Rosiglitazone modifies the adipogenic potential of human muscle satellite cells
- First Online:
- Cite this article as:
- De Coppi, P., Milan, G., Scarda, A. et al. Diabetologia (2006) 49: 1962. doi:10.1007/s00125-006-0304-6
- 726 Downloads
Satellite cells are responsible for postnatal skeletal muscle regeneration. It has been demonstrated that mouse satellite cells behave as multipotent stem cells. We studied the differentiation capacities of human satellite cells and evaluated the effect of the insulin sensitiser rosiglitazone, a well known peroxisome proliferative activated receptor gamma (PPARG) agonist, on their adipogenic conversion.
Subjects, materials and methods
We obtained human satellite cells from human muscle biopsies of healthy subjects by single-fibre isolation and cultured them under myogenic, osteogenic and adipogenic conditions. Moreover, we compared the morphological features and the adipose-specific gene expression profiling, as assessed by quantitative PCR, between adipocytes differentiated from human satellite cells and those obtained from the stromal vascular fraction of human visceral fat.
We proved by morphological analysis, mRNA expression and immunohistochemistry that human satellite cells are able to differentiate into myotubes, adipocytes and osteocytes. The addition of rosiglitazone to the adipogenic medium strongly activated PPARG expression and enhanced adipogenesis in human satellite cells, but did not in itself trigger the complete adipogenic programme. Moreover, we observed a decrease in wingless-type MMTV integration site family member 10B and an upregulation of growth differentiation factor 8 expression, both being independent of PPARG activation.
Human satellite cells possess a clear adipogenic potential that could explain the presence of mature adipocytes within skeletal muscle in pathological conditions such as obesity, type 2 diabetes and ageing-related sarcopenia. Rosiglitazone treatment, while enhancing adipogenesis, induces a more favourable pattern of adipocytokine expression in satellite-derived fat cells. This could partially counteract the worsening effect of intermuscular adipose tissue depots on muscle insulin sensitivity.
KeywordsAdipocytokines Adipogenesis Ectopic fat depot Insulin resistance Muscle Rosiglitazone Satellite cells
adiponectin, C1q and collagen domain containing
CCAAT/enhancer binding protein alpha
Electronic supplementary material
fatty acid-binding protein 4
fetal bovine serum
growth differentiation factor 8, gene encoding for myostatin
leptin, murin obesity homologue
myogenic differentiation 1
paired box gene 7
pre-B-cell colony enhancing factor 1, gene encoding visfatin
peroxisome proliferative activated receptor gamma
runt-related transcription factor 2
wingless-type MMTV integration site family member 10B
Lipotoxicity could result from NEFA overflow, altered adipose tissue topography, impaired adipocyte function and increase of fat cells in non-adipose organs. Particularly at the skeletal muscle level, both intramyocellular lipids and intermuscular adipose tissue depots directly correlate with the degree of insulin resistance in obese subjects and patients with type 2 diabetes [1, 2, 3]. What is the origin of fat cells surrounding muscle bundles? They could derive from different progenitors normally present in the adult skeletal muscle: mesenchymal stem cells, muscle-derived stem cells or satellite cells . After birth, muscle regeneration is mostly mediated by satellite cells, a unique population of committed stem cells located adjacent to the plasma membrane of myofibres . Several studies carried out in animal models have shown that satellite cells give rise to new myotubes, but they can also differentiate into other cells of mesenchymal origin. In particular, it has been shown that in vitro mouse satellite cells can directly differentiate into adipocytes, osteocytes and chondrocytes [6, 7]. Moreover, thiazolidinediones treatment induces C2C12 myoblasts and satellite cells isolated from normal mice to differentiate into adipocyte-like cells .
There is evidence that a balance exists between myogenic and adipogenic differentiation programmes in muscle satellite stem cells and that this balance could be altered in particular pathological conditions. It remains to be determined whether tuning of this regulation contributes to the increased adiposity in muscle.
