, Volume 46, Issue 6, pp 487–494

Temporary Increase of PPAR-γ and Transient Expression of UCP-1 in Stromal Vascular Fraction Isolated Human Adipocyte Derived Stem Cells During Adipogenesis


  • Seong Jin Jo
    • Department of DermatologySeoul National University College of Medicine
  • Won Woo Choi
    • Wells Dermatology Clinic
  • Eun Seong Lee
    • Center for NanoBio ConvergenceKorea Research Institute of Standards and Science
  • Jae Yong Lee
    • Center for NanoBio ConvergenceKorea Research Institute of Standards and Science
  • Hyun Sun Park
    • Department of DermatologySeoul National University College of Medicine
  • Dae Won Moon
    • Center for NanoBio ConvergenceKorea Research Institute of Standards and Science
  • Hee Chul Eun
    • Department of DermatologySeoul National University College of Medicine
    • Department of DermatologySeoul National University College of Medicine
Original Article

DOI: 10.1007/s11745-011-3525-5

Cite this article as:
Jo, S.J., Choi, W.W., Lee, E.S. et al. Lipids (2011) 46: 487. doi:10.1007/s11745-011-3525-5


In this study, cells from the stromal vascular fraction of human subcutaneous tissues were induced to differentiate toward adipose cells in vitro for 2 weeks. During adipogenic differentiation, we followed the chronological changes in their morphology with Coherent anti-Stokes Raman scattering (CARS) microscopy and checked the PPAR-γ and UCP-1 expression with RT-PCR. On day 4 after inducing adipogenic differentiation, CARS imaging showed multiple small lipid droplets (LD) distributed peripherally along the cellular membrane. PPAR-γ began to express at this time and increased until day 14 at a steady rate. On day 7, the cells appeared as brown adipocytes with numerous small LD throughout the cytoplasm, and the mRNA level of UCP-1 rose abruptly by 6- to 7-fold. After an additional 7 days, CARS imaging showed the development of a large LD, which is characteristic of white adipocytes, and the mRNA level of UCP-1 slumped significantly. These results demonstrate the possibility that ADSC pass through a brown adipocyte-like stage while differentiating into white adipocytes.


AdipogenesisAdipose-derived stem cellUCP-1



Adipose-derived stem cell


Brown adipose tissue


Coherent anti-Stokes Raman scattering


Dulbecco’s modified Eagle’s medium


Fetal bovine serum




Lipid droplet


Phosphate-buffered saline


Peroxisome proliferator-activated receptor-γ


Stromal vascular fraction


Mitochondria-uncoupling protein-1


White adipose tissue


Adipose tissue has been classified into two distinct types in mammals: WAT and BAT [1]. WAT is mainly involved in energy storage [2], while BAT forms heat using accumulated lipids [3] and the thermogenic capability depends on the presence of UCP-1 which is unique and specific to brown adipocytes [4, 5]. Although both of them are important to control the energy balance, they are regarded as distinct tissues and thought to develop independently for several reasons. One of the reasons is that white adipocytes do not express UCP-1 during adipogenesis, which was supported indirectly by an experiment using double transgenic mice [6].

Adipogenesis is a complicated process, which eventually results in the expression of various adipocyte-associated genes as well as an increased capacity of the cell for lipid-filling [7, 8]. For the past 20 years, many studies of adipogenesis have been conducted using preadipocyte clonal lines from rodents [810]. Although the results of studies using established cell lines have been invaluable, they are limited in their applicability to an in vivo human context because of their aneuploidy, unipotency, site-restriction and the species–species differences between humans and mice [9, 11]. Recently, a new era has started after the identification of human ADSC, mesenchymal stem cells isolated from adipose tissue. ADSC have a self-renewal capacity and multi-lineage potential toward adipocytes, osteoblast, chondrocytes, myoblasts and neuronal cells [12, 13]. Additionally, they are easily processed from lipoaspirated fat and can provide a significant quantity of multipotent stem cells [14]. Compared to clonal cell lines from mice, human ADSC may be much more appropriate for studies of adipocyte differentiation because they can offset the disadvantages of previous models and can better reflect the in vivo human context.

CARS microscopy is an advanced technique for the real-time imaging of live cells. It requires neither fixation nor a fluorescent probe. In the CARS process, a “pump” beam at a higher frequency (ωp), a “Stokes” beam at a lower frequency (ωs) and a “probe” beam (\( \omega^{\prime}_{\text{p}} \)) at the same frequency as the pump beam interact with a sample to generate an “anti-Stokes” beam, presented in the form of the equation 2ωp − ωs. These signals are maximized by tuning the frequency difference (ωp − ωs) to the vibration of specific chemical bonds, which provides chemical selectivity in CARS microscopy [15]. In particular, CARS microscopy shows high sensitivity to the C–H vibration of lipid-rich molecules. Thus, CARS microscopy has been used to visualize lipids, axonal myelin sheaths and lipid-rich cells [16, 17]. Moreover, it was shown to be effective for imaging LD, in which the fatty acyl group is highly accumulated, in live cells without fixation [18].

In this study, we isolated the stromal-vascular fraction (SVF) cells from human subcutaneous tissue which contains ADSC and induced the cells to differentiate toward adipose tissue in vitro. To characterize the morphologic and molecular change of the cells during adipogenic differentiation, we serially imaged live cells with CARS microscopy and examined the PPAR-γ and UCP-1 expression with RT-PCR. On day 7 of adipogenesis, these cells transiently appeared multilocular with numerous small LD with an abrupt increase of UCP-1, thus suggesting that stem cells passes through a brown adipocytes-like stage while differentiating to white adipocytes.

Materials and Methods

Isolation and Culture of Human ADSC

Before the study, the protocols for fat collection were reviewed and approved by the institutional research board of the Seoul National University Hospital Clinical Research Institute, and informed consent was obtained from the human subjects. Liposuction aspirates from abdominal subcutaneous tissue were acquired from three healthy female donors undergoing elective procedures under tumescent anesthesia. The lipoaspirates were washed with PBS and were subsequently finely minced. The extracellular matrix was digested with 0.075% collagenase type I (Sigma-Aldrich, St. Louis, MO). Enzyme activity was neutralized with DMEM containing 10% FBS and the tissue was centrifuged for 10 min at 1,200g at room temperature. The supernatant, containing mature adipocytes, was aspirated. The pelleted SVF cells were collected and then plated in 175 cm2 flasks (Becton–Dickinson AG, Basel. Switzerland) in a control medium which consisted of DMEM, 10% FBS, and 1% penicillin/streptomycin 5,000 U/ml (Invitrogen AG, Bagel, Switzerland). After that, they were incubated overnight at 37 °C in 5% CO2. The resulting cell population includes human ADSC. [12, 13]. This initial passage was referred to as passage 0 (P0). After they achieved 90% confluence, the cells were passaged repeatedly until P8 by trypsin (0.05%, Invitrogen AG, Basel, Switzerland) in the same way.

Flow Cytometry

Cells at passage 4 were harvested and labeled with the following anti-human antibodies: anti-CD29-FITC (Santa Cruz Biotechnology, Santa Cruz, CA), anti-CD44-FITC (Dako, Carpinteria, CA), anti-CD31-FITC, and anti-CD34-FITC (Becton–Dickinson, San Diego, CA). Mouse isotype antibodies served as controls. Cells were analyzed using a FACS Calibur flow cytometer (Becton–Dickinson) [19].

Multilineage Differentiation of Human ADSC in Vitro

Human ADSC at passage 4 were analyzed to confirm their capacity to differentiate toward the adipogenic, osteogenic, and chondrogenic lineages. To induce the respective differentiations, ADSC were cultured with previously described lineage-specific induction media [12, 20, 21], as detailed in Table 1. To induce adipogenic and osteogenic differentiation, cultures in respective lineage-specific media were maintained at 37 °C in 5% CO2 for 2 weeks. Chondrogenic differentiation was determined using a slight modification of a previously described pellet culture technique [22]. Briefly, ADSC were placed in 15 ml polypropylene conical tubes, centrifuged at 1,500 rpm for 5 min, and then cultured in chondrogenic medium (CM) at 37 °C in 5% CO2 for 4 weeks. ADSC maintained in the control medium were analyzed as negative controls. During the culturing process, the medium was changed every 2 days.
Table 1

Media supplementation for lineage-specific differentiation





Adipogenic (AM)


10% FBS

0.5 mM isobutyl-methylxanthine (IBMX), 1 μM dexamethasone, 10 μM insulin, 200 μM indomethacin, 1% antibiotic/antimycotic

Osteogenic (OM)


10% FBS

0.1 μM dexamethasone, 50 μM ascorbate-2-phosphate, 10 mM β-glycerophosphate, 1% antibiotic/antimycotic

Chondrogenic (CM)


10% FBS

1.08 μM insulin, 0.23 nMl TGF-β1, 50 nM ascorbate-2-phosphate, 1% antibiotic/antimycotic

Histological Analysis

Differentiated ADSC were confirmed using histological assays [12]. Oil red O stain was used for adipogenesis, alkaline phosphatase (AP) activity and von Kossa stain for osteogenesis, and Safranin O stain for chondrogenesis. For Oil red O stain, cells were fixed with a 3.7% solution of formaldehyde, washed with distilled water, and stained with Oil red O solution for 10 min. AP activity was detected by fixing the cells with a 3.7% solution of formaldehyde and staining with AP solution including 1% naphthol ABSI phosphate. For von Kossa staining, the cells were fixed with 1% paraformaldehyde for 15 min and overlaid for 15 min with a 5% silver nitrate solution in the dark at room temperature. Then, they were left under UV light for 20 min, followed by incubation with sodium thiosulfate for 2 min. For Safranin-O staining, the tissue sections were deparaffinized with xylene and ethanol. Then 1% aqueous Safranin-O was added for 30 min and 0.2% fast green was added for 3 min. After that, the sections were washed with distilled water and with serial concentrations of 70, 80, and 95% ethanol.

Coherent Anti-Stokes Raman Scattering (CARS) Microscopy

For the CARS image measurements, the pump and the Stokes beams were generated from two synchronized 76-MHz near-infrared mode-locked lasers. An Nd:Vanadate laser with a pulse duration of 7 ps (PicoTRAIN, High Q Laser Production GmbH, Hohenems, Austria) was used as the Stokes beam at 1,064 nm, and a pump beam with a duration of 6 ps at 776 nm was generated from an intracavity doubled optical parametric oscillator (Levante, APE GmbH, Berlin, Germany) that is synchronously pumped by the Nd:vanadate laser. The power levels of the pump and the Stokes beams were kept around 50 and 25 mW, respectively. A multiplexed pump beam of 30 nm bandwidth at 817.2 nm was produced from a femtosecond Ti:sapphire laser (Micra-10, Coherent Inc., Santa Clara, CA) with an average power of 900 mW, whose output pulse train was synchronized with that of the 1,064 nm picosecond laser by using an electronic cavity feedback module (SynchroLock-AP, Coherent Inc.). The Raman shift value set in the laser wavelengths was nearly in resonance with the CH-stretching vibrational modes of the sample. The synchronized three laser beams were then collinearly combined and sent to and inverted optical microscope (IX81, Olympus, Tokyo, Japan). The CARS excitation beams were then focused onto a sample by a 1.2-NA water immersion objective lens (UPLSAPO/IR 60X, Olympus). The forward CARS signal in the lipid window (644–683 nm) was collected by a 0.55-NA condenser lens and directed to the photomultiplier tube (R3896, Hamamatsu Photonics, Hamamatsu, Japan). The CARS images having a maximum field of view of 250 × 250 μm2 were acquired by performing raster scans of the laser beams on the sample.

Reverse Transcription-Polymerase Chain Reaction (RT-PCR) Analysis

Adipogenic differentiated cells were collected at defined time points (day 1, 4, 7 and 14), and non-induced cells were also examined as a negative control. Total RNA from each cell group was extracted by TRIzol reagent (Invitrogen) and was reverse transcribed using the TaqMan Gold RT-PCR kit (Applied Biosystems, Foster City, CA, USA). All real-time PCR measurements were performed using an ABI 7000 real-time PCR system (Applied Biosystems) according to the standard temperature cycling protocol for the relative quantification assays. The expression of PPAR-γ and UCP-1 was quantitated and the primer pairs used were as follows: PPAR-γ: 5′-ATGACAGCGACTTGGCAA-3′ (forward primer) and 5′-AATGTTGGCAGTGGCTCA-3′ (reverse primer); UCP-1: 5′-AGAGCCATCTCCACGGAA-3′ (forward primer) and 5′-CCAGGATCCAAGTCGCAA-3′ (reverse primer). The expression of human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used to normalize the gene expression levels.

Image Analysis

We used the IMT morphology program (iMT technology, Bucheon, Korea) to count the number of LD and to estimate the size of each LD.

Statistical Method

Statistical significance was determined using the Student’s t-test and a p value of <0.05 was considered significant. Statistical analysis was performed using SPSS version 17.0 (SPSS, Chicago, IL, USA).


Primary Culture and Phenotypic Characterization of SVF Cells

SVF cells including ADSC extracted from the lipoaspirates adhered to the tissue culture dish and grew into spindle-shaped cells, while non-adherent cells such as red blood cells were removed by changing the media. SVF cells proliferated rapidly and exhibited a relatively consistent population doubling rate from passage 0–8 (data not shown). At passage 4, these cells appeared to have a fibroblast-like shape. To examine the immunophenotype of these cells, we characterized the cell population according to its CD marker profile using flow cytometry. The results demonstrated that cells at passage 4 expressed CD29 (99.5%) and CD44 (97.6%) but did not express hematopoietic lineage markers CD31 (6.1%) and CD34 (5.4%) (Fig. 1).
Fig. 1

Phenotypic characterization of human SVF cells. Cells at passage 4 were harvested and flow cytometric analyses of the expression of CD29, CD31, CD34 and CD44 were performed

Multi-lineage Differentiation of ADSC

To verify the ADSC, we tested the multilineage capacity of the cultured cells. Cells were differentiated toward the adipogenic, osteogenic, and chondrogenic lineages using lineage-specific induction media, and the instances of differentiation were assessed by histology (Fig. 2). Our cells were cultured in AM for 2 weeks and developed lipid-containing droplets which were stained by Oil Red O, which is consistent with the phenotype of mature adipocytes. Cells treated with OM formed a dense and extensive network. They expressed the increased AP activity by which an osteoblast was characterized and calcification as detected by von Kossa staining. Finally, Safranin-O staining showed distinct proteoglycan production, which indicated chondrogenic differentiation, in cells cultured in CM by the pellet technique. These indicated that ADSC composed the SVF cells from human lipoaspirates.
Fig. 2

Multi-lineage differentiation of SVF cells in vitro. SVF cells (a) were cultured in lineage-specific media respectively: adipogenic media (AM), osteogenic media (OM), and chondrogenic media (CM). Adipogenic differentiation was confirmed by Oil red O staining (b), osteogenic differentiation by alkaline phosphatase activity (c) and von Kossa staining (d), and chondrogenic differentiation by Safranin O staining (e). These indicated that the SVF cells contained ADSC

Serial Observation of Human ADSC in Adipogenic Differentiation

SVF cells including ADSC were cultured in control medium and then switched to AM at passage 4. Using Oil red O stain and CARS microscopy, we observed the chronological change of morphology during the adipogenic differentiation of human ADSC on day 0, 1, 7, and 14 after induction. Treatment with AM generally resulted in enlarged cell morphology and a time-dependent increase in intracellular LD (Fig. 3). Details are as follows: on day 0, the cells were elongated and fibroblast-like. Oil red O staining was negative, and the CARS image showed a centered nucleus with low signal intensity and a ground glass-appearing cytoplasm with multiple medium signal intensity spots.
Fig. 3

