Inhibition of adipocyte differentiation by mechanical stretching through ERK-mediated downregulation of PPARγ₂

Obesity often promotes a range of cardiovascular diseases, including atherosclerosis, hypertension and type 2 diabetes (Sowers and Haffner, 2002). Recent advances in adipocyte biology have established that fat tissue not only serves as a means of energy storage in the form of triglycerides but also exerts secretory/endocrine gland functions, producing various secretory molecules such as hormones and cytokines (Kahn and Flier, 2000). In view of the preventative and therapeutic aspects of the abovementioned diseases, most intensive clinical interventions have been primarily directed at decreasing excessive amounts of fat tissue by changing the balance between intake and expenditure of energy (Yamauchi et al., 2003); such changes are typically effected via daily exercise and diet control (Spiegelman and Flier, 2001). Mechanical stimuli such as stretching and rubbing of fat and skeletal muscle using gymnastic exercises or massage are believed to decrease obesity as well. Surprisingly, there is still relatively little known about the role of mechanical stimulation in decreasing obesity, because few studies of the cellular and molecular mechanisms of action have been conducted to date.

Adipocytes are the major cellular component in fat tissue and excessive growth, differentiation and hypertrophy of adipocytes are fundamental processes of obesity. Maturation of adipocytes can occur among cells from a pre-existing pool of adipocyte progenitor cells (preadipocytes) that are present irrespective of age (Gregore et al., 1998). Therefore, from a pathophysiological point of view, both the proliferation and differentiation of preadipocytes into mature adipocytes remain important issues in this context.

A range of mammalian non-sensory cells such as fibroblasts (Kanda and Matsuda, 1993; MacKenna et al., 1998), cardiomyocytes (Sadoshima and Izumo, 1997), vascular endothelial cells (Azuma et al., 2000; Davies and Tripathi, 1993; Kanda and Matsuda, 1993; Saito et al., 2003; Tanabe et al., 2000), smooth-muscle cells (Kanda and Matsuda, 1993; Li and Xu, 2000; Oldenhof et al., 2002; Tanabe et al., 2000), osteoblasts (Duncan, 1995) and skeletal muscle cells (Csukly et al., 2002) can respond to mechanical stimuli, including stretching. Interestingly, most of these cells, as well as adipocytes, are considered to be of mesodermal origin. To the best of our knowledge, no study has yet been conducted on the direct effects of mechanical stimulation on adipocyte differentiation. Here, we demonstrate that cyclic mechanical stretching inhibited the differentiation of mouse 3T3-L1 cells into adipocytes; this effect was mostly attributable to the reduced expression of the peroxisome proliferator-activated receptor (PPAR) γ₂, an adipocyte-specific nuclear hormone receptor/adipogenic transcription factor (Tontonoz et al., 1994a; Tontonoz et al., 1994b). This reduced expression was mediated via the activation of an extracellular-signal-regulated-protein kinase/mitogen-activated-protein kinase (ERK/MAPK) pathway. The modulation of adipocyte differentiation in response to stretching might provide further insight into the physiological significance of the local application of mechanical stress to fat tissues with respect to the inhibition of differentiation and the renewal of adipocytes.

