MiR-21 Regulates Adipogenic Differentiation through the Modulation of TGF-β Signaling in Mesenchymal Stem Cells Derived from Human Adipose Tissue

Abstract

A better understanding of the molecular mechanisms that govern human adipose tissue-derived mesenchymal stem cells (hASCs) differentiation could improve hASCs-based cell therapy and provide new insights into a number of diseases, including obesity. In this study, we examined the roles of microRNA-21 (miR-21) in adipogenic differentiation of hASCs. We found that miR-21 expression was transiently increased after induction of adipogenic differentiation, peaked at 3 days, and returned to the baseline level 8 days. Lentiviral overexpression of miR-21 enhanced adipogenic differentiation. Overexpression of miR-21 decreased both protein and mRNA levels of TGFBR2. The expression of TGFBR2 was decreased during adipogenic differentiation of hASCs in concordance with an increase in the level of miR-21. In contrast, inhibiting miR-21 with 2′-O-methyl-antisense microRNA increased TGFBR2 protein levels in hASCs, accompanied by decreased adipogenic differentiation. The activity of a luciferase construct containing the miR-21 target site from the TGFBR2 3′UTR was lower in LV-miR21-infected hASCs than in LV-miLacZ infected cells. TGF-β-induced inhibition of adipogenic differentiation was significantly decreased in miR-21 overexpressing cells compared with control lentivirus-transduced cells. RNA interference-mediated downregulation of SMAD3, but not of SMAD2, increased adipogenic differentiation. Overexpression and inhibition of miR-21 altered SMAD3 phosphorylation without affecting total levels of SMAD3 protein. Our data are the first to demonstrate that the role of miR-21 in the adipogenic differentiation of hASCs is mediated through the modulation of TGF-β signaling. This study improves our knowledge of the molecular mechanisms governing hASCs differentiation, which may underlie the development of obesity or other metabolic diseases. STEM CELLS 2009;27:3093–3102

INTRODUCTION

MicroRNAs (miRNAs, miR's) are endogenous 22-nucleotide RNAs, some of which play important regulatory roles in animals by post-transcriptionally regulating gene expression. miRNAs have been implicated in many processes in invertebrates, including cell proliferation and apoptosis [1, 2], fat metabolism [1], neuronal patterning [3], and tumorigenesis [4]. Recent evidence has shown that miRNAs influence stem cell functions, including differentiation, through negatively regulating gene expression at the post-transcriptional level [5, 6].

It has been reported that miR-143 increases adipocyte differentiation through the target gene ERK5 in human preadipocytes [7]. However, it remains possible that additional different miRNAs play important roles in adipogenic differentiation processes. Three years ago, miR-21 was identified as the miRNA most commonly and strongly upregulated in human brain tumor glioblastoma [8]. Since then, miR-21 has been studied in the contexts of development, oncology, stem cell biology, and aging [9]. miR-21 expression is increased in hepatocellular carcinomas [10], gastric cancer [11], ovarian cancer [12, 13], cervical carcinoma [14], multiple head and neck cancer cell lines [15], papillary thyroid carcinoma [16], and certain solid tumors. It is also upregulated in several mouse models of cardiac hypertrophy [17–19] and in vascular walls after balloon injury [20]. Furthermore, oncogenic Ras-induced dedifferentiation of the thyroid cell line FRTL-5 has been shown to upregulate miR-21 [21]. These findings suggest that miR-21 promotes cell proliferation and induces cell dedifferentiation. In some cases, however, miR-21 expression was induced during differentiation in cell lines expressing low or undetectable levels of miR-21, such as mouse embryonic stem cells, neuroblastoma, and the myeloid line HL-60 [8, 21–23].

A recent study showed that miR-21 targets genes related to the TGF-β pathway, such as TGFBR2, TGFBR3, and DAXX, in glioblastoma cells [24]. TGF-β is known to inhibit adipose differentiation of preadipocyte cell lines and primary cultures [25–28]. TGF-β also blocks adipogenesis in vivo. Transgenic overexpression of TGF-β1 in adipose tissue severely reduces both white and brown adipose tissue masses, as adipocytes fail to differentiate [29].