Chronological observation of adipogenic differentiation with Oil red O staining (a) and CARS microscopy (b). Lipid droplets (LD) began to be stained sparsely on day 4 and showed a continuous increment of staining by Oil red O (a). CARS imaging also demonstrated that multiple 1–3 μm LD were distributed at the periphery of the cells on day 4, which then enlarged to 4–6 μm on day 7 with a centrally placed nucleus. On day 14, a few 10 μm LD pushing the nucleus to the periphery developed (b), and this was clearer in the magnified figure (c). The number of LD per cell decreased on day 14 compared to that on day 7 (d), while the mean size of LD increased (e). LD lipid droplet, Nu nucleus, a.u. arbitrary unit). *p < 0.05 and **p < 0.01

On Day 1, there appeared to be no significant changes in the morphology of the cells. Cells were still elongated and not stained with Oil red O. The CARS finding was somewhat similar to that on day 0. The characteristic changes began to be observed only after 4 days. The ADSC had become rounder, and Oil red O staining showed sparse positively stained LD. Additionally, multiple 1–3 μm sized high signal globules were distributed in a lace-like pattern peripherally along the cell border in the CARS image.

On day 7, slightly enlarged, rounder cells were observed. They had several enlarged and positively stained LD with Oil red O. The CARS image demonstrated a number of 4–6 μm high signal globules that filled the entire cytoplasm. The nucleus was spared and positioned in the center or on the slight periphery. These morphologic features in CARS image are similar to those of the brown adipocytes [1].

Treatment of AM for additional 7 days enlarged the cells much more. On day 14, Oil red O staining showed large and densely positive LD. On CARS microscopy, the ADSC resembled white adipocytes with a large globule approximately 10 μm in size that had been generated from aggregation of adjacent LD. The nucleus was distorted and leaned onto the cytoplasmic membrane. The number of LD per cell decreased compared to that on day 7.

Monitoring Adipogenesis with Real-Time qPCR

Real-time qPCR demonstrated that PPAR-γ was expressed at an essentially steady rate since it was detected initially on day 4 (Fig. 4a). The mRNA level of UCP-1, a specific marker of brown adipocytes, rose 6- to 7-fold abruptly on day 7 (Fig. 4b), however, slumped significantly on day 14.
Fig. 4

Assessment of gene expression in adipogenesis. The mRNA level of PPAR-γ increased steadily since it was detected initially on day 4 (a). UCP-1 showed an abrupt six–sevenfold increment on day 7 though this regressed on day 14 (b). *p < 0.05 and **p < 0.01 versus day 0


To reflect an in vivo human context better, we undertook this study using cells from human subcutaneous tissue. Before inducing these cells to differentiate toward adipose tissue, we showed that SVF cells exhibit a stable proliferation rate and a self-renewal capacity, and were positive for mesenchymal cell markers CD29, CD44 but negative for CD31, CD34. This indicates that our cells were not contaminated with progenitors from bone marrow or hematopoietic lineage cells from blood vessels. In addition, we demonstrated the multipotentiality of our cells at passage 4 through differentiating them into osteoblasts, adipocytes, and chondrocytes using lineage-specific induction media. Thus, we concluded that the cultured cells included previously known ADSC. Afterward, the cells including ADSC were induced to differentiate into adipocytes as described previously [12, 20]. In the adipogenic media, IBMX and dexamethasone turn on the transcriptional factor PPAR-γ to direct the transcription of the lipid synthesis gene, and insulin facilitates glucose uptake and promotes adipocyte differentiation of ADSC [23]. We continuously used this media for 2 weeks.

During adipogenic differentiation, we used CARS microscopy to image ADSC. Traditionally, LD, characteristic structures in adipocytes, can be labeled with Oil red O and imaged using fluorescent microscopy [24]. However, these methods are only applicable to a fixed sample; they disturb LD structures significantly when organic solvents such as ethanol and acetone are used for fixation [25]. Even with formalin fixation, aggregation, fusion, deformation and growth in size of LD arise [18, 26], preventing adipogenesis from being precisely evaluated. In contrast, CARS microscopy allows selective imaging of LD in unstained live cells with a very high contrast, as LD are aggregates of neutral lipids, mainly triglycerides rich in C–H bonds. Consequently, compared to conventional methods, we were able to observe adipogenic differentiation of live ADSC in real-time without fixing and without perturbing the cells.

