Cell culture media, serum, buffers, and trypsin were purchased from Lonza Verviers (Verviers, Belgium) and cell culture reagents from Sigma-Aldrich Chimie (Saint-Quentin Fallavier, France). Rosiglitazone was purchased from BertinPharma (Montigny le bretonneux, France).

The establishment and characterization of the multipotency and self-renewal capacity of hMADS cells have been described (Rodriguez et al., 2004, 2005b; Elabd et al., 2007, 2009). In the experiments reported herein hMADS-1, 2, and 3 cells, established respectively, from the umbilical fat pad of a 31-month-old female donor, from the pubic region fat pad of a 5-year-old male donor and the prepubic fat pad of a 4-month-old male, were used between passages 16 and 35 corresponding from 35 to 100 population doublings. Cells were seeded at a density of 4500 cells/cm² in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS, 2.5 ng/ml hFGF2, 60 μg/ml penicillin, and 50 μg/ml streptomycin. The medium was changed every other day and hFGF2 was removed when cells reached confluence and were triggered for differentiation at day 2 post-confluence (designated as day 0). Cells were then maintained in DMEM/Ham’s F12 media supplemented with 10 μg/ml transferrin, 0.85 μM insulin, 0.2 nM triiodothyronine, 1 μM dexamethasone, 500 μM isobutyl-methylxanthine. Three days later, the medium was changed (dexamethasone and isobutyl-methylxanthine were omitted) and 100 nM rosiglitazone were added for the indicated periods. Media were then changed every other day and cells used at the indicated days. Glycerol-3-phosphate dehydrogenase (GPDH) activity measurements and Oil Red O staining were performed as described previously (Negrel et al., 1978; Bezy et al., 2005).

Total RNA was extracted using TRI-Reagent kit (Euromedex, Souffelweyersheim, France) according to the manufacturer’s instructions. Quality control for purity and integrity of RNA were tested by OD (260/280 nM) measurements and ethidium bromide-stained agarose analysis. Reverse transcription-polymerase chain reaction (RT-PCR) analysis was conducted as described previously (Rodriguez et al., 2004; Bezy et al., 2005; Zaragosi et al., 2006; Elabd et al., 2007). Primer sequences, designed using Primer Express software (Applied Biosystems, Courtaboeuf, France), are listed in Table 1 and were tested for their specificity, efficiency, reproducibility, and dynamic range. For quantitative PCR, final reaction volume was 20 μl using SYBR green master mix (Eurogentec, Angers, France) and assays were run on an ABI Prism 7700 real-time PCR machine (PerkinElmer Life and Analytical Sciences, Boston, MA, USA). An aliquot of PCR products was analyzed on 2% ethidium bromide-stained agarose. The expression of selected genes was normalized to that of TATA-box binding protein (TBP) gene and quantified using the comparative-ΔCt method. TBP expression did not vary along the adipocyte differentiation of hMADS cells; we used also 36B4 and POLR2A genes as housekeeping genes which gave similar data. Human skeletal muscle RNA extracts were obtained from a previous study (Pisani et al., 2010).

Living cells were submitted to 100 nM MitoTracker Red FM (Invitrogen) for 45 min at 37°C. Cells were finally washed with prewarmed culture medium and visualized with an Axiovert microscope (Carl Zeiss, Le Pecq, France) and pictures were captured and treated with AxioVision software (Carl Zeiss).

Cells were disrupted using a polytron in 10 mM Tris pH 8, 1 mM EDTA, and 0.25 M Sucrose containing protease inhibitors. Cell lysate was centrifuged at 750 g for 10 min with a pellet corresponding to a nuclear-enriched fraction. The supernatant was then centrifuged at 10 000 g for 20 min, the pellet corresponding to a mitochondrial-enriched fraction. The mitochondrial-enriched fraction was used to determine Cytochrome c Oxidase activity according to manufacturer’s instructions (Cytochrome c Oxidase Assay kit, Sigma).

Data are expressed as mean values ± SEM and are analyzed using the 2-tailed Student’s t-test. Differences were considered statistically significant at P ≤ 0.05.