Adipose tissue-derived mesenchymal stem cells (ASCs) share many of the characteristics of their counterparts in bone marrow, including extensive proliferative potential and the ability to differentiate toward adipogenic, osteogenic, chondrogenic, and myogenic lineages [30–32]. Understanding the molecular events involved in adipocyte differentiation is of interest for the development of therapeutics for metabolic diseases such as obesity and diabetes.

Therefore, in this study, we examined the role of miR-21 in adipogenic differentiation of human ASCs (hASCs) and identified several molecular targets of miR-21. Our data are the first to demonstrate that miR-21, which is transiently upregulated during the adipogenic differentiation of hASCs, promotes differentiation by binding to target sequences in the 3′-untranslated region of TGFBR2.

MATERIALS AND METHODS

Cell Culture

All protocols involving human subjects were approved by the Institutional Review Board of Pusan National University. Superfluous materials were collected from four individuals undergoing elective abdominoplasty after informed consent was given by each individual. The hASCs were isolated according to the methods described in previous studies [33]. We used four different hASC samples for each experiment and performed the duplicate experiment per each sample.

Viral Vector Construction and Transduction

The engineered pre-miRNA sequence was cloned into the cloning site of a BLOCK-iT Pol II miR RNAi Expression vector (Invitrogen, CA, http://www.invitrogen.com) that is flanked on either side with sequences from hsa-miR-21 (mature sequence: 5′-uagcuuaucagacugauguuga) to allow proper processing of the engineered pre-miRNA sequence. Other subcloning and viral transduction procedures were performed as described [34]. We used virus titers ranging from 5 × 10⁵ up to 1 × 10⁷ transducing units (TU)/ml.

Induction of Differentiation

ASCs were plated at 3 × 10⁴ cells/cm² and cultured in complete culture media for 2-3 days in 12-well culture dishes. Then, adipogenic differentiation was induced for 7-10 days in an adipogenic medium (10% fetal bovine serum (FBS), 1 μM dexamethasone, 0.5 mM/ml 3-isobutyl-1-methylxanthine, and 200 μM indomethacin in α-MEM and assessed with Oil Red O stain, which is an indicator of intracellular lipid accumulation. In order to obtain quantitative data, 1 ml of isopropyl alcohol was added to the stained culture dish. After 5 minutes, the absorbance of the extract was assayed by a spectrophotometer at 510 nm after dilution to a linear range. Osteogenic differentiation was induced via the culturing of the cells for 10 days in osteogenic medium (10% FBS, 0.1 μM dexamethasone, 10 mM β-glycerophosphate, and 50 μM ascorbic acid in α-MEM); extracellular matrix calcification was estimated with Alizarin red S stain. Osteogenic differentiation was quantified via the measurement of the Alizarin red-stained area and density in 12-well dishes with Image Gauge v.3.1, Fuji, Japan.

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

Total cellular RNA was isolated from hASCs and reverse transcribed using conventional protocols. The primer sequences used in the experiment were as follows: GAPDH: 5′-TCC ATG ACA ACT TTG GTA TCG-3′, 5′-TGT AGC CAA ATT CGT TGT CA-3′; PPARG: 5′- GCT GTT ATG GGT GAA ACT CTG-3′, 5′-ATA AGG TGG AGA TGC AGG TTC-3′; C/EBP-α: 5′-GCA AGG CCA AGA AGT CGG TGG AC -3′, 5′-TGC CCA TGG CCT TGA CCA AGG AG-3′; aP2: 5′-GGT GGT GGA ATG CGT CATG-3′, 5′-CAA CGT CCC TTG GCT TAT GC-3′; TGFB1: 5′-AGC GAC TCG CCA GAG TGG TTA-3′, 5′- GCA GTG TGT TAT CCC TGC TGT CA-3′; TGFBR2 5′-ACG TGT TGA GAG ATC GAGG-3′, 5′- CCC AGC ACT CAG TCA ACG TC-3′; SMAD2: 5′-GTT CCT GCC TTT GCT GAG AC-3′, 5′- TCT CTT TGC CAG GAA TGC TT-3′; SMAD3: 5′-GCC AGT TAC CTA CTG CGA GC-3′, 5′-CTC CGA TGT AGT AGA GCC GC-3′; SMAD4: 5′-CCA TTT CCA ATC ATC CTG CT-3′, 5′-ACC TTT GCC TAT GTG CAA CC-3′. All primer sequences were generated from established GenBank sequences.

