Human Wharton's Jelly Mesenchymal Stem Cells Show Unique Gene Expression Compared with Bone Marrow Mesenchymal Stem Cells Using Single-Cell RNA-Sequencing

Abstract

Human Wharton's jelly stem cells (hWJSCs) isolated from the human umbilical cord are a unique population of mesenchymal stem cells (MSCs) with significant clinical utility. Their broad differentiation potential, high rate of proliferation, ready availability from discarded cords, and prolonged maintenance of stemness properties in culture make them an attractive alternative source of MSCs with therapeutic value compared with human bone marrow MSCs (hBMMSCs). We aimed to characterize the differences in gene expression profiles between these two stem cell types using single-cell RNA sequencing (scRNA-Seq) to determine which pathways are involved in conferring hWJSCs with their unique properties. We identified 436 significantly differentially expressed genes between the two cell types, playing roles in processes, including immunomodulation, angiogenesis, wound healing, apoptosis, antitumor activity, and chemotaxis. Expression of immune molecules is particularly high in hWJSCs compared with hBMMSCs. These differences in gene expression may help to explain many of the advantages that hWJSCs have over hBMMSCs for clinical application. Although cell surface protein marker expression indicates that isolated hWJSCs and hBMMSCs are both homogenous populations, using scRNA-Seq we can clearly identify extreme variability in expression levels between individual cells within a certain cell type. If the cells are examined as bulk populations, it is not possible to appreciate that a single cell may be making a major unique contribution to the apparent overall expression level. We demonstrated how the fine tuning of expression within hWJSCs and hBMMSCs may be achieved by expression of molecules with opposing function between two cells. We hypothesize that a greater understanding of these differences in gene expression between the two cell types may aid in the development of new therapies using hWJSCs.

Introduction

Traditionally, mesenchymal stem cells (MSCs) are derived from various sources such as bone marrow, fetal and adult organs, and peripheral blood. These sources have several limitations, including painful invasive harvest, low cell number, diminishing stem cell behavior with patient age, short-lived stemness in vitro, and ethical sensitivities [1,2]. Recently, primitive MSCs from various compartments of the human umbilical cord (amnion, subamnion, perivascular region, and Wharton's jelly) have been characterized [3]. Cells from the gelatinous Wharton's jelly compartment (human Wharton's jelly stem cells [hWJSCs]) offer the best potential for clinical application compared with MSCs from the other compartments for several reasons: they (1) show the richest stemness characteristics; (2) are a homogeneous, defined cell population with minimal contamination of other cell types; (3) can be generated in large numbers with minimal expansion avoiding changes in phenotype, even following freezing; (4) their derivation is quick and easy to standardize; and (5) they have a broad differentiation potential [3 –5]. In addition, hWJSCs are hypoimmunogenic with high tolerance in donor settings [5,6], are safe, and do not induce tumor formation in mice [7], nonhuman primates [8], or in humans [9,10]. It has been demonstrated that human bone marrow mesenchymal stem cells (hBMMSCs) can transform into tumor-associated fibroblasts (TAFs) in the vicinity of breast and ovarian cancer cells, enhancing tumor growth and metastasis [11,12], whereas hWJSCs do not form TAFs [13]. Thus, with these advantages, hWJSCs appear to be an ideal substitute for other sources of MSCs and an adjunct to cord blood haematopoietic stem cell (HSC) storage in cord blood banks for cell-based therapies and regenerative medicine studies.

Clinical trials using hWJSCs to treat a variety of diseases have shown early success [9,14,15]. One area of interest for hWJSCs is their use in wound healing [16]. A nanoscaffold impregnated with hWJSCs was shown to be able to stimulate wound healing, most likely through paracrine signaling [17]. Interestingly, a number of publications reported that hWJSCs also have unique anticancer properties [18 –29], which may be related to the high expression of tumor suppressor and proapoptotic genes observed in the transcriptome profiles of hWJSCs compared with other stem cell types [30,31]. hWJSC extracts [conditioned medium (hWJSC-CM) and lysate (hWJSC-CL)] inhibited the growth of breast and ovarian adenocarcinomas, osteosarcomas, and lymphoma cells in vitro and halted the growth of breast tumors in xenograft animal models [7,18,21,27,32 –34]. Since hWJSC-CM and hWJSC-CL can mediate the tumoricidal effects of hWJSCs, it is probable that the actions of these cells are, in part, due to the action of secreted molecules. The broad repertoire of cytokines expressed by MSCs, known as the MSC secretome, varies between different MSC types. A previous comparison of hWJSCs and hBMMSCs found that hWJSCs had a better capacity to induce neural differentiation, protect against ischemia, and induce angiogenesis in endothelial cells [35]; these effects were postulated to be due to differences in the secretome.

To understand what confers unique properties on hWJSCs, it is important to thoroughly characterize their gene expression. A comparison of gene expression profiles in early passaged populations of hWJSCs with human embryonic stem cells (hESCs), human embryonal carcinoma cells, and nonstem stromal cells (skin fibroblasts) using DNA microarray showed that there was no overlap in the profile of highly expressed genes [30]. A meta-analysis suggested that hWJSCs showed upregulation of 14 cytokines, including IL8, IL6, IL1B, and IL2A, and 13 cell surface markers, compared with just 2 cytokines and 1 cell surface marker in hBMMSCs [30]. The major drawback to using microarrays is that coverage is dependent on the probe sets available for hybridization on the microarray. Genome-wide “tiling” microarrays have been widely used, but due to spacing constraints are still unable to cover the entire transcriptome; in addition, data produced can be “noisy” since reliant on a fluorescence signal. RNA sequencing (RNA-Seq), on the contrary, provides a less biased readout of complementary DNA (cDNA) sequence generated from an RNA sample, and it is not necessary to have prior knowledge of the transcripts to be profiled.

Traditionally, alterations in gene expression in tissues or organs in response to a stimulus have been examined in terms of very subtle fold changes, using population-level observations on bulk material. However, although cells are arranged into tissues or organs, each cell may behave in a unique manner, and even within a purified population of cells there exist large differences in expression. Transcription within cells is a stochastic process, occurring in “bursts” [36], with differences due to random transcriptional noise, differences in genetic and epigenetic properties, and stage of cell cycle, as well as external factors such as cell-to-cell contact or extracellular signaling molecules. Some cells may be particularly sensitive to a given stimulus, while others remain unresponsive. Bulk analysis of cells from a particular system may mask the unique characteristics of individual cells. Development of methods to analyze individual cells leads to a much more accurate representation of cell-to-cell variation. It has become technically feasible to sequence the RNA from single cells, using a process known as single-cell RNA sequencing (scRNA-Seq), enabling elucidation of cell-specific transcriptional programs. Despite the fact that, as mentioned previously, transcription occurs in bursts, and there is a potential for technical noise, profiling of sufficient numbers of cells will allow us to overcome these issues using bioinformatic techniques. scRNA-Seq has been used to characterize the transcriptomes of individual cells in many systems, including the pancreas, brain, lung, and dendritic cells [37 –40]. Cellular heterogeneity is an intrinsic feature of many cellular processes, including cell division, apoptosis, and the induction of pluripotent stem cells.

Insights into the biology of single hWJSCs may have important implications for a better understanding of their functions and unique characteristics. In this study, we report the characterization of differential gene expression between hWJSCs and hBMMSCs at the single-cell level, focusing particularly on expression of genes, which may contribute to the unique phenotype of hWJSCs and help to explain why they may be superior to hBMMSCs for therapeutics.

Materials and Methods

Our study was carried out with ethical approval from the Institutional Domain Specific Review Board (DSRB-D/08/358), National Healthcare Group (NHG), Ministry of Health Singapore. All experimental methods were performed in accordance with relevant guidelines and regulations.

