IFATS Collection: Adipose Stromal Cell Differentiation Is Reduced by Endothelial Cell Contact and Paracrine Communication: Role of Canonical Wnt Signaling


  • Gangaraju Rajashekhar,

    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Department of Cellular and Integrative Physiology, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Dmitry O. Traktuev,

    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • William C. Roell,

    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Department of Cellular and Integrative Physiology, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Brian H. Johnstone,

    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Stephanie Merfeld-Clauss,

    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Bruce Van Natta,

    1. Meridian Plastic Surgery Center, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Elliot D. Rosen,

    1. Department of Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Keith L. March,

    Corresponding author
    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Department of Cellular and Integrative Physiology, Indiana University School of Medicine, Indianapolis, Indiana, USA
    3. Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    • Keith L. March, Keith L. March, M.D., Ph.D., Indiana Center for Vascular Biology and Medicine, 975 West Walnut Street IB442, Indianapolis, Indiana 46202, USA. Telephone: 317-278-0130; Fax: 317-278-0089===

      Matthias Clauss, Matthias Clauss, Ph.D., Indiana Center for Vascular Biology and Medicine, 975 West Walnut Street IB433, Indianapolis, Indiana 46202, USA. Telephone: 317-278-2837; Fax: 317-278-0089===

    Search for more papers by this author
  • Matthias Clauss

    Corresponding author
    1. Indiana Center for Vascular Biology and Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Department of Cellular and Integrative Physiology, Indiana University School of Medicine, Indianapolis, Indiana, USA
    • Keith L. March, Keith L. March, M.D., Ph.D., Indiana Center for Vascular Biology and Medicine, 975 West Walnut Street IB442, Indianapolis, Indiana 46202, USA. Telephone: 317-278-0130; Fax: 317-278-0089===

      Matthias Clauss, Matthias Clauss, Ph.D., Indiana Center for Vascular Biology and Medicine, 975 West Walnut Street IB433, Indianapolis, Indiana 46202, USA. Telephone: 317-278-2837; Fax: 317-278-0089===

    Search for more papers by this author


Adipose stromal cells (ASC) are multipotential mesenchymal progenitor cells that are readily induced to undergo adipogenic differentiation, and we have recently demonstrated them to have functional and phenotypic overlap with pericytes lining microvessels in adipose tissues. In this study we addressed the hypothesis that modulation of ASC fate within this perivascular niche can occur via interaction with endothelial cells (EC), which serve to modulate the adipogenic potential of ASC. To this end, we investigated contact as well as paracrine effects of EC on ASC adipogenesis, in two-dimensional coculture and via conditioned medium and analyzed mutual gene expression changes by real-time reverse transcription polymerase chain reaction (PCR). A significant decrease in adipogenic differentiation was observed in ASC when they were cocultured with EC but not control fibroblasts. This endothelial cell-specific effect was accompanied by increased expression of factors involved in Wnt signaling, most prominently Wnt1, Wnt4, and Wnt10a, which are well-known inhibitors of adipogenesis. Suppression of Wnt1 but not Wnt 10a or scrambled control short interfering RNA in cocultures partially reversed the endothelial cell effect, thus increasing adipogenic differentiation, suggesting a plausible role of Wnt1 ligand in modulation of adipogenesis by the vasculature. Furthermore, addition of recombinant Wnt ligand or the Wnt signaling agonist inhibited adipogenic differentiation of ASC in the absence of EC. In conclusion, these data define the relationship in adipose tissue between ASC and EC in the perivascular niche, in which the latter act to repress adipogenesis, thereby stabilizing vasculature. It is tempting to speculate that abnormal endothelial function may be associated with pathologic derepression of adipogenesis.

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


Author contributions: G.R.: collection and/or assembly of data, data analysis and, interpretation, conception and design, manuscript writing; D.O.T., W.C.R., and S.M.-C.: collection and/or assembly of data; B.H.J. and E.D.R.: data analysis and interpretation; B.V.N.: provision of study material or patients; K.L.M. and M.C.: conception and design, manuscript writing; K.L.M. and M.C. contributed equally to this study.

Microvascular endothelium is covered by specialized smooth muscle cells (SMC) commonly referred as pericytes. Pericytes share many of the same markers as SMC, which are found in much more abundance in the mural layers of larger vessels but are distinguishable by SMC-α-actin and platelet-derived growth factor receptor expression levels [1]. They are believed to function in vascular stabilization through providing nourishment to endothelial cells (EC), blocking inflammatory processes, and controlling angiogenic stimuli. Recently, we have shown that in adipose tissues these mural cells share cell surface markers and properties with adipose stromal or stem cells [2]. It has been demonstrated that both pericytes and adipose stromal cells (ASC) are pluripotent mesenchymal stem cells and that adipose-resident ASC/pericytes readily undergo adipogenesis in response to stimuli [2, [3], [4], [5]–6].