Peroxisome proliferative activated receptor gamma (PPARG) is the major transcription factor regulating adipogenesis, which results in the acquisition of the mature adipocyte phenotype with the presence of secretory factors collectively called adipokines [9, 10].
It has been demonstrated that PPARG overexpression in fibroblasts and the myoblast cell line-G8 was sufficient to develop an adipocyte-like morphology [11, 12], further confirming the fundamental role of PPARG not only in adipogenesis, but also as a factor determining mesodermal cell fate .
Other factors have recently been suggested as hypothetical master switches controlling myogenic and adipogenic differentiation programmes. Wingless-type MMTV integration site family member 10B (WNT10B) has been described as a molecular key that blocks the adipogenic programme in mouse preadipocytes, inhibiting the induction of CCAAT/enhancer binding protein alpha (CEBPA) and PPARG [14, 15]. Moreover, the disruption of WNT signalling causes the transdifferentiation of myoblasts into adipocytes in vitro .
Myostatin (also known as growth differentiation factor 8 [GDF8]) has been shown to inhibit myogenesis  and muscle growth by blocking the activation of satellite cells , and to promote the adipogenic differentiation of a multipotent mesenchymal cell line . Moreover, loss of myostatin reduces body fat accumulation , although other studies reported that myostatin treatment reduces fat mass in vivo  and inhibits 3T3-L1 adipogenic differentiation by decreasing PPARG expression [19, 20].
To verify whether muscle adult satellite stem cells are able to differentiate into various mesenchymal lineages in humans, we isolated a population of satellite cells from single myofibres and cultured them to induce in vitro myogenesis, osteogenesis and adipogenesis.
We studied the effect of rosiglitazone treatment on the adipogenic differentiation capacity of human satellite cells and compared the phenotype of fully differentiated adipocytes obtained from human adipose tissue stromal vascular fraction with those from human satellite cells.
Subjects, materials and methods
Human satellite cells
Cell source and treatment
Single fibres were isolated from biopsies of human rectus abdominis muscles from three patients (two men, one woman; age 10–25 years) who underwent minor abdominal surgery. We performed one set of cell culture experiments from each subject.
Donor subjects for muscle and adipose tissue biopsies were otherwise healthy, normal weight, free from endocrine-metabolic alterations and with regular physical activity. Institutional review board approval and informed consent were obtained; the study was carried out in accordance with the Declaration of Helsinki.
Immunofluorescence was performed using human satellite cells maintained in plating medium for two or three passages, in order to avoid cell fusion and myotube generation. One million human satellite cells were detached using citrate buffer solution (Sigma-Aldrich), washed in PBS and stained for 1 h at 4°C with fluorescein-5-isothiocyanate or phycoerythrin-conjugated antibodies against CD3, CD4, CD31, CD44, CD45, CD56, HLA-ABC, HLA-DR (Immunotech, Marseille, France). Analyses were performed with a Coulter EPICS-XL using EXPO 32 ADC Software (Beckman Coulter, Fullerton, CA, USA).
Five days after seeding, the plating medium was replaced with myogenic medium (MM), consisting of DMEM supplemented by 20% fetal bovine serum (FBS) (GIBCO), 10% horse serum and 0.5% chicken embryo extract. Where indicated (MM+rosiglitazone) 10 μmol/l rosiglitazone (a kind gift from GlaxoSmith Kline) was added to cultures.
To induce adipogenic differentiation human satellite cells were cultured in adipogenic medium (AM) containing DMEM/F12 supplemented by 10% FBS, 33 μmol/l biotin, 17 μmol/l pantothenate, 10 μg/ml human transferrin (Sigma-Aldrich), 66 nmol/l insulin (Lilly Research, Indiana, IN, USA), 100 nmol/l dexamethasone, 1 nmol/l triiodothyronine and 0.25 mmol/l 3-isobutyl-1-methylxanthine (Sigma-Aldrich). Where indicated (AM + rosiglitazone) 10 μmol/l rosiglitazone was added to cultures.