Cell culture

Derivation and propagation of hWJSCs

Human umbilical cords were obtained with informed patient consent. hWJSCs were derived from three human umbilical cords (W0013F, W0026F, and W0030M) based on our previously published protocol [41]. The hWJSCs were cultured in hWJSC medium that comprised Dulbecco's modified Eagle's medium (DMEM) high glucose supplemented with 20% fetal bovine serum (Hyclone), 1% nonessential amino acids, 2 mM l-glutamine, 0.1 mM β-mercaptoethanol, 1% insulin/transferrin/selenium (ITS), antibiotic/antimycotic mixture (Invitrogen), and 16 ng/mL basic fibroblast growth factor (bFGF; Merck Millipore) and were then frozen for subsequent experiments.

Human bone marrow mesenchymal stem cells

Commercial adult hBMMSCs were purchased from Lonza (Allendale, NJ) and approval for their use in this project was obtained from the National University of Singapore Institutional Review Board (NUS-IRB). Primary cultures of hBMMSCs were expanded in the commercial medium supplied with the cells by the manufacturer and then frozen for subsequent experiments.

Fluorescence-activated cell sorting analysis of hWJSCs

The hWJSCs were disassociated for 3–5 min using TrypLE^™ Express (Invitrogen). Cells were washed with PBS⁽⁻⁾ (phosphate-buffered saline) and blocked with 10% normal goat serum (NGS; Invitrogen) for 20 min to prevent nonspecific binding. Cells were incubated with primary antibodies for MSC CD markers (CD34, CD73, CD90, and CD105; BioLegend, San Diego, CA) for 45 min, followed by incubation with the Alexa Fluor^® 488 (1:500) secondary antibody (Invitrogen) for 30 min. Cells were washed with PBS⁽⁻⁾ and resuspended in PBS⁽⁻⁾. Finally, cells were filtered using a 70 μm nylon strainer (BD Biosciences) to remove cell clumps and were analyzed using a CyAn^™ ADP Analyzer (Beckman Coulter, Fullerton, CA).

Picking single cells using a micromanipulator

The hWJSCs and hBMMSCs were disassociated using TrypLE Express (Invitrogen) for 3–5 min and washed with PBS⁽⁺⁾. Cells were resuspended in PBS⁽⁺⁾ and 300–500 μL was separately transferred to the center of 100 mm sterile plastic dishes (Becton Dickinson, BD). Single cells were picked using a biopsy pipette (The Pipette Company) attached to a micromanipulator system (Olympus, Tokyo, Japan) and released directly into a cell lysis buffer [0.2% Triton X-100 (Sigma-Aldrich) containing 2 U/μL RNAse inhibitor (Clontech)] in sterile 0.2 mL polymerase chain reaction (PCR) tubes. Cells were stored at −80°C for subsequent experiments.

Extraction of RNA and cDNA synthesis from hWJSC and hBMMSC samples

Cells were grown to 80% confluence in T-75 flasks, and 1 mL of TRIzol reagent (Invitrogen) was added to each for 5 min. Cells were scraped from the surface of the T75 flasks, pipetted up and down to mix, and removed into 1.5 mL microcentrifuge tubes. Two hundred microliters of chloroform was added to each tube, the samples were vortexed and incubated at room temperature for 3 min, followed by centrifugation at 4°C for 15 min. The aqueous phase was transferred to a fresh 1.5 mL tube, and 450 μL of 70% ethanol was added. After vortexing, 700 μL was added to an RNeasy column and centrifuged at 8,000g for 15 s. The RNeasy kit was used from this point onward, following the manufacturer's instructions and including a DNAseI digestion to remove contaminating genomic DNA. Synthesis of cDNA from 5 μg of RNA per sample was performed using the Tetro cDNA synthesis kit (Bioline) according to the manufacturer's instructions. All cDNAs were diluted 10-fold using TE buffer.

Quantitative real-time PCR

Quantitative real-time PCR (qPCR) was performed for 10 genes using the ABI PRISM 7500 Real-Time PCR System (Applied Biosystems), and data were analyzed using Sequence Detection software (version 1.4; Applied Biosystems). For each gene, 2 μL of cDNA was mixed with 10 μL Fast SYBR Green Master Mix (Thermo Fisher Scientific), and 300 nM each of forward and reverse primers, with water added to a final volume of 20 μL. Thermal cycling was performed as follows: 95°C for 20 s, and 40 cycles of 95°C for 3 s and 60°C for 30 s; a melt curve was added at the end of amplification to confirm the specificity of qPCR. β-2-microglobulin was used as a housekeeping gene for normalization of data. All reactions were performed in triplicate. Primer sequences are listed in Supplementary Table S1 (Supplementary Data are available online at www.liebertpub.com/scd).

cDNA synthesis and Nextera library preparation for NGS

Cell lysis, RNA extraction, and cDNA synthesis were carried out using the Smart-seq2 method, specifically targeting polyA+ messenger RNA (mRNA) [42]. Briefly, first-strand synthesis and template switching were performed, followed by 25 cycles of preamplification. cDNAs were purified using AMPure XP magnetic beads (Agencourt). Ninety-six-plex libraries were constructed and amplified using the Nextera XT library preparation kit (Illumina), and the libraries were then individually purified using AMPure XP magnetic beads. The libraries were quantified using the Qubit 3.0 fluorometer (Thermo Fisher Scientific), diluted to 10 nM, and then pooled to give 96 samples per lane. Quality of the final libraries was assessed using the Agilent bioanalyzer (Genomax) before sequencing at the Genome Institute of Singapore using 101-bp paired end reads on a HiSeq2000 (Illumina).

Data analysis

Paired-end sequencing FASTQ files were processed using Bowtie2 2.3.0 [43] and then aligned to the H. Sapiens NCBI GRCh38 reference genome using Spliced Transcripts Alignment to a Reference (STAR) software [44], followed by SAMtools to extract the aligned reads. Gene expression counts were estimated using root square mean error [45]. For quality control, samples with low expression (fewer than 8 million reads) and genes with fewer than 10 counts were removed. DESeq2 [46] was used for differential expression analysis, including normalization of the library size and removal of batch effects. Four hundred and thirty-six highly differentially expressed genes were selected for further analysis, having a fold change of 4 or greater, and false discovery rate (FDR) adjusted P values of <0.01. Cells were visualized in the feature space using principal component analysis (PCA) to informatively display genetic distance and relatedness between the cell populations.

Genes that were overexpressed in hWJSCs and hBMMSCs were separately examined using the Fishes test in “topGO” package of Bioconductor package to identify enriched Gene Ontology (GO) biological process pathways [47]. Multiple testing-corrected FDR were generated using the Benjamini-Hochberg procedure (BH method) [48]. Due to the hierarchical structure of GO pathways, we focused on more specific ones with pathway size <100. The top 436 differentially expressed genes were also analyzed using the Canonical and Hallmark data sets [49] from the Molecular Signatures Database (MSigDB) [50], and heatmaps were generated for the different pathways containing these genes.

Data availability

The data sets generated during and/or analyzed during the current study are available in the NCBI Gene Expression Omnibus (GEO) repository, under accession number GSE110791 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE110791).

Results

Data quality

A total of 103 single hWJSCs from 3 umbilical cords and 63 single hBMMSCs from 2 different donors were examined. We obtained a median of 2,365,954 reads per cell for the hWJSCs [interquartile range (IQR): 1,505,028–3,000,851] and 2,178,995 for the hBMMSCs (IQR: 1,934,689–2,582,680). After alignment to the NCBI GRCh38 reference genome, this resulted in a median of 1,771,069 reads (IQR: 1,164,104–2,286,098) for the hWJSCs, and a median of 1,786,084 reads (IQR: 1,533,992–2,103,281) for the hBMMSCs. Median read coverage was 80.2% (IQR: 75.6%–83.8%) and 82.1% (IQR: 78.4%–83.8%) for hWJSCs and hBMMSCs, respectively. Data for each cell are summarized in Supplementary Table S2. Following removal of samples with fewer than 8 million reads and genes with fewer than 10 counts, 9,875 genes expressed in 161 samples remained (Supplementary Fig. S1). Although the cell types appear homogeneous following fluorescence-activated cell sorting (FACS) and are described as “pure” MSC populations, there is a wide variation in the total number of genes expressed in individual cells within a cell type. The range in levels of expression of genes with ≥10 counts per cell is ∼10-fold in hWJSCs (range: 990–9,074), and 12-fold in hBMMSCs (range: 744–9,161). There was no significant difference in the number of expressed genes between the two cell types (P = 0.39).

Expression of surface markers

Confirmation that the hWJSCs were relatively pure populations of undifferentiated cells was revealed by FACS analysis of cell surface markers. MSCs are expected to have high expression of NT5E (CD73), THY1 (CD90), and ENG (CD105) and absent expression of CD34 [51]. The hWJSCs from all three cords in this study expressed the MSC markers in accordance with this scheme (Supplementary Fig. S2). Expression of a range of MSC-specific and endothelial markers showed that there is a population of hWJSCs with high expression of most of these markers, the remaining hWJSCs having marker expression similar to that of hBMMSCs (Fig. 1). Expression of hematopoietic and other lineage-specific markers (CD33, CD34, CD45, CD14, CD79A, CD19, ITGAM, PECAM1, CDH5, KDR, and CSFIR) in both cell types was low/absent, indicating that neither type of MSCs contained differentiated cells. FLT1 and MME are overexpressed in hWJSCs, whereas ANPEP and ENG are overexpressed in hBMMSCs; there was no significantly differential expression of the remaining markers observed between the two cell types.

FIG. 1.

Expression of cell surface markers on hWJSCs and hBMMSCs. Sixteen cell surface markers specific for MSCs and endothelial cells were examined using hierarchical clustering. MSC-specific markers are indicated using a yellow bar along the top of the heatmap, and endothelial-specific markers with a pink bar. The cell type is indicated along the left-hand side of the heatmap—hWJSCs are coded as red, and hBMMSCs are black. hBMMSCs, human bone marrow mesenchymal stem cells; hWJSCs, human Wharton's jelly stem cells; MSCs, mesenchymal stem cells. Color images available online at www.liebertpub.com/scd

Differential gene expression between the WJSCs and hBMMSCs

Following removal of batch effects using DESeq2, 436 genes were found to be differentially expressed with log fold change ≥2, counts ≥10, and an FDR <0.05 (Supplementary Table S3 and Supplementary Fig. S3). Two hundred and twenty-two genes are upregulated in hWJSCs, and 214 are upregulated in hBMMSCs. Principal component analysis and hierarchical clustering revealed that the hWJSCs and hBMMSCs can be clustered in 2 distinct groups using these 436 differentially expressed genes (Fig. 2a, b). The top 10 most highly differentially overexpressed genes in hWJSCs compared with hBMMSCs are CXCL8 (IL8), CXCL1, IL1B, DSG2, ST6GALNAC5, PCDH10, CXCL6, KRTAP7-1, DSC3 (Desmocollin 3), and NEFL. We selected 10 genes from the top 25 most highly overexpressed genes and performed qPCR for each on cDNA generated from the 3 hWJSC samples and the 3 hBMMSC samples. This confirmed that expression for each gene was significantly greater in hWJSCs than in hBMMSCs (Fig. 3).

FIG. 2.

Comparison of global gene expression between hWJSCs and hBMMSCs. (a) Principal component analysis of hWJSCs (blue) and hBMMSCs (pink) after batch effects were removed. (b) Heatmap of log-transformed normalized gene expression following removal of batch effects. Four hundred and thirty-six highly differentially expressed genes with adjusted P value <0.01 and fourfold change are included. Two hundred and twenty-two genes are overexpressed in hWJSCs (group indicated in red) and 214 are overexpressed in hBMMSCs (group indicated in black). Color images available online at www.liebertpub.com/scd

FIG. 3.

Confirmation of high relative expression of 10 genes in hWJSCs compared with hBMMSCs using qPCR. Expression of 10 genes shown to be highly differentially overexpressed in hWJSCs using single-cell sequencing was also assessed in each individual bulk sample using qPCR. It was shown that in all cases, expression was low or absent in hBMMSCs compared with hWJSCs. Vertical bars indicate SEM. qPCR, quantitative real-time polymerase chain reaction; RQ, relative quantification; SEM, standard error of the mean.

A great many of the genes upregulated in hWJSCs play key roles in processes contributing to the success of hWJSCs as a therapeutic source of stem cells, including immunomodulation, angiogenesis, wound healing, apoptosis, antitumor activity, and chemotaxis. Table 1 lists the most important genes upregulated in hWJSCs and hBMMSCs in relation to each of these processes.

Table 1.

Differentially Expressed Genes That May Contribute to the Differences in Functional Properties Between Human Wharton's Jelly Stem Cells and Human Bone Marrow Mesenchymal Stem Cells

Process	hWJSCs	hBMMSCs
Immune regulation	CXCL1, CXCL2, CXCL3. CXCL5, CXCL6, CXCL8, IL1A, IL1B, IL32, IL33, CCL2, C3, CD200	CXCL12, HGF, HLA-C, HLA-DMA
Chemotaxis	CXCL1, CXCL2, CXCL3. CXCL5, CXCL6, CXCL8	CXCL12, PDGFRA, PDGFRB
Angiogenesis	CXCL6, F3, HDAC9, FOXF1, FGF2, CCL2, THBS1, MMP1, MMP3, CTGF	SERPINB1, SERPINB7, SERPINE2, SERPINF1, SERPING2, PDGFRA, PDGFRB, TIMP3
Proliferation	FGF2, INHBA, GREM1, THSB1, IGFBP2	IGFBP3, IGFBP6
Coagulation	MMP1, MMP3, C3, DPP4, F3, FGF2	TIMP3
Tumorigenesis and apoptosis	IL1A, IL1B, BIRC3	TIMP3, IGFBP3
Neurogenesis	NEFL, SOX11, CDH2, FLT1, NTF3, IL1B	IGFBP6

The genes are separated into pathways they have been previously reported to function in; many genes are involved in multiple pathways.

CTGF, connective tissue growth factor; FGF2, fibroblast growth factor-2; hBMMSCs, human bone marrow mesenchymal stem cells; hWJSCs, human Wharton's jelly stem cells; THBS1, thrombospondin 1; TIMPs, tissue inhibitors of metalloproteinases.

Functional significance of differentially expressed genes

To better understand the functional differences between the two cell types, gene set enrichment analysis (GSEA) was performed using the MSigDB. We examined Hallmark pathways and Canonical pathways. Enriched pathways with an FDR >0.05 are listed in Supplementary Table S4; pathways that were selected for further examination are shaded in gray. Differentially expressed genes from each pathway were used to generate heatmaps and PCA plots to determine whether they could separate hWJSCs from hBMMSCs. Pathways giving very good separation of the two cell types include the matrisome core and associated genes (Fig. 4a–d), the angiogenesis pathway (Fig. 4e, f), and the Reactome immune pathway (Fig. 4g, h). Other significant pathways include the coagulation pathway (Supplementary Fig. S4a, b) and the epithelial/mesenchymal transition (EMT) genes (Supplementary Fig. S4c, d). Enriched GO terms for hWJSCs are shown in Fig. 5 and the full list is presented in Supplementary Table S5. hWJSCs are enriched for GO terms, including positive regulation of vascular endothelial growth factor production, macrophage activation, response to interleukin 1, reactive oxygen species biosynthesis, neutrophil migration, myeloid leukocyte-mediated immunity, endodermal cell differentiation, and positive regulation of cytokine secretion. The most enriched GO terms for hBMMSCs include kidney vascular development and bone mineralization (Supplementary Fig. S5 and Supplementary Table S6).