Real-Time PCR

The isolation of small species-enriched RNA was performed as per the manufacturer's instructions (mirVana miRNA isolation kit, Ambion, Austin, TX, http://www.ambion.com)). miRNA was reverse-transcribed with an Ncode miRNA first-strand cDNA synthesis kit (Invitrogen, CA) according to the manufacturer-specified guidelines. Forward primer sequences were designed as the corresponding mature miRNA sequences, and 5S rRNA was used as a normalizing control. Real-time quantitation was performed with the LightCycler assay, using a fluorogenic SYBR Green I reaction mixture for PCR carried out in a LightCycler Instrument (Roche, Germany, http://www.roche-applied-science.com). Data analyses were performed according to the methods described in previous studies [33].

Northern Blot

Oligonucleotides complementary to mature miRNAs were end-labeled with a T7 promoter sequence (miRVana miRNA probe construction kit (Ambion)) and used as probes. Probe sequences were as follows: miR-21: 5′-TAG GTA GTT TCA TGT TGT TGG cctgtctc-3′; U6, 5′-GCA GGG GCC ATG CTA ATC TTC TCT GTA TCG cctgtctc-3′. Other northern blot procedures were performed as described [34].

Western Blot Analysis

Confluent hASCs were treated under the appropriate conditions and lysed, after which the protein content of the lysate was determined using a protein assay kit (Bio-Rad Laboratories, Hercules, CA, http://www.bio-rad.com). The proteins were loaded on 10% SDS polyacrylamide gels, transferred to nitrocellulose membranes (Hybond-ECL, Amersham Pharmacia Biotech, Piscataway, NJ, http://www.amersham.com), and probed with polyclonal antibodies (anti-TGFBR2, anti-phospho-SMAD2, anti-SMAD2, anti-phospho-SMAD3, anti-SMAD3, and anti-β-actin; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com). Immunoreactive bands were detected with anti-rabbit or anti-mouse peroxidase-conjugated secondary antibodies (Amersham Pharmacia Biotech) and visualized via enhanced chemiluminescence (ECL detection kit; Amersham Pharmacia Biotech).

RNA Transfection

Anti-miR miRNA inhibitors (anti-miRs) and a scrambled RNA oligomer were purchased from Ambion. These were transfected into hASCs at a final concentration of 50 nM using DharmaFECT Transfection Reagent, as per the manufacturer's instructions. Small interfering RNA (siRNA) duplex oligos (on-TARGET plus SMART pool, Dharmacon, CO, http://www.dharmacon.com) targeting SMAD2, SMAD3, SMAD4, and TGFBR2 mRNA or a nontargeting duplex oligo (negative control) were transfected into cells using the DharmaFECT Transfection Reagent.

Reporter Vectors and DNA Constructs

A putative miRNA 21-recognition element (single copy) from the TGFBR2 gene was cloned into the 3′-untranslated region (UTR) of a firefly luciferase reporter vector, according to the manufacturer-specified guidelines. The oligonucleotide sequences were designed to carry the HindIII and SpeI sites at their extremities to allow ligation into the HindIII and SpeI sites of pMIR-Report (Ambion). The oligonucleotides used in these studies were pMIR-TGFBR2: 5′-CTAGT TGAC ATTGTCATAGGATAAGCTGA-3′, 5′-AGCTT CAGCTTATCCTATGACAATGTCA A-3′, pMIR-Mut-TGFBR2, 5′-CTAGT TGACATTGTCTTAGGAAATGGTG A-3′, 5′-AGCTT CACCATTTCCTAAGACAATGTCA A-3′.

Reporter Gene Assay

All transient transfections were conducted using Lipofectamine Plus (Invitrogen). The pMIR-report, pMIR-TGFBR2, pMIR-mut-TGFBR2, and pMIR-β-gal plasmids were used as reporter constructs. Cells were harvested 48 hours after transfection and harvested in CCLR buffer and were subsequently assayed for luciferase activity (Luciferase Assay System, Promega, WI, http://www.promega.com). The transfections were performed in duplicate and all experiments were repeated several times. Luciferase expression was normalized to β-galactosidase activity in all cases.

