CpG Methylation Profiles of Endothelial Cell-Specific Gene Promoter Regions in Adipose Tissue Stem Cells Suggest Limited Differentiation Potential Toward the Endothelial Cell Lineage


  • Andrew C. Boquest,

    1. Institute of Basic Medical Sciences, Faculty of Medicine, Department of Biochemistry, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Agate Noer,

    1. Institute of Basic Medical Sciences, Faculty of Medicine, Department of Biochemistry, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Anita L. Sørensen,

    1. Institute of Basic Medical Sciences, Faculty of Medicine, Department of Biochemistry, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Kristin Vekterud,

    1. Institute of Basic Medical Sciences, Faculty of Medicine, Department of Biochemistry, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Philippe Collas Ph.D.

    Corresponding author
    1. Institute of Basic Medical Sciences, Faculty of Medicine, Department of Biochemistry, University of Oslo, Oslo, Norway
    • Institute of Basic Medical Sciences, Department of Biochemistry, University of Oslo, PO Box 1112 Blindern, 0317 Oslo, Norway. Telephone: 47-22851066; Fax: +47-22851058
    Search for more papers by this author


In vivo endothelial commitment of adipose stem cells (ASCs) has scarcely been reported, and controversy remains on the contribution of ASCs to vascularization. We address the epigenetic commitment of ASCs to the endothelial lineage. We report a bisulfite sequencing analysis of CpG methylation in the promoters of two endothelial-cell-specific genes, CD31 and CD144, in freshly isolated and in cultures of ASCs before and after induction of endothelial differentiation. In contrast to adipose tissue-derived endothelial (CD31+) cells, freshly isolated ASCs display a heavily methylated CD31 promoter and a mosaically methylated CD144 promoter despite basal transcription of both genes. Methylation state of both promoters remains globally stable upon culture. Endothelial stimulation of ASCs in methylcellulose elicits phenotypic changes, marginal upregulation of CD31, and CD144 expression and restrictive induction of a CD31+CD144+ immunophenotype. These events are accompanied by discrete changes in CpG methylation in CD31 and CD144 promoters; however, no global demethylation that marks CD31+ cells and human umbilical vein endothelial cells occurs. Immunoselection of CD31+ cells after endothelial stimulation reveals consistent demethylation of one CpG immediately 3′ of the transcription start site of the CD31 promoter. Adipogenic or osteogenic differentiation maintains CD31 and CD144 methylation patterns of undifferentiated cells. Methylation profiles of CD31 and CD144 promoters suggest a limited commitment of ASCs to the endothelial lineage. This contrasts with the reported hypomethylation of adipogenic promoters, which reflects a propensity of ASCs toward adipogenic differentiation. Analysis of CpG methylation at lineage-specific promoters provides a robust assessment of epigenetic commitment of stem cells to a specific lineage.

Disclosure of potential conflicts of interest is found at the end of this article.


Stromal stem cells found in mesenchymal tissues have received intense research interest due to their perceived potential application in regenerative medicine [1, 2]. Mesenchymal stem cells (MSCs) seem to be primarily restricted to form only mesoderm-specific cell types such as adipocytes, osteoblasts, chondrocytes, and myocytes. However, rare MSC subsets identified in bone marrow have been shown to form tissues of all three germ layers and have challenged the perceived restrictive nature of somatic stem cells [3].

Adipose tissue constitutes a rich and easily accessible source of MSCs [2, 4]. Human adipose stem cells (ASCs) with a CD34+CD105+CD45CD31 immunophenotype can be obtained in large quantities and with high purity (∼99%) from the stromal vascular fraction of liposuction material [5]. ASCs display a gene expression profile and surface antigen phenotype similar to MSCs from bone marrow [5, [6]–7]. ASCs exhibit primarily mesodermal differentiation capacities in vitro and can promote osteogenic [8], hepatic [9], or neuronal functions [10] and reconstitution of the immune system [11] in vivo. The transcriptome of ASCs reveals expression of genes extending across the three germ layers, suggestive of a differentiation potential toward nonmesodermal lineages [5]. Nevertheless, whether ASCs can form functional tissues of these lineages in vivo remains to be fully elucidated.

ASCs may also possess the ability to differentiate toward the endothelial cell (EC) lineage. Culture of mouse or human ASCs in low serum-containing medium supplemented with insulin-like growth factor 1 and vascular endothelial growth factor [12, [13]–14] or in a methylcellulose matrix [15, 16] promotes the development of a cell network that expresses EC markers. Freshly isolated undifferentiated ASCs express basal levels of transcripts (but not proteins) for EC markers including CD144 (also called vascular endothelium cadherin or CDH5) and CD31 (also called platelet endothelial cell adhesion molecule-1 or PECAM1) [5], suggesting a potential for EC differentiation. Subsets of these cells also express the EC marker von Willebrand factor (vWF) [5]. Human ASCs can also release angiogenic factors [14, 17, [18], [19]–20], a property enhanced under hypoxia [14]. Following transplantation in vivo, ASCs promote revascularization of ischemic tissue [12, 13, 15, 20, [21]–22]. However, how MSCs achieve this remains debated [23]. Contribution of transplanted ASCs to vasculature repair may be direct [13, 15, 21] or mediated through paracrine mechanisms [14, 20]. An in-depth understanding of EC differentiation at the molecular level is needed to shed light on the potential of ASCs to form functional ECs.