FIG. 4.

Hierarchical clustering and PCA of differential gene expression between hWJSCs and hBMMSCs give insight into the unique properties of hWJSCs. Pathways examined include matrisome core genes (a, b), matrisome-associated genes (c, d), angiogenesis (e, f), and the immune Reactome (g, h). Hierarchical clustering (a, c, e, g) was performed using gene expression values z-transformed by gene (forcing each gene to have zero mean and standard deviation = 1). Bone marrow cells are indicated by a black bar along the y axis, and hWJSCs are in red. Upregulated genes are colored red in the heatmap and downregulated genes are blue. Principal component analysis (b, d, f, h) indicates the clustering of cells (hWJSCs in red and hBMMSCs in black). PCA, principal component analysis. Color images available online at www.liebertpub.com/scd

FIG. 5.

Selected enriched GO terms for genes overexpressed in hWJSCs, associated with unique properties of hWJSCs. ES (enrichment score) represents the -log(FDR) of GO pathways enriched in 222 overexpressed genes from hWJSCs. FDR, false discovery rate; GO, gene ontology. Color images available online at www.liebertpub.com/scd

Expression of genes involved in wound healing

Wound healing is a complex process, requiring chemotaxis of molecules to the site of injury, clotting of the wound, remodeling of the extracellular matrix (ECM), angiogenesis, and re-epithelialization. These pathways were identified using GSEA and GO term analysis as containing a significant number of differentially expressed genes between hWJSCs and hBMMSCs.

The two cell types can be well separated by expression of the matrisome genes (Fig. 4a–d). The “matrisome” is a term used to describe the proteins that make up the ECM, and comprises core matrisome proteins (encoded by 278 genes) and matrisome-associated factors (encoded by 778 genes) [52]. Twenty-nine matrisome “core” genes and 48 matrisome-associated genes were differentially expressed between hWJSCs and hBMMSCs. ECM regulators, including the matrix metalloproteinases MMP1 and MMP3, and the disintegrin metalloproteinases such as ADAM19, ADAM23, and ADAMTS12, are essential for tissue remodeling and can be inhibited by tissue inhibitors of metalloproteinases (TIMPs). hBMMSCs express high levels of TIMP3, which may prevent them from being as effective as hWJSCs for promoting wound healing.

The ECM plays a critical role in another process essential for wound healing: the formation of new blood vessels through angiogenesis. Many important proangiogenic molecules are upregulated in hWJSCs (Fig. 4e), including fibroblast growth factor-2 (FGF2), MMP1, MMP3, connective tissue growth factor (CTGF), and FOXF1. In addition to these proangiogenic factors, the potent inhibitor of angiogenesis, thrombospondin 1 (THBS1), is also upregulated in hWJSCs. The antiangiogenic SERPINs (SERPINB7, -B1, -B6, -E2, -F1, and -G1) are expressed at much lower levels in hWJSCs than in hBMMSCs.

Clustering analysis of genes involved in the coagulation pathway (Supplementary Fig. S4a, b) indicates that as well as overexpressing MMP1 and MMP3, hWJSCs also overexpress C3, DPP4, and importantly, the clotting factor F3. F3 plays an important role in coagulation, angiogenesis, and tissue repair. Many genes found in the EMT pathway are significantly differentially expressed between hWJSCs and hBMMSCs (Supplementary Fig. S4c, d). EMT results in changes to cell phenotypes, transforming them from stationary differentiated cells to mobile dedifferentiated cells [53]. EMT is essential for re-epithelialization following a wound, causing keratinocytes to migrate to the site of injury and proliferate [54]. These migrating keratinocytes remodel the ECM surrounding the area.

We found significant overexpression of the CXCL molecules in hWJSCs. These molecules are involved in multiple wound-healing pathways, including chemotaxis and angiogenesis. IL33, recently found to be important for wound healing in mice [55], is also upregulated in hWJSCs. Analysis of GO terms identified hWJSCs as being involved in positive regulation of cytokine secretion, macrophage activation, and neutrophil migration. Ingenuity Pathway Analysis confirmed the important role of many of the differentially expressed genes in the recruitment, adhesion, and accumulation of phagocytes and recruitment of granulocytes (Fig. 6).

FIG. 6.

IPA showing upstream regulator effect analysis of the cell migration pathway. Molecules known to influence recruitment, adhesion, and activation of phagocytes and recruitment of granulocytes are upregulated in hWJSCs, supporting the role for this cell type in wound healing. The top row indicates the master regulators, and the middle row indicates the intermediate regulators found in our data set, which in turn exert their effects on the cell migration functions indicated on the bottom row. Predicted activation of master regulators is color coded in orange, and predicted repression is in blue. Upregulated intermediate regulators are indicated in red and downregulated molecules are in green, with intensity correlating with log fold change. Causal relationships consistent with the IPA database are shown using orange and blue lines, whereas those inconsistent are indicated with a yellow line. IPA, ingenuity pathway analysis. Color images available online at www.liebertpub.com/scd

Expression of immune response genes

Aside from the previously mentioned overexpression of chemokines in hWJSCs, there are other important differences in expression of immune response genes between hWJSCs and hBMMSCs (Fig. 4g, h), affecting pathways, including immune response, cellular proliferation, differentiation, and apoptosis. The highly overexpressed interleukins, IL1A and IL1B, are involved in immunomodulation. IL32 and IL33 are also highly overexpressed. IL32 can induce TNF-α expression from macrophages [56] and induces expression of CXCL8 and CXCL2, both also highly expressed in hWJSCs. CD200 plays a role in immunosuppression, contributes to allograft tolerance, and is already known to be upregulated in hWJSCs [57].

Tumorigenic properties

Antitumor properties of hWJSCs are related to expression of tumor suppressors and proapoptotic genes [30]. Five oncogenes (BCL11A, BIRC3, CCND2, LCP1, and HOXA11) were highly expressed in hWJSCs, and another five oncogenes were more highly expressed in hBMMSCs (CD74, ETV1, PDGFRA, PDGFRB, and PRRX1). No differential expression of tumor suppressors was noted using GSEA, but CADM1, a tumor suppressor in nonsmall-cell lung cancer [58], was overexpressed 18-fold in hWJSCs. IL1A and IL1B are the most highly overexpressed apoptosis-related genes in hWJSCs compared with hBMMSCs. BIRC3, an antiapoptotic protein, was overexpressed 7.3-fold in hWJSCs compared with hBMMSCs.

Differences in gene expression related to proliferation

Several genes essential for promotion of self-renewal of embryonic stem cells (hESCs) and their maintenance in an undifferentiated state are also upregulated in hWJSCs, including FGF2 and its targets, INHBA, GREM1, and THSB1. Two related molecules displaying negative effects on proliferation, and underexpressed in hWJSCs compared with hBMMSCs, are the insulin-like growth factor binding proteins, IGF3BP and IGFBP6.

There was significant overexpression of DSG2 and DSC3 (353-fold and 231-fold overexpression, respectively). DSG2 (desmoglein 2) is associated with desmosome assembly, mediating intercellular adhesion. Expression is required for embryonic stem cell proliferation [59], and overexpression results in abnormally increased proliferation through activation of mitogenic signaling pathways, including PI3K/Akt and JAK/Stat3 [60]. DSC3 is found primarily in epithelial cells, along with the desmogleins, and is also essential for cell adhesion and desmosome formation.