We showed previously that hMADS cells were able to differentiate into white and brown adipocytes depending on the length of activation of PPARγ by rosiglitazone (Elabd et al., 2009). For hMADS-1 cells, originally established from the umbilical fat pad of a 31-month-old female donor, a 6-day exposure to rosiglitazone (between days 3 and 9) proved to be optimal for the cells to become white adipocytes. As shown in Figures 1A,B, further exposure of the cells to rosiglitazone did not alter the overall adipogenesis level per se since treatment between days 3 and 16 brought no change in triglyceride accumulation (Oil Red O staining) and GPDH activity. Under this condition, as shown in Figure 1C, a dramatic increase in the expression of UCP1 mRNA was observed. Interestingly, when hMADS-1 cells were first differentiated as above into white adipocytes and rosiglitazone being removed for the next 5 days, a further 2-day rosiglitazone exposure (between days 14 and 16) was sufficient to stimulate the expression of UCP1 (Figure 1C). In order to extend these observations, hMADS-2 and hMADS-3 cells, established respectively from the pubic fat pad of a 5-year-old male donor and the prepubic fat pad of a 4-month-old male, were used and gave similar results (Figure 1C). Under these conditions, UCP1 protein could be detected by immunoblotting analysis (data not shown).

We then measured for the expression of two other members of the UCP family, UCP2 and UCP3. The expression of UCP2 mRNA was induced during adipogenesis of hMADS-1, 2, and 3 cells, and its expression was partially but not significantly enhanced by the duration of rosiglitazone treatment (Figure 1D). Interestingly, UCP3 mRNA was barely detected (100–200 times lower as compared to skeletal muscle cells), was not induced during adipogenesis and remained insensitive to rosiglitazone treatment (Figure 1E).

Among members of the thiazolidinedione family, rosiglitazone is known as a high-affinity ligand of PPARγ. We therefore carried out a comparative study with other thiazolidinediones, i.e., troglitazone and ciglitazone, with respect to the induction of UCP1 gene expression. For this purpose, hMADS-2 cells were first differentiated into white adipocytes as above, i.e., rosiglitazone treatment between days 3 and 9 followed by its removal for the next 5 days, were then exposed to increasing concentrations of various thiazolidinediones between days 14 and 16. The levels of UCP1 mRNA were analyzed by quantitative RT-PCR at day 16. As shown in Figure 2A, troglitazone and ciglitazone were able to induce UCP1 expression in a dose-dependent manner. Clearly, the potency of rosiglitazone was already maximal at a dose where the effects of troglitazone and ciglitazone were weak. In order to gain further insights into the critical role of PPARγ, we tested whether a specific PPARγ antagonist (GW9662) was able to abolish the effect of rosiglitazone on UCP1 expression. As above, hMADS-2 cells were first differentiated into white adipocytes, then treated or not with 100 nM rosiglitazone in the presence or absence of the PPARγ antagonist. As shown in Figure 2B the rosiglitazone-induced UCP1 mRNA expression was abolished in a dose-dependent manner by GW9662. Of note, GW9662 showed no effect on untreated hMADS cells. Similar data were observed using hMADS-3 cells (data not shown). Altogether, these observations demonstrated that induction of UCP1 was dependent on PPARγ activation and that rosiglitazone was the most efficient agonist in this process.

As abundance of mitochondria is higher in brown as compared to white fat cells, we tested whether mitochondriogenesis was affected in the “browning” process of hMADS cells. Clearly, more mitochondria were observed in brown-like adipocytes as compared to white adipocytes (Figure 3A). This observation was further supported by quantitative RT-PCR analysis of CPT1B, a fatty acid transporter of the outer mitochondrial membrane (Figure 3B). Under these conditions, CPT1B mRNA levels were 10-fold higher in brown-like adipocytes as compared to white adipocytes. Furthermore, the increase of CPT1B mRNA levels upon rosiglitazone treatment was similar to that observed for UCP1 (Figure 1C).

