed in the future.

5. Conclusions

In this study, single-cell transcriptome and trajectory analyses revealed a clear developmental cliff between MSCs and stem cells. Based on this distinction between MSCs and stem cells, we propose using 8 hub self-renewal genes SOX2, NANOG, POU5F1, SFRP2, DPPA4, SALL4, ZFP42 and MYCN as “negative” markers, and 5 functional genes of MSCs, TMEM119, FBLN5, KCNK2, CLDN11 and DKK1 as “positive” markers to distinction of MSCs from stem cells. AD-MSCs, relative to MSCs from other tissues, would have more therapeutic potentials and less safety concerns.

Corresponding author

Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding author.

CRediT authorship contribution statement

Kaijing Yan: Writing – original draft, Project administration, Conceptualization. Fei Ma: Writing – original draft, Methodology, Formal analysis, Data curation. Xiaoxi Song: Software. Huizhen Wang: Software. Pengchong Liu: Writing – original draft, Software. Jinlai Zhang: Resources. Xin Jin: Investigation, Conceptualization. Pengfei Han: Conceptualization. Xiao Zuo: Conceptualization. Y. James Kang: Writing – review & editing, Supervision, Project administration, Conceptualization.

Materials availability

AD-MSCs generated in this study will be made available on request, but may require a completed Materials Transfer Agreement.

Data and code availability

Sources associated with sequencing data are referenced in the Materials and Methods and the data details are listed in Table 1. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Fei Ma, Xiaoxi Song, Huizhen Wang, Pengchong Liu, Jinlai Zhang, Xin Jin, Pengfei Han, Xiao Zuo reports financial support, administrative support, article publishing charges, equipment, drugs, or supplies, statistical analysis, travel, and writing assistance were provided by Tasly Pharmaceutical Company Ltd. Y. James Kang has patent #202410395707.0 pending to Tasly Stem Cell Biology Laboratory, Tasly Group. Y. James Kang has patent #202411513300.X pending to Tasly Stem Cell Biology Laboratory, Tasly Group. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the supplementary data to this article: