Functional Module Analysis Reveals Differential Osteogenic and Stemness Potentials in Human Mesenchymal Stem Cells from Bone Marrow and Wharton's Jelly of Umbilical Cord

Abstract

Mesenchymal stem cells (MSCs) found in bone marrow (BM)-MSCs are an attractive source for the regeneration of damaged tissues. Alternative postnatal, perinatal, and fetal sources of MSCs are also under intensive investigation. MSCs from the Wharton's jelly matrix of umbilical cord (WJ)-MSCs have higher pancreatic and endothelial differentiation potentials than BM-MSCs, but the underlying mechanisms are poorly understood. We compared the gene expression profiles, enriched canonical pathways, and genetic networks of BM-MSCs and WJ-MSCs. WJ-MSCs express more angiogenesis- and growth-related genes including epidermal growth factor and FLT1, whereas BM-MSCs express more osteogenic genes such as RUNX2, DLX5, and NPR3. The gene expression pattern of BM-MSCs is more similar to osteoblasts than WJ-MSCs, suggesting a better osteogenic potential. In contrast, WJ-MSCs are more primitive because they share more common genes with embryonic stem cells. BM-MSCs are more sensitive to environmental stimulations because their molecular signatures altered more significantly in different culture conditions. WJ-MSCs express genes enriched in vascular endothelial growth factor and PI3K-NFκB canonical pathways, whereas BM-MSCs express genes involved in antigen presentation and chemokine/cytokine pathways. Drylab results could be verified by wetlab experiments, in which BM-MSCs were more efficient in osteogenic and adipogenic differentiation, whereas WJ-MSCs proliferated better. WJ-MSCs thus constitute a promising option for angiogenesis, whereas BM-MSCs in bone remodeling. Our results reveal systematically the underlying genes and regulatory networks of 2 MSCs from unique ontological and anatomical origins, as well as the resulted phenotypes, thereby providing a better basis for cell-based therapy and the following mechanistic studies on MSC biology.

Introduction

S omatic stem cells as well as embryonic stem cells (ESCs) have attracted much attention because of their unique biological behaviors and potential clinical usages. Mesenchymal stem cells (MSCs) found in many tissues are attractive somatic stem cell sources for the regeneration of damaged tissues. Although bone marrow (BM) is a primary source of multipotent MSCs, the procurement of BM is a highly invasive procedure and the number, differentiation potential, and, further, the frequency as well as their proliferative potential decline with increasing age [1]. Alternative sources from which MSCs can be isolated are therefore a subject of intensive investigation. Currently, MSCs can be isolated from a variety of tissues, including BM, umbilical cord blood, adipose tissue, and muscle [2 –4]. Another attractive alternative source of MSCs is the umbilical cord, from which MSCs can be obtained noninvasively [5,6].

The umbilical cord vessels and the surrounding mesenchyme [including the Wharton's jelly (WJ) connective tissue matrix] are derived from extraembryonic mesoderm and embryonic mesoderm. Several reports showed that MSCs with pluripotent differentiation potential can be isolated from the matrix part of WJ [5,6]. Postnatal stem cells derived from WJ can give rise to various lineages of cells, including adipocytes, chondrocytes, osteoblasts, and cardiomyocytes [6 –8]. In addition, fibroblast-like cells of WJ can be induced into neural-like cells expressing neuron-specific enolase, as well as other neural cell markers [5]. Several groups further showed that WJ-derived MSCs (WJ-MSCs) can be induced to dopaminergic neurons in vitro and have the potential for the treatment of Parkinson's disease [5,9,10] or spinal cord trauma [11]. WJ-MSCs are also capable of pancreatic and endothelial differentiation [12,13]. WJ-MSCs form larger islet-like cell clusters than BM-MSCs [12], and endothelial differentiated WJ-MSCs have significantly higher total tubule length, diameter, and area than differentiated BM-MSCs [13]. Also, WJ-MSCs have higher proliferative potential than BM-MSCs [13]. Although WJ-MSCs can differentiate into cells of the mesenchymal lineage [5,6], their osteogenesis ability has not been compared with that of BM-MSCs on a side-by-side manner.

Although the immunophenotypic profiles of MSCs from different tissues are very similar [2,5,6], the disparate characteristics of MSCs from different anatomic locations should be considered to better apply MSCs clinically, because the origin or source of MSCs will determine their fate and functional characteristics. Monitoring genome-wide characters offers an opportunity to discover unique property and underlying regulatory networks in stem cells and holds great promise for the future of regenerative medicine. Comparative analysis between MSCs derived from BM, umbilical cord blood, and adipose tissue has been conducted [4,14]. The gene expression profiles, as well as their comparisons, of various human MSCs have also been reported by quite a few studies [3,15 –18]. Differences in gene expression profiles and cellular function between human WJ-MSCs and BM-MSCs remain yet to be explored. A very recent stem cell marker analysis on long-term cultured MSCs showed that human WJ-MSCs express more ESC markers (such as NANOG, DNMT3B, and GABRB3) and some early endodermal markers at both early and late passages [19]. WJ-MSCs thereby possess properties of true stem cells, which they retain even after extended in vitro culturing.

In this study we applied systems biology tools to investigate the global transcriptome as well as the functional differences between BM-MSCs and WJ-MSCs and to provide the responsible genes and functional genetic networks. We show by functional module analysis why WJ-MSCs may differentiate into endothelial cells more efficiently, and drylab analyses further suggest BM-MSCs to be a better source for osteoblasts. The basic biological potentials (such as osteogenic differential potential) of different MSCs remain unchanged in 2 tested culture conditions, although the expression of certain genes may change in response to in vitro culture conditions. Drylab conclusions can be verified by wetlab differentiation assays. Our data thereby contribute new insights into a refined molecular picture and a better understanding of these 2 MSC subtypes. Results obtained will benefit cell-based therapy, as well as subsequent mechanistic studies in angiogenesis and osteogenesis.

Materials and Methods

Isolation and cultivation of human MSCs

This research follows the tenets of the Declaration of Helsinki and informed consent was obtained from the donor patients. All human MSCs used for experiment were cultured for <8 passages in the MesenCult™ medium (StemCell Technologies) in the presence of 5% CO₂, or in Dulbecco's modified Eagle's medium (DMEM; Cat. No. 12100-061; Gibco-BRL) supplemented with 10% fetal bovine serum (FBS; Cat. No. 12003; JRH Bioscience). For BM-MSC isolation, the mononuclear fraction of BM was isolated by Histopaque-1077 gradient (1077-1; Sigma) and plated in the culture medium. Nonadherent cells were discarded on next day. For array analysis, BM-MSCs were isolated from 2 male and 2 female donors with age under 40, and those used in reverse transcription polymerase chain reaction (RT-PCR) and differentiation experiments were isolated from both male and female donors with age 40–70.

MSCs from WJ were collected as published [9,20]. Briefly, umbilical cords (UCs) were processed within 24 h after delivery. WJ tissues were isolated from UCs and then digested with collagenase and trypsin. Cells were washed twice with phosphate-buffered saline (PBS) and plated on regular culture dishes in fresh culture medium; nonadherent cells were discarded on next day. Approximately 4 days to 1 week after explantation, fibroblast-like adherent cells were detached using trypsin–ethylenediaminetetraacetic acid solution and then reseeded in fresh culture medium without dilution for further expansion [passage 1 (P1)].

Array probe preparation and data processing

Osteoblast, adipocyte, and chondrocyte array data were downloaded from the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/; GSE9451). ESC array data were from our previous publication [21]. Total RNA collection, cRNA probe preparation, array hybridization, and data analysis were done as previously described [21,22]. Heatmap was created by the dChip software (http://biosun1.harvard.edu/complab/dchip/). Classical multidimensional scaling (MDS) was performed using the standard function of the R program to provide a visual impression of how the various sample groups are related. Gene annotation was performed by our ArrayFusion web tool (http://microarray.ym.edu.tw/tools/arrayfusion/) [23], and gene enrichment analysis was performed by the DAVID Bioinformatics Resources 2008 (http://david.abcc.ncifcrf.gov/) [24]. The average linkage distance was employed to assess similarity between 2 groups of gene expression profiles, as described previously [22]. The difference in distance between 2 groups of sample expression profiles to a third was assessed via the comparison of corresponding average linkage distances [the mean of all pairwise distances (linkages) between members of the 2 groups concerned]. The error on such a comparison was estimated by combining the standard errors (the standard deviation of pairwise linkages divided by the square root of the number of linkages) of the average linkage distances involved [21,22].

Genetic network analysis

Differential gene expression profiles were imported into the Ingenuity Pathways Analysis (IPA) software (Ingenuity Systems; www.ingenuity.com/) to obtain functional regulatory networks and canonical pathways. IPA uses a curated database to construct functional regulatory networks from a list of individual genes and determines a statistical score for each network according to the fit of the network to the set of focus genes. The score is the negative log of P and denotes the likelihood of the focus genes in the network being found together by chance. The biological functions assigned to each network are ranked according to the significance of that biological function to the network. A Fisher's exact test was used to calculate P values to determine the probability that the biological function assigned to a network is explained by chance alone. The significance/P value calculations were based on the hypergeometric distribution calculated via the right-tailed Fisher's exact test for 2 × 2 contingency tables. Right-tailed refers to the fact that IPA shows only overrepresented functional/pathway annotations.

Immunofluorescent assay and flow cytometry

For flow cytometry, 5 × 10⁵ MSCs in PBS were incubated with monoclonal antibodies of anti-CD29 (Cat. No. 303004; Biolegend), CD34 (Cat. No. MCA547PE; Serotec), CD44 (Cat. No. 312306; Biolegend), CD45 (Cat. No. 304006; Biolegend), CD73 (Cat. No. 550257; BD Pharmingen), or anti-CD90 (Cat. No. MCA90F; Serotec) on ice for 1 h. Unstained MSCs were used as a negative control.

RNA isolation and real-time PCR

Total mRNA was extracted using RNeasy mini kit (Cat. No. 74106; Qiagen) and 100 ng to 1 μg of total RNA was reverse transcribed using First cDNA Synthesis Kit (Cat. No. K1612; Fermentas). For quantitative real-time PCR analysis, human premessenger RNA sequence was obtained from the NCBI AceView program (www.ncbi.nlm.nih.gov/AceView/). All primers were designed cross introns in the Primer3 web site (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi/) or Primer Express software (Applied Biosystems), and thermodynamics and primer specificity analysis were performed by the Vector NTI suite (Invitrogen) and the NCBI reverse electronic PCR program (www.ncbi.nlm.nih.gov/sutils/e-pcr/reverse.cgi/). Real-time PCR reactions were performed using Maxima™ SYBR Green qPCR Master Mix (Cat. No. K0222; Fermentas), and specific products were detected and analyzed using a StepOne™ sequence detector (Applied Biosystems). The expression level of each gene was normalized to the expression level of glyceraldehyde 3-phosphate dehydrogenase. All the primer sequences are given in Supplementary Table S1 (available online at www.liebertonline.com/scd).

Proliferation assay

BM-MSCs and WJ-MSCs were started at passage 3 and plated at a density of 2 × 10³ cells/cm² in MesenCult medium. Cells were subcultured every 5 days and counted using a hemocytometer. Cumulative population doublings were calculated as published [25].

Multilineage differentiation assays

Osteogenic differentiation

MSCs were plated at 5 × 10³ cells/cm² in 6-well plates and treated with osteogenic medium for 4 weeks. Osteogenic medium consists of 0.1 μM dexamethasone, 10 mM β-glycerolphosphate, and 50 μM ascorbic acid in DMEM supplemented with 10% FBS (all chemicals were purchased from Sigma-Aldrich). Osteogenic differentiation was determined by alkaline phosphatase staining (Cat. No. 86R-1KT; Sigma). For evaluation of mineralized matrix, differentiated cells were treated with 2% silver nitrate (Cat. No. S6506; Sigma) under UV lamp for 1 h. Cells were then fixed with 2.5% sodium thiosulphate (Cat. No. S7026; Sigma) for 5 min and washed with water. Finally, cells were counterstained with 1% neutral red (Cat. No. N2880; Sigma) for 1 min.

Adipogenic differentiation

About 5 × 10³ MSCs/cm² were induced in adipocytic differentiation medium (1 μM dexamethasone, 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), 10 μg/mL insulin, and 100 μM indomethacin in DMEM with 10% FBS). The medium was changed twice a week. Adipogenic differentiation was determined by Oil Red O staining. The cells were fixed with 10% formalin in PBS for 5 min at room temperature. After washing with 60% isopropanol, the cells were stained by Oil Red O solution (Cat. No. O0625; Sigma) (0.2% Oil Red O in 60% isopropanol) for 10 min at room temperature. All differentiation experiments were repeated 3 times using 3 batches of MSCs from 3 different donors. Around 90% of BM-MSCs could be differentiated into adipogenic cells at week 3 in each batch of cells.

Immunoblotting

Proteins from total cell lysates were resolved by sodium dodecyl sulphate–polyacrylamide gel electrophoresis, transferred to polyvinylidene fluoride (PVDF) membrane, and then blocked with 5% nonfat milk in PBS/Tween-20 or 5% bovine serum albumin in Tris-buffered saline–Tween-20, according to the instructions for antibodies. For immunoblotting, the following antibodies were used: phosphor-NFκB-p65 (Ser536) (No. 3031; Cell Signaling), NFκB p65 (GTX107678; GeneTex), IκB-α (C-21) (sc-371; Santa Cruz), and glyceraldehyde 3-phosphate dehydrogenase (MAB374; Millipore). Band intensity was measured by the ImageQuant software.

Macrophage preparation and coculture with MSCs

Human CD14+ monocytes were purified from peripheral blood mononuclear cells and monocyte-derived macrophages were cultured as described previously [26]. About 3 × 10⁴ MSCs (upper inserts) and 4–5 × 10⁵ macrophages (lower chamber) isolated from unrelated donors were cocultured using 12-well Transwell plates (No. 3460; Corning) for 2 days, and 1 μg/mL E.coli lipopolysaccharide (LPS; Sigma) was then added to the lower chamber for activating macrophages. Sixteen hours later, tumour necrosis factor-α and interleukin 6 pro-inflammatory cytokines expressed by macrophages were analyzed by quantitative (q)RT-PCR.

Results

Characterization of MSCs from WJ or human BM

To access molecular signatures for MSCs from different sources, primary MSCs from human BM or umbilical cord WJ were collected. Both MSC types shared a similar fibroblastoid-shaped morphology (Fig. 1A). Flow cytometry analyses of cells less than passage 8 revealed that both BM-MSCs and WJ-MSCs expressed known MSC markers such as CD29 and CD73 (SH3/4), but were negative for hematopoietic CD14 marker and CD34 (Fig. 1B, C). Both MSCs were also positive for CD44 but negative for CD45 (not shown).

FIG. 1.

Characterization of 2 mesenchymal stem cell (MSC) subtypes. (A) Morphology of cultured bone marrow- (BM-) or Wharton's jelly (WJ)-MSCs. (B, C) Flow cytometry analyses of WJ-MSCs (B) and BM-MSCs (C). Cells were cultured for 5–8 passages and then labeled with CD29, CD73, CD14 and CD34 antibodies. Filled areas represent controls and green peaks represent the specific binding for indicated antigens. (D) Oil Red O staining shows the adipogenic differentiation of BM- and WJ-MSCs during differentiation. Different MSCs in the adipogenic induction show varying degrees of adipogenesis (staining positive by Oil Red O) at 3 weeks (wks) of culture. (E) qPCR analyses of adipocyte marker genes. Total RNA was analyzed by qRT-PCR. Mean expression levels of target genes, including PPARγ, FABP4, LPL and Leptin (LEP), were compared with that of glyceraldehyde 3-phosphate dehydrogenase control. Each bar represents a different individual. Results were expressed as the mean ± standard deviation. qPCR, quantitative polymerase chain reaction.

As a pilot study to test whether there is any functional difference between these 2 MSC subtypes, we tested their adipogenic potentials. Adipogenic differentiation was judged as Oil Red O-positive cells and was also analyzed by qPCR, in which the adipocytic marker genes PPARγ and FABP4 were expressed in the adipogenic formula-treated cells. Differentiated adipogenic cells could be observed in both MSCs at week 1, but lipid accumulation was faster in BM-MSCs (Fig. 1D). More differentiated BM-MSCs stained positive for Oil Red O at weeks 2 and 3 (Fig. 1D). Four examined adipogenic markers, PPARγ, FABP4, LPL, and Leptin, were also more abundantly expressed in BM-MSCs than in WJ-MSCs during differentiation (Fig. 1E).

Functional transcriptome analysis on different MSCs

To provide more insights into the functional differences of different MSCs, we performed transcriptome analysis on them. Genes differentially expressed between each cell type (the molecular signatures) were identified according to a statistical pipeline used [21,22]. There were 762 probe sets (corresponding to 454 unique genes) abundantly overexpressed in WJ-MSCs, whereas another 334 probe sets (175 genes) in BM-MSCs [with a positive false discovery rate (pFDR) threshold q < 0.001]. An MDS plot using these genes illustrated the grouping power of them (Fig. 2A). The top 50 known genes most strongly expressed in each MSC are listed in Table 1. In WJ-MSCs, known neurogenic genes such as somatostatin receptor 1 [27], NRXN3 (neurexin 3) [28], RTN1 (reticulon 1) [29] and spondin 2 (M-spondin) [30] are expressed by those 50 genes (Table 1A, underlined). Also in the list is natriuretic peptide precursor B (NPPB), which is expressed in the early embryonic nervous system and may play region- and stage-specific roles during the development of the peripheral and central nervous systems [31]. Genes involved in cell proliferation, including cyclin D2 (CCND2), epidermal growth factor (EGF), and FLT1 (Fms-related tyrosine kinase 1; vascular endothelial growth factor receptor), are also enriched in WJ-MSCs. ICAM1 (intercellular adhesion molecule 1; CD54), an endothelial marker involved in cell adhesion, is also overexpressed in WJ-MSCs (Table 1A, underlined). Genes involved in blood vessel development, such as FOXF1 and NPPB, are enriched in WJ-MSCs.

Table 1A.

Top 50 Known Genes in Wharton's Jelly-Mesenchymal Stem Cells

Probe set ID	UniGene ID	Gene title	Gene symbol	Location	Folds
235591_at	Hs.248160	Somatostatin receptor 1	SSTR1	chr14q13	149.603
209031_at	Hs.370510	Immunoglobulin superfamily, member 4	IGSF4	chr11q23.2	91.403
202274_at	Hs.516105	Actin, gamma 2, smooth muscle, enteric	ACTG2	chr2p13.1	85.825
203485_at	Hs.368626	Reticulon 1	RTN1	chr14q23.1	79.755
206801_at	Hs.219140	Natriuretic peptide precursor B	NPPB	chr1p36.2	63.558
209008_x_at	Hs.533782	Keratin 8	KRT8	chr12q13	53.848
217901_at	Hs.412597	Desmoglein 2	DSG2	chr18q12.1	51.387
206825_at	Hs.2820	Oxytocin receptor	OXTR	chr3p25	49.379
206032_at	Hs.41690	Desmocollin 3	DSC3	chr18q12.1	42.102
237206_at	Hs.567641	Myocardin	MYOCD	chr17p11.2	41.499
203434_s_at	Hs.307734	Membrane metalio-endopeptidase (neutral endopeptidase, CD10)	MME	chr3q25.1-q25.2	40.762
205433_at	Hs.420483	Butyrylcholinesterase	BCHE	chr3q26.1-q26.2	39.831
225864_at	Hs.124951	Family with sequence similarity 84, member B	FAM84B	chr8q24.21	37.770
1556329_a_at	Hs.192859	Protocadherin 10	PCDH10	chr4q28.3	35.919
201925_s_at	Hs.527653	CD55 molecule, decay accelerating factor for complement	CD55	chr1q32	34.856
229649_at	Hs.368307	Neurexin 3	NRXN3	chr14q31	33.708
37512_at	Hs.524513	Hydroxysteroid (17-beta) dehydrogenase 6	HSD17B6	chr12q13	31.797
203424_s_at	Hs.369982	Insulin-like growth factor binding protein 5	IGFBP5	chr2q33-q36	31.687
202638_s_at	Hs.515126	Intercellular adhesion molecule 1 (CD54), human rhinovirus receptor	ICAM1	chr19p13.3-p13.2	31.578
207717_s_at	Hs.164384	Plakophilin 2	PKP2	chr12p11	31.341
225720_at	Hs.480615	Synaptopodin 2	SYNPO2	chr4q26	31.197
212094_at	Hs.147492	Paternally expressed 10	PEG10	chr7q21	31.071
235228_at	Hs.117136	Coiled-coil domain containing 85A	CCDC85A	chr2p16.1	30.964
204363_at	Hs.62192	Coagulation factor III (thromboplastin, tissue factor)	F3	chr1p22-p21	30.291
200953_s_at	Hs.376071	Cyclin D2	CCND2	chr12p13	28.068
201596_x_at	Hs.406013	Keratin 18	KRT18	chr12q13	25.223
205935_at	Hs.155591	Forkhead box F1	FOXF1	chr16q24	23.643
204224_s_at	Hs.86724	GTP cyclohydrolase 1 (dopa-responsive dystonia)	GCH1	chr14q22.1-q22.2	23.466
213094_at	Hs.318894	G protein-coupled receptor 126	GPR126	chr6q24.1	22.824
204702_s_at	Hs.404741	Nuclear factor (erythroid-derived 2)-like 3	NFE2L3	chr7p15-p14	22.098
211538_s_at	Hs.432648	Heat shock 70-kDa protein 2	HSPA2	chr14q24.1	21.945
226497_s_at	Hs.507621	Fms-related tyrosine kinase 1 (vascular endothelial growth factor receptor)	FLT1	chr13q12	21.088
207426_s_at	Hs.181097	Tumor necrosis factor (ligand) superfamily, member 4	TNFSF4	chr1q25	20.043
211478_s_at	Hs.368912	Dipeptidyl-peptidase 4 (CD26, adenosine deaminase complexing protein 2)	DPP4	chr2q24.3	19.881
200636_s_at	Hs.272062	Protein tyrosine phosphatase, receptor type, F	PTPRF	chr1p34	18.001
218638_s_at	Hs.302963	Spondin 2, extracellular matrix protein	SPON2	chr4p16.3	17.468
218885_s_at	Hs.47099	UDP-N-acetyl-alpha-d-galactosamine N-acetylgalactosaminyltransferase 12	GALNT12	chr9q22.33	17.208
206254_at	Hs.419815	Epidermal growth factor (beta-urogastrone)	EGF	chr4q25	17.129
229400_at	Hs.123070	Homeobox D10	HOXD10	chr2q31.1	17.010
203889_at	Hs.156540	Secretogranin V (7B2 protein)	SCG5	chr15q13-q14	16.893
204072_s_at	Hs.591225	Furry homolog (Drosophila)	FRY	chr13q13.1	16.507
209723_at	Hs.104879	Serpin peptidase inhibitor, clade B (ovalbumin), member 9	SERPINB9	chr6p25	16.317
205534_at	Hs.570785	BH-protocadherin (brain-heart)	PCDH7	chr4p15	14.681
209369_at	Hs.480042	Annexin A3	ANXA3	chr4q13-q22	14.672
203642_s_at	Hs.470457	COBL-like 1	COBLL1	chr2q24.3	14.579
213256_at	Hs.132441	Membrane-associated ring finger (C3HC4) 3	MARCH3	chr5q23.2	14.512
223734_at	Hs.154140	Ovary-specific acidic protein	OSAP	chr4q31.1	14.354
235489_at	Hs.525389	ras homolog gene family, member J	RHOJ	chr14q23.2	13.785
203951_at	Hs.465929	Calponin 1, basic, smooth muscle	CNN1	chr19p13.2-p13.1	13.564
214974_x_at	Hs.89714	Chemokine (C-X-C motif ) ligand 5	CXCL5	chr4q12-q13	12.546

FIG. 2.

BM-MSCs express more genes involved in osteogenesis. (A) An MDS plot using differentially expressed genes (1097 probe sets, q < 0.001) between BM- and WJ-MSCs. Each spot represents a single array. (B) Predominant osteogenic potential of BM-MSCs cultured in MesenCult™ medium. This heat map shows BM-MSC-specific genes and their corresponding expression levels in the “skeletal development” Gene Ontology category (the P value is shown). Genes in red: increased expression; in blue: decreased. Underlined: conserved in different culture conditions. (C) Influence of cell culture conditions on the gene expression patterns of different MSCs. This MDS plot using the whole transcriptome shows the relationships between MSC subtypes cultured in MesenCult medium (M) or Dulbecco's modified Eagle's medium + 10% FBS (S). (D) Conserved osteogenic potential of BM-MSCs cultured in different culture media. (E) An MDS plot using the whole transcriptome shows the relationships between different MSCs, osteoblasts, and another 2 mesenchymal lineage cells. (F) Transcriptome distance analysis for different MSCs and osteoblasts. Average lineage distances between each transcriptome were calculated using whole transcriptome. FBS, fetal bovine serum; MDS, multidimensional scaling.

In BM-MSCs, known angiogenic genes (such as ANGPT1 and PGF) are present (Table 1B, underlined). CXCL12 (also known as stromal cell-derived factor-1, SDF1) is involved in the migration and trafficking of BM-MSCs, endothelial progenitor cells, and muscle progenitors [32,33]. It is known that migration of BM and cord blood MSCs in vitro is regulated by the SDF1-CXCR4 and the hepatocyte growth factor-c-Met axes [33]. PRRX1 (paired related homeobox 1; alias PHOX1) is a homeobox-containing protein that is expressed in the mesenchyme of facial, limb, and vertebral skeletal precursors during embryogenesis. PRRX1 is required for early events of skeletogenesis as well as in molar tooth morphogenesis [34,35]. Also enriched in BM-MSCs is NPR3 (natriuretic peptide receptor C or guanylate cyclase C), which is an important anabolic regulator of endochondral bone growth [36,37]. The presence of these known genes in our gene lists supports the reliability of our results.

Table 1B.

Top 50 Known Genes in Bone Marrow-Derived Mesenchymal Stem Cells

Probe set ID	UniGene ID	Gene title	Gene symbol	Location	Folds
209687_at	Hs.522891	Chemokine (C-X-C motif ) ligand 12 (stromal cell-derived factor 1)	CXCL12	chr10q11.1	72.421
213791_at	Hs.339831	Proenkephalin	PENK	chr8q23-q24	67.221
209160_at	Hs.78183	Aldo-keto reductase family 1, member C3	AKR1C3	chr10p15-p14	57.018
209392_at	Hs.190977	Ectonucleotide pyrophosphatase/phosphodiesterase 2 (autotaxin)	ENPP2	chr8q24.1	56.265
221019_s_at	Hs.464422	Collectin subfamily member 12/	COLEC12	chr18pter-p11.3	52.558
219789_at	Hs.237028	Natriuretic peptide receptor C/guanylate cyclase C	NPR3	chr5p14-p13	36.483
209312_x_at	Hs.534322	Major histocompatibility complex, class II, DR beta 1	HLA-DRB1	chr6p21.3	33.167
211990_at	Hs.347270	Major histocompatibility complex, class II, DP alpha 1	HLA-DPA1	chr6p21.3	29.788
205542_at	Hs.61635	Six transmembrane epithelial antigen of the prostate 1	STEAP1	chr7q21	29.007
208894_at	Hs.520048	Major histocompatibility complex, class II, DR alpha	HLA-DRA	chr6p21.3	28.492
233533_at	Hs.534499	Keratin associated protein 1–5	KRTAP1-5	chr17q12-q21	27.269
212226_s_at	Hs.405156	Phosphatidic acid phosphatase type 2B	PPAP2B	chr1pter-p22.1	25.121
201858_s_at	Hs.1908	Proteoglycan 1, secretory granule	PRG1	chr10q22.1	24.322
207191_s_at	Hs.513022	Immunoglobulin superfamily containing leucine-rich repeat	ISLR	chr15q23-q24	24.223
212188_at	Hs.109438	Potassium channel tetramerization domain containing 12	KCTD12	chr13q22.3	23.398
227242_s_at	Hs.591374	Early B-cell factor 3	EBF3	chr10q26.3	23.371
211653_x_at	Hs.567256	Aldo-keto reductase family 1, member C2	AKR1C2	chr10p15-p14	17.218
218625_at	Hs.103291	Neuritin 1	NRN1	chr6p25.1	17.198
214761_at	Hs.530930	Zinc finger protein 423	ZNF423	chr16q12	16.402
1569003_at	Hs.444569	Transmembrane protein 49	TMEM49	chr17q23.1	15.844
227300_at	Hs.449718	Transmembrane protein 119	TMEM119	chr12q24.11	15.536
238332_at	Hs.374774	Ankyrin repeat domain 29	ANKRD29	chr18q11.2	15.437
229357_at	Hs.58324	ADAM metallopeptidase wit thrombospondin type 1 motif, 5	ADAMTS5	chr21q21.3	14.834
209619_at	Hs.591258	CD74 molecule	CD74	chr5q32	12.663
201422_at	Hs.14623	Interferon, gamma-inducible protein 30	IFI30	chr19p13.1	12.175
205856_at	Hs.101307	Solute carrier family 14 (urea transporter), member 1	SLC14A1	chr18q11-q12	12.063
202435_s_at	Hs.154654	Cytochrome P450, family 1, subfamily B, polypeptide 1	CYP1B1	chr2p21	11.571
228347_at	Hs.539969	Sine oculis homeobox homolog 1 (Drosophila)	SIX1	chr14q23.1	11.531
207316_at	Hs.57697	Hyaluronan synthase 1	HAS1	chr19q13.4	11.465
242138_at	Hs.407015	Distal-less homeobox 1	DLX1	chr2q32	10.679
205991_s_at	Hs.632475	Paired related homeobox 1	PRRX1	chr1q24	10.568
203854_at	Hs.312485	Complement factor 1	CFI	chr4q25	10.033
205608_s_at	Hs.369675	Angiopoietin 1	ANGPT1	chr8q22.3-q23	9.838
220794_at	Hs.98206	Gremlin 2, cysteine knot superfamily, homolog (Xenopus laevis)	GREM2	chr1q43	9.296
205066_s_at	Hs.527295	Ectonucleotide pyrophosphatase/phosphodiesterase 1	ENPP1	chr6q22-q23	9.184
228407_at	Hs.12923	Signal peptide, CUB domain, EGF-like 3	SCUBE3	chr6p21.3	9.184
202855_s_at	Hs.500761	Solute carrier family 16, member 3 (monocarboxylic acid transporter 4)	SLC16A3	chr17q25	9.100
223315_at	Hs.201034	Netrin 4	NTN4	chr12q22-q23	9.100
228933_at	Hs.201623	Nance-Horan syndrome (congenital cataracts and dental anomalies)	NHS	chrXp22.13	9.068
223796_at	Hs.521495	Contactin associated protein-like 3	CNTNAP3	chr9p13.1	8.417
217478_s_at	Hs.77522	Major histocompatibility complex, class II, DM alpha	HLA-DMA	chr6p21.3	8.311
201752_s_at	Hs.501012	Adducin 3 (gamma)	ADD3	chr10q24.2-q24.3	8.306
220976_s_at	Hs.900145	Keratin-associated protein 1-1	KRTAP1-1	chr17q12-q21	8.287
206377_at	Hs.484423	Forkhead box F2	FOXF2	chr6p25.3	8.107
232204_at	Hs.308048	Early B-cell factor	EBF	chr5q34	8.009
218899_s_at	Hs.533446	Brain and acute leukemia, cytoplasmic	BAALC	chr8q22.3	7.963
219295_s_at	Hs.8944	Procollagen C-endopeptidase enhancer 2	PCOLCE2	chr3q21-q24	7.473
234880_x_at	Hs.534495	Keratin associated protein 1-3	KRTAP1-3	chr17q12-q21	7.181
213435_at	Hs.516617	SATB family member 2	SATB2	chr2q33	6.964
209652_s_at	Hs.252820	Placental growth factor, vascular endothelial growth factor related	PGF	chr14q24-q31	6.876

To gain more insights into the functional consequences from differential gene expression and to provide quantitative evidence, signature genes were subjected to the Gene Ontology (GO) database search [38] to find statistically overrepresented functional groups within gene lists. The DAVID 2008 web-based tool, a graph theory evidence-based method to agglomerate gene/protein identifiers [24], was applied for this task. A P value was assigned for each category based on the EASE cooccurrence scoring system, a Fisher's exact-based method that identifies functional categories overrepresented in a gene list [24]. Given that the whole human transcriptome was represented in a microarray, this analysis was not biased toward the coverage of the microarray. We addressed whether there were genes enriched in the differentiation categories according to the GO database and found BM-MSCs significantly enriched with genes involved in skeletal development and osteoblast differentiation (P = 0.00012; Fig. 2B). Fourteen BM-MSC genes (ALX1, ANKH, DLX5, ENPP1, FOXC1, EST1, EST2, NPR3, POSTN, RUNX2, SIX1, SOX9, SRGN, and TOB1) were present in this category (Fig. 2B).

Conserved biological properties, but not gene expression patterns, in MSCs cultured in different conditions

The above gene profiling data suggest differential biological potentials of WJ- and BM-MSCs cultured in the defined MesenCult medium. As culture conditions may affect cell properties dramatically and many laboratories culture their MSCs in DMEM supplemented with 10% FBS, we tested whether different culture media may alter the biological potentials of MSCs. BM- and WJ-MSCs cultured in DMEM+ 10% FBS for less than 6 passages were collected, and their gene expression patterns were again analyzed. When the whole transcriptome of each cell type was applied for MDS relationship analysis, we found that different culture conditions caused only minor changes in the gene expression patterns of WJ-MSCs (Fig. 2C). In contrast, BM-MSCs were more sensitive to environment cues as their molecular patterns were significantly altered by varied cellular conditions (Fig. 2C).

We then explored the possible functional consequences of altered gene expression patterns. A total of 984 probe sets were relatively abundant in WJ-MSCs cultured in 10% FBS, whereas another 808 probe sets were found in BM-MSCs (q < 0.001). These probe sets were again subjected to the GO database analysis to find statistically enriched functional groups. Genes involved in skeletal development were still predominant in BM-MSCs (P = 0.00026; Fig. 2D). DLX5, FOXC1, NPR3 and SIX1 were present in BM-MSCs cultured in both conditions (Fig. 2B, D, underlined).

The overexpression of osteogenic genes in BM-MSCs indicates they may possess a more similar transcriptome pattern to that of osteoblast. We then included osteoblasts, adipocytes, and chondrocytes array data in the MDS analysis. BM-MSCs cultured in either MesenCult medium (Fig. 2E) or DMEM + 10% FBS (not shown) showed a closer relationship to osteoblasts. To provide more quantitative evidences, we calculated the distances between osteoblasts and different MSC subtypes by employing an average linkage distance analysis to assess the similarity between 2 groups of gene expression profiles as described previously [21]. The distance between osteoblasts and BM-MSCs was smaller than that between osteoblasts and WJ-MSCs (Fig. 2F), supporting a superior osteogenic potential of BM-MSCs.

Pluripotent marker analysis on perinatal and adult MSCs

UC-MSCs were shown to have higher proliferative and colony-forming unit-fibroblast-forming ability than BM-MSCs in 2-dimentional monolayer cultures, and perinatal MSCs were suggested to be more primitive than adult ones [39]. To analyze whether this hypothesis could be reflected on a genomic level, we compared WJ- and BM-MSC transcriptomes with that of ESC. As shown in Fig. 3A, WJ-MSCs and ESCs shared a closer relationship, and such impression could be verified by average linkage distance analysis (Fig. 3B). Unsupervised hierarchical clustering also grouped WJ-MSCs and ESCs in the same leaf, while BM-MSCs and all 3 other mesenchymal cells in another (Fig. 3C).

FIG. 3.

Comparative analyses of BM-MSCs, WJ-MSCs, and ESCs. (A) An MDS plot using the differentially expressed genes between BM- and WJ-MSCs (q < 0.001) shows the relationships between BM-, WJ-MSCs and ESCs. (B) Linkage distance analysis shows the transcriptome relationships of BM- and WJ-MSCs to ESCs. Distance were calculated using genes differentially expressed between different MSCs. (C) A dendrogram plot of different MSCs, ESC, and differentiated mesenchymal cells. (D, E) A Venn diagram shows that there are 137 genes highly expressed in both WJ-MSCs and ESCs (D), and these genes are represented by a heat map (E). (F, G) Proliferative potentials of WJ-MSCs and BM-MSCs. Total cell numbers of cultured WJ-MSCs and BM-MSCs were counted every 5 days till day 20 after plating (F), and from these values, the cumulative population doublings were calculated (G). Experiments were performed in triplicate and bar represents mean ± standard deviation. *P < 0.05. ESCs, embryonic stem cells.

We then investigated genes shared between ESCs and WJ-MSCs and checked the possible outcomes of the preferential ESC gene pattern in WJ-MSCs. A total of 137 genes were found commonly in both cell types when compared with BM-MSCs (Fig. 3D). Enriched among those 137 genes were those involved in cell differentiation (P = 0.0000284), positive regulation of cell proliferation (P = 0.0244), insulin receptor signaling pathway (P = 0.00188), and growth (P = 0.0494) (Fig. 3E). Also enriched in these 137 genes are those involved in antiapoptosis (FOXO1, SOCS2, IGF1R, SNCA, and SERPINB9; P = 0.029), cell adhesion (KAL1, PTPRF, LAMA1, CXADR, JUP, CADM1, DDR1, CYFIP2, DSC2, DSG2, CDK6, PCDH7, MLLT4, and F11R; P = 0.0024), and striated muscle development (CUGBP2, CXADR, CSRP2, and MYEF2; P = 0.032) (not shown).

The above analysis suggested a more multipotent status and preferential growth advantage of WJ-MSCs over BM-MSCs. We examined this hypothesis by culturing cells for 15 days. Cell number was counted every 5 days, and WJ-MSCs were found to proliferate faster than BM-MSCs (Fig. 3F). After day 5, WJ-MSCs increased their rate of growth relative to BM-MSCs, which corresponded to a significant increase in cumulative population doublings (Fig. 3G). Throughout the assay period, both cells types maintained similar cell morphology (ie, polygonal and fibroblastoid phenotypes) (data not shown).

Activated canonical pathways in different MSCs

It is likely that MSCs of different origins express different genes by utilizing unique canonical pathways. We then employed an additional tool, the IPA software, for investigating the biological pathways that may be differentially activated between 2 MSCs. The database behind IPA contains more canonical pathways culled from the literature. A 1-tailed Fisher's exact test was applied in IPA to determine the probability if the pathway assigned to the dataset is explained by chance alone (see Materials and Methods section). The canonical pathways associated to each dataset are ranked according to the significance of that pathway to the queried dataset.

Supplementary Fig. S1A, B (available online at www.liebertonline.com/scd) summarized the statistically significant themes that were identified by the IPA. Genes involved in IGF-1, PI3K/AKT, and ERK5 signaling pathways were more abundant in WJ-MSCs (Supplementary Fig. S1A). Cardiovascular system-related nitric oxide signaling pathway were more active in WJ-MSCs. CAMK4, PRKAR2B, ITPR2, PRKCD, FLT1, PRKAR1B, NOS3, and PRKG1 were higher in WJ-MSCs (Supplementary Fig. S1A). Axonal guidance genes (including ACTR2, NTF3, ITSN1, NRP2, ARHGEF7, EGF, NGFB, ROBO1, DDR1, MYL9, MICAL1, PRKAR2B, PRKCD, PRKAR1B, and ADAM19) were enriched in WJ-MSCs, correlating with the strong neurogenic differentiation ability of this MSC type [5].

In BM-MSCs, more genes were involved in immune response, such as antigen presentation pathway (the top 1 enriched pathway) and chemokine, IL4/IL9/IL22, and GM-CSF signaling pathways (Supplementary Fig. S1B). Antigen presentation genes, including HLA-DMA, HLA-DRB4, HLA-DRA, HLA-DRB1, CD74, and HLA-DPA1, were abundant in BM-MSCs (Table 1B). Ephrin receptor and BMP signaling pathways were also active (Supplementary Fig. S1B). In both WJ- and BM-MSCs, the ILK signaling pathway was active (Supplementary Fig. S1).

Genetic networks of different MSCs

Increasing evidence showed that genes do not act as individuals but collaborates in genetic networks. To better understand how genes enriched in different stem cells are related and, most importantly, how those interactions are related to cellular function, we performed genetic network analysis for signature genes. Filtrated probe sets were inputted into the IPA software to construct network modules. The knowledge base behind IPA summarized known molecular interactions evidenced in published literature. The term “network” in IPA is not the same as a biological or canonical pathway with a distinct function (ie, angiogenesis), but rather is a reflection of all interactions of a given protein as defined in the literature.

In WJ-MSC a major network consisting 137 genes was identified (Fig. 4A). This network includes several growth factors or cytokines: CLEC11A, CXCL2, CXCL5, EGF, HBEGF, NGFB, and NTF3. Among those genes, EGF and HBEGF can enhance the migration of human and rabbit MSCs [40], and EGF has been also reported to stimulate the differentiation of human MSCs into bone-forming cells [41]. This network also revealed genes with significant biological roles in WJ-MSCs. Some genes have higher connectivity to others (regarded as “hubs”) or resided in a position among submodules in the major network (“bridge” genes) (Fig. 4A, panels 1–5). Deregulation of hub or bridge genes may eventually lead to the disruption of the genetic network and the malfunction of cells [42]. ICAM1, NFKB1 (NFκB), NGFB (nerve growth factor, beta polypeptide), and EGF are bridges connecting different submodules in the major network component (Fig. 4A, panels 1–4). Central to the network there are significant hubs, including ACTG2, CDK6, CCND2, CDKN1A, COL18A1, FOXO1A, FOXO3A, IGFBP5, KSR1, MYH9, NOS3, NFKBIA, PRKCD, and SERPINE1 (Fig. 4A, panel 5). Of these hub genes, CDK6, CCND2, COL18A1, IGFBP5, and SERPINE1 are all involved in enhancing cell growth. Two forkhead transcription factors (FOXO1A and FOXO3A) involved in brain development [43] are also present.

FIG. 4.

Interaction network analysis as a framework for the interpretation of MSC biology. (A) A functional genetic network composed of multiple WJ-MSC genes. This network is displayed graphically as nodes (gene products) and edges (biological relationships between nodes) mapped by the Ingenuity™ Pathway Analysis tool. The intensity of the node color indicates the degree of upregulation. Panels 1–5 are selected core regions of this WJ-MSC gene network, highlighting several hub genes. (B) A functional genetic network composed of BM-MSC genes. Key hub genes are indicated with arrows. Nodes are displayed using various shapes that represent the functional class of the gene product (right panel).

When a similar functional network analysis was conducted on BM-MSC genes, a major network consisting 57 genes was identified (Fig. 4B). Extracellular proteins ANGPT1, CXCL12, DCN, ELN, IL28A, KLK4, LGALS3, MMP16, PCOLCE, PENK, PGF, PLAU, and POSTN are present in this network. HLADRA, CREB1, and RUNX2 link submodules to the main network (Fig. 4B). Also crucial in this network (ie, crucial in BM-MSCs) are transcription factors STAT1, STAT5B, CDC42, and cytokine CXCL12 (Fig. 4B).

Verification of in silico analysis results

The differential expression of genes involved in these functional groups was examined by qRT-PCR and immunoblotting in Fig. 5. Cell proliferation genes such as FLT1 and EGF were more abundant in WJ-MSCs (Fig. 5A). The dominant expression of TEK, a PI3K-AKT pathway gene, in WJ-MSCs could be verified by qPCR (Fig. 5B). As NFκB is one of the major downstream effectors of PI3K signaling, we examined by immunoblotting the levels of NFκB and IκB-α proteins in different MSCs. Clearly, WJ-MSCs expressed less IκB-α but possessed more active form p65 NFκB (phosphorylated at S536) (Fig. 5B). Genes involved in neurogenesis, such as RTN1, NPPB, and NRP2, were also more abundant in WJ-MSCs (Fig. 5C).

FIG. 5.

Validation of array data by qRT-PCR (A–E) and immunoblotting (B). Functional or signaling groups containing these genes are also shown. Genes related to cell proliferation (A), PI3K-NFkB signaling pathway (B), and neurogenesis (C), which include FLT1, EGF, TEK, RTN1, NPPB, and NRP2, were up-regulated in WJ-MSCs than in BM-MSCs. DLX5 and NPR3 involved in skeleton development (D) and immune response-related genes, HLA-DRA and CXCL12 (E), were highly expressed in BM-MSCs. Folds of relative band intensity in immunoblotting are indicated (B). (F) Immunosuppressive effect of different MSCs. MSCs were cocultured with macrophages differentiated from peripheral blood mononuclear cell CD14+ monocytes as illustrated. Interleukin 6 and tumor necrosis factor-α mRNA levels in macrophages were examined by qRT-PCR.