To induce osteogenic differentiation, human satellite cells were cultured in osteogenic medium (OM) consisting of DMEM low glucose, supplemented by 10% FBS, 1% antibiotics (GIBCO), 100 nmol/l dexamethasone, 10 mmol/l β-glycerophosphate (Sigma-Aldrich) and 0.05 mmol/l ascorbic acid-2-phosphate (Wako Chemicals, Irving, TX, USA).
The differentiation media were changed three times per week.
Abdominal visceral fat was obtained from subjects (n=5) undergoing elective surgery for minor abdominal diseases. One set of cell culture experiments was performed from each subject.
The stromal vascular fraction was isolated from adipose tissue by collagenase type II digestion (1 mg/ml) (Sigma-Aldrich) in DMEM at 37°C for at least 1 h. The sedimented stromal cells obtained by 350×g centrifugation were resuspended in an erythrocyte-lysing buffer, washed and seeded in DMEM/F12 supplemented by 10% FBS (0.7×106 cells/well in 24-well plates). After 16 to 20 h for cell attachment, cultures were re-fed with a serum-free AM with or without 10 μmol/l rosiglitazone. 3-Isobutyl-1-methylxanthine and rosiglitazone were removed after 3 days; medium was changed three times per week. Preadipocytes maintained in DMEM/F12 alone were used as controls.
Histochemical staining and immunofluorescence analyses
Alkaline phosphate activity was determined histologically with a leucocyte alkaline phosphatase kit (Sigma-Aldrich).
Lipid droplets were identified by Oil-Red-O staining. Cells were fixed in 10% formalin (Sigma-Aldrich) for 1 h at 4°C and stained with a 0.5% solution of Oil-Red-O (Sigma-Aldrich) in 60% isopropanol for 15 min at room temperature.
Human satellite cells were characterised with anti-MyoD (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and anti-Troponin I (Chemicon International, Temecula, CA, USA) antibodies diluted in PBS 2% BSA. Primary antibody binding was visualised with fluorochrome-conjugated secondary antibodies (Molecular Probes, Invitrogen Labeling & Detection, Eugene, OR, USA). Cells were counterstained with 4′6-diamidino-2-phenylindole and mounted in aqueous fluorescent mounting media (DakoCytomation, Glostrup, Denmark).
RNA extraction and quantitative PCR determinations
Total RNA was extracted using a kit (RNeasy Mini; Qiagen, Hilden, Germany) following the supplier’s instructions. One microgram of RNA was treated with DNase treatment and removal reagents (Ambion, Austin, TX, USA) and reverse-transcribed for 1 h at 37°C in a 50 μl reaction containing 1× RT buffer, 150 ng random hexamers, 0.5 mmol/l deoxynucleotide triphosphates, 20 U of RNAsin ribonuclease inhibitor and 200 U of M-MLV RT (Promega Corporation, Madison, WI, USA). Primers sequences and reaction conditions are reported in the Electronic Supplementary Material (ESM) (ESM Table 1). PCR was carried out using a DNA Engine (Opticon 2 Continuous Fluorescence Detection System; MJ Research, Waltham, MA, USA). Reactions were performed two times with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) and 5 to 10 ng of cDNA as previously described .
Standard curves were obtained using cDNAs from: (1) skeletal muscle tissue for myogenic differentiation 1 (MYOD1), myogenin (MYOG), growth differentiation factor 8 (GDF8) and wingless-type MMTV integration site family member 10B (WNT10B); (2) adipose tissue for PPARG2, leptin (LEP), adiponectin (ADIPOQ), fatty acid-binding protein 4 (FABP4) and pre-B-cell colony enhancing factor 1 (PBEF1); (3) osteocyte cultures for runt-related transcription factor 2 (RUNX2); and (4) peripheral blood mononuclear cells for resistin (RETN).
Results were normalised by 18S rRNA or hydroxymethylbilane synthase mRNA content and reported as arbitrary units ratio.