Statistical Analysis

All results are presented as mean ± SEM. Comparisons between groups were analyzed via 2-sided t-tests or ANOVA for experiments with more than two subgroups. Post hoc range tests and pairwise multiple comparisons were conducted using the 2-sided t-test with Scheffe adjustments. P values < 0.05 were considered to be statistically significant.

RESULTS

Changes in MiR-21 Levels During Adipogenic Differentiation of hASCs

We chose the microRNAs that showed high levels of expression in preliminary microRNA microarray study and targeted differentiation-related molecules in Target scan database. We then determined their expression during the adipogenic differentiation of hASCs by real-time PCR. From this study, we found that miR-21 expression was increased at one day after the induction of adipogenic differentiation of hASCs and peaked at 2-3 days after differentiation. After 3 days, the expression of miR-21 was gradually decreased (Fig. 1A). However, the expression of miR-196a, which regulates osteogenic differentiation in hASCs [34], was not changed. Northern blot analysis confirmed the changes of miR-21 expression during adipogenic differentiation (Fig. 1B). Esau et al. (2004) reported that miR-143 is involved in the regulation of adipogenic differentiation of human preadipocytes [7]. Therefore, we compared the levels of endogenous miR-21 and miR-143 in hASCs by real-time PCR and northern blot analysis. The level of miR-21 expression was 10-fold higher than that of miR-143 expression in naïve hASCs (Figs. 1C, 1D).

Figure 1. miR-21 expression increases during adipogenic differentiation of human adipose tissue-derived mesenchymal stem cells (hASCs). (A, B) Expression levels of miR-21 and miR-196a in hASCs after 8 days of adipogenic differentiation induction. miRNAs were prepared at the indicated times. Relative quantities of amplified products were determined with an image analyzer. (C, D) Expression levels of miR-21 and miR-143 in hASCs were determined by real-time polymerase chain reaction and northern blot. Data represent mean ± SEM of the relative ratio of miRNA signal of the corresponding samples (n = 4). *, p < .05 compared with control hASCs at the indicated days. Abbreviations: miR, MicroRNA; miRNA, MicroRNA.

The Effects of MiR-21 on the Adipogenic Differentiation of hASCs

To determine the role of miR-21 in hASCs functions, we used a lentivirus (LV) overexpressed miR-21 in hASCs. Because the lentivirus vector contains green fluorescent protein (GFP) coding sequence of which expression is driven by cytomegalovirus promoter, the efficiency of lentivirus transduction can be determined by a fluorescent microscope. Single transductions with the lentivirus showed that most hASCs (>90%) expressed GFP and that miR21-overexpressing cells were morphologically similar to naïve hASCs or LV-miLacZ-infected hASCs (Fig. 2A). Real-time PCR analysis and northern blot analysis with a miR-21 probe confirmed that hASCs transduced with the miR-21 lentivirus exhibited increased miR-21 expression (Figs. 2B, 2C).

Figure 2. miR-21 enhances the adipogenic differentiation of hASCs. (A) Lentivirus transduction did not affect the morphology of human adipose tissue-derived mesenchymal stem cells (hASCs). A representative phase contrast image of each cell line is shown (100× magnification). (B, C) miRNA isolated from lentivirus-transduced hASCs was subjected to real-time polymerase chain reaction (PCR) (B) and northern blot (C). (D–G) Lentivirus-transduced hASC were grown to confluence, and adipogenic or osteogenic differentiation was induced in differentiation media. (D) Osteogenic differentiation was determined by Alizarin Red S staining to visualize calcification deposits within the cell monolayer. (E) Adipogenic differentiation was determined by Oil Red O staining to visualize intracellular lipid accumulation. (F) Adipogenic and osteogenic differentiation were quantified by determining the density and area of Alizarin Red S staining with an image analysis program (Multi Gauge V3.0, FUJIFILM), or by measuring the optical density of Oil Red O staining in isopropanol extracts, respectively. (G) Reverse transcription-PCR (RT-PCR) analysis of the adipogenic genes C/EBP-α, PPARG, and aP2 during adipogenic differentiation (at each indicated day). Relative quantities of amplified products were determined with an image analyzer. PCR data represent mean ± SEM of the relative ratio of the gene signal to the GAPDH signal. *, n = 4, p < .05 compared to LV-miLacZ-transduced. Abbreviations: aP2, fatty acid binding protein; C/EBP-α, CCAAT/enhancer binding protein alpha; miR, MicroRNA; miRNA, MicroRNA; OM, osteogenic medium; PPARG, peroxisome proliferator-activated receptor gamma.

To investigate the impact of miR-21 overexpression on the differentiation of hASCs, we induced miR-21-overexpressing hASCs to differentiate along adipogenic or osteogenic lineages with the appropriate media. After incubating hASC with osteogenic media for 7 days, cells were stained with Alizarin Red S, showing no evident difference between conditions. This result suggests that miR-21 may not play a critical role during osteogenic differentiation (Figs. 2D, 2F). However, miR-21-overexpressing hASCs significantly increased adipogenic differentiation, as indicated by oil red O staining (Figs. 2E, 2F). RT-PCR analysis of adipogenesis-related genes revealed that LV-miR21-transduced hASCs exhibited higher levels of peroxisome proliferator-activated receptor gamma (PPARG) mRNAs than did control lentivirus transduced cells. We also evaluated the changes in PPARG, CCAAT/enhancer binding protein alpha (C/EBP-α), and fatty acid binding protein (aP2) expression during adipogenic differentiation. C/EBP-α expression was detected at 2 days after differentiation, while aP2, a late marker of adipogenic differentiation, was detected at 3 days after differentiation in LV-miLacZ-infected hASCs. LV-miR21-transduced hASCs expressed C/EBP-α, PPARG, and aP2 earlier than did LV-miLacZ-transduced cells (Fig. 2G).

To investigate the effect of miR-21 inhibition on hASCs differentiation, we transfected hASCs with a specific miRNA inhibitor. Oil red O staining and RT-PCR analysis of adipogenic marker genes clearly indicated that downregulation of miR-21 expression decreased the adipogenic differentiation of hASCs (Fig. 3).

Figure 3. Inhibition of miR-21 decreases adipogenic differentiation of human adipose tissue-derived mesenchymal stem cells (hASC). (A, B) miR-21 levels were determined in control- (anti-miR Cont) or miR-21 inhibitor (anti-miR21) transfected-hASC using real-time polymerase chain reaction (PCR) (A) and northern blot analysis (B). (C, D) Oligonucleotide-transfected-hASC were grown to confluence, and adipogenic differentiation was induced for 7 days. (D) Reverse transcription-PCR (RT-PCR) analysis of the adipogenic genes 3 days after induction. The PCR products were quantified as in Figure 2. *, n = 4, p < .05 compared with anti-miR Cont-transfected-hASC. Abbreviations: aP2, fatty acid binding protein; C/EBP-α, CCAAT/enhancer binding protein alpha; miR, MicroRNA; miRNA, MicroRNA; PPARG, peroxisome proliferator-activated receptor gamma.

MiR-21 Targets the 3′ UTR of TGFBR2 mRNA

Targetscan 5.1 database and molecular signatures database (MSigDB) presented 105 genes and 118 genes as the putative targets of miR-21, respectively. The number of overlapped genes between two data sets was 43. TGFBR2 among the overlapped candidates was the best candidate for the target of miR-21 for explaining miR-21-induced increase of adipogenic differentiation in hASCs, because TGF-β signaling inhibits adipogenic differentiation and the starting point of TGF-β signal is TGFBR2. Therefore, we investigated TGFBR2 expression levels during adipogenic differentiation. Real-time PCR analysis showed that TGFBR2 mRNA levels gradually decreased and reached the lowest level at 2-3 days after the induction of differentiation and then recovered to the basal levels (Fig. 4A). The protein levels of TGFBR2 also showed similar changes with its mRNA levels with a lag period of 1-2 days (Fig. 4B). To further confirm the relationship between miR-21 and TGFBR2, we determined the mRNA and protein levels of TGFBR2 in hASCs overexpressing miR-21 or transfected with a miR-21-specific inhibitor. Real-time PCR and western blot analyses showed that miR-21-overexpressing hASCs exhibit less TGFBR2 expression and that anti-miR-21-transfected hASCs exhibit increased TGFBR2 expression (Figs. 4C, 4D).