The mouse preadipocyte cell line 3T3-L1 was obtained from the Human Science Research Resources Bank (Osaka, Japan). The cells were propagated in Dulbecco's modified Eagle's medium (DMEM) plus 25 mM glucose (Sigma) supplemented with heat-inactivated foetal bovine serum (FBS) (Sigma) to 10% (v/v) in a humidified 5% CO₂ incubator at 37°C. For adipocyte differentiation on an elastic silicon membrane, 7×10⁵ 3T3-L1 cells were seeded onto an elastic rectangular chamber (4×2×1.5 cm length×width×height) coated with a mixture of collagen types I and IV (10:1) (each purchased from Wako Pure Chemicals, Osaka, Japan); the cells were then incubated for an additional period of 2 days to allow the cells to reach confluence. The cells were then exposed to induction medium containing 10% FBS, 10 μg ml^–1 human recombinant insulin (INS), 0.25 μM dexamethasone (DEX) and 0.5 mM 3-isobutyl-1-methylxanthine (MIX) (each from Wako Pure Chemicals) for up to 45 hours with or without cyclic stretching (induction period). The application of uniaxial cyclic stretching to 3T3-L1 cells during the induction period was carried out in a manner similar to that described previously (Tanabe et al., 2000). Briefly, the 3T3-L1 cells were subjected to cyclic stretching to 110-175% of the initial length with a frequency of 1 Hz for the entire period (0-45 hours), for the first 15-hour period (early phase, 0-15 hours) and for the final 15-hour period (late phase, 30-45 hours) of the 45-hour induction period. In some experiments, PD98,059 [2-(2-amino-3-methoxyphenyl)-4H-1-benzopyran-4-one; Wako Pure Chemicals] [a MAPK/ERK-kinase 1 (MEK1) inhibitor] and troglitazone ({±}-5-{4-[(6-hydroxy-2,5,7,8-tetramethylchroman-2-yl)methoxy]benzyl}-2,4-thiazolidinedione; CS-045; Sankyo, Tokyo, Japan), a synthetic ligand for PPARγ, were added to the induction medium. After washing with PBS without Ca²⁺ or Mg²⁺ [PBS(–)], the cells were again fed with maintenance medium consisting of 10% FBS and 5 μg ml^–1 INS in DMEM and the cells were left undisturbed for an additional 9 days, with a change of medium every 2 days (maturation period). The cells were washed, fixed in 4% paraformaldehyde and exposed to filtered 0.3% Oil-Red-O (Sigma) in 60% (v/v) 2-propanol in water for 1 hour followed by extensive washing with PBS(–).

3T3-L1 cells under the various conditions were rinsed twice with PBS(–), resuspended in 600 μl 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM phenylmethylsulfonylfluoride (PMSF), 1 mg ml^–1 pepstatin, and 1 mg ml^–1 leupeptin, and then lysed by sonication at 4°C. An aliquot of the homogenate of the cells was cleared by centrifugation at 12,800 g for 5 minutes at 4°C, and the supernatant was subjected to glycerol-3-phosphate dehydrogenase (GPDH) assay according to a previously reported method (Wise and Green, 1979); the other aliquot of the homogenate was mixed with an equal volume of chloroform/methanol mix (2:1), mixed vigorously for 10 minutes and centrifuged at 12,800 g for 10 minutes at 4°C. The resulting chloroform layer was dried, resuspended in 1% Triton X-100 and subjected to a triglyceride assay using a commercially available kit (Triglyceride G-test WAKO, Wako Pure Chemicals).

3T3-L1 cells of various conditions were rinsed twice with ice-cold PBS(–), resuspended, and lysed in a lysis buffer consisting of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% sodium dodecyl sulfate (SDS), 1% NP-40, 1 mM PMSF, 1 mg ml^–1 pepstatin, 1 mg ml^–1 leupeptin at 4°C. Cell extract containing an equal amount of protein (20 μg) was resolved by 10% SDS-polyacrylamide-gel electrophoresis (SDS-PAGE) under reducing conditions. Proteins was transferred to a Biotrace PVDF membrane (Pall Gelmann Laboratory, Ann Arbor, MI) and the membrane was blocked with 5% w/v bovine serum albumin (BSA) in 20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 0.1% Tween-20 (TBST-500). Western-blot analysis was then carried out using polyclonal anti-ERK1/2 antibody (sc-94, Santa Cruz Biotechnology, Santa Cruz, CA, USA), monoclonal anti-phosphorylated-ERK antibody (sc-7383, Santa Cruz Biotechnology) or polyclonal anti-PPARγ antibody (sc-7196, Santa Cruz Biotechnology). Control analyses were also carried out using mouse and rabbit IgG (Inter-Cell Technologies, Hopewell, NJ, USA). Each primary antibody (1:5000-1:10,000) and horseradish-peroxidase-conjugated secondary antibody (1:5000) were diluted with TBST-500 containing 5% BSA. Immunoreactive signals were visualized using the ECLplus system (Amersham Pharmacia Biotech, Buckinghamshire, UK).

Total RNA (2 μg) extracted from 3T3-L1 cells was reverse-transcribed with oligo dT_12-18 (Amersham Biosciences, Uppsala, Sweden) and M-MLV reverse transcriptase (Wako Pure Chemicals) at 37°C for 90 minutes. The resultant cDNA cocktail was heat inactivated and diluted to 400 μl with water. The reaction cocktail (10 μl) consisted of 1 μl of the diluted cDNA (corresponding to 5 ng of input total RNA), 12.5 pmol each of the following sense and antisense oligonucleotide pairs (PPARγ₁: 5′-AGAAGTCACACTCTGACAGG-3′ and 5′-CAATCGGATGGTTCTTCGGA-3′; PPARγ₂: 5′-ACTGCCTATGAGCACTTCAC-3′ and 5′-CAATCGGATGGTTCTTCGGA-3′; C/EBPα: 5′-TGGACAAGAACAGCAACGAG-3′ and 5′-AATCTCCTAGTCCTGGCTTG-3′; C/EBPβ: 5′-ACTACGGTTACGTGAGCCTC-3′ and 5′-CAGCTGCTTGAACAAGTTCC-3′; C/EBPδ: 5′-ACCTCTTCAACAGCAACCAC-3′ and 5′-TTCTGCTGCATCTCCTGGTT-3′; β-actin: 5′-GAGACCTTCAACACCCCAGC-3′ and 5′-CACGGAGTACTTGCGCTCAG-3′), 2 nmol of dNTP, and 0.25 U of Taq DNA polymerase in a buffer supplied by the manufacturer. In addition, dimethylsulfoxide (DMSO) was also included in the reaction cocktail to a 10% (v/v) concentration for amplification of C/EBPβ and C/EBPδ cDNAs. The protocols for the temperature cycling of the polymerase chain reaction (PCR) were 30 seconds at 94°C, 30 seconds at 58°C, and 1 minute at 72°C for 22 cycles (β-actin), 24 cycles (C/EBPβ, C/EBPδ) or 27 (C/EBPα, PPARγ₁, PPARγ₂) cycles. The number of cycles was determined to ensure quantitative amplification. Serially diluted (four times) cDNA cocktail, corresponding 20 ng, 5 ng, 1.25 ng and 0.3125 ng of input total RNA of reverse transcription reaction, of both differentiated and undifferentiated 3T3-L1 cells was used to generate a standard curve in each PCR experiment. The resulting PCR products were electrophoresed through 1.5% agarose, blotted onto a Biodyne B membrane (Pall Gelmann), and then hybridized with digoxygenin (DIG)-labelled cDNA probes for each target. During the revision of this work, real-time PCR was also performed on ABI GeneAmp 5700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) using QuantiTect SYBR Green PCR kit (Qiagen, Valencia, CA, USA) to confirm the quantitative results of PPARγ₂, PPARγ₁ and β-actin mRNAs.

Chemiluminescent images of western- or Southern-blot analyses on X-ray film were acquired using a flatbed scanner with a transparent film adapter unit (CanoScan FB1200S with FAU-S10, Canon, Tokyo, Japan) and measured by NIH Image software (version 1.62, National Institutes of Health, Bethesda, MD, USA).

The data are expressed as the mean±s.e.m. Statistical analyses were performed with Fisher's protected least significant difference test or Scheffe's F-test after analysis of variance using StatView, version 4.5 (Abacus Concepts, Berkeley, CA, USA). Differences were considered to be significant when P was less than 0.05.

According to the standard protocol for the differentiation of 3T3-L1 cells into adipocytes (Wise and Green, 1979), confluent cells were exposed to an induction medium containing DEX, MIX, INS and FBS in DMEM for 45 hours (induction period), and then the cells were maintained in a medium containing only INS and FBS for an additional period of 9 days (maturation period). Cyclic stretching (≤130% of the initial length) with a frequency of 1 Hz was applied to 3T3-L1 cells undergoing adipocyte differentiation; the cells were cultured in a collagen-coated silicon chamber (Tanabe et al., 2000). Representative phase-contrast images of 3T3-L1 cells with or without cyclic stretching (130%) during the period of differentiation are shown in Fig. 1A. After 45 hours of induction, the cells that had been subjected to cyclic stretching were oriented perpendicular to the axis of the stretching (Fig. 1Ad), whereas the cells that had not undergone cyclic stretching were randomly oriented (Fig. 1Ab). The number of viable cells subjected to cyclic stretching was examined by the trypan-blue dye-exclusion method. Although the number of cells subjected to cyclic stretching in the chambers (130%, 1 Hz) was lower (79.8±15.7%, n=4) than the number of cells in the resting state (100±12%, n=4), no remarkable decrease in the viability of cells was observed (97.7±0.4% with cyclic stretching vs 99.3±0.3% without stretching). The unidirectionally oriented response of the cells completely disappeared, and the accumulation of lipid droplets stained with Oil-Red-O was significantly inhibited after the 3T3-L1 cells had matured (Fig. 1Ac,e).

The effects of cyclic stretching on differentiation markers for adipocytes (Wise and Green, 1979) were also examined. Both the expression of cytosolic GPDH activity and the accumulation of intracellular triglycerides were significantly reduced when the 3T3-L1 cells were stretched at a frequency of 1 Hz within a range 110-175% of the original length during induction (Fig. 1B,C). Because the maximal inhibitory effect of cyclic stretching on the expression of these differentiation markers was obtained when the cells reached 130% of their original length, an optimal cyclic stretching percentage of 130% at 1 Hz was applied to the cells in the following experiments.

Three members of the C/EBP family (C/EBPα, C/EBPβ and C/EBPδ) and γ-isoforms of the PPAR family (PPARγ₁ and PPARγ₂) play important roles in the regulation of adipocyte differentiation (Cao et al., 1991; Tontonoz et al., 1994a; Tontonoz et al., 1994b; Yeh et al., 1995). Thus, the effect of cyclic stretching on the expression of these transcription factors was assessed in the present study by quantitative reverse-transcription PCR (RT-PCR).

During the induction period, C/EPBβ, C/EPBδ and PPARγ₁ were rapidly induced as early as 30 minutes (Fig. 2A,C-E); C/EBPα and PPARγ₂ (Fig. 2A,B,F) were also induced but much later during the induction period (∼45 hours). The cyclic stretching (130%, 1 Hz) significantly reduced the expression of C/EBPδ, PPARγ₁ and PPARγ₂ (Fig. 2D-F). By contrast, cyclic stretching showed no appreciable effect on the expression of C/EBPβ mRNA (Fig. 2C) and only a slightly facilitative effect on the expression of C/EBPα mRNA (Fig. 2B).

Cascade and/or cross-regulation has been reported between adipogenic transcription factors, including C/EBPs and PPARγs (Cao et al., 1991; Tanaka et al., 1997; Tontonoz et al., 1994a; Tontonoz et al., 1994b; Wu et al., 1995; Wu et al., 1999; Yeh et al., 1995). It is possible that the reduced expression of PPARγ₂ mRNA was due to an earlier suppression of a transient burst of C/EBPδ mRNA expression in response to cyclic stretching. To address the effects of cyclic stretching on the sequential expression of C/EBPs and PPARγs, the cells that underwent the commitment of differentiation were stretched during the early phase (S_E15; 0-15 hours) or the late phase (S_L15; 30-45 hours) of the induction period (Fig. 3A). After 9 days of maturation, cytosolic GPDH activity and triglyceride accumulation were measured and compared with those observed under the resting (R₄₅; 0-45 hours) and stretching (S₄₅; 0-45 hours) conditions throughout the induction period.

The GPDH activity and triglyceride content was significantly reduced when the cells were stretched throughout the induction period (S₄₅) and when they were stretched during the late phase of induction (S_L15). By contrast, neither GPDH activity nor triglyceride accumulation was affected by cyclic stretching during the early phase of induction (S_E15) (Fig. 3B,C) when the expression of C/EBPδ mRNA peaked in resting cells 3 hours after the onset of induction.

The effect of cyclic stretching in the late phase of the induction period on the expression of C/EBPs and PPARγ mRNAs was also examined. The expression of PPARγ₁ and PPARγ₂ mRNA decreased in response to the stretching during the late phase of the induction period, and no appreciable change on the expression of C/EBPα, C/EBPβ and C/EBPδ isoforms was observed (Fig. 3D). These results indicated that only the application of cyclic stretching during the late phase of induction could inhibit the adipocyte differentiation of 3T3-L1 cells, which was accompanied by the downregulation of PPARγ₁ and PPARγ₂ without any changes in the expression of C/EBP mRNA.