Adipose tissue development from multipotent mesenchymal stem cells is believed to proceed through a sequence of determination, commitment to the adipocyte lineage, generation of preadipocytes, and terminal differentiation to mature adipocytes. Although the molecular regulation of the determination step is not well characterized, several pathways that affect adipogenesis have been recently uncovered, many involving negative regulation of adipogenic fate determination concomitant with positive regulation of alternative cell fates [7]. For example, the transcription factor GATA2 reduces adipogenesis while promoting hematopoiesis and urogenital development [8, 9]. Studies addressing differentiation from predetermined preadipocytes to adipocytes have predominately used the murine 3T3-L1 preadipocytic cell line. A major role for Wnt signaling in modulating adipogenesis has been demonstrated in these cells [7]. Importantly, forced overexpression of Wnt1 in 3T3-L1 preadipocytes maintains the cells in an undifferentiated state by inhibiting the adipogenic master regulators peroxisome proliferator-activated receptor γ (PPARγ) and CCAAT/Enhancer-Binding Protein, Alpha (CEBPα) [10].

Given the emerging recognition of mesenchymal stem cells as related to pericytes [3, [4], [5]–6], it is of interest to determine mechanisms controlling their differentiation, particularly in the context of their mutually stabilizing interactions with endothelium. Here we address the hypothesis that vascular EC create a periendothelial niche for mural ASC in adipose tissues, regulating adipogenic differentiation of ASC/pericytes. In this study we provide evidence that EC, via Wnt signaling, actively restrain the adipogenic differentiation potential of ASC in favor of a pericytic phenotype.

Materials and Methods

Subjects Included in the Study

Human subcutaneous adipose tissue samples were obtained from lipoaspiration/liposuction procedures from patients visiting the Meridian Plastic Surgery Center. Approximately six to eight female patients, with a mean age of 36.5 ± 5.1 and with a mean body mass index (BMI) 26.1 ± 2.8 (mean ± SEM), were included in the study. This study was approved by Indiana University School of Medicine Institutional Review Board.

Isolation and Cultivation of Human ASC

Human subcutaneous adipose tissue samples obtained from lipoaspiration/liposuction procedures were processed to isolate ASC as described previously [2]. In brief, the fat tissue was digested in collagenase type I solution (Worthington Biochemical, Lakewood, NJ, http://www.worthington-biochem.com) under agitation for 2 hours at 37°C and centrifuged at 300g for 8 minutes to separate the stromal cell fraction (pellet) from adipocytes. The pellet was resuspended in Dulbecco's modified Eagle's medium/Ham's F-12 medium containing 10% fetal bovine serum (FBS; HyClone, Logan, UT, http://www.hyclone.com), filtered through 250-μm Nitex filters (Sefar America Inc., Kansas City, MO, http://www.sefar.com), and centrifuged at 300g for 8 minutes. The cell pellet was treated with red-cell lysis buffer (154 mmol/l NH4Cl, 10 mmol/l KHCO3, 0.1 mmol/l EDTA) for 10 minutes. The final pellet was resuspended in EBM-2/5% FBS or EGM2MV (Cambrex, Baltimore, http://www.cambrex.com).

Flow Cytometric Characterization of Human ASC

ASC cultured for 2 days on culture plastic were analyzed for surface marker expression using a FACSCalibur cytometer analyzer and Cell QuestPro software (BD Biosciences, San Diego, http://www.bdbiosciences.com). Day 2 cells were harvested with 2 mmol/l EDTA/phosphate-buffered saline (PBS). All of the following steps were performed on ice. Cell pellets were incubated for 20 minutes with primary antibodies or matching isotype controls (5 μg/ml). The primary antibodies used were CD31-phycoerythrin (PE), CD45-fluorescein isothiocyanate, CD34-allophycocyanin, CD90-PE, CD140a-PE, CD140b-PE, and CD144-PE (Chemicon, Temecula, CA, http://www.chemicon.com). Labeled cells were washed with 2% FBS/PBS, fixed with 2% paraformaldehyde and analyzed.

Endothelial Cells

All studies were performed using human umbilical cord blood-derived EC, which were cultured and maintained in EGM2 medium with SingleQuots supplements (Lonza, Walkersville, MD, http://www.lonza.com) and 10% FBS as previously described [11]. Results were confirmed by using adult dermal microvascular EC (Clonetics; Lonza), which were cultured and maintained in EGM2MV medium. Cells between passage 3 and passage 7 were used for the assays. For conditioned medium experiments, approximately 2 × 105 cells per well of a 12-well plate were incubated with EBM-2/5% FBS medium for 72 hours, the supernatant was collected and centrifuged (1,000 rpm), and cell-free conditioned medium was frozen at −80°C until further analysis.