Figure 4. miR-21 targets the 3′UTR of TGFBR2 mRNA (A) Human adipose tissue-derived mesenchymal stem cells (hASCs) were induced to differentiate in adipogenic media for 8 days. Total RNA was prepared at the indicated times and subjected to real-time PCR. (B) TGFBR2 expression was analyzed by western blot with β-actin as a loading control. TGFBR2 expression levels were quantified with an image analysis program. *, n = 4, p < .05 compared with control hASCs at the indicated days. (C, D) TGFBR2 expression in hASCs infected with LV or transfected with oligonucleotide was analyzed by real-time PCR (C) and western blot (D). (E, F) Empty, pMIR-TGFBR2, or pMIR-mutTGFBR2 luciferase constructs were transfected into control or miR21 lentivirus-transduced hASCs (E) and anti-miR Cont or miR-21 transfected hASCs (F). Data represent mean ± SEM (n = 4), * p < .05, compared with LV-miLacZ-transduced hASCs, # p < .05 compared with anti-miR Cont-transfected hASCs. Abbreviations: Cont, control; LV, lentivirus; miR, MicroRNA; miRNA, MicroRNA.

A luciferase reporter assay was used to demonstrate that miR-21 directly decreased TGFBR2 expression. We aligned the miR-21 sequence with TGFBR2 3′UTR insert and used the resulting construct to transfect hASCs. Transfection of LV-miR21-infected hASCs with the parental luciferase construct (without the TGFBR2 3′UTR) or a scrambled construct did not significantly change the expression of the reporter. However, cells transfected with a luciferase construct in which the miR-21 target site from the TGFBR2 3′UTR (pMIR-TGFBR2) was inserted exhibited significantly lower luciferase activity in LV-miR21-infected hASCs than in control lentivirus-infected cells. Cotransfection with an anti-miR21 oligo increased the luciferase activity of pMIR-TGFBR2 in hASCs relative to cotransfection with a control oligo (Figs. 4E, 4F).

The Effect of TGF-β1 on the Adipogenic Differentiation of hASCs

To test whether modulation of TGFBR2 by miR-21 affects TGF-β action on the adipogenic differentiation of hASCs, we treated hASCs with different concentrations of TGF-β1 under adipogenic conditions. TGF-β1 treatment inhibited the adipogenic differentiation of lentiviral vector-overexpressing hASCs in a dose-dependent manner with maximal inhibition at 2 ng/ml of TGF-β1. The inhibitory action of TGF-β1 on adipogenic differentiation of hASCs was significantly weakened by the overexpression of miR-21. In vector-overexpressing cells, 0.2 ng/ml of TGF-β1 was sufficient to inhibit adipogenic differentiation, whereas 0.8 ng/ml of TGF-β1 was required to inhibit differentiation in miR-21 overexpressing cells (Fig. 5A).

Figure 5. TGF-β regulates the adipogenic differentiation of Human adipose tissue-derived mesenchymal stem cells (hASCs). (A) Lentivirus-transduced hASCs were treated with various concentrations (0.2-2 ng/ml) of TGF-β during adipogenic differentiation. n = 4, * p < .05 compared with LV-miLacZ-transduced hASCs in the absence of TGF-β; # p < .05 LV-miR21 transduced hASCs in the absence of TGF-β. (B) TGF-β (1 ng/ml)-treated hASCs were incubated in adipogenic media for the indicated number of days. Adipogenic differentiation was determined by Oil Red O staining to visualize intracellular lipid accumulation and quantified by measuring the optical density of Oil Red O staining in isopropanol extracts, respectively. *, n = 4, p < .05 compared with control hASCs in the absence of TGF-β. Abbreviations: LV, lentivirus; miR, MicroRNA.

The data in Figure 1 and Figure 4 show that the expression of miR-21 and TGFBR2 is transiently decreased during adipogenic differentiations, suggesting that the inhibitory action of TGF-β1 on adipogenic differentiation is confined to its early phase. To test this possibility, hASCs were incubated in adipogenic media with TGF-β1 for the days indicated in Figure 5B. Exposure to TGF-β1 during the first 3 days inhibited adipogenic differentiation, but treatment with TGF-β1 at 3 days after the induction of differentiation or later failed to inhibit the process.