In addition to the images of high-signal-intensity LD from day 4, CARS images on day 7 were reminiscent of brown adipocytes with numerous scattered small LD showing a multilocular appearance with a very recognizable nucleus, which was either central or somewhat displaced at the periphery of the cell in accordance with the amount of LD. These features were strikingly similar to those of brown adipocytes in human newborns [27]. However, on day 14, a single large LD was eventually formed perhaps through the coalescence of numerous small LD, pushing the nucleus to the side, which is a typical morphology of mature white adipocytes. The result from the real-time qPCR of UCP-1, which is considered to be the single and best characterized means to distinguish clearly both cell types thus far [5, 6, 28], was consistent with the morphological change of ADSC in adipogenesis; UCP-1 showed a sudden peak on day 7 with a brown adipocyte-like appearance of ADSC, slumping on day 14 with a white adipocyte-like appearance.

As described previously, WAT and BAT has been thought to develop independently for several reasons. First, the developmental patterns of both adipose cells are different, as BAT occurs during late gestation and possesses all the features of mature tissue at birth whereas WAT development takes place mainly after birth [1]. Second, it has been assumed that functional BAT is absent in healthy adults. Third, in vitro precursor cells isolated from WAT or BAT are already committed and therefore differentiate primarily into white or brown adipocytes, respectively [2931]. However, recent studies have shown that brown adipocytes are dispersed throughout human adipose tissue and are metabolically active [32]. It was also found that BAT activity occurs in healthy men during exposure to cold [33]. Additionally, large depots of BAT development were observed in a patient with a pheochromocytoma that secretes catecholamines [34], although it remains undetermined whether this was reactivation of remnant BAT from neonatal depots or conversion from WAT. Regarding the conversion of white adipocyte to brown adipocyte tissue, Tiraby et al. [35] suggested that it is possible to induce a metabolic shift in white fat cells with PPAR-γ coactivator 1α (PGC-1α) experimentally. Furthermore, upon chronic exposure to rosiglitazone, a specific PPAR-γ agonist, ADSC-derived white adipocytes are able to switch to brown adipocytes by expressing UCP-1 [28]. Our results also support that WAT and BAT do not develop independently, suggesting that white adipocytes undergo a momentary brown adipocyte-like stage during their development. Even if this is not the case, it remains possible that a certain association exists between the developmental processes of both adipocytes.

This study has some limitations. The SVF cells usually represent a very heterogeneous population, which means that every cell we dealt with was not a multipotent stem cell. However, ADSC seems to be mostly responsible for the changes during adipogenic differentiation including transient increase of UCP-1 because cells other than ADSC do not show adipogenic differentiation when cultured in AM. Another limitation is that the morphology of ADSC on day 14 during adipogenesis was not completely identical to that of mature lipid cells. There still remained several small LD although a distinctively large globule developed. However, we think that ADSC would show unilocular LD finally if observation were extended a few days more.

In summary, we observed and characterized the chronological stages of SVF cells including ADSC during adipogenic differentiation. On day 7 of adipogenesis, cells appeared multilocular with numerous small LD according to CARS imaging and the level of UCP-1 increased significantly. However, on day 14, they showed a unilocular appearance with a large single LD and the level of UCP-1 decreased markedly. Based on this morphological change and on the transient expression of UCP-1, we suggest that ADSC passes through a brown adipocyte-like stage while differentiating to white adipocytes.


This work was supported by the Next-Generation New-Technology Development Program of MKE, and by the Bio-signal Analysis Technology Innovation Program (No. 2009-0084137) and the National Research Foundation of Korea grant (No. 2009-0092835) of MEST, Republic of Korea.

Conflict of Interest


Copyright information

© AOCS 2011