Adipose Stromal Cell Differentiation

To test whether EC modulate the adipogenic potential of ASC in the context of direct contact, 2.5 × 105 EC per well were plated in a 12-well plate, allowing 4 hours for attachment. An equal number of ASC were plated directly onto the EC monolayer, allowing 4 hours for attachment. Medium was changed to 2:1 preadipocyte differentiation medium (Promocell, Heidelberg, Germany, http://www.promocell.com, or Zen-Bio, Research Triangle Park, NC, http://www.zen-bio.com)/EBM-2/5% FBS medium. As a positive (non-coculture) control, ASC were cultured alone, whereas human dermal fibroblast cells (Lonza) instead of EC were cocultured with ASC to provide a nonendothelial cell control. Adipose differentiation was monitored for 2–14 days by real-time PCR analysis of LPL and PPARγ and by lipid accumulation, as demonstrated with Nile red staining.

Nile Red Staining and Lipid Quantification Analysis

To obtain a quantitative measure of the intracellular triacylglycerol accumulation, the cells were stained with Nile red [12]. Briefly, cells were fixed postdifferentiation with 1% paraformaldehyde, washed, and stained with a solution of 100 nM Nile red (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), 100 μg/ml RNase1 (Sigma-Aldrich) in PBS. After a 30-minute incubation at room temperature, cells were washed with PBS, and the fluorescence was measured using Spectramax fluorescent plate reader (Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com) to determine Nile red intensity per well. Using a mixture of 66% methanol, 0.1% Triton X-100, and 10 μg/ml Hoechst 33342 in PBS, the lipid was extracted and incubated for a further 15 minutes at room temperature to label the nuclei. After determining the Hoechst 33342 intensity per well, the results were computed as ratio of total lipid fluorescence to total nuclear fluorescence.

Effect of Conditioned Media on Wnt Gene Expression

To test whether the conditioned media from EC upregulate canonical Wnt signaling molecules in ASC, approximately 2.5 × 105 ASC were plated per well of a 12-well plate and stimulated with the conditioned medium (CM) mixed with EBM-2/5% FBS basal medium (BM) in several ratios (1:0, 1:1, and 2:1 ratios of CM to BM). At 4, 24, and 48 hours total RNA was isolated from the individual wells and processed as mentioned below for mRNA analysis.

Endothelial Cell-ASC Coculture

For coculture experiments, generally a 1:4 ratio of ASC to EC was used, which reflects a mean (reportedly 10%–50%) of pericyte coverage of endothelium in tissues [1] and was used in our previous publication to build functional vessels in vivo [2]. For adipogenic differentiation, experiments were performed using a 1:1 ratio of EC to ASC to optimize the PCR-based detection of transcriptional correlates of adipogenic differentiation. Briefly, approximately 2.2 × 106 EC were plated on a collagen-coated 100-mm dish, and cells were allowed 4 hours to attach. Approximately 0.55 × 106 ASC were layered on the confluent EC and allowed to attach for a further 4 hours. Media on cocultures were changed to EBM-2/5% FBS with 10 ng/ml vascular endothelial growth factor (Sigma-Aldrich) and incubated for a total period of 72 hours. As a control, EC and ASC were cultured separately for the same time period. Experiments were performed from ASC cells isolated from at least 6–8 patients in duplicate.

Fluorescence-Activated Cell Sorting Analysis

After 72 hours of coculture, cells were trypsinized and collected. Monoclonal antibodies to CD31 (APC-labeled) for EC isolation and CD140b (PE-labeled) for ASC isolation or isotype matched antibodies were used as per the standard procedures. After 20 minutes cells were washed three times with PBS and processed for cell sorting using the FACSAria cell-sorting system (BD Biosciences) on the basis of two distinct populations expressing CD31+ (R2) and CD140b+ (R3). Data were further processed with the use of WinMDI 2.8 software (TSRI Cytometry Software, The Scripps Research Institute, La Jolla, CA, http://facs.scripps.edu).