The Role of SMADs in TGF-β-Induced Inhibition of Adipogenic Differentiation

To determine whether miR-21-induced TGFBR2 downregulation affected pathways downstream of TGF-β, we examined the roles of TGF-β signaling related SMADs in adipogenesis. We found that the addition of TGF-β1 increased SMAD2 and -3 phosphorylation (Fig. 6A). To further investigate the roles of SMAD2 and -3 in adipogenic differentiation, we downregulated the expression of these genes with siRNAs (Figs. 6B, 6C). Transfection of SMAD2 and SMAD3 siRNAs significantly inhibited SMAD expression. However, only SMAD3 downregulation increased adipogenic differentiation in the absence or presence of TGF-β1; SMAD2 downregulation had no effect on it. To further confirm the relationship between TGF-β signaling and miR-21, we determined SMAD3 phosphorylation in miR-21 overexpressed hASCs and anti-miR-21-transfected hASCs. Western blot analyses showed that miR-21-overexpressing hASCs exhibit less SMAD3 phosphorylation than LV-transduced cells and that anti-miR-21-transfected hASCs exhibit increased SMAD3 phosphorylation compared with control oligo transfected cells (Fig. 6D).

Figure 6. The TGF-β signaling mechanism in miR-21-induced adipogenic differentiation. (A) Human adipose tissue-derived mesenchymal stem cells (hASCs) were incubated with or without TGF-β (1 ng/ml) for 30 minutes, and lysates from these cells were used in western blots. (B) SMAD2 and -3 mRNA levels were determined in control- (siCont), SMAD2, and SMAD3 oligonucleotide (siSMAD2 and 3)-transfected hASCs with RT-PCR. (C) Oligonucleotide-transfected hASCs (siCont, siSMAD2, and siSMAD3) were treated with adipogenic differentiation media in the presence or absence of TGF-β for 7 days. Data represent mean ± SEM (n = 4) * p < .05 compared with control oligonucleotide-transfected hASCs in the absence of TGF-β; # p < .05 control oligonucleotide-transfected hASCs in the presence of TGF-β. (D) Effect of miR-21 overexpression and inhibition on SMAD3 phosphorylation in hASCs. SMAD3 phosphorylation was analyzed by western blot. Data represent mean ± SEM (n = 4) ** p < .05 compared with LV-miLacZ-transduced hASCs, ## p < .05 compared with anti-miR Cont-transfected hASCs. Abbreviations: Cont, control; LV, lentivirus; miR, MicroRNA.

The Effect of TGFBR2 Downregulation on the Adipogenic Differentiation of hASCs

To determine the role of TGFBR2 in the differentiation of hASCs, we suppressed TGFBR2 in hASCs with an RNA interference technique using oligo transfection. RT-PCR analysis confirmed that RNAi effectively inhibited TGFBR2 expression in hASCs (Fig. 7A). To investigate the effect of TGFBR2 downregulation on the differentiation of hASCs, we induced adipogenic differentiation in the TGFBR2 oligonucleotide-transfected hASCs. Oil red O staining and RT-PCR analysis for adipogenic marker genes clearly indicated that the downregulation of TGFBR2 by RNAi increased the adipogenic differentiation of hASCs (Figs. 7B, 7C).

Figure 7. The effect of TGFBR2 RNAi on the adipogenic differentiation of human adipose tissue-derived mesenchymal stem cells (hASCs). Silencing of the TGFBR2 gene enhanced the adipogenic differentiation of hASCs. (A) TGFBR2 mRNA levels were determined in control- (siCont) or TGFBR2 oligonucleotide (siTGFBR2)-transfected hASCs using reverse transcription-polymerase chain reaction (RT-PCR). (B, C) Oligonucleotide-transfected hASCs were grown to confluence and adipogenic differentiation was induced and quantified as in Figure 2F. (C) RT-PCR analysis of the expression of the adipogenic genes C/EBP-α, peroxisome proliferator-activated receptor gamma (PPARG), and aP2 was quantified as in Figure 2. *, n = 4, p < .05 compared with control oligonucleotide-transfected hASCs. Abbreviations: aP2, fatty acid binding protein; C/EBP-α, CCAAT/enhancer binding protein alpha; PPAR, peroxisome proliferator-activated receptor.