In BM-MSCs, genes involved in skeletal development were more abundant (Fig. 5D), consistent with drylab analysis data (Fig. 2). Also abundant in BM-MSCs were genes involved in immune response, such as CXCL12 and HLADRA (Fig. 5E), indicating WJ-MSCs maybe a better choice in allogeneic transplantation. The superior allogeneic transplantation potential of WJ-MSCs was verified by examining their immunosuppressive effects over allogenic macrophages. WJ-MSCs prevent cocultured macrophages from producing inflammatory cytokines tumor necrosis factor-α and IL6 more significantly (Fig. 5F).

Osteogenic differentiation potentials of WJ- and BM-MSCs

Genomics analyses suggested different osteogenesis characters of different MSCs as well as the involved genes. We examined drylab data by wetlab experiments. The osteogenic differentiation potentials of both MSCs cultured in either defined medium or DMEM + 10% FBS were tested by culturing the cells in osteogenic medium. Under osteogenic induction, BM-MSCs cultured in the defined MesenCult medium expressed higher alkaline phosphatase (ALP) activity than WJ-MSCs (Fig. 6A, B). ALP was expressed in more than 95% of BM-MSCs at week 3 but in only few WJ-MSCs even at week 4 (Fig. 6B). Mineralization of differentiated cells was further tested by von Kossa staining, which showed a greater extent of mineralization with more detectable bone nodules in differentiated BM-MSCs (Fig. 6C). Bone nodule could be observed clearly at week 3 in differentiating BM-MSCs but could not be observed even at week 4 in differentiated WJ-MSCs (Fig. 6C). The expression of 2 other osteogenic markers [osteopontin (SPP1) and RUNX2] was examined by qPCR. As shown in Fig. 6D, these marker genes were more abundantly expressed in BM-MSCs than in WJ-MSCs early during osteogenic differentiation. Similar results were found when MSCs were cultured in 10% FBS: ALP, osteopontin, and RUNX2 were expressed more dominantly in differentiated BM-MSCs (Fig. 6E). ALP staining data from osteogenic differentiated MSCs also showed that when MSCs were cultured in 10% FBS, BM-MSCs still held a better osteogenic ability (Supplementary Fig. S2, available online at www.liebertonline.com/scd).

FIG. 6.

Different osteogenic capacities of WJ-MSCs and BM-MSCs. (A, B) Alkaline phosphatase (ALP) staining of differentiated MSCs at indicated differentiation time points. Representative images of osteogenic cultures stained for ALP were shown (B). (C) Mineralization abilities of differentiated cells by von Kossa staining. Bone nodules were typically observed in BM-MSCs, but not in WJ-MSCs, cultured under osteogenic conditions by week 4. (D, E) Validation of osteogenic genes, including SPP1, RUNX2 and ALP, by qRT-PCR. MSCs cultured in MesenCult (D) or 10% fetal bovine serum–Dulbecco's modified Eagle's medium (E) were induced osteogenesis and total RNA was collected at indicated time points. Three osteocyte markers were analyzed.

Discussion

Stem cells have drawn immense research interest because they hold great promise for future therapeutic applications. Although much progress has been made in stem cell biology, the clinical application of these cells is stymied by an inadequate or insufficient understanding of the functional integration of complex interactive genetic processes. Further characterization of these stem cells via a genomics approach will be critical for a better understanding of the molecular events involved in stem cell properties as well as the utilization of these cells. The differential gene expression patterns was indicated by immunophenotype analysis that unlike BM-MSCs, small populations of WJ-MSCs express endoglin (SH2, CD105) and CD49e at passage 8 [10]. Here we identify global transcriptome differences and functional variations between WJ- and BM-MSCs, and some of the functions revealed may correlate to clinical applications. We confirm drylab data by wetlab experiments and found that MSCs from BM are a better source for osteogenesis, whereas WJ-MSCs are more primitive and more similar to ESCs. These findings shed light on the myth of different MSCs, provide the underlying mechanisms, and will contribute in cell-based therapy.

The gene expression profiles, as well as their comparisons, of various human MSCs have been reported by quite a few groups [3,4,14 –18]. Nevertheless, interpreting genomics data into biological significance is always an overwhelming task. Analyzing gene expression data by assessing the biological significance of filtrated gene sets, rather than individual genes, has become a main approach in microarray data analysis and this has promisingly derived new biological interpretations of microarray data. In principle, this approach measures the gene–phenotype correlations of every gene in a predefined gene set, usually according to the GO [38] categories or the sets of genes representing biological pathways in the cell, and then give a score to represent its changes associated with the phenotype. The DAVID database we used and the Gene set enrichment analysis [44], for example, are among the most mentioned and used gene set-based approaches. In addition to dividing filtrated genes into functional subgroups, systems biology approaches (such as connecting signature genes into networks) also reveal genetic modules inside cells. In systems biology the overall molecular events are viewed as a biological system for a given biological process, where we can find important hub and betweenness controller genes, which could not be readily extracted from 1-dimensional gene list analysis [21,45,46]. By applying these 2 approaches (functional grouping and genetic network construction), we successfully generated hypotheses from bioinformatics analyses. For example, we suggested that BM-MSCs may hold a better osteogenic potential, whereas WJ-MSCs may proliferate quicker. These hypotheses could be verified by the following wetlab experiments (Figs. 3F, G and 6).

The functional comparison between WJ- and BM-MSCs has been performed by several groups [12,13]. However, the underlying mechanisms and responsible genes are yet unclear. The greater proliferative potential of WJ-MSCs may be due to the abundant expression of telomerase [5] or embryonic stem cell (hES) genes such as NANOG, DNMT3B, and GABRB3 [19]. Additionally, genes (including GAB2, NFKBIA, FOXO1A, YWHAB, FOXO3A, CDKN1A, MAP3K8, NFKB1, NOS3, DDR1 and TEK) involved in the phosphoinositide 3-kinase (PI3K)-AKT survival/proliferation pathway are also higher in WJ-MSCs (Supplementary Fig. S1, Figs. 4A and 5A, B). NFκB is also more active in WJ-MSCs (Fig. 5B), which partly explains why WJ-MSCs divide faster than those from human BM (Fig. 3F, G). One of several effector arms of PI3K-AKT signaling are the highly related members of the mammalian FoxO transcription factor family, FoxO1, FoxO3, and FoxO4 [47]. FOXO1 (FOXO1A) and FOXO3A are both key hub genes maintaining the stability of the whole genetic network in WJ-MSCs (Fig. 4A). RUNX2, a transcription factor crucial in osteogenesis, is a hub gene in BM-MSCs (Fig. 4B), suggesting a dominant osteogenic ability of BM-MSCs. We could confirm this by wetlab experiments (Fig. 6). Our transcriptome analysis on these 2 MSC subtypes therefore provides a basic roadmap for future in-depth exploration of the molecular mechanisms governing multilineage differentiation and stemness of MSCs.

On top of providing molecular mechanisms for reported phenotypes, our functional genomics analyses further provide new hypotheses. For example, WJ-MSCs might induce less rejection response in allogeneic transplantation because they express less antigen presentation genes (eg, HLA-DRA; Supplementary Fig. S1 and Fig. 5E) and suppress more host inflammatory cytokines (Fig. 5F). HLA-DRA is a member of MHC class II and is involved in immune recognition. Lower expression of HLA-DRA in WJ-MSCs may decrease the risk of transplantation rejection. Another hypothesis is that WJ-MSCs, although still can, may possess less osteogenic ability than BM-MSCs. We proved this point by wetlab osteogenic differentiation experiments (Fig. 6). A very recent study suggests that WJ-MSCs predominantly expresses genes engaged in osteogenic differentiation, and the tendency of WJ-MSCs to differentiate into osteoblast is similar to BM-MSCs [48]. However, a side-by-side functional comparison between WJ-MSCs and BM-MSCs was not performed. It was therefore unclear whether WJ-MSCs shares the same, or less, osteogenic potential than BM-MSCs. Our study thus provides the first parallel mesenchymal lineage differentiation analyses for these 2 MSC subtypes. WJ-MSCs and BM-MSCs have been compared for their cartilage tissue engineering potential in a 3-dimentional culture scaffold: type II collagen staining was more intense for hBM-MSCs than hUC-MSCs, suggesting a worse chondrogenic differentiation potential of WJ-MSCs [49]. As to adipogenic differentiation, Karahuseyinoglu et al. compared the adipogenesis abilities of WJ-MSCs and BM-MSCs and found that WJ-MSC-derived adipocytes were less mature compared with those from BM-MSCs [50]. This conclusion is consistent with our observations (Fig. 1D, E). WJ-MSCs may, thus, constitute a promising option for angiogenic or neural cells, whereas BM-MSCs in bone remodeling in regenerative medicine.