Real-Time Quantitative Reverse Transcription PCR Analysis of mRNA Expression of ASC and Endothelial Cell Cocultures

Each individual sorted population was collected and isolated total RNA using the Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany, http://www1.qiagen.com). Approximately 50 ng of RNA was mixed with Sybergreen mix containing gene-specific primers as per the manufacturer's instructions (Verso SYBR Green 1-Step quantitative RT-PCR kit; ABgene, Epsom, U.K., http://www.abgene.com). Samples were analyzed in a Chromo4 Opticon analyzer (Bio-Rad, Hercules, CA, http://www.bio-rad.com) in a total reaction volume of 20 μl. The thermal cycling program consisted of an initial 15 minutes of cDNA synthesis at 50°C followed by 15 minutes of Thermo-start activation. The amplification included denaturation at 95°C, followed by 40 cycles of denaturation at 95°C for 15 seconds, annealing at 58°C for 30 seconds, and extension at 72°C for 30 seconds. To compare the levels of Wnt gene transcripts between the ASC and EC that had been cocultured, we used the comparative CP method for relative quantification as described previously [13]. The amount of target gene transcript, normalized to the elongation factor 1α (eF1α) endogenous housekeeping gene transcript and relative to the calibrator, was computed by 2inline image, where inline image = ΔCP (unknown target gene) − ΔCP (calibrator), and ΔCP of target or calibrator is the CP of the target gene subtracted from the CP of the housekeeping gene.

Topflash Reporter Gene Assays

Transfection experiments were carried out using electroporation (Bio-Rad). Briefly, cells were trypsinized and counted. Approximately 2 × 105 cells per 2-mm cuvette were taken with TCF7-Topflash and pRL-TK Renilla plasmids (1 μg each) and subjected to electroporation at 900 V for 0.9 milliseconds and two pulses at a 5-second interval. After electroporation the cells were left for 10 minutes in a 37°C, 5% CO2 incubator and plated in full medium in each well of a 12-well plate. After overnight attachment, cells were stimulated for 48 hours with 1 μM Wnt agonist or 10 ng/ml Wnt3a, and cell lysates were assayed for luminescence using the Dual Luciferase Assay system (Promega, Madison, WI, http://www.promega.com). The percentage of activity of Topflash was normalized to Renilla luciferase and plotted as percentage activity of Topflash per milligram of total protein to normalize for total protein content per lysate. All transfections were carried out in triplicate on at least three independent occasions.

Wnt Gene Silencing in ASC and Endothelial Cell Cocultures with Small Interfering RNA

RNA interference (RNAi) is an endogenous post-transcriptional gene silencing pathway that uses short interfering RNAs (siRNAs) to target specific transcripts for degradation. Gene knockdown by RNAi can be mediated by synthetic molecules (e.g., chemically synthesized siRNAs), and in this regard we used Dharmacon SMARTpool reagents (Dharmacon, Inc., Lafayette, CO, http://www.dharmacon.com) that combine four SMARTselection-designed siRNAs into one highly effective siRNA pool. This combination of siRNAs mimics the natural silencing pathway and has been proven to reduce off-target effects and false negatives by targeting four different mRNA regions at once. Approximately 2 × 105 ASC and EC (1:1) were transferred into a 2-mm electroporation cuvette along with an exogenous siRNA against either glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (positive control) or Wnt1 or Wnt10a (On-Target Plus reagents; Dharmacon) and electroporated using optimized square wave pulse conditions (900 V, 0.9 milliseconds, two pulses, 5-second interval). As a negative control a scrambled siRNA control or an equal amount of cells that had been mock-transfected were used. After electroporation, cells were plated in 12-well plates in full media and left overnight for attachment. Media on cocultures were changed to preadipocyte differentiation media:EBM-2/5% FBS (2:1) and incubated for a total period of 48 hours. Total RNA was isolated from the individual wells and processed as mentioned earlier for mRNA analysis of genes involved in adipocyte differentiation.

Statistical Analysis

For ASC/EC coculture, data are expressed as mean ± SEM of a representative experiment from a total of 6–8 patients, with duplicates, and as mean ± SEM for each group from quantitative RT-PCR experiments performed in triplicate for mRNA expression profiles and repeated at least two additional times. For conditioned medium experiments, data are shown as mean ± SEM of three independent experiments, with duplicates, that were repeated with at least three patient endothelial cells. For silencing experiments, data are shown from a representative experiment of three independent experiments. Statistical significance was determined by Student's t test using the Microsoft Excel statistical package (Microsoft, Redmond, WA, http://www.microsoft.com). A probability value of p < .05 was considered statistically significant.