Although gene expression profiles and signaling pathways underlying stromal stem cell properties are being unraveled [6, 24], little is known on the role of epigenetics in MSC function. Epigenetic modifications of DNA and histones largely contribute to regulation of gene expression [25]. Methylation of the 5 position of a cytosine in a CpG dinucleotide is a heritable modification that favors genomic integrity, ensures correct regulation of transcription, and is involved in long-term gene silencing [26, 27]. Interestingly, DNA methylation regulates the activity of the endothelial nitric-oxide synthase promoter by modulating binding of transcriptional regulators in endothelial and nonendothelial cells [28]. Alterations in DNA methylation may also accompany precursor or stem cell differentiation, but only sporadic evidence has been reported [29, [30]–31]. Recent data reveal mosaic DNA methylation at selected promoters in intestinal crypt stem cells [32, 33] and precursor cells from adipose tissue [34], and a propensity of adipogenic promoters to be hypomethylated in undifferentiated ASCs [35].

This study addresses the epigenetic commitment of ASCs, at the DNA methylation level, to EC differentiation. We report a bisulfite sequencing analysis of CpG methylation of EC-specific promoters in freshly isolated uncultured ASCs and in undifferentiated and EC-differentiated polyclonal and clonal ASC cultures. In contrast to adipogenic promoters [35], the CD31 and CD144 promoters display hypermethylation in uncultured ASCs. Despite limited upregulation of CD31 and CD144 expression after endothelial stimulation, the maintenance of a largely methylated state of these promoters reflects a limited commitment of ASCs to the EC lineage in vitro.

Materials and Methods

Adipose Stem Cell Isolation and Clonal Culture

Cells with a CD34+CD105+CD45CD31 (ASCs) and with a CD34+CD105+CD45CD31+ (ECs) phenotype were isolated from the stromal vascular fraction of human adipose tissue as described earlier [5]. In short, subcutaneous lipoaspirates obtained from the hip/thigh regions of female donors after informed consent were washed in Hanks' balanced saline solution and digested for 2 hours with 0.2% collagenase and 10 ng/ml DNase I, and buoyant adipocytes were separated from stromal vascular cells after centrifugation at 400g. After lysis of erythrocytes and sedimentation, stromal cell pellets were suspended in Hanks' balanced saline solution containing 2% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) and strained through 100-μm and 40-μm cell sieves. CD45+ cells were removed with paramagnetic beads conjugated to anti-human CD45 monoclonal antibodies (Miltenyi Biotec, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com) using a SuperMACS magnet (Miltenyi Biotech). Remaining CD45 cells were incubated with fluorescein isothiocyanate (FITC)-conjugated anti-human CD31 antibodies (Serotec Ltd., Oxford, U.K., http://www.serotec.com; 10 μl per 106 cells) for 15 minutes at 4°C. Cells were washed and incubated with anti-FITC MicroBeads (Miltenyi Biotech) for 15 minutes at 4°C. CD31 and CD31+ cells were separated using an LS Column (Miltenyi Biotech). CD31 cells were re-exposed to a new LS Column to eliminate remaining CD31+ cells. Flow cytometry analysis of each cell subset showed a purity ≥98% (see Results). Aliquots of CD31 and CD31+ cells were cultured (CD31 cells only) or snap-frozen in liquid nitrogen for DNA and RNA isolation. No positive selection for CD34 and CD105 surface expression was carried out; we have previously reported that CD45CD31 cells isolated as described here are CD34+CD105+ [5].

Clones of ASCs were generated as described [5]. Single freshly isolated CD31 cells were cultured in each well of 48-well plates in Dulbecco's modified Eagle's medium (DMEM)/F12 medium containing 50% FBS and antibiotics. After ∼16 hours, the medium was replaced by DMEM/F12/20% FBS. After 3 weeks, colonies containing over 100 cells were passaged by trypsinization and expanded. Typically, 5%–10% of single-plated cells gave rise to expandable cell lines. Up to six ASC clonal lines derived from two donors were analyzed in this study. Clones A1, A2, and A3 were from one donor, and clones B1, B2, and B3 were from another.

Endothelial Cell Differentiation

EC differentiation was performed as described [15]. Cells cultured in DMEM/F12/20% FBS were trypsinized and plated at 2 × 105 cells per milliliter in 3 ml of methylcellulose (MethoCult GF H4434; Stem Cell Technologies, Vancouver, BC, Canada, http://www.stemcell.com). Cells were cultured at 37°C, 5% CO2 in air for 7 days. Cells were immunolabeled using FITC-conjugated anti-human CD31 antibodies (Serotec) for surface protein expression assessment by flow cytometry.

Adipogenic and Osteogenic Differentiation

For adipogenic differentiation, cells were cultured to confluency in DMEM/F12/10% FBS and stimulated for 3 weeks with 0.5 mM 1-methyl-3 isobutylxanthine, 1 μM dexamethasone, 10 μg/ml insulin (Novo Nordisk, Copenhagen, Denmark, http://www.novonordisk.com), and 200 μM indomethacin (Dumex-Alpharma, Copenhagen, Denmark, http://www.alpharma.com) [5]. To visualize lipid droplets, fixed cells were stained with oil red O. For osteogenic differentiation, cells were cultured in DMEM/F12/10% FBS containing 100 nM dexamethasone, 10 mM β-glycerophosphate, and 50 μM l-ascorbate-2-phosphate for 3 weeks [5]. Mineralization of the extract cellular matrix was visualized with Alizarin red.

Quantitative Reverse Transcription-Polymerase Chain Reaction

Reverse transcription-polymerase chain reaction (RT-PCR) was carried from 500 ng of total RNA using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Quantitative RT-PCRs were performed in triplicates on a MyiQ Real-Time PCR Detection System using iQ SYBR Green (Bio-Rad). RT-PCR primers used are listed in Table 1. SYBR Green polymerase chain reaction (PCR) conditions were 95°C for 4.5 minutes and 40 cycles of 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds using GAPDH as a normalization control.

Table Table 1.. Real-time reverse transcription-polymerase chain reaction primers used in this study
original image

Bisulfite Sequencing Analysis

We analyzed genes encoding two endothelial markers, CD31 and CD144 [36, [37]–38]. The CD31 promoter region examined [39] encompassed nucleotides 1,095–1,480 (GenBank X96848) and included 18 potentially methylated cytosines in CpG dinucleotides within nucleotides −352 to +34 relative to the ATG translation initiation side. The CD144 promoter region examined [40] was from nucleotides 39,011–39,303 (GenBank AC132186) and spanned 15 CpGs starting 12,414 base pairs upstream of the ATG (due to initiation of translation in exon 2 of the CD144 gene). Putative binding sites for known transcriptional regulators in the CD31 and CD144 promoter regions examined were identified using the Transcription Elements Search System software (http://www.cbil.upenn.edu/cgi-bin/tess/tess; supplemental online Fig. 1). The functional relevance of the regions examined was illustrated by their differential methylation patterns in ECs and in CD31 cells (see Results).

DNA was purified by phenol-chloroform-isoamylalcohol extraction. Cells were lysed in 10 mM Tris-HCl, pH 8, 100 mM EDTA, and 0.5% SDS and digested with 0.1 mg/ml proteinase K. Bisulfite conversion was performed using the MethylEasy DNA Bisulfite Modification Kit (Human Genetic Signatures, Sydney, Australia, http://www.geneticsignatures.com). Converted DNA was amplified by seminested PCR using CD31- and CD144-specific primers available from Human Genetic Signatures [35]. Conditions were, for each PCR, 95°C for 3 minutes and 30 cycles of 95°C for 1 minute, 50°C for 2 minutes, and 72°C for 2 minutes followed by 10 minutes at 72°C. PCR products were directly sequenced or cloned into sequencing vectors using the TOPO TA Cloning kit (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and sequenced (MWG Biotech, Ebersberg, Germany, http://www.mwg-biotech.com).

Methylation data are represented as filled circles (methylated CpG) and empty circles (unmethylated CpG) for each bacterial clone obtained. Each row of circles represents the methylation pattern obtained from the sequence of one PCR product (i.e., one genomic allele) in a bacterial clone. Percentages of global CpG methylation in a given promoter in indicated sample groups were performed using paired t tests. Extent of methylation of a given CpG between two cell populations or donors was compared using a paired t test. Significance was expressed using two-tailed p values.


CD31+ and CD31 Cells from Adipose Tissue Express the EC Marker Genes CD31 and CD144

We previously reported the isolation of fresh ASCs with a CD31 immunophenotype from the stromal vascular fraction of human lipoaspirates [5]. We now show that CD31 cells can be consistently isolated with a high (>98%) degree of purity by two rounds of negative selection against the CD31 surface antigen (Fig. 1A). The EC-specific marker CD31 was expressed at high levels in adipose tissue-derived ECs (CD31+ cells) from three donors as judged by endpoint RT-PCR (Fig. 1B) and by real-time RT-PCR analysis (Fig. 1C). CD31 was also expressed in CD31 cells of each donor (Fig. 1B), albeit at a much lower level, despite the lack of CD31 antigens on the cell surface. EC-specific CD144 transcripts were also detected both in CD31+ and CD31 cells (Fig. 1B) but also at a reduced level in CD31 cells (Fig. 1C). Despite transcriptional activation of CD144, the CD144 antigen was not detected on the surface of CD31 or CD31+ cells (data not shown). Furthermore, variations in CD31 and CD144 expression levels in CD31+ versus CD31 cells were detected between donors (Fig. 1C; p < .001; paired t tests). Thus, the EC marker genes CD31 and CD144 are transcribed in presumably a low proportion of ASCs (CD31 cells) as well as in CD31+ cells. The data also extend earlier observations that ASCs may express lineage-specific mRNAs despite the absence of protein expression [5].

Figure Figure 1..

Expression of CD31 and CD144 in CD31 adipose stem cells and CD31+ endothelial cells isolated from the stromal vascular fraction of lipoaspirates. (A): Flow cytometry analysis of CD31 protein expression in CD31 and CD31+ cells (shown here for donor 1) after separation. Similar profiles were obtained for the two other donors (not shown). (B): Reverse transcription-polymerase chain reaction (RT-PCR) analysis of mean ± SD expression of CD31 and CD144 in CD31+ and CD31 cells from three donors. GAPDH expression is shown as loading control. (C): Real-time RT-PCR analysis of CD31 and CD144 expression in CD31+ cells relative to CD31 cells (level 1). Asterisk, p < .001 between donors (t test).

Differential CpG Methylation at the CD31 and CD144 Promoters in CD31 and CD31+ Cells

To extend our previous analysis of DNA methylation in ASCs [35], we determined by bisulfite sequencing the methylation profile in the promoter regions of CD31 and CD144 in uncultured CD31 and CD31+ cells (Fig. 2A; supplemental online Fig. 1). Among CD31 cells, CD31 was overall highly methylated in donors 1 and 2 (85.0% ± 5.13% vs. 87.8% ± 3.45% methylation, respectively; p = .588; paired t test; Fig. 2B). Regions covering CpGs numbers 14–18 displayed mosaicism between and within donors (Fig. 2B). Only two CpGs (numbers 5 and 18) were differentially methylated between donors 1 and 2 (p < .01 for each; paired t test; Fig. 2B). The CD144 promoter also displayed similar methylation profiles between donors in CD31 cells but was less methylated than CD31 (Fig. 2B). Furthermore, whereas the first 8 CpGs in CD144 were uniformly unmethylated, CpGs numbers 9–15 were more, and mosaically, methylated between and within donors (Fig. 2B). Mosaicism in DNA methylation was consistent with that reported for adipogenic promoters in ASCs [35] and may be attributed to stochastic methylation errors resulting from replication or aging [41].

Figure Figure 2..

CpG methylation in the CD31 and CD144 promoters in freshly isolated CD31 and CD31+ cells. (A): CpG dinucleotide-containing regions examined in the human CD31 and CD144 promoters. Numbers refer to nucleotide number in relation to the ATG translation initiation site. CpG dinucleotides are underlined. The CD31 transcription start site is shown at position −204. (B, C): Bisulfite analysis of CD31 and CD144 methylation in (B) CD31 and (C) CD31+ cells from two donors. Percentages of methylated CpGs (•) are shown, and CpG numbers are indicated, with CpG number 1 being the 5′-most CpG examined in the sequence. (D): Proportions of individual CpG methylation in the CD31 (left) and CD144 (right) promoters in CD31 and CD31+ cells for donors 1 and 2. (E): Bisulfite analysis of CD31 and CD144 promoter methylation in passage-2 HUVECs. Percentages of methylated CpGs are shown. Abbreviations: bp, base pairs; D1, donor 1; D2, donor 2; HUVEC, human umbilical vein endothelial cell; TSS, transcription start site.

In CD31+ cells, the CD31 promoter displayed two methylation patterns, with ∼50% of the alleles (or cells, see below) being nearly fully methylated in donor 1, whereas ∼50% were completely unmethylated (Fig. 2C, 2D). A similar trend was observed for donor 2 despite greater mosaicism and higher global methylation than donor 1 (50.9% ± 2.2% vs. 67.8% ± 3.3% methylation, respectively; p = .0004; paired t test; Fig. 2C, 2D). Overall, CD31 was clearly hypermethylated in CD31 relative to CD31+ cells both for donor 1 (85.0% ± 5.1% vs. 50.9% ± 5.2% methylation, respectively; p < .0001; paired t test) and donor 2 (87.8% ± 3.5% vs. 67.8% ± 3.3% methylation, respectively; p < .0001; paired t test). The CD144 promoter retained a methylation profile similar to that of CD31 cells in 40% of the alleles in donor 1, whereas the rest of the alleles were completely unmethylated or largely methylated (Fig. 2C, 2D). Donor 2 was characterized by complete demethylation of CD144 in ∼80% of the alleles (Fig. 2C, 2D) and, overall, CD144 was hypomethylated in donor 2 relative to donor 1 (41.3% ± 4.6% vs. 19.3% ± 1.8% methylation, respectively; p = .0001; paired t test).