Results are presented as mean±SD. Statistical analysis was assessed using unpaired t-tests or by ANOVA when required.
Characterisation of human satellite cells
Satellite cells migrated out of the original fibre and proliferated in culture generating 50 to 200 cells surrounding the fibres within 3 to 4 days (Fig. 1b). After 7 to 10 days in culture, when cell density had increased, cells fused to form myotubes (Fig. 1c).
We further characterised the phenotype of human satellite cells by cytofluorimetric analysis. We found that they expressed CD44 and CD56, both well-known markers of satellite cells, and HLA-ABC, and were also negative for specific haematopoietic antigens such as CD3, CD4, CD45 and HLA-DR. CD31-negative staining ruled out the possibility of any endothelial cell contamination (Fig. 1d–k).
Human satellite cells may enter different mesenchymal differentiation pathways
To investigate the differentiation capability of human satellite cells, we cultured them in appropriate media to induce myogenic (in MM), adipogenic (in AM) or osteogenic (in OM) phenotypes.
Human satellite cells cultured in AM progressively changed their morphology into a round shape. After 14 days in culture, more than 20% of the cells were filled up with lipid droplets, which positively stained with Oil-Red-O (Fig. 2b).
Light microscopy analysis showed that morphologically 50% of human satellite cells cultured for 4 days in OM were completely changed, developing an osteoblast-like appearance with finger-like excavations into the cytoplasm; after 14 days they were intensely positive for alkaline phosphate chemical staining (Fig. 2c).
To track the specific mesenchymal development, we quantified MYOD1 and MYOG (an early and late marker of myogenic differentiation), LEP and ADIPOQ (adipose-specific genes) and RUNX2 (a gene upregulated in osteoblast and hypertrophic chondrocytes) expression by quantitative PCR in all satellite cell cultures at different time points.
ESM Fig. 1a,d shows that myogenesis-related genes were upregulated in the three culture conditions with a significantly higher expression (4.16-fold increase for MYOD1 and 6.07-fold increase for MYOG by day 14) in the human satellite cells cultured in AM than in those maintained under MM or OM conditions. On the other hand, adipogenic markers were induced in the human satellite cells only under adipogenic conditions, while RUNX2 expression increased only in OM medium (ESM Fig. 1b,e,c).
Effect of rosiglitazone on human satellite cell differentiation
Upon incubation of human satellite cells with AM, 5 to 10% of the cells became adipocyte-like cells, an effect that was clearly evident after 7 days. The addition of rosiglitazone greatly accelerated the appearance of adipocyte-like cells, this being detectable after only 4 days of treatment, and strongly increased the number of human satellite cells that differentiated into mature adipocytes to 40 to 50% of the total number of cells, as depicted in Fig. 3.
Among adipocytokines involved in insulin resistance, we evaluated gene expression of RETN (gene encoding resistin) and PBEF1 (gene encoding visfatin). RETN expression was undetectable in all culture conditions analysed. On the other hand, we measured a high level of PBEF1 mRNA in human satellite cells, comparable to that measured in the same amount of cDNA obtained from adipose tissue and used as standard. This high level was found, without changes during myogenic or adipogenic differentiation, and both with and without rosiglitazone (Fig. 4e).
Adipose-specific gene expression profiling of adipocytes obtained from adipose tissue
RETN expression was very low in cells isolated from the human adipose tissue stromal vascular fraction (cycle threshold [Ct] 37–40) compared with immune cells used as standard (Ct 34–35). On the other hand, PBEF1 expression was as elevated in adipocytes cultured in vitro (Ct 20–24) as it was in adipose tissue specimens (Ct 25). The expression of both these adipocytokines was higher in preadipocytes and declined during adipogenic differentiation, as reported in Fig. 5e,f.
WNT10B and GDF8 expression in human satellite cells
With regard to GDF8, a slight increase in expression was observed in MM (Fig. 6b), which was independent of the presence of rosiglitazone. GDF8 mRNA was significantly higher in human satellite cells cultured in AM than in those cultured in MM.