Adipose Stromal Cell Differentiation Is Reduced by Endothelial Cell Cocultivation

On the basis of our recent finding that ASC exhibit phenotypic and functional characteristics of microvascular pericytes in adipose tissue, residing in a periendothelial niche [2], we tested the hypothesis that EC are specifically capable of modifying the adipogenic potential of ASC. We initially investigated the effects of EC on ASC adipogenesis in a two-dimensional coculture model with direct EC-ASC contact. Lipid accumulation in adipogenic differentiation medium with PPARγ agonist was markedly reduced in ASC cocultured with EC versus ASC monoculture (Fig. 1A–1D; supplemental online Fig. 1). Specificity of this effect was tested by coculturing ASC with an equal number of fibroblasts, which demonstrated no effect on adipogenesis. Furthermore, mRNA analysis of PPARγ and LPL, key indicators of adipogenic differentiation, confirmed significant repression by EC cocultivation (Fig. 1E). Of note, EC maintained in a monolayer throughout the period of the study in coculture with ASC in the presence of differentiation medium, as demonstrated with DiI-prelabeled EC (supplemental online Fig. 1). Finally, the endothelial monolayer preserved endothelial functionality in ASC cultures in the presence of differentiation medium, as the endothelial-specific adhesion molecule vascular cell adhesion molecule-1 could still be induced with tumor necrosis factor-α (supplemental online Fig. 2).

Figure Figure 1..

Coculture of ASC and endothelial cells results in decreased adipocyte differentiation. ASC were cultured either alone (A) or in a monolayer over human FB (B) or EC (C) in a 1:1 ratio. Adipogenic differentiation was monitored in basal or differentiation medium for up to 14 days. Quantification of intracellular triacylglycerol with Nile red staining of lipids, and Hoechst 33342 (D), as well as mRNA analysis (E), for PPARγ and LPL showed that ASC, when cocultured with EC but not when cocultured with FB, demonstrated a significant decrease in adipocyte differentiation. Data are shown from an experiment, with duplicates, that was repeated at least three additional times. Abbreviations: ASC, adipose stromal cells; EC, endothelial cells; FB, fibroblasts; PPAR, peroxisome proliferator-activated receptor.

Cocultivation of ASC and EC Upregulates Canonical Wnt Signaling Molecules

Because Wnt signaling is well known to be involved in modulation of adipogenesis [7, 14], we tested whether the corresponding gene expression levels were affected by culture with EC. Expression of multiple Wnt family and Wnt pathway genes were consistently increased by 2–3-fold in mixed cultures of ASC and EC, compared with the transcript levels for each in the sum of equivalent total numbers of ASC and EC in monoculture. Multiple molecules known to be involved in Wnt signal transduction were found to be strongly upregulated in conditions of coculture, including the Wnt receptor Fz3, the intracellular signaling molecules Dvl1 (2.5 ± 0.5), TCF-7, and genes known to be expressed downstream of Wnt signaling, such as Sox-17 (4.2 ± 0.9) and Wisp1 (Fig. 2). In contrast, Wnt ligand coreceptors LRP5 and LRP6; membrane-bound Wnt receptors Fz1, Fz2, and soluble frizzled-related protein1 (sFRP1); and the secreted Wnt ligand binding protein Wnt-inhibitory factor-1 (WIF-1) were not upregulated (Fig. 2; data not shown). These data suggest that Wnt signaling pathways are activated upon contact of ASC and EC.

Figure Figure 2..

Wnt genes are upregulated in cocultures of ASC and endothelial cells. Endothelial cells were cultured with ASC for up to 72 hours, and isolated RNA was analyzed for mRNA expression of Wnt signaling-related genes by reverse transcription polymerase chain reaction. A number of Wnt ligands, receptors, and intracellular signaling proteins were upregulated, whereas the antagonistic soluble receptors were downregulated in cocultures compared with cells that had been cultured alone (p < .05). Data shown comprise an average from at least three independent experiments, with duplicates, that were repeated at least three times. Abbreviations: ASC, adipose stromal cells; EC, endothelial cells; v/s, versus.

Increased Expression of Canonical Wnt Signaling Molecules by Coculture Occur in Both ASC and EC

To evaluate the cell-intrinsic modulation of Wnt levels in ASC and EC respectively, coculture was performed as above, followed by fluorescence-activated cell sorting isolation of ASC and EC populations and RNA purification. Sorting was conducted taking CD140b+/CD31− to define the ASC population (R3; lower right quandrant in Fig. 3A) and CD140b−/CD31+ to define the endothelial cell population (R2; upper left quadrant). These two populations accounted for >95% of the cell events analyzed. A third population of cells expressing CD31 and CD140b was identified but was excluded from RNA analysis. Figure 3B shows that expression of the endothelial-specific CD144 (VE-Cadherin) gene occurred exclusively in the EC population and was identified in EC from coculture conditions, whereas chondroitin sulfate proteoglycan (NG2), which is selectively produced in pericytic cells, was identified only in ASC, confirming the purity and identity of the two sorted populations (Fig. 3B; supplemental online Fig. 4). A significant elevation of Wnt1 gene expression occurred upon coculture in both ASC and EC but was more pronounced in EC (Fig. 3C). Similarly, both cell types responded by enhancing other canonical Wnt signaling proteins, such as Wnt4, 8a, 10a, and 10b. Furthermore, genes downstream of canonical Wnt signaling, such as TCF7 and Wisp1, were also increased in both EC and ASC. Although the membrane-bound Fz1 and Fz2 were not increased significantly in either cell population, sFRP2 was strongly upregulated only in EC (Fig. 3C).