These results indicate that the CD31 promoter region examined is strongly methylated in ASCs but displays complete demethylation in ∼50% of alleles or cells among the CD31+ cell population. Because bisulfite sequencing does not allow the distinction between single cells and single alleles, we cannot at present state whether the CD31 promoter is hemimethylated in CD31+ cells or whether 50% of the CD31+ cells harbor a fully unmethylated CD31 promoter. CD144 displays a consistent methylation profile within and between donors in ASCs but exhibits mosaicism in CD31+ cells. This mosaicism may reflect the presence of several subpopulations in the CD31+ stromal cell pool or a minor contamination of the CD31+ cell population with CD31 cells. Of note, the CpG methylation profile of the CD31 promoter in CD31+ cells was similar to that of cultured passage-2 human umbilical ECs (human umbilical vein endothelial cells [HUVECs]) (Fig. 2E). The CD144 promoter, however, displayed a methylation pattern distinct from CD31+ cells (apart from the unmethylation of 40% of the alleles detected in both cell types; Fig. 2E). We have previously shown that CD31+ cells from adipose tissue expressed EC marker genes (including CD31 and vWF) and major histocompatibility complex class II [5]. On that basis, together with the methylation profile of CD31 and CD144, we argue that CD31+ cells isolated from adipose tissue stroma are of endothelial type.

Culture of ASCs Maintains CpG Methylation Profiles in CD31 and CD144 Promoters

Bisulfite analysis of freshly isolated ASCs from two donors and of a pool of two cultured ASC clones at passages 4–5 revealed overall stability of global methylation in the CD31 and CD144 promoters upon clonal culture (Fig. 3A). No single CpG in either promoter was found to be affected by culture (p > .20 for any cytosine examined; paired t tests; Fig. 3A, 3B). To support these observations, analysis of two polyclonal cultures of ASCs from two separate donors confirmed the hypermethylated state of the CD31 promoter region examined (Fig. 3C). These data corroborate our previous observations on the stability of CpG methylation of adipogenic promoters upon clonal culture of ASCs [35].

Figure Figure 3..

CpG methylation profiles in the CD31 and the CD144 promoter in uncultured and cultured adipose stem cells (ASCs). (A):CD31 promoter, (B):CD144 promoter. Mean ± SEM methylation of individual CpGs in freshly isolated ASCs (uncultured, purified from two donors) and in two undifferentiated cultured ASC clones. (C): Bisulfite sequencing analysis of CD31 promoter methylation in two polyclonal cultures of ASCs.

EC Differentiation of Clonal Cultures of ASCs

Single cell clonal cultures of ASCs seeded on methylcellulose formed within 7 days interconnected and elongated cells characteristic of ECs (Fig. 4A). Both CD31 and CD144 transcripts were upregulated upon methylcellulose culture, albeit to variable levels (Fig. 4B). CD31 was upregulated 4–10-fold relative to nonstimulated cells, whereas a 3–30-fold increase in CD144 mRNA was detected depending on the cell clone (Fig. 4B). Flow cytometry analysis indicated that a small proportion of the cells expressed CD31 (4.4%–6.1%) and CD144 (10.8%–29.1%) upon methylcellulose culture (Fig. 4C; gates were set using an FITC-conjugated irrelevant control antibody). Similar results were obtained with polyclonal cultures of ASCs (data not shown). Thus, methylcellulose culture of ASCs causes morphological alterations suggestive of induction of EC differentiation. Of note, however, transcriptional and translational changes detected were moderate relative to mRNA (Fig. 1C) and protein levels reported for CD31+ cells [5], indicating a limited commitment to the EC lineage in vitro.

Figure Figure 4..

Endothelial cell differentiation of adipose stem cells (ASCs). (A): Three ASC clones were induced to differentiate for 7 days in methylcellulose and examined by phase contrast microscopy. Note the formation of a network of elongated cells. Bar, 50 μm. (B): Real-time reverse transcription-polymerase chain reaction analysis of CD31 and CD144 expression in ASC clones after methylcellulose culture as in (A), relative to expression level in the same but undifferentiated clones. Samples were analyzed in triplicates in triplicate experiments (mean ± SD; see text for p values). (C): Flow cytometry analysis of CD31 and CD144 protein expression in undifferentiated ASCs (one representative polyclone) and in clones B1, B2, and B3 after endothelial differentiation. Gates were set using a fluorescein-isothiocyanate-conjugated unspecific isotype control antibody. Percentages of cells expressing CD31 or CD144 are shown. Abbreviations: FL1-H, fluorescence channel 1; FSC-H, forward scatter; Undiff., undifferentiated.

Methylcellulose Culture Elicits Discrete Changes in CpG Methylation in the CD31 and CD144 Promoters

Bisulfite analysis of CD31 and CD144 promoters in two ASC clones before and after culture in methylcellulose revealed subtle changes in CpG methylation. The CD31 promoter was 92%–99% mosaically methylated in undifferentiated cells and remained globally methylated after 1 week in methylcellulose (p = .357 and p = 1.000 for clones B1 and B3, respectively; paired t tests; Fig. 5A, 5B). Partial demethylation or methylation of specific CpGs (numbers 14, 15, 18) was observed in clone B3 but not in clone B1; however, extensive hypomethylation, such as that seen in CD31+ cells or HUVECs, was not detected (Fig. 2C, 2E).

Figure Figure 5..

Bisulfite sequencing analysis of CpG methylation in the CD31 and CD144 promoters in adipose stem cell (ASC) clones before and after induction of endothelial cell (EC) differentiation. (A): Bisulfite analysis and (B) percentage of methylation of individual CpGs in the CD31 promoter in clones B1 and B3 in undifferentiated and stimulated cells (total populations). (C): Bisulfite analysis and (D) percentage of individual CpG methylation in the CD144 promoter. CpG number 1 is the 5′-most CpG examined in the sequence (see Fig. 1A). (E): A polyclone of ASCs was induced to differentiate toward the EC lineage for 7 days. CD31+ cells were immunoisolated and CD31 methylation profile determined. Note the consistent demethylation of CpG number 8 (arrow). Abbreviation: diff., differentiated.

CD144 promoter methylation was also mosaic between clones B1 and B3 in undifferentiated ASCs but rather homogenous within clones (Fig. 5C). Upon culture in methylcellulose, the percentage of methylation remained unaltered in clone B1 (28.7% ± 9.8% vs. 33.3% ± 10.7% before and after methylcellulose culture, respectively; p = .204; paired t test) and B3 (32.0% ± 10.0% vs. 18.8% ± 7.8%, respectively; p = .077; paired t test) (Fig. 5C, 5D). At the individual CpG level, methylation profile of clone B1 remained remarkably stable, whereas clone B3 underwent demethylation at CpGs numbers 11 (70%–10% methylation; p < .001; paired t test) and 14 (90%–0% methylation; p < .001; paired t test). As for the CD31 promoter, hypomethylation of CD144 seen in CD31+ cells or HUVECs was not observed. Methylcellulose culture, therefore, provokes a discrete, clone-specific DNA methylation response in the CD144 and CD31 promoters. However, no global demethylation such as that reported for CD31+ cells or HUVECs occurs, indicating that ASCs have not epigenetically committed to EC differentiation.

To determine whether selection of in vitro differentiated CD31 expressing cells would enable detection of demethylation events, EC differentiation was induced in a polyclonal culture of ASCs, and CD31+ cells were immunoisolated. Surprisingly, we only found a 7–8-fold upregulation of CD31 mRNA in these cells compared with nonstimulated cells. This was similar to what we reported for unsorted cells (Fig. 4B), suggesting that virtually all cells upregulate CD31 to some level upon induction of EC differentiation. Furthermore, bisulfite sequencing revealed complete demethylation of CpG number 8 in the CD31 promoter, immediately 3′ of the transcription start site (TSS) (Fig. 5E, arrow; compare with data from Fig. 3C, upper panel). Thus, induction of CD31 protein expression correlates with demethylation of CpG 8 near the CD31 promoter TSS in the CD31+ cells. Nevertheless, we did not detect extensive CD31 demethylation such as that of HUVECs or CD31+ cells isolated from adipose tissue, corroborating the limited ability of ASCs to epigenetically commit to EC differentiation.

CD31 and CD144 Promoter Methylation After Adipogenic and Osteogenic Differentiation

To determine the lineage-specificity of CpG methylation changes or lack thereof observed upon methylcellulose culture, methylation at the CD31 and CD144 promoters was monitored after differentiation of the same ASC clones (B1–B3) toward the adipogenic or osteogenic pathway. Adipogenic- and osteogenic-specificity of differentiation was assessed by (a) oil red O and Alizarin red staining, respectively (Fig. 6A, upper and lower panels, shown here for clone B3), (b) upregulation of fatty acid binding protein 4 (FABP4) and osteomodulin (OMD), markers of adipogenesis and osteogenesis, respectively (Fig. 6B), and (c) the absence of upregulation of CD31 or CD144 expression (Fig. 6C), whose transcript levels remained close to the detection limit by real-time RT-PCR.