Several reports seem to report contradicting results that MSCs purified from human umbilical cord exhibited better osteogenic potential than BM-MSCs [39,51]. However, the MSCs they purified were actually from the perivascular portion of the umbilical cord, but not the WJ part. We used WJ-MSCs in comparative genomics research, and data obtained therefore are different. Several reports have shown that, except from the matrix part of WJ, MSCs or somatic stem cells with pluripotent differentiation potentials can also be isolated from the perivascular portion of the umbilical cord [52], or from the vasculature endothelial surface of the umbilical cord [53,54] (reviewed by Troyer et al. [55]). It has been shown that MSCs from WJ possess the least ability of osteogenesis when compared with those from umbilical veins or arteries [20]. Moreover, in the absence of osteogenic induction, MSCs from arteries also expressed osteocytic marker alkaline phosphatase [20]. Additionally, MSCs from WJ and perivascular regions have different marker expressions. For examples, pancytokeratin (an ecto/endodermal marker) and α-smooth muscle actin are only expressed in hUCPVCs, which are thought to be more differentiated myofibroblasts, whereas cells in WJ are more immature [50,56].

Using a knowledge-based database, only 29.1% of WJ-MSC genes and 32.6% of BM-MSC genes was found to be involved in networks, respectively. This is due to the fact that we applied a knowledge-based strategy to construct genetic networks, so that only genes with known interactions will be included. Stemness and differentiation are most complex processes governed by a highly coordinated regulation of distinct genetic programs, and biological function of genes enriched in different MSCs remains mostly unknown. The genetic networks revealed are therefore still incomplete. More comprehensive, integrated studies enabling the determination of all interactions will offer additional insights into how such a complex interaction map may contribute to unique stem cell behaviors. The global genetic networks among stemness genes are therefore needed to be further constructed via bioinformatics algorithms or data-driven approaches [21].

An improved understanding of MSCs of different sources can help the applications of these cells and the control of their differentiation directions in vivo, thereby furthering their potentials in clinical applications. Studying genes or pathways differentially expressed in various MSCs may eventually decipher the mechanisms of neurogenesis, neuroplasticity, angiogenesis, and osteogenesis. Thus, this study provides novel strategies for further functional or mechanistic studies on stem cells. The approaches taken here can also serve as a model for the comparison of other complex biological systems, or the study of the function of a given gene.

Footnotes

Acknowledgments

The authors thank Dr. Szu-Ting Chen from Prof. Shie-Liang Hsieh's laboratory for providing macrophages, and Dr. Tom K. Kuo for his critical reading of the manuscript. The authors acknowledge the technical services provided by the Microarray and Gene Expression Analysis Core Facility of the National Yang-Ming University VGH Genome Research Center. The Gene Expression Analysis Core Facility was supported by the National Research Program for Genomic Medicine, National Science Council (NSC). This work was supported by the NSC (NSC97-3111-B-010-004 and NSC98-2320-B-010-020-MY3), the Yen Tjing Lin Medical Foundation (CI-96-11), the National Health Research Institutes (NHRI-EX97-9704BI), the Mackay Memorial Hospital (MMH-HB-98-04), Department of Health (Cancer Excellence Center Plan), and a grant from the Yang-Ming University (Ministry of Education, Aim for the Top University Plan).

Author Disclosure Statement

No competing financial interests exist.

References

Mueller

, Glowacki

. 2001. Age-related decline in the osteogenic potential of human bone marrow cells cultured in three-dimensional collagen sponges. J Cell Biochem, 82:583–590.

Lee

, Kuo

, Chen

, Lee

, Hsieh

, Chen

. 2004. Isolation of multipotent mesenchymal stem cells from umbilical cord blood. Blood, 103:1669–1675.

Panepucci

, Siufi

, Silva

Jr. , Proto-Siquiera

, Neder

, Orellana

, Rocha

, Covas

, Zago

. 2004. Comparison of gene expression of umbilical cord vein and bone marrow-derived mesenchymal stem cells. Stem Cells, 22:1263–1278.

Chang

, Shih

, Tseng

, Hsieh

, Lee

, Hwang

. 2006. Disparate mesenchyme-lineage tendencies in mesenchymal stem cells from human bone marrow and umbilical cord blood. Stem Cells, 24:679–685.

Mitchell

, Weiss

, Mitchell

, Martin

, Davis

, Morales

, Helwig

, Beerenstrauch

, Abou-Easa

, Hildreth

, Troyer

, Medicetty

. 2003. Matrix cells from Wharton's jelly form neurons and glia. Stem Cells, 21:50–60.

Wang

, Hung

, Peng

, Huang

, Wei

, Guo

, Fu

, Lai

, Chen

. 2004. Mesenchymal stem cells in the Wharton's jelly of the human umbilical cord. Stem Cells, 22:1330–1337.

Kadner

, Hoerstrup

, Tracy

, Breymann

, Maurus

, Melnitchouk

, Kadner

, Zund

, Turina

. 2002. Human umbilical cord cells: a new cell source for cardiovascular tissue engineering. Ann Thorac Surg, 74:S1422–S1428.

Kadner

, Zund

, Maurus

, Breymann

, Yakarisik

, Kadner

, Turina

, Hoerstrup

. 2004. Human umbilical cord cells for cardiovascular tissue engineering: a comparative study. Eur J Cardiothorac Surg, 25:635–641.

, Cheng

, Lin

, Cheng

, Chu

, Chou

, Shih

, Ko

, Sung

. 2006. Conversion of human umbilical cord mesenchymal stem cells in Wharton's jelly to dopaminergic neurons in vitro: potential therapeutic application for Parkinsonism. Stem Cells, 24:115–124.

10.

Weiss

, Medicetty

, Bledsoe

, Rachakatla

, Choi

, Merchav

, Luo

, Rao

, Velagaleti

, Troyer

. 2006. Human umbilical cord matrix stem cells: preliminary characterization and effect of transplantation in a rodent model of Parkinson's disease. Stem Cells, 24:781–792.

11.

Yang

, Shih

, Ko

, Hsu

, Cheng

, Fu

. 2008. Transplantation of human umbilical mesenchymal stem cells from Wharton's jelly after complete transection of the rat spinal cord. PLoS ONE, 3:e3336.

12.

, Wang

, Liu

, Wei

. 2009. Differentiation of Wharton's jelly primitive stromal cells into insulin-producing cells in comparison with bone marrow mesenchymal stem cells. Tissue Eng Part A, 15:2865–2873.

13.

Chen

, Lie

, Li

, Wei

. 2009. Endothelial differentiation of Wharton's jelly-derived mesenchymal stem cells in comparison with bone marrow-derived mesenchymal stem cells. Exp Hematol, 37:629–640.

14.

Kern

, Eichler

, Stoeve

, Kluter

, Bieback

. 2006. Comparative analysis of mesenchymal stem cells from bone marrow, umbilical cord blood or adipose tissue. Stem Cells, 24:1294–1301.

15.

Wagner

, Wein

, Seckinger

, Frankhauser

, Wirkner

, Krause

, Blake

, Schwager

, Eckstein

, Ansorge

, Ho

. 2005. Comparative characteristics of mesenchymal stem cells from human bone marrow, adipose tissue, and umbilical cord blood. Exp Hematol, 33:1402–1416.

16.

Song

, Webb

, Song

, Tuan

. 2006. Identification and functional analysis of candidate genes regulating mesenchymal stem cell self-renewal and multipotency. Stem Cells, 24:1707–1718.

17.

Wagner

, Feldmann

Jr. , Seckinger

, Maurer

, Wein

, Blake

, Krause

, Kalenka

, Burgers

, Saffrich

, Wuchter

, Kuschinsky

, Ho

. 2006. The heterogeneity of human mesenchymal stem cell preparations—evidence from simultaneous analysis of proteomes and transcriptomes. Exp Hematol, 34:536–548.

18.

Tsai

, Hwang

, Chen

, Lee

, Hsu

, Chang

, Wang

, Peng

, Chang

, Chao

, Chang

, Lee

, Wang

, Soong

. 2007. Functional network analysis of the transcriptomes of mesenchymal stem cells derived from amniotic fluid, amniotic membrane, cord blood, and bone marrow. Stem Cells, 25:2511–2523.

19.

, Nekanti

, Rao

, Bahirvani

, Jan

, Totey

. 2009. Long-term expansion and pluripotent marker array analysis of Wharton's jelly-derived mesenchymal stem cells. Stem Cells Dev, 19:117–130.

20.

Ishige

, Nagamura-Inoue

, Honda

, Harnprasopwat

, Kido

, Sugimoto

, Nakauchi

, Tojo

. 2009. Comparison of mesenchymal stem cells derived from arterial, venous, and Wharton's jelly explants of human umbilical cord. Int J Hematol, 90:261–269.

21.

Huang

, Hsieh

, Wu

, Jen

, Tsuang

, Chiou

, Partanen

, Anderson

, Jaatinen

, Yu

, Wang

. 2008. Functional network reconstruction reveals somatic stemness genetic maps and dedifferentiation-like transcriptome reprogramming induced by GATA2. Stem Cells, 26:1186–1201.