Differentiation

Several neurogenesis-related factors were upregulated in hWJSCs. The neurogenic transcription factor SOX11, as well as receptors associated with this process, including CDH2 and FLT1, has previously been reported to be highly expressed in hWJSCs compared with hBMMSCs [35]. NTF3 (involved in neurogenesis, neuron survival, and differentiation) was shown to be upregulated, and the neurogenic marker, NEFL, is very highly overexpressed in hWJSCs in our study (216-fold), supporting a role for hWJSCs in induction of neurogenesis. IGFBP6, which is overexpressed in hBMMSCs, negatively regulates neurogenesis [61]. hBMMSCs express higher levels of the transcription factor FOS, which has been shown to play a role in osteogenesis [62].

Discussion

Differences in gene expression between hWJSCs and hBMMSCs have previously been reported, but this is the first study to examine differences between the two cell types at the single-cell level. We found differential expression of 436 genes between the two cell types at a level of fourfold difference or greater, given that there is expression of ∼9,000 genes with >10 copies per cell, which means that ∼5% of the genes are differentially regulated. Although cell surface marker expression indicates that isolated hWJSCs and hBMMSCs are both homogenous populations, we can clearly identify that the expression levels are extremely variable from cell to cell. Therefore, although a particular MSC type may be predicted by cell surface markers to be a “pure” population of cells, there is more heterogeneity than previously predicted, with many cells undergoing “bursts” of expression. For example, in our data set, a single hWJSC expressed particularly high levels of CXCL1, while another has exceedingly high MMP1. If the cells are examined as bulk populations, it is not possible to appreciate that a single cell may be making a major contribution to the apparent overall expression level.

We have demonstrated that a great number of genes differentially expressed between hWJSCs and hBMMSCs could be predicted to play roles in the pathways contributing to the unique properties of hWJSCs. Significant among these were the genes making up the ECM, immune pathway, matrisome, and angiogenesis pathways.

As well as being a fundamental component of eukaryotic cells, providing architectural structure to define tissue boundaries, and a substrate for cells to migrate along, the dynamic composition of the ECM and its interacting proteins provides regulatory cues influencing cell behavior. ECM components sequester and release cytokines, including growth factors, and transmit signals essential for apoptosis, adhesion, proliferation, and differentiation. Many of these molecules are released by protease activation, allowing rapid cellular response without the need for de novo protein synthesis. Alterations in the ECM are associated with disorders such as cancer, fibrosis, and skeletal disorders. It is surprising that there are so many structural differences between the two MSC types, given their common origin as MSCs, but the significant differences in expression of the matrisome proteins may be key to the differences in function of hWJSCs and hBMMSCs.

Expression of immune molecules was particularly high in hWJSCs. Our finding that there are numerous cytokines overexpressed in hWJSCs compared with hBMMSCs supports work previously reported by several other groups [30,63]. The high expression of chemokines by hWJSCs promotes wound healing by enhancing recruitment of neutrophils, granulocytes, and macrophages, as well as potentially recruiting skin MSCs to the wound site [17]. Chemokines are released at the site of injury, promoting migration of neutrophils and macrophages into the area by chemotaxis along a chemical gradient. The neutrophils and macrophages clean the area, by ingesting debris and bacteria, and then macrophages and platelets release platelet-derived growth factor (PDGF). Release of PDGF attracts dermal fibroblasts to the region, resulting in remodeling of the ECM.

The overrepresentation of vascular endothelial processes supports previous data using microarray analysis comparing hWJSCs and hBMMSCs [35]. FGF2, an ECM-affiliated protein, is one of the most important regulators of angiogenesis, inducing endothelial cell growth. MMP1 and MMP3, both highly expressed in hWJSCs, are essential for degradation of existing vessel ECM, allowing newly formed endothelial cells to migrate out, along with secreted factors. CTGF is an important molecule for tissue remodeling and promotes VEGF-dependent angiogenesis [64]. The transcription factor, FOXF1, is required for angiogenesis. Expression of antiangiogenic SERPINs was much lower in hWJSCs than hBMMSCs. Our results indicate that hWJSCs express more pro- than antiangiogenic factors, suggesting that they may be more suitable for wound healing applications.

Within a cell type, there may be expression in some cells of molecules, which are activators of a particular process, while other cells express repressive molecules for the same pathway. For example, there are many proangiogenic genes overexpressed in hWJSCs, but also THBS1, an antiangiogenic molecule, is overexpressed in some hWJSCs. It is possible that this may support the concept of an “angiogenic switch,” originally proposed by Hanahan and Folkman in 1996 [65]; it may be that there are subpopulations of cells within the WJSCs, which are primed to induce angiogenesis and others that are inhibitory. The histone deacetylase, HDAC9, has a proangiogenic function [66] and is overexpressed in hWJSCs compared with hBMMSCs. It has been shown to exert its effects by repressing the miR17-92 cluster, which in endothelial cells has been found to be antiangiogenic [66]. However, in tumors, the miR17-92 cluster promotes angiogenesis by repression of its targets, the antiangiogenic THSB1 and proangiogenic CTGF. Preliminary data from our group have shown that the miR17-92 cluster is upregulated in hWJSCs compared with hBMMSCs, and so, it is unexpected to also find upregulation of HDCA9.

One of the key advantages that hWJSCs have over hBMMSCs is their increased rate of proliferation. We showed that FGF2 and its targets, INHBA, GREM1, and THSB1, are all expressed at high levels in hWJSCs compared with hBMMSCs. It could be that the upregulation of these proteins and the subsequent activation of the TGF-β signaling pathway are, at least in part, responsible for increased proliferation in hWJSCs. INHBA has been shown to regulate the growth and differentiation of many different cell types through both autocrine and paracrine mechanisms [67]. FGF2 can be induced in hESCs by an autocrine mechanism following addition of serum containing FGF2, growth medium for hWJSCs contains FGF2, whereas the medium for hBMMSCs does not, and so it is possible that the increased expression of FGF2, along with its downstream targets, is a consequence of the growth medium. It has long been known that IGFP3, which is upregulated in hBMMSCs, has an antiproliferative function, binding IGF molecules and reducing bioavailability for their receptors [68]. IGF/IGFBP complexes can be cleaved by MMPs, releasing IGF at the site of activity; cleavage of IGFBP3 by both MMP1 [69] and MMP3 [70] has been demonstrated. One possible mechanism for the comparatively low proliferation rate of hBMMSCs would be that their reduced levels of MMP1 and MMP3 mean that IGFBP3 remains uncleaved and bound to IGF; this sequestration could prevent binding of IGF to its receptors. IGFBP3 can also exert antiproliferative functions in an IGF-independent manner [71], and so, there is more than one possible route for the effects of increased IGFBP3 in hBMMSCs. Oh et al. demonstrated that one of these IGF-independent mechanisms is through apoptosis via activation of caspases 7 and 8 [71].

A recent study showed that DSG2 and DSC3 were expressed at higher levels in placental MSCs than in hBMMSCs [72]; similarly, umbilical cord-derived MSCs [73] and ESCs [74] were found to have a higher expression of DSG2 and DSC3 than hBMMSCs. It has been suggested that as well as playing a vital role in cell adhesion, these molecules are also signaling regulators playing a role in diverse processes, including morphogenesis, wound healing, differentiation, and homeostasis [75]. Changes in desmosomal cadherin levels play a role in tumorigenesis [76]; loss of expression of DSG in tumor cells has been suggested as a mechanism for evasion of apoptosis. Changes in the level of desmosomal cadherins may alter levels of binding sites for signaling molecules.

Immunogenic cell death of cancer cells can be induced using concentrated CM from hWJSCs [77], and it has been proposed that this may be, at least in part, due to the high levels of immune molecules secreted by hWJSCs into the CM, including CXCL8 and ICAM, both of which are highly upregulated in hWJSCs compared with hBMMSCs. Overexpression of these immune molecules in hWJSCs compared with hBMMSCs indicates that hWJSCs may have superior antitumor capabilities.