Figure Figure 6..

CpG methylation of CD31 and CD144 upon adipogenic and osteogenic differentiation. (A): Adipose stem cells (ASCs) (clone B3) were induced to differentiate for 3 weeks toward adipogenic or osteogenic pathways and stained with, respectively, oil red O (top panels) and Alizarin red (bottom panels). Arrows point to oil-red-O-stained lipid droplets. Dark spots in the lower right panel are extracellular calcium phosphate precipitates stained by Alizarin red. (B): Real-time reverse transcription-polymerase chain reaction (PCR) analysis of expression of FABP4 and OMD in clones B1–B3 after adipogenic and osteogenic differentiation, relative to expression in the same but undifferentiated clones. (C): Expression of CD31 and CD144 in undifferentiated (level 1) and in adipogenic- or osteogenic-differentiated clones (see text for p values). (D): Bisulfite analysis of CD31 and CD144 in six ASC clones after adipogenic or osteogenic differentiation by direct sequencing of PCR products. Right panel shows sequence analysis of eight bacterial clones of CD144 PCR products of clone B3 after adipogenic differentiation. For arrow and bar, see text. Abbreviations: Ad, adipogenic; Adipo, adipogenic; Os and Osteo, osteogenic; Un, undifferentiated.

Bisulfite analysis by direct sequencing of PCR products showed stability of CpG methylation of the CD31 and CD144 promoters after adipogenic and osteogenic differentiation (Fig. 6D). Similar results were obtained for three additional ASC clones (A1–A3; Fig. 6D). Moreover, sequencing of eight bacterial clones of PCR products from clone B3, which displayed CD144 CpG 11 and 14 demethylation upon EC differentiation, showed, however, maintenance of CpG 14 methylation (Fig. 6D, right panel, arrow) and demethylation of CpG 11 and 12 after adipogenic differentiation (Fig. 6D, right panel, bar). These data suggest that, at least for clone B3, endothelial stimulation in methylcellulose triggers specific CpG (number 14) demethylation, whereas induction of differentiation, irrespective of the lineage, seems to promote demethylation of other CpGs (e.g., numbers 11 and 12).


This work provides the first assessment of CpG methylation in EC-related gene promoter regions in relation to transcription and EC-lineage differentiation potential in MSCs. Bisulfite sequencing analysis of freshly isolated, cultured undifferentiated, and cultured differentiated (both polyclonal and clonal cell lines) human ASCs reveals consistent hypermethylation and mosaic methylation patterns in CD31 and CD144 promoters. This finding is in contrast to adipogenic promoters, which are largely hypomethylated in undifferentiated and differentiated ASCs [35], but is consistent with the methylation of the myogenic-specific locus MYOG [35]. Mosaic methylation profiles between and within ASC donors concurs with heterogeneity in methylation patterns at adipogenic promoters in ASCs [35] and in stem cells from single intestinal crypts [32, 33] and may arise from stochastic methylation, which accumulates independently in different cells.

The CpG methylation profiles of EC-specific gene promoter regions in ASCs presented here shed light from a new angle on the controversy surrounding MSC plasticity, or lack thereof, toward the EC lineage. The methylated nature of the CD31 and CD144 promoters in ASCs, together with their relatively hypomethylated state in ECs such as CD31+ cells and HUVECs, suggests that MSCs from adipose tissue are restricted in their capacity to form ECs.

Increasing evidence argues that transplantation of MSCs offers a promising approach for restoration of tissue vasculature after ischemic events. Abundance and easy accessibility make ASCs an appealing source of MSCs for transplantation. Recent in vivo studies have shown improvements of blood flow and capillary density in ischemic-tissue-receiving ASCs [13, 15, 20]. Two studies attributed these effects to direct incorporation of ASCs into the vasculature [13, 15]. Incorporated cells expressed EC-specific markers, suggesting that they underwent EC differentiation in vivo—a conclusion reinforced by in vitro differentiation data. However, another study using ASCs found no such contribution and attributed vasculature improvement to paracrine mechanisms [20]. This conclusion is consistent with the finding that injection of ASC-conditioned media stimulates angiogenesis in ischemic tissue [14]. Furthermore, a recent study using extensively cultured porcine bone marrow MSCs showed that a minor proportion (0.3%) of cells expressed CD31 and vWF after injection into infarcted hearts of nonobese diabetic/Scid mice, even though such treatment led to improved cardiac function [42]. The authors concluded that the small number of cells expressing EC-specific markers ruled out the possibility that the cells significantly contributed structurally to the vasculature; rather, positive effects of injecting MSCs were likely due to paracrine mechanisms [42]. This restrictive ability to express CD31 in vivo is in agreement with our evidence for heavy methylation of the CD31 promoter in MSCs from adipose tissue.