The experimental procedure we used led to the isolation of a cell population from human skeletal muscle, which possesses all phenotypic markers of satellite stem cells that give rise to different mesenchymal lineages, such as fully differentiated adipocytes, osteocytes and myofibres.
Myotubes, as well as adipocytes and osteocytes, obtained from human satellite cells disclosed tissue-specific morphological features and gene expression profiles. Thus, the multipotency of human satellite cells could be a possible mechanism to explain the ectopic bone and fat formation that has been described within skeletal muscle [3, 24].
When the AM was added to the human satellite cells, the myogenic markers dramatically increased, together with the activation of adipogenic-specific genes. This behaviour could be explained in part by the simultaneous growth of both myotubes and fat cells and in part by the presence in the AM of hormones and substrates that are well recognised to work also as muscle growth factors [25, 26].
Adipocytes coming from the human adipose tissue stromal vascular fraction and those coming from human satellite cells share very similar morphological characteristics and adipose-specific gene expression during the different stages of adipogenic differentiation. This observation led us to conclude that properly differentiated adipocytes can be derived from human satellite cells.
PPARG is a master transcription factor in adipogenesis; its expression increases during the conversion of preadipocytes to adipocytes and it is upregulated by rosiglitazone .
PPARG mRNA content was negligible in human satellite cells even when rosiglitazone was added to the MM and the formation of fully differentiated myotubes was unaffected as previously reported . Moreover, the addition of rosiglitazone to human satellite cell cultures did not lead to triglyceride accumulation or the expression of specific adipogenic markers. Other authors have reported that chronic treatment of muscle cell cultures with troglitazone induced expression of adipose-specific genes, but they failed to detect any morphological feature characteristic of a fat cell, further indicating that thiazolidinediones do not induce transdifferentiation [27, 28]. However, in the present study, under adipogenic conditions, human satellite cells expressed PPARG and this expression was further upregulated by rosiglitazone treatment in parallel with the enhancement of their adipogenic differentiation.
Taken together, these findings suggest that rosiglitazone is not sufficient in itself to convert human satellite cells to adipocytes, in contrast with previous observations in a mouse myoblast cell line and in mouse satellite cells .
On the other hand, our results underline the fundamental role of PPARG activation in triggering and regulating the adipogenic programme also in human satellite cells. Previous studies reported that G8 myoblasts, when infected with both PPARG and CEBPA retroviral vector cultured in AM, transdifferentiate into mature adipocytes . Moreover, adipogenesis did not occur in PPARG-deficient embryonic stem cells in vivo or in vitro , thus suggesting that PPARG is probably the master gene that commits multipotent mesenchymal stem cells to the adipogenic lineage.
The addition of rosiglitazone to the AM also increased the expression of genes transcriptionally activated by PPARG, namely FABP4 and ADIPOQ, in human satellite cells.
ADIPOQ mRNA was undetectable in human satellite cells under myogenic conditions, but its expression may be turned on by PPARG activation, as seen when human satellite cells were cultured in AM and to a greater extent when rosiglitazone was added to the AM.
In contrast, rosiglitazone did not further increase LEP mRNA levels, which would have been expected from the great acceleration of adipogenesis. This is in keeping with findings showing that PPARG activation inhibits LEP gene transcription in 3T3-L1 preadipocytes  and human adipocytes .
We showed that human satellite cells express PBEF1 at similar levels in all culture conditions used. Moreover, the lack of effect of rosiglitazone on gene encoding for visfatin expression could suggest that this adipokine is not modulated by PPARG activation in contrast to what was recently reported in rosiglitazone-treated OLETF rats .
PBEF1 and RETN expression was higher in human preadipocytes than in mature adipocytes differentiated from the human adipose tissue stromal vascular fraction, with the former slightly decreasing its expression during differentiation, and the latter being quite undetectable in mature fat cells.