A major limitation of using single-cell sequencing to compare two cell types is the amount of handling of the cells required, which may affect the levels of gene expression in each individual cell. It has recently been shown that primary hBMMSCs differentially express ∼3,000 genes compared with cultured hBMMSCs either maintained as spheroids or as adherent cultures [78]. However, gene expression between early- and late-passage hWJSCs was demonstrated to be minimal [30], and no phenotypic changes were seen following prolonged passaging [79]. It is possible that the hWJSCs used in this study were reliably representative of primary hWJSCs, whereas the hBMMSCs were already altered by their brief time in culture.

Identification of low-quality data and removal of outliers at the start of the data analysis process are crucial to obtaining robust data. Using DESeq2, batch effects on expression between the different donors were removed. Amplification biases constitute another possible source of technical noise, which may mask more subtle changes in expression from cell to cell. We used the SMART-Seq2 method described by Picelli et al. [42] for our library preparation, allowing us to examine full-length transcripts. It would be possible to generate data on the splice variants if sequencing was performed at a higher depth, adding an extra layer of detail to the profile of individual cells. To fully characterize the differences within a population of hWJSCs, it would be necessary to examine far greater numbers of cells. Distinct subtypes of cells can be identified within a tissue using as few as 50,000 reads [80], which would enable us to run a greater number of single cells on a flow cell than used in this study, increasing the throughput. However, it would not be expected that the differences between cells within the hWJSC type would be sufficient to obtain meaningful data using such low-coverage sequencing. In addition, the manual method that we have used is not amenable to handling more than 100 cells at a single time. Various high-throughput technologies have been developed over recent years, with droplet and bead methods such as Drop-seq [81] and inDrop [82] having the capacity to examine thousands of cells at a time. We have only examined the 436 most highly differentially expressed genes in this study, and so, it is possible that there are more subtle differences in expression between the two cell types, which may also help to explain the unique properties of hWJSCs.

Conclusion

Many of the advantages that hWJSCs have over hBMMSCs for clinical application can be explained by the differences in their gene expression. They have greatly increased expression of immune regulators, helping to explain their capacity to cause immunogenic cell death (ICD); they express higher levels of molecules required for all stages of wound healing, from recruitment of macrophages to clotting, and migration of keratinocytes for re-epithelialization; many genes involved in proliferation are also increased in expression in hWJSCs. This increased understanding of the differences between the MSC types may help with the development of new treatments using these cells.

Footnotes

Acknowledgment

This work was supported by the Singapore National University Health System (NUHS) Aspiration Fund (New Idea; R-174-000-155-720) research grant.

Author Disclosure Statement

The authors have no conflicts of interests to declare.

References

Baksh

, Yao

and Tuan

. (2007). Comparison of proliferative and multilineage differentiation potential of human mesenchymal stem cells derived from umbilical cord and bone marrow. Stem Cells, 25:1384–1392.

Rao

and Mattson

. (2001). Stem cells and aging: expanding the possibilities. Mech Ageing Dev, 122:713–734.

Subramanian

, Fong

, Biswas

and Bongso

. (2015). Comparative characterization of cells from the various compartments of the human umbilical cord shows that the Wharton's jelly compartment provides the best source of clinically utilizable mesenchymal stem cells. PLoS One, 10:e0127992.

Fong

, Subramanian

, Biswas

and Bongso

. (2016). Freezing of fresh Wharton's jelly from human umbilical cords yields high post-thaw mesenchymal stem cell numbers for cell-based therapies. J Cell Biochem, 117:815–827.

Prasanna

and Jahnavi

. (2011). Wharton's jelly mesenchymal stem cells as off-the-shelf cellular therapeutics: a closer look into their regenerative and immunomodulatory properties. Open Tissue End Regen Med J, 4:28–38.

Fan

, Zhang

and Zhou

. (2011). Therapeutic potentials of mesenchymal stem cells derived from human umbilical cord. Stem Cell Rev, 7:195–207.

Gauthaman

, Fong

, Suganya

, Subramanian

, Biswas

, Choolani

and Bongso

. (2012). Extra-embryonic human Wharton's jelly stem cells do not induce tumorigenesis, unlike human embryonic stem cells. Reprod Biomed Online, 24:235–246.

Wang

, Han

, Ma

, Zuo

, Geng

, Gong

, Sun

, Li

, Wang

, et al. (2012). A toxicity study of multiple-administration human umbilical cord mesenchymal stem cells in cynomolgus monkeys. Stem Cells Dev, 21:1401–1408.

, Chan

, Tsai

, Chang

, Sieber

, Chiu

, Ho

, Peng

, Wu

and Huang

. (2011). Effective treatment of severe steroid-resistant acute graft-versus-host disease with umbilical cord-derived mesenchymal stem cells. Transplantation, 91:1412–1416.

10.

Chen

, Zhang

, Li

, Guo

, Wang

, Yang

and Gao

. (2012). Human umbilical cord Wharton's jelly stem cells: immune property genes assay and effect of transplantation on the immune cells of heart failure patients. Cell Immunol, 276:83–90.

11.

Spaeth

, Dembinski

, Sasser

, Watson

, Klopp

, Hall

, Andreeff

and Marini

. (2009). Mesenchymal stem cell transition to tumor-associated fibroblasts contributes to fibrovascular network expansion and tumor progression. PLoS One, 4:e4992.

12.

Zhang

, Lee

, Rui

, Cheng

, Jiang

and Li

. (2013). Bone marrow-derived mesenchymal stem cells promote growth and angiogenesis of breast and prostate tumors. Stem Cell Res Ther, 4:70.

13.

Subramanian

, Shu-Uin

, Kae-Siang

, Gauthaman

, Biswas

, Choolani

, Bongso

and Chui-Yee

. (2012). Human umbilical cord Wharton's jelly mesenchymal stem cells do not transform to tumor-associated fibroblasts in the presence of breast and ovarian cancer cells unlike bone marrow mesenchymal stem cells. J Cell Biochem, 113:1886–1895.

14.

Zhang

, Lin

, Shi

, Xu

, Fu

, Lv

, Chen

, Lv

, Li

, et al. (2012). Human umbilical cord mesenchymal stem cells improve liver function and ascites in decompensated liver cirrhosis patients. J Gastroenterol Hepatol, 27(Suppl 2):112–120.

15.

, Zhang

, Liu

, Qiuwaxi

, Ashwood

, Cho

, Huan

, Ge

, Chen

, et al. (2013). Transplantation of human cord blood mononuclear cells and umbilical cord-derived mesenchymal stem cells in autism. J Transl Med, 11:196.

16.

Fong

, Tam

, Cheyyatraivendran

, Gan

, Gauthaman

, Armugam

, Jeyaseelan

, Choolani

, Biswas

and Bongso

. (2014). Human Wharton's jelly stem cells and its conditioned medium enhance healing of excisional and diabetic wounds. J Cell Biochem, 115:290–302.

17.

Tam

, Cheyyatraviendran

, Venugopal

, Biswas

, Choolani

, Ramakrishna

, Bongso

and Fong

. (2014). A nanoscaffold impregnated with human wharton's jelly stem cells or its secretions improves healing of wounds. J Cell Biochem, 115:794–803.

18.

Ayuzawa

, Doi

, Rachakatla

, Pyle

, Maurya

, Troyer

and Tamura

. (2009). Naive human umbilical cord matrix derived stem cells significantly attenuate growth of human breast cancer cells in vitro and in vivo. Cancer Lett, 280:31–37.

19.

Chao

, Yang

and Chen

. (2012). Human umbilical cord mesenchymal stem cells suppress breast cancer tumourigenesis through direct cell-cell contact and internalization. J Cell Mol Med, 16:1803–1815.