Inconsistencies in the studies outlined above are likely attributed to the composition of injected cell populations. In work demonstrating ASC incorporation into the vasculature [13, 15], injected cells had only been cultured for a brief period (days) after isolation without serial passaging. Therefore, injected cell suspensions of ASCs were likely to also contain ECs or EC-specific progenitors, as early passages of plated cells from the stromal vascular fraction of human adipose tissue were reported to be contaminated with EC cells [4]. Furthermore, in the present study, we detected transcripts for CD31 and CD144 in purified populations of CD31 ASCs, and subsets of these express vWF [5]. This implies the presence of EC progenitors in the CD31 cell population. Therefore, the possible contribution of EC progenitors to the vasculature and in cultures differentiating toward the EC lineage in vitro cannot be discounted in the previous studies. The plasticity of multipotent ASCs toward the EC pathway can only definitively be assessed by employing clonal cell lines thereof.

To our knowledge, the present study is the first to utilize clonal cell lines of ASCs to examine their potential to differentiate into ECs. These cell lines meet the key stem cell criteria of extensive self-renewal capacity and ability to differentiate into more than one cell type [5]. Most importantly, they exclude the possibility of cultures contaminated with progenitors committed to solely the EC lineage. Intriguingly, upon EC differentiation in methylcellulose, only a subpopulation of cells within each clonal and polyclonal cell line expresses the cell surface CD31 and CD144 markers, which correlates with marginal increases in the transcriptional level of these genes. In contrast, overexpression of these genes was significantly higher in uncultured CD31+ ECs compared with CD31 ASCs, which further highlights the low ability of cultured cells to express the EC-specific genes studied.

The correlation between this restrictive in vitro differentiation capacity was also observed recently [20] and is consistent with hypermethylation of CD31 and mosaic methylation of CD144 in freshly isolated, cultured, and EC-differentiated ASCs (only cytosine- and cell-clone-specific demethylation events were detected). We cannot at present, however, exclude the possibility that (a) additional CpGs in the regulatory regions of CD31 and CD144 would not show more consistent methylation changes upon in vitro differentiation and (b) that a small subset of cells undergo complete CD31 or CD144 demethylation upon EC differentiation in vitro, which may be detected by sequencing a large number of PCR products.

Nonetheless, our analysis of CD31+ and HUVECs illustrates the functional relevance of the promoter regions examined here, a contention supported by the identification of binding sites for transcriptional regulators including Sp1, GATA2, E2F-1, and p300 (supplemental online Fig. 1) [36, 43]. Furthermore, the relatively lower methylation state of CD144 compared with CD31 correlates with higher CD144 expression levels. Apart from the slight increase in CD31 methylation, similarities in the methylation signatures of freshly isolated ASCs and of clonal and polyclonal cell lines suggest preservation of epigenetic fidelity of these two promoters during culture. This is consistent with our previous finding that methylation profiles of adipogenic-related gene promoters are maintained in cultured ASCs [35].

We cannot conclude from the current study that DNA methylation plays a causative role in the restrictive expression of the endothelial-related genes studied (particularly CD31). Rather, bisulfite analysis of freshly isolated, clonal, and polyclonal cell lines originating from several donors shows a substantial correlation between expression propensity and methylation. We also show that selection of CD31+ cells after in vitro EC induction of cultured ASCs coincides with complete demethylation of a specific CpG immediately upstream of the TSS in the CD31 promoter. Further work relying on DNA demethylating agents may shed light on the causative role DNA methylation plays in transcriptional regulation of these genes.

The distinct methylation patterns of EC-related loci in ASCs compared with HUVECs also highlight the limited ability of ASCs to differentiate towards the EC lineage. Moreover, CD31 and CD144 methylation profiles of HUVECs and CD31+ cells are highly similar, strongly suggesting the commitment of CD31+ cells to the EC lineage. Therefore, CD31+ ECs and CD31 ASCs are likely to coexist as unrelated cell populations within the stromal vascular niche of adipose tissue. The methylated state of CD31 and CD144 in undifferentiated ASCs contrasts with methylation signatures of adipogenic-related genes, which show striking resemblance to those of fully differentiated cultured human adipocytes [35]. We found that this hypomethylated state is even preserved after EC and osteogenic differentiation (A.C.B., A.N., A.L.S., and P.C., unpublished data), strengthening the natural propensity of ASCs to form adipocytes. Thus, CpG methylation profiles in undifferentiated MSCs may dictate lineage-specificity of differentiation.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


We thank L. Stijac and C.T. Freberg for technical assistance and A. Shahdadfar and S. Naderi for help with flow cytometry. This work was supported by the Research Council of Norway (FUGE, YFF, and STORFORSK programs), the Norwegian Cancer Society, and the Norwegian Stem Cell Network.