Finally, we measured the Cytochrome c oxidase activity in this fraction as an index of respiratory chain activity. As expected, Cytochrome c oxidase activity was higher in UCP1-containing fractions from hMADS cells chronically treated as compared to cells treated between days 3 and 9 (Figure 3C). Thus, hMADS adipocytes expressing UCP1 display a significant increase in mitochondriogenesis accompanied by an increase in their respiratory activity; both indicating the acquisition of a brown-like phenotype. As brown-like adipocytes arise from white adipocytes, brown-like adipocytes can be defined at first sight under our conditions as brite adipocytes.

Recently, studies of mouse model have described genes with differential expression in white vs. brown adipocytes (Vernochet et al., 2009; Petrovic et al., 2010). Therefore we aimed at analyzing the expression of a large set of these genes in white vs. brown-like hMADS adipocytes.

First, we analyzed the expression of Myf5 known to be a molecular signature common to brown-fat and skeletal muscle cells (Seale et al., 2008). Myf5 expression was not detectable in proliferating or differentiated hMADS cells originating from three donors (data not shown).

Table 2 summarized the expression values (ΔCt referred to TBP) of different genes in hMADS cells at day 9 of adipocyte differentiation as well as day 16 of white and brite differentiation of hMADS-2 and 3 cells. White and brite hMADS adipocytes expressed classical adipogenic markers (FABP4, ADIPOQ, RARRES2, PANK3, PPARγ, and low level of LXR) at similar levels. In addition, hMADS brite adipocytes expressed UCP1, CPT1B, CIDEA, and ELOVL3, four genes found to be highly expressed in mouse brite adipocytes (Petrovic et al., 2010). Conversely, these cells did not express ZIC1 and LHX8, two genuine brown adipocyte specific genes (Vernochet et al., 2009). Interestingly, hMADS cells expressed PRDM16, MEOX2, and AGT at the same level in both white and brite adipocytes. With respect to white adipocyte specific genes, it is noticeable that leptin (LEP) and DPT but not HOXC9 levels decreased with brite adipocyte formation.

Co-activators (PGC1α, PGC1β) and co-repressors (CtBP1, CtBP2, NRIP1) of PPARγ were also expressed, demonstrating the ability of these cells to modulate PPAR signaling pathways. hMADS cells also expressed other classical components of brown adipocyte signaling such as DiO2 (thyroid hormone pathway), COX1, COX2, PGTiS, and PGTiR (arachidonic acid pathway) and interestingly, an increase in the bile acid receptor, GPBAR1.

Altogether, these observations clearly confirmed the “brite” phenotype of hMADS adipocytes as they display both an origin and a molecular signature distinct from those of genuine brown adipocytes.

As conversion of hMADS cells into brite adipocytes is dependent upon PPARγ activation, we became interested in deciphering pathways that are independent of rosiglitazone treatment. It has been shown in mouse models that Hedgehog and BMP pathways modulated the formation of brown adipocytes (Lee et al., 2008; Pospisilik et al., 2010; Schulz et al., 2011). We analyzed UCP1 expression after activation of the Hedgehog pathway, using two known activators of smoothen, i.e., purmorphamine and smoothened agonist (SAG). hMADS cells that were differentiated in the presence of rosiglitazone either between day 3 and 9 or between day 3 and 16 were exposed to the two compounds and analyzed at day 16. As expected, the expression of Gli1 mRNA (Figure 4A), a marker of the activation of Hedgehog signaling (Hooper and Scott, 2005), was observed. However, neither purmorphamine nor SAG were able to substitute to rosiglitazone and to modulate the expression of UCP1 or FABP4.

In a similar way we tested whether BMPs could induce the “browning” of hMADS cells. hMADS cells, in the presence or absence of rosiglitazone between days 3 and 9, were treated with or without 10 nM BMP2 or BMP7. As shown in Figure 4B, neither BMP2 nor BMP7 affected adipogenesis of hMADS cells, as illustrated by the expression of FABP4 and only induced weakly the formation of brite adipocytes as illustrated by the expression of UCP1 and CIDEA.