22.

Wang

, Trotter

, Lagos

, Bourboulia

, Henderson

, Makinen

, Elliman

, Flanagan

, Alitalo

, Boshoff

. 2004. Kaposi sarcoma herpesvirus-induced cellular reprogramming contributes to the lymphatic endothelial gene expression in Kaposi sarcoma. Nat Genet, 36:687–693.

23.

Yang

, Chang

, Lin

, Hsu

, Wang

. 2006. ArrayFusion: a web application for multi-dimensional analysis of CGH, SNP and microarray data. Bioinformatics, 22:2697–2698.

24.

Dennis

Jr. , Sherman

, Hosack

, Yang

, Gao

, Lane

, Lempicki

. 2003. DAVID: Database for annotation, visualization, and integrated discovery. Genome Biol, 4:P3.

25.

Cristofalo

, Allen

, Pignolo

, Martin

, Beck

. 1998. Relationship between donor age and the replicative lifespan of human cells in culture: a reevaluation. Proc Natl Acad Sci USA, 95:10614–10619.

26.

Chang

, Hsu

, Lin

, Chio

, Chiu

, Chen

, Lin

, Hsieh

. 2004. Modulation of macrophage differentiation and activation by decoy receptor 3. J Leukoc Biol, 75:486–494.

27.

Binaschi

, Bregola

, Simonato

. 2003. On the role of somatostatin in seizure control: clues from the hippocampus. Rev Neurosci, 14:285–301.

28.

Graf

, Zhang

, Jin

, Linhoff

, Craig

. 2004. Neurexins induce differentiation of GABA and glutamate postsynaptic specializations via neuroligins. Cell, 119:1013–1026.

29.

, Lu

, Qahwash

, Hu

, Chang

, Yan

. 2004. Reticulon family members modulate BACE1 activity and amyloid-beta peptide generation. Nat Med, 10:959–965.

30.

Adams

, Tucker

. 2000. The thrombospondin type 1 repeat (TSR) superfamily: diverse proteins with related roles in neuronal development. Dev Dyn, 218:280–299.

31.

DiCicco-Bloom

, Lelievre

, Zhou

, Rodriguez

, Tam

, Waschek

. 2004. Embryonic expression and multifunctional actions of the natriuretic peptides and receptors in the developing nervous system. Dev Biol, 271:161–175.

32.

Yamaguchi

, Kusano

, Masuo

, Kawamoto

, Silver

, Murasawa

, Bosch-Marce

, Masuda

, Losordo

, Isner

, Asahara

. 2003. Stromal cell-derived factor-1 effects on ex vivo expanded endothelial progenitor cell recruitment for ischemic neovascularization. Circulation, 107:1322–1328.

33.

Son

, Marquez-Curtis

, Kucia

, Wysoczynski

, Turner

, Ratajczak

, Janowska-Wieczorek

. 2006. Migration of bone marrow and cord blood mesenchymal stem cells in vitro is regulated by stromal-derived factor-1-CXCR4 and hepatocyte growth factor-c-met axes and involves matrix metalloproteinases. Stem Cells, 24:1254–1264.

34.

Martin

, Bradley

, Olson

. 1995. The paired-like homeo box gene MHox is required for early events of skeletogenesis in multiple lineages. Genes Dev, 9:1237–1249.

35.

Mitchell

, Hicklin

, Doughty

, Hicklin

, Dickert

Jr. , Tolbert

, Peterkova

, Kern

. 2006. The Prx1 homeobox gene is critical for molar tooth morphogenesis. J Dent Res, 85:888–893.

36.

Jaubert

, Jaubert

, Martin

, Washburn

, Lee

, Eicher

, Guenet

. 1999. Three new allelic mouse mutations that cause skeletal overgrowth involve the natriuretic peptide receptor C gene (Npr3) Proc Natl Acad Sci USA, 96:10278–10283.

37.

Agoston

, Khan

, James

, Gillespie

, Serra

, Stanton

, Beier

. 2007. C-type natriuretic peptide regulates endochondral bone growth through p38 MAP kinase-dependent and -independent pathways. BMC Dev Biol, 7:18.

38.

Harris

, Clark

, Ireland

, Lomax

, Ashburner

, Foulger

, Eilbeck

, Lewis

, Marshall

, Mungall

, Richter

, Rubin

, Blake

, Bult

, Dolan

, Drabkin

, Eppig

, Hill

, Ni

, Ringwald

, Balakrishnan

, Cherry

, Christie

, Costanzo

, Dwight

, Engel

, Fisk

, Hirschman

, Hong

, Nash

, Sethuraman

, Theesfeld

, Botstein

, Dolinski

, Feierbach

, Berardini

, Mundodi

, Rhee

, Apweiler

, Barrell

, Camon

, Dimmer

, Lee

, Chisholm

, Gaudet

, Kibbe

, Kishore

, Schwarz

, Sternberg

, Gwinn

, Hannick

, Wortman

, Berriman

, Wood

, de la Cruz

, Tonellato

, Jaiswal

, Seigfried

, White

. 2004. The Gene Ontology (GO) database and informatics resource. Nucleic Acids Res, 32:D258–D261.

39.

Zhang

, Teoh

, Chong

, Schantz

, Fisk

, Choolani

, Chan

. 2009. Superior osteogenic capacity for bone tissue engineering of fetal compared with perinatal and adult mesenchymal stem cells. Stem Cells, 27:126–137.

40.

Ozaki

, Nishimura

, Sekiya

, Suehiro

, Kanawa

, Nikawa

, Hamada

, Kato

. 2007. Comprehensive analysis of chemotactic factors for bone marrow mesenchymal stem cells. Stem Cells Dev, 16:119–129.

41.

Kratchmarova

, Blagoev

, Haack-Sorensen

, Kassem

, Mann

. 2005. Mechanism of divergent growth factor effects in mesenchymal stem cell differentiation. Science, 308:1472–1477.

42.

Hanahan

, Weinberg

. 2000. The hallmarks of cancer. Cell, 100:57–70.

43.

Choe

, Horvath

, Cloughesy

, Crosby

, Seligson

, Palotie

, Inge

, Smith

, Sawyers

, Mischel

. 2003. Analysis of the phosphatidylinositol 3′-kinase signaling pathway in glioblastoma patients in vivo. Cancer Res, 63:2742–2746.

44.

Subramanian

, Tamayo

, Mootha

, Mukherjee

, Ebert

, Gillette

, Paulovich

, Pomeroy

, Golub

, Lander

, Mesirov

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

45.

Barabasi

, Oltvai

. 2004. Network biology: understanding the cell's functional organization. Nat Rev Genet, 5:101–113.

46.

Janes

, Yaffe

. 2006. Data-driven modelling of signal-transduction networks. Nat Rev Mol Cell Biol, 7:820–828.

47.

Paik

, Kollipara

, Chu

, Ji

, Xiao

, Ding

, Miao

, Tothova

, Horner

, Carrasco

, Jiang

, Gilliland

, Chin

, Wong

, Castrillon

, DePinho

. 2007. FoxOs are lineage-restricted redundant tumor suppressors and regulate endothelial cell homeostasis. Cell, 128:309–323.

48.

Ciavarella

, Dammacco

, De Matteo

, Loverro

, Silvestris

. 2009. Umbilical cord mesenchymal stem cells: role of regulatory genes in their differentiation to osteoblasts. Stem Cells Dev, 18:1211–1220.

49.

Wang

, Tran

, Seshareddy

, Weiss

, Detamore

. 2009. A comparison of human bone marrow-derived mesenchymal stem cells and human umbilical cord-derived mesenchymal stromal cells for cartilage tissue engineering. Tissue Eng Part A, 15:2259–2266.

50.

Karahuseyinoglu

, Cinar

, Kilic

, Kara

, Akay

, Demiralp

, Tukun

, Uckan

, Can

. 2007. Biology of stem cells in human umbilical cord stroma: in situ and in vitro surveys. Stem Cells, 25:319–331.

51.

Baksh

, Yao

, 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.

52.

Sarugaser

, Lickorish

, Baksh

, Hosseini

, Davies

. 2005. Human umbilical cord perivascular (HUCPV) cells: a source of mesenchymal progenitors. Stem Cells, 23:220–229.

53.

Romanov

, Svintsitskaya

, Smirnov

. 2003. Searching for alternative sources of postnatal human mesenchymal stem cells: candidate MSC-like cells from umbilical cord. Stem Cells, 21:105–110.

54.

Kadivar

, Khatami

, Mortazavi

, Shokrgozar

, Taghikhani

, Soleimani

. 2006. In vitro cardiomyogenic potential of human umbilical vein-derived mesenchymal stem cells. Biochem Biophys Res Commun, 340:639–647.

55.

Troyer

, Weiss

. 2008. Wharton's jelly-derived cells are a primitive stromal cell population. Stem Cells, 26:591–599.

56.

Nanaev

, Kohnen

, Milovanov

, Domogatsky

, Kaufmann

. 1997. Stromal differentiation and architecture of the human umbilical cord. Placenta, 18:53–64.

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

0.32 MB