These results are in keeping with reports in which RETN expression is only occasionally observed in human adipose tissue biopsies and undetectable in human mature adipocyte cultures [36, 37, 38], and they lead us to conclude that resistin should not be considered as a marker of fully differentiated adipocytes.
A number of genes have been implicated in the commitment of mesenchymal stem cells to a myogenic or adipogenic lineage. Among them, WNT proteins are important in muscle development [39, 40], and in particular the WNT10B isoform seems to play a crucial role in impeding adipogenesis, partly by inhibiting PPARG expression in preadipocytes and in adult myoblasts [14, 15, 41]. The role of WNT10B in the pathophysiology of obesity was recently highlighted by the identification in a severely obese subject of a missense variant, which abrogated the ability of this protein to block adipogenesis .
Human satellite cells cultured under adipogenic conditions showed a biphasic pattern with an early increase followed by a drop in WNT10B mRNA content, this effect being independent of PPARG activation obtained by rosiglitazone treatment. This is consistent with an early acceleration of myogenesis, proved by high levels of MYOD1 and MYOG expression, which was overcome by the preferential conversion of human satellite cells into adipose cells. In contrast, human satellite cells cultured to form myotubes continued to express WNT10B during their differentiation process. These data are consistent with the idea that WNT signalling is implicated in the initiation of myogenesis and the maintenance of muscle phenotype, and that it is an effective inhibitor of adipogenesis in human satellite cells. In this regard, Vertino et al. recently reported that myoblasts isolated from Wnt10b null mice showed increased adipogenic potential .
In our experiments, GDF8 expression slightly increased with myofibre formation, but was clearly upregulated in human satellite cells cultured in AM, independently of the presence of rosiglitazone. These observations confirm the hypothesis that myostatin could regulate the fate of muscle stem cells, promoting their commitment to the adipogenic lineage and inhibiting myogenesis as previously shown in a multipotent mesenchymal cell line .
Therefore, even if a complex balance of stimulatory and inhibitory signals determines the progression of human satellite cells through a particular lineage, our data suggest that PPARG activation could represent the central event of human satellite cell adipogenic differentiation, while WNT10B and myostatin possibly only act as regulators of this process. Nevertheless, we are aware that measurements of the gene expression profile do not necessarily imply similar changes in protein production and activity.
In the present work, we have clearly shown the adipogenic potential of human satellite cells, and this could explain the origin of mature adipocytes within myofibres or within the intermuscular space. Obesity, type 2 diabetes [1, 2, 3] and ageing-related sarcopenia  are characterised by ‘fatty muscle’. All of these pathological conditions are associated with insulin resistance and it is possible to hypothesise that the presence of fat cells close to muscle cells may directly influence their metabolism and insulin sensitivity through the delivery of metabolites, adipocytokines and inflammatory factors.
Rosiglitazone treatment in itself did not induce any fat conversion in human satellite cells, but greatly enhanced their adipogenic potential triggered by the addition of a specific medium permissive of adipogenesis. However, rosiglitazone is able to induce a more favourable adipocytokine expression profile (higher ADIPOQ, lower LEP, elevated PBEF1 and no RETN) in adipocytes derived from human satellite cells. This effect could modify muscle insulin sensitivity in a paracrine manner. Thus thiazolidinediones, which are able to increase fatty acid disposal and oxidation in skeletal muscle , could counteract the worsening effect both of increased intramyocellular triglyceride content and of the presence of fat cells within muscle fibres.
The excellent technical assistance of M. Tormene and S. Leandri is greatly appreciated. This work was supported by the following grants: (1) from the Ministry of Education, Universities and Research (MIUR): RBNE01KCX4_008 MIUR-FIRB, 2003061834_006 MIUR-PRIN; (2) from Regione Veneto: 115/03 to R. Vettor; (3) from the Città della Speranza Foundation, Muscle Project: Pr04/02 to P. De Coppi. The indispensable contribution of S. Galuppo to the processing of adipose tissue and preadipocytes cultures is hugely acknowledged.
Duality of interest statement
The authors declare that they have no duality of interest.