Figure Figure 3..

Coculture of ASC and EC followed by flow cytometric cell sorting. ASC were cultured in a monolayer over EC for 72 hours in EBM-2, 5% fetal bovine serum medium in the presence of 10 ng/ml vascular endothelial growth factor. (A): Based on cell surface markers, flow cytometric sorting revealed two individual populations, CD31+ (R2) and Cd140b+ (R3). (B): In these individual populations, mRNA was assessed for expression of NG2 and CD144, two additional markers for ASC/pericytic and EC, respectively. (C): Wnt signaling genes were upregulated both in cocultured EC and ASC. Data shown comprise an average from at least three independent experiments, with duplicates, that were repeated at least three times. Abbreviations: ASC, adipose stromal cells; EC, endothelial cells; FSC, forward scatter; SSC, side scatter.

Endothelial-conditioned Medium Upregulates Canonical Wnt Signaling Molecules in ASC

To determine whether the increase of Wnts and Wnt signaling seen in coculture of EC and ASC depended on direct contact of these two cell types, we tested for the upregulation of canonical Wnts in ASC cultures after exposure to CM from EC for 48 hours. As shown in Figure 4, endothelial-CM induced multiple canonical Wnt ligands, their receptors, and intracellular signaling molecules such as TCF7, whereas competitively antagonistic soluble Wnt ligand receptors, including sFRP1, were not upregulated. Surprisingly, the intracellular Wnt signaling repressor Axin was strongly upregulated, suggesting that paracrine interactions of CM can modulate agonistic and antagonistic Wnt signaling. Of note, unconditioned EBM-2/5% FBS did not show any effect on Wnt gene expression in ASC (data not shown).

Figure Figure 4..

Conditioned medium from endothelial cells upregulates Wnt gene expression in adipose stromal cells (ASC). Conditioned medium from endothelial cells was added to ASC, which were incubated for the indicated times. RNA was extracted for reverse transcription polymerase chain reaction analysis. Several Wnt ligands, receptors, and intracellular signaling proteins were significantly (*, p < .05) upregulated at 48 hours compared with ASC that had been left untreated. Data shown are an average of three independent experiments, with duplicates.

Targeted Modulation of Canonical Wnt Signaling Molecules Affects the Adipogenic Potential of ASC

On the basis of the reported ability of Wnt activation to reduce differentiation of 3T3-L1 preadipocytes to adipocytes [10], we hypothesized that the observed canonical Wnt signaling contributed to the negative regulation of ASC adipogenesis. Thus, we examined the specific effect of canonical Wnt signaling stimulation by either the Wnt agonist ([(2-amino-4-(3,4-(methylenedioxy)benzylamino)-6-(3-methoxyphenyl) pyrimidine)]; Calbiochem, San Diego, http://www.emdbiosciences.com) or recombinant Wnt3a protein (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) on both downstream Wnt pathway activation and features of adipogenesis. Activation of the β-catenin-TCF/LEF transcriptional coactivator was tested using the TCF-responsive luciferase reporter construct (Topflash). As shown in Figure 5A, promoter activity was increased by approximately 60% in ASC when stimulated with the Wnt agonist compared with unstimulated cells. When Wnt signaling was similarly activated during ASC differentiation induction, adipogenic differentiation, assessed by the adipogenic markers PPARγ and LPL, was markedly reduced (Fig. 5B). These data demonstrate the sufficiency of canonical Wnt signaling to repress adipogenic differentiation of ASC.

Figure Figure 5..

Effect of canonical Wnt signaling on adipogenic differentiation of adipose stromal cells (ASC). (A): ASC were treated with either Wnt signaling agonist (1 μM; Calbiochem) or recombinant Wnt protein (Wnt3a; 10 ng/ml) for 3 days prior to analysis of activation of canonical Wnt signaling using the TCF7-driven Topflash reporter assay. Note a 60% increase in Topflash activity with Wnt agonist stimulation. (B): Adipogenic differentiation of ASC after 3 days using gene expression analysis of PPARγ and LPL by reverse transcription polymerase chain reaction. A significant (*, p < .03) decrease was noted in adipogenic differentiation upon stimulation with Wnt agonist or a Wnt3a recombinant protein. Abbreviations: BM, basal medium; DM, differentiation medium; PPAR, peroxisome proliferator-activated receptor.