DISCUSSION

In this study, we first demonstrated that miR-21 expression was transiently changed during the adipogenic differentiation of hASCs. We further showed that this change positively regulates the adipogenic differentiation of hASCs with miR-21 overexpression and transfection of an miR-21 inhibitor. Esau et al. (2004) reported that miR-143 levels increased in differentiating human adipocytes [7]. However, in our study, real-time PCR and northern blot analysis showed that the expression level of miR-143 in hASCs was much lower than that of miR-21 (less than 10-fold), suggesting that miR-21 may be a major regulator of adipogenic differentiation in MSCs.

A search with Targetscan 5.1 revealed that the miR-21 miRNA is complementary to sites in the 3′ untranslated region (3′ UTR) of TGFBR2. Here, we find that lentiviral overexpression of miR-21 and transduction of miR-21 inhibitors have opposing effects on TGFBR2 expression and that TGFBR2 expression is changed during adipogenic differentiation of hASCs in concordance with the changes in miR-21 level. An assay with a firefly luciferase reporter plasmid containing the predicted TGFBR2 target gene sequences indicated that TGFBR2 is a direct target of the endogenous miR-21 in hASCs. It is well known that TGF-β1 inhibits adipose differentiation of preadipocyte cell lines and primary cultures [25–28] and also blocks adipogenesis in vivo. Transgenic overexpression of TGF-β1 in adipose tissue severely reduces both white and brown adipose tissue masses, due to the failure of adipocyte differentiation [29]. Factors in the TGF-β family bind to a heteromeric cell-surface complex of two type II and two type I receptors [35, 36]. Both receptor types are transmembrane serine/threonine kinases, and ligand binding induces phosphorylation and activation of the type I receptors by type II receptor kinases. The Smad proteins act as effectors in this signaling pathway; upon COOH-terminal phosphorylation by type I TGF-β receptors, Smad2 and/or Smad3 heteromerize with Smad4 and are translocated into the nucleus [37–39]. The importance of TGFBR2 in TGF signaling has been demonstrated in mouse knockout experiments [35]. A recent study demonstrated that the expression level of TGFBR2 is directly correlated with TGF-β response [36]. Therefore, downregulation of TGFBR2 expression by miR-21 enhances adipogenic differentiation of hASCs by blocking TGF-β signal transduction. In fact, our data demonstrated that miR-21-overexpressing cells were less sensitive to the inhibitory action of TGF-β on adipogenic differentiation (Fig. 5A).

The transient changes in miR-21 and TGFBR2 levels during the adipogenic differentiation of hASCs observed in this study suggest that TGF-β1 may act in the early phase of adipogenic differentiation and that increases of miR-21 in this phase block the inhibitory action of TGF-β on adipogenic differentiation through the downregulation of TGFBR2 expression. This possibility was supported by the finding that treatment with TGF-β1 during the first 3 days after induction of differentiation inhibited adipogenic differentiation whereas treatment with TGF-β1 at 5 days after differentiation or later had no effect (Fig. 5B). These findings also provide a reasonable explanation of how mature adipose tissues can contain high levels of TGF-β despite its strong inhibitory action on adipogenic differentiation.

The diverse functions of TGF-β are known to be mediated by various Smad isoforms [40–42]. Choy et al. (2000) showed that TGF-β-induced inhibition of adipogenic differentiation is mediated by the activation of SMAD3 in a mouse preadipocyte cell line [43]. In this study, we also demonstrated that downregulation of SMAD3, but not SMAD2, by siRNA increased the adipogenic differentiation of hASCs (Fig. 6C), indicating a central role for SMAD3 in TGF-β-induced inhibition of adipogenic differentiation. Our finding that SMAD3 phosphorylation was correlated with the modulation of miR-21 in hASCs (Fig. 6D) further supports the conclusion that the downregulation of TGFBR2 by miR-21 is sufficient to modulate TGF-β signaling in the adipogenic differentiation of hASCs.

Here, we have characterized the roles of miR-21 in adipogenic hASCs differentiation and elucidated the mechanisms of TGF-β action in this process. These findings present miR-21 as a potential target for adipose tissue engineering through miRNAs and in the management of obesity and other metabolic diseases.

Acknowledgements

This study was supported by a grant (A080359) from the Ministry of Health and Welfare.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.