20.

Dzobo

, Vogelsang

, Thomford

, Dandara

, Kallmeyer

, Pepper

and Parker

. (2016). Wharton's jelly-derived mesenchymal stromal cells and fibroblast-derived extracellular matrix synergistically activate apoptosis in a p21-dependent mechanism in WHCO1 and MDA MB 231 cancer cells in vitro. Stem Cells Int, 2016:4842134.

21.

Gauthaman

, Fong

, Arularasu

, Subramanian

, Biswas

, Choolani

and Bongso

. (2013). Human Wharton's jelly stem cell conditioned medium and cell-free lysate inhibit human osteosarcoma and mammary carcinoma cell growth in vitro and in xenograft mice. J Cell Biochem, 114:366–377.

22.

Kalamegam

, Pushparaj

, Khan

, Sait

, Anfinan

and Al-Qahtani

. (2015). Primary ovarian cancer cell inhibition by human Wharton's Jelly stem cells (hWJSCs): mapping probable mechanisms and targets using systems oncology. Bioinformation, 11:529–534.

23.

Lin

, Fong

, Biswas

, Choolani

and Bongso

. (2014). Human Wharton's jelly stem cells, its conditioned medium and cell-free lysate inhibit the growth of human lymphoma cells. Stem Cell Rev, 10:573–586.

24.

Liu

, Han

, Liu

and Qin

. (2013). Suppression of cholangiocarcinoma cell growth by human umbilical cord mesenchymal stem cells: a possible role of Wnt and Akt signaling. PLoS One, 8:e62844.

25.

, Hao

, Zhang

and Zhang

. (2012). The in vitro and in vivo effects of human umbilical cord mesenchymal stem cells on the growth of breast cancer cells. Breast Cancer Res Treat, 133:473–485.

26.

Maurya

, Doi

, Kawabata

, Pyle

, King

, Wu

, Troyer

and Tamura

. (2010). Therapy with un-engineered naive rat umbilical cord matrix stem cells markedly inhibits growth of murine lung adenocarcinoma. BMC Cancer, 10:590.

27.

Rachakatla

, Marini

, Weiss

, Tamura

and Troyer

. (2007). Development of human umbilical cord matrix stem cell-based gene therapy for experimental lung tumors. Cancer Gene Ther, 14:828–835.

28.

Sun

, Yu

, Bhandari

, Jung

, Kang

and Kang

. (2010). Human umbilical cord blood mesenchymal stem cell-derived extracellular matrix prohibits metastatic cancer cell MDA-MB-231 proliferation. Cancer Lett, 296:178–185.

29.

, Ju

, Du

, Zhu

and Liu

. (2013). Microvesicles derived from human umbilical cord Wharton's jelly mesenchymal stem cells attenuate bladder tumor cell growth in vitro and in vivo. PLoS One, 8:e61366.

30.

Fong

, Chak

, Biswas

, Tan

, Gauthaman

, Chan

and Bongso

. (2011). Human Wharton's jelly stem cells have unique transcriptome profiles compared to human embryonic stem cells and other mesenchymal stem cells. Stem Cell Rev, 7:1–16.

31.

Nekanti

, Rao

, Bahirvani

, Jan

, Totey

and Ta

. (2010). Long-term expansion and pluripotent marker array analysis of Wharton's jelly-derived mesenchymal stem cells. Stem Cells Dev, 19:117–130.

32.

Ganta

, Chiyo

, Ayuzawa

, Rachakatla

, Pyle

, Andrews

, Weiss

, Tamura

and Troyer

. (2009). Rat umbilical cord stem cells completely abolish rat mammary carcinomas with no evidence of metastasis or recurrence 100 days post-tumor cell inoculation. Cancer Res, 69:1815–1820.

33.

Lin

, Bongso

, Gauthaman

, Biswas

, Choolani

and Fong

. (2013). Human Wharton's jelly stem cell conditioned medium enhances freeze–thaw survival and expansion of cryopreserved CD34⁺ cells. Stem Cell Rev, 9:172–183.

34.

Gauthaman

, Yee

, Cheyyatraivendran

, Biswas

, Choolani

and Bongso

. (2012). Human umbilical cord Wharton's jelly stem cell (hWJSC) extracts inhibit cancer cell growth in vitro. J Cell Biochem, 113:2027–2039.

35.

Hsieh

, Wang

, Chang

, Liao

, Lee

, Lin

, Wu

, Lin

and Cheng

. (2013). Mesenchymal stem cells from human umbilical cord express preferentially secreted factors related to neuroprotection, neurogenesis, and angiogenesis. PLoS One, 8:e72604.

36.

Raj

and van Oudenaarden

. (2008). Nature, nurture, or chance: stochastic gene expression and its consequences. Cell, 135:216–226.

37.

, Klughammer

, Farlik

, Penz

, Spittler

, Barbieux

, Berishvili

, Bock

and Kubicek

. (2016). Single-cell transcriptomes reveal characteristic features of human pancreatic islet cell types. EMBO Rep, 17:178–187.

38.

Treutlein

, Brownfield

, Wu

, Neff

, Mantalas

, Espinoza

, Desai

, Krasnow

and Quake

. (2014). Reconstructing lineage hierarchies of the distal lung epithelium using single-cell RNA-seq. Nature, 509:371–375.

39.

Xin

, Kim

, Ni

, Wei

, Okamoto

, Lee

, Adler

, Cavino

, Murphy

, et al. (2016). Use of the Fluidigm C1 platform for RNA sequencing of single mouse pancreatic islet cells. Proc Natl Acad Sci U S A, 113:3293–3298.

40.

Shalek

, Satija

, Shuga

, Trombetta

, Gennert

, Lu

, Chen

, Gertner

, Gaublomme

, et al. (2014). Single-cell RNA-seq reveals dynamic paracrine control of cellular variation. Nature, 510:363–369.

41.

Fong

, Subramanian

, Biswas

, Gauthaman

, Srikanth

, Hande

and Bongso

. (2010). Derivation efficiency, cell proliferation, freeze–thaw survival, stem-cell properties and differentiation of human Wharton's jelly stem cells. Reprod Biomed Online, 21:391–401.

42.

Picelli

, Björklund

ÅK

, Faridani

, Sagasser

, Winberg

and Sandberg

. (2013). Smart-seq2 for sensitive full-length transcriptome profiling in single cells. Nat Methods, 10:1096–1098.

43.

Langmead

and Salzberg

. (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods, 9:357–359.

44.

Dobin

, Davis

, Schlesinger

, Drenkow

, Zaleski

, Jha

, Batut

, Chaisson

and Gingeras

. (2013). STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 29:15–21.

45.

and Dewey

. (2011). RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics, 12:323.

46.

Love

, Huber

and Anders

. (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol, 15:550.

47.

Alexa

and Rahnenfuhrer

. (2016). topGO: Enrichment Analysis for Gene Ontology. R package version 2.30.0. https://bioconductor.statistik.tu-dortmund.de/packages/3.6/Boix/html/topGO.html accessed February 1, 2018 .

48.

Benjamini

, Drai

, Elmer

, Kafkafi

and Golani

. (2001). Controlling the false discovery rate in behavior genetics research. Behav Brain Res, 125:279–284.

49.

Liberzon

, Birger

, Thorvaldsdóttir

, Ghandi

, Mesirov

and Tamayo

. (2015). The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst, 1:417–425.

50.

Subramanian

, Tamayo

, Mootha

, Mukherjee

, Ebert

, Gillette

, Paulovich

, Pomeroy

, Golub

, Lander

and Mesirov

. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A, 102:15545–15550.

51.