Gene Silencing of Wnt1 in Cocultures of ASC and EC Partially Derepresses ASC Adipogenic Differentiation

To determine whether particular Wnt factors, which were found to be upregulated in EC-ASC cocultures, were indeed responsible for the observed suppression of adipogenic differentiation in ASC-EC cocultures with EC (Fig. 1), specific Wnt factor genes were silenced with small inhibitory RNA pools (siRNA). We chose to test silencing of two candidate genes, Wnt1 and Wnt10a, because of their strong upregulation in both cocultured EC and ASC. Exposure to the respective siRNA pools was confirmed to cause 75% ± 10% reduction of Wnt1 or Wnt10a gene expression but not unspecific silencing, as demonstrated by assessing GAPDH expression (supplemental online Fig. 5). Adipogenic differentiation in the ASC-EC coculture assessed by the expression of PPARγ revealed a significant increase upon silencing of Wnt1 but not Wnt10a compared with scrambled siRNA (Fig. 6). These data provide the first evidence for the hypothesis that upregulation of specific Wnt factors in ASC, and Wnt1 in particular, may be a necessary mechanism for the negative regulation of adipogenic differentiation by exposure to EC.

Figure Figure 6..

Effect of Wnt1 gene silencing in endothelial cell/adipose stromal cell (ASC) cocultures. Silencing of Wnt1 in endothelial cells and ASC affected ASC adipogenesis in cocultures that had been transfected with Wnt siRNA or a scrambled siRNA control 24 hours prior to cultivation in adipocyte DM or in the BM. A significant (*, p < .03) increase in PPARγ but not LPL (#, p > .05) was observed with Wnt1 targeted siRNA versus scrambled siRNA. Of note, Wnt10a silencing did not significantly increase LPL and PPARγ (@, p > .05) expression compared with scrambled siRNA-transfected cells. The data shown are from a representative experiment, performed in triplicate, of three independent experiments. Abbreviations: BM, basal medium; DM, differentiation medium; PPAR, peroxisome proliferator-activated receptor; siRNA, short interfering RNA.


This is the first demonstration that interactions between EC and ASC negatively regulate the adipogenic potential of ASC. Repression of adipogenesis in EC-ASC cocultures was observed both with PPARγ agonist-containing (Fig. 1) and PPARγ agonist-free (supplemental online Fig. 3) adipogenic differentiation media. This suppression of adipogenesis correlated with the upregulation of several canonical Wnt-signaling molecules, as shown by gene expression analysis (Fig. 2). Strikingly, the upregulation of secreted, as well as intracellular, canonical Wnt signaling molecules was observed both in EC and in ASC (Fig. 3). This, in addition to specific differences noted in the program of ASC gene expression when in contact with EC versus EC-conditioned media, suggests that bidirectional communication between ASC and EC leads to the precise pattern of Wnt signaling molecule regulation in the periendothelial niche. The importance of Wnt signaling in general was made apparent by the potent reduction in adipogenesis that was induced by the addition of recombinant canonical Wnt protein or Wnt-signaling agonist. Conversely, derepression of adipogenesis in EC-ASC coculture was noted upon silencing the expression in ASC of at least one Wnt gene (Wnt1), which was upregulated in coculture. However, although Wnt1 silencing caused an increase in the expression of PPARγ, an early marker and essential mediator of adipogenesis, a less prominent enhanced expression of LPL, a marker of more mature adipocytes, was observed. The latter observation may be possibly explained by the short induction periods chosen (2 days), in the context of transient siRNA transfection. The ability of Wnt1 silencing to increase PPARγ and LPL as markers of adipogenic differentiation in the EC-ASC coculture model is consistent with a previous demonstration that overexpression of Wnt1 in 3T3-L1 cells can abolish adipogenesis [10]. This effect was explained by the ability of Wnt1-induced canonical signaling to inhibit the adipogenic transcription factors C/EBPα and PPARγ. Of note, not all Wnt proteins expressed in human ASC are expressed in the murine 3T3-L1 line and vice versa, including Wnt3a, which is involved in adipogenic differentiation of 3T3-L1 cells [15].

Surprisingly, silencing of Wnt10a, which was also upregulated in cocultures, did not show significant effects on the adipogenic transcription factor PPARγ in the cocultures. There are several possible explanations for this observation. These include the possibility that control of Wnt10a protein production occurs at the levels of translation or secretion rather than the assessed transcriptional levels. This is difficult to address because of the lack of antibodies specific for human Wnt proteins. Another possibility could be that there is inhibition of Wnt10a by specific binding proteins; the specificity of interactions among the respective families of Wnt ligands and Wnt binding proteins is largely unexplored to date. This could be addressed by silencing soluble receptors or binding proteins of Wnt proteins, including the Dickkopfs (Dkks), WIF [16, [17]–18], or sFRP2, the only soluble frizzled-related protein gene that we observed to be strongly upregulated in EC in coculture. Interestingly, sFRP2 not only is a Wnt antagonist but also was shown to induce cellular resistance to apoptosis and may act as a survival factor through Wnt-independent mechanisms [19]. An alternative explanation of our finding that Wnt1 is more important in coculture-mediated suppression of adipogenesis than Wnt10a may be that sFRP2 displays a higher affinity for Wnt10a than for Wnt1, and thus tonically and selectively ameliorates the activity of Wnt10a [20] in the context of ASC-EC interaction. Finally, noncanonical signaling Wnt proteins, which were not addressed in this study, may interfere with canonical signaling, as recently demonstrated for Wnt5b [15].