Dominici

, Le Blanc

, Mueller

, Slaper-Cortenbach

, Marini

, Krause

, Deans

, Keating

, Prockop

and Horwitz

. (2006). Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy, 8:315–317.

52.

Naba

, Clauser

, Hoersch

, Liu

, Carr

and Hynes

. (2012). The matrisome: in silico definition and in vivo characterization by proteomics of normal and tumor extracellular matrices. Mol Cell Proteomics, 11:M111.014647.

53.

Kalluri

and Neilson

. (2003). Epithelial-mesenchymal transition and its implications for fibrosis. J Clin Invest, 112:1776–1784.

54.

Leopold

, Vincent

and Wang

. (2012). A comparison of epithelial-to-mesenchymal transition and re-epithelialization. Semin Cancer Biol, 22:471–483.

55.

Oshio

, Komine

, Tsuda

, Tominaga

, Saito

, Nakae

and Ohtsuki

. (2017). Nuclear expression of IL-33 in epidermal keratinocytes promotes wound healing in mice. J Dermatol Sci, 85:106–114.

56.

Kim

, Han

, Azam

, Yoon

and Dinarello

. (2005). Interleukin-32: a cytokine and inducer of TNFalpha. Immunity, 22:131–142.

57.

Najar

, Raicevic

, Jebbawi

, De Bruyn

, Meuleman

, Bron

, Toungouz

and Lagneaux

. (2012). Characterization and functionality of the CD200-CD200R system during mesenchymal stromal cell interactions with T-lymphocytes. Immunol Lett, 146:50–56.

58.

Kuramochi

, Fukuhara

, Nobukuni

, Kanbe

, Maruyama

, Ghosh

, Pletcher

, Isomura

, Onizuka

, et al. (2001). TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer. Nat Genet, 27:427–430.

59.

Eshkind

, Tian

, Schmidt

, Franke

, Windoffer

and Leube

. (2002). Loss of desmoglein 2 suggests essential functions for early embryonic development and proliferation of embryonal stem cells. Eur J Cell Biol, 81:592–598.

60.

Brennan

, Hu

, Joubeh

, Choi

, Whitaker-Menezes

, O'Brien

, Uitto

, Rodeck

and Mahoney

. (2007). Suprabasal Dsg2 expression in transgenic mouse skin confers a hyperproliferative and apoptosis-resistant phenotype to keratinocytes. J Cell Sci, 120:758–771.

61.

Barkho

, Song

, Aimone

, Smrt

, Kuwabara

, Nakashima

, Gage

and Zhao

. (2006). Identification of astrocyte-expressed factors that modulate neural stem/progenitor cell differentiation. Stem Cells Dev, 15:407–421.

62.

Closs

, Murray

, Schmidt

, Schön

, Erfle

and Strauss

. (1990). c-fos Expression precedes osteogenic differentiation of cartilage cells in vitro. J Cell Biol, 111:1313–1323.

63.

Yoo

, Jang

, Lee

, Kim

, Yang

, Eom

, Lee

, Kim

, Yang

, et al. (2009). Comparison of immunomodulatory properties of mesenchymal stem cells derived from adult human tissues. Cell Immunol, 259:150–156.

64.

Liu

, Chuang

, Hsu

, Tsai

, Wang

and Tang

. (2014). CTGF increases vascular endothelial growth factor-dependent angiogenesis in human synovial fibroblasts by increasing miR-210 expression. Cell Death Dis, 5:e1485.

65.

Hanahan

and Folkman

. (1996). Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell, 86:353–364.

66.

Kaluza

, Kroll

, Gesierich

, Manavski

, Boeckel

, Doebele

, Zelent

, Rössig

, Zeiher

, et al. (2013). Histone deacetylase 9 promotes angiogenesis by targeting the antiangiogenic microRNA-17-92 cluster in endothelial cells. Arterioscler Thromb Vasc Biol, 33:533–543.

67.

Reis

, Luisi

, Carneiro

, Cobellis

, Federico

, Camargos

and Petraglia

. (2004). Activin, inhibin and the human breast. Mol Cell Endocrinol, 225:77–82.

68.

Wetterau

, Moore

, Lee

, Shim

and Cohen

. (1999). Novel aspects of the insulin-like growth factor binding proteins. Mol Genet Metab, 68:161–181.

69.

Rajah

, Nachajon

, Collins

, Hakonarson

, Grunstein

and Cohen

. (1999). Elevated levels of the IGF-binding protein protease MMP-1 in asthmatic airway smooth muscle. Am J Respir Cell Mol Biol, 20:199–208.

70.

Fowlkes

, Thrailkill

, Serra

, Suzuki

and Nagase

. (1995). Matrix metalloproteinases as insulin-like growth factor binding protein-degrading proteinases. Prog Growth Factor Res, 6:255–263.

71.

, Müller

, Lamson

and Rosenfeld

. (1993). Insulin-like growth factor (IGF)-independent action of IGF-binding protein-3 in Hs578T human breast cancer cells. Cell surface binding and growth inhibition. J Biol Chem, 268:14964–14971.

72.

Hart

, Rusch

, Kaupp

, Nieselt

and Aicher

. (2017). Expression of desmoglein 2, desmocollin 3 and plakophilin 2 in placenta and bone marrow-derived mesenchymal stromal cells. Stem Cell Rev, 13:258–266.

73.

Klingemann

, Matzilevich

and Marchand

. (2008). Mesenchymal stem cells—sources and clinical applications. Transfus Med Hemother, 35:272–277.

74.

Billing

, Ben Hamidane

, Dib

, Cotton

, Bhagwat

, Kumar

, Hayat

, Yousri

, Goswami

, et al. (2016). Comprehensive transcriptomic and proteomic characterization of human mesenchymal stem cells reveals source specific cellular markers. Sci Rep, 6:21507.

75.

Johnson

, Najor

and Green

. (2014). Desmosomes: regulators of cellular signaling and adhesion in epidermal health and disease. Cold Spring Harb Perspect Med, 4:a015297.

76.

Huber

and Petersen

. (2015). 150th Anniversary Series: desmosomes and the Hallmarks of Cancer. Cell Commun Adhes, 22:15–28.

77.

Lin

, Biswas

, Choolani

, Fong

and Bongso

. (2017). Induction of immunogenic cell death in lymphoma cells by Wharton's jelly mesenchymal stem cell conditioned medium. Stem Cell Rev, 13:801–816.

78.

Ghazanfari

, Zacharaki

, Li

, Ching Lim

, Soneji

and Scheding

. (2017). Human primary bone marrow mesenchymal stromal cells and their in vitro progenies display distinct transcriptional profile signatures. Sci Rep, 7:10338.

79.

Bharti

, Shivakumar

, Park

, Ullah

, Subbarao

, Park

, Lee

, Park

and Rho

. (2018). Comparative analysis of human Wharton's jelly mesenchymal stem cells derived from different parts of the same umbilical cord. Cell Tissue Res, 372:51–65.

80.

Pollen

, Nowakowski

, Chen

, Retallack

, Sandoval-Espinosa

, Nicholas

, Shuga

, Liu

, Oldham

, et al. (2015). Molecular identity of human outer radial glia during cortical development. Cell, 163:55–67.

81.

Macosko

, Basu

, Satija

, Nemesh

, Shekhar

, Goldman

, Tirosh

, Bialas

, Kamitaki

, et al. (2015). Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell, 161:1202–1214.

82.

Klein

, Mazutis

, Akartuna

, Tallapragada

, Veres

, Li

, Peshkin

, Weitz

and Kirschner

. (2015). Droplet barcoding for single-cell transcriptomics applied to embryonic stem cells. Cell, 161:1187–1201.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.42 MB

0.40 MB

0.42 MB

0.58 MB

0.07 MB

0.14 MB

0.08 MB

0.21 MB

0.09 MB

0.19 MB

0.06 MB