At least a portion of the EC-mediated reduction of adipogenesis of ASC was due to paracrine effects of factors secreted into the medium, as demonstrated by decreased Nile red staining in ASC treated with CM from EC (unpublished observation). This decreased adipogenic differentiation induced by CM can at least in part be explained by the concomitant enhanced levels of Wnts in ASC (Fig. 4), to levels similar to those observed in cocultures (Fig. 2). There were, however, also several striking differences between conditioned medium treatment and coculture experiments. First, Wnt1 became the most prominent of the induced Wnt ligands in ASC in conditioned medium-treated ASC. Another difference is that sFRP2 was strongly upregulated in coculture conditions in EC only and was not induced in ASC by EC-conditioned medium, thus demonstrating the effect of ASC to modulate EC factor secretion in ways that may in turn feed back to alter effects on ASC. Conversely, the intracellular Wnt signaling repressor, Axin, was upregulated only in conditioned medium-treated ASC (Fig. 4) but not in either cocultured ASC or EC. Another antagonist, WIF, is downregulated by contact in ASC but not in ASC treated with EC-conditioned medium. Comparison of these results from coculture and conditioned medium experiments suggest that both physical proximity and paracrine interactions can modulate agonistic and antagonistic Wnt signaling. However, functional and expression data indicate that the agonistic pathways prevail in both cases, at least in cells derived from healthy individuals.

Candidate factors produced by EC that may be responsible for controlling Wnt expression by ASC may be proteins that previously have been shown to inhibit adipogenesis in preadipocytes and are produced or activated by EC, such as platelet-derived growth factor, Ang-1, and transforming growth factor-β (TGFβ) [7]. In this context, TGFβ has been demonstrated both to cooperate with canonical Wnt signaling [21] and to evoke Wnt-signaling responses [22, 23]. Of note, in preliminary data we have observed induction of Wnt signaling proteins in ASC stimulated with TGFβ, although it was less pronounced than that observed with conditioned medium from EC (unpublished observation).

In this study we used ASC cells from a minimum of six different donors (nonobese, with BMI <30; average BMI 26.1 ± 2.8) to minimize variations due to genetic background and environmental influences. To have a stable source of presumably nondysfunctional EC, we used clones from cord blood-derived EC (derived from at least three independent donors). We speculate that there may be important differences in the effects on ASC of EC derived from aged or metabolically abnormal individuals. It may be that depending on the “history” of such donors, EC, as well as ASC, will exhibit alterations that may become epigenetic; in the event of a strongly activated endothelium, the antiadipogenic properties of the endothelium may be lost or even reverted. Future experiments will test these hypotheses by coculturing ASC and EC from donors of specifically differing metabolic status, as well as ASC with cytokine-activated endothelium.

Together, our data suggest that Wnt signaling pathways are activated by interaction of ASC and EC and that they may play a key role in ASC fate regulation in the context of the association of ASC/pericytes with EC in a perivascular niche. In consequence, other differentiation pathways, also, are expected to depend on Wnt signaling upregulation in this coculture. One candidate is myogenesis based on the switch in differentiation observed after inhibition of Wnt signaling in c2c12 myoblasts [10].


We have shown that endothelial cells can repress adipogenic differentiation of ASC in vitro and that this repression is at least in part mediated via Wnt1 dependent Wnt signaling. These findings are consistent with the previous description of Wnt factors as suppressors of adipogenesis and suggest a novel regulatory role of vascular endothelial cells in tissue differentiation processes. Future studies will reveal progressively more detailed insight into the mechanisms by which endothelial cells may modulate and control differentiation of nearby progenitor cells in the context of normal tissue maintenance, as well as in tissue growth and repair.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


This work was supported by NIH Grant R01-HL077688-04 (to K.L.M.), NIH Training Grant T32 HL 079995-02 (to D.O.T.), Eli Lilly and Co. (to K.L.M., W.C.R., M.C., and G.R.), and the Cryptic Masons Medical Research Foundation (to K.L.M.). W.C.R. is employed by Eli Lilly and Co.