Several lines of evidence have suggested that ERK1/2 is involved in the signal transduction pathway in a range of cell types, including vascular endothelial cells, in response to mechanical stimuli (MacKenna et al., 1998; Li and Xu, 2000; Oldenhof et al., 2002). In order to clarify whether cyclic stretching activates ERK1/2 in differentiating 3T3-L1 cells, we performed western-blot analysis using anti-phosphorylated-ERK antibody and anti-ERK antibody to examine the effects of stretching on the phosphorylation of ERK1/2 during the induction period (Fig. 4A). No significant differences were observed in the total ERK levels among the cells, regardless of the status of differentiation or of the mechanical stimulus (Fig. 4B). Nevertheless, the phosphorylation of ERK1/2 in these cells had increased significantly by the end of the induction period in response to cyclic stretching (Fig. 4C). The stretch-induced phosphorylation of ERK1/2 was abolished by the presence of PD98,059 (Fig. 4D), a selective inhibitor of MEK1 (an isoform of MAPK/ERK kinase), at 20 μM, a concentration previously shown to be highly specific to the MEK-ERK pathway (Oldenhof et al., 2002).

We next investigated whether the activation of ERKs was involved in the stretch-induced decrease in the expression of mRNA for C/EPBδ, PPARγ₁ and PPARγ₂. PD98,059 was applied to 3T3-L1 cells during the induction period with and without cyclic stretching, and the expression levels of mRNA for C/EBPδ, PPARγ₁ and PPARγ₂ were examined by RT-PCR (Fig. 5A). The expression of both C/EBPδ and PPARγ₂, which had been reduced in response to cyclic stretching, was restored by PD98,059 (Fig. 5B,C). However, the expression of PPARγ₁, which had been lowered in response to cyclic stretching, was not restored by PD98,059 (Fig. 5D).

The expression of PPARγ at the protein level was also examined. The PPARγ protein was significantly increased during the differentiation (Fig. 6A). The cyclic stretching significantly reduced PPARγ expression at the protein level, which was also restored by PD98,059 (Fig. 6B,C). These results indicated that the stretch-induced downregulation of PPARγ mRNA was consistent with the decrease in PPARγ protein.

We also studied the effect of PD98,059 on the stretch-induced inhibition of the differentiation of 3T3-L1 cells. PD98,059 (20 μM) was administered to cells with and without cyclic stretching during the 3T3-L1 cell induction period. The unidirectionally orienting response of the cells was abolished by PD98,059 when it was administered during the induction period (Fig. 7Ab,d). The cells were then washed and cultured for 9 days of maturation without being subjected to stretching. The number of differentiated adipocytes carrying droplets of cytoplasmic triglycerides in the cells that had been subjected to the stretching in the presence of PD98,059 was similar to that of cells without the stretching (Fig. 7Ag,h). Likewise, the reduced expression of cytoplasmic GPDH activity and decreased intracellular accumulation of triglyceride observed after the maturation period was restored to the control level by PD98,059 (Fig. 7B,C). These results suggest that the inhibitory effect of cyclic stretching on the adipocyte differentiation of 3T3-L1 cells is mediated by the activation of the MEK-ERK pathway.

To examine whether the inhibitory effect of cyclic stretching on the differentiation of 3T3-L1 cells can be restored by the activation of PPARγ, the cells were treated with troglitazone, a synthetic PPARγ ligand. The addition of troglitazone at 30 μM, a concentration previously shown to act as a full agonist of PPARγ in 3T3-L1 cells (Camp et al., 2000), offset the decreased number of lipid-droplet-laden differentiated adipocytes in response to the stretching (Fig. 8Ad,e). The reduced cytoplasmic GPDH activity (Fig. 8B) and the decreased intracellular accumulation of triglyceride (Fig. 8C) were overcome by troglitazone. Thus, the inhibitory effect of the cyclic stretching on the differentiation of 3T3-L1 cells could be restored through activation of PPARγ by the synthetic ligand. Administration of troglitazone together with DEX-MIX-INS cocktail during induction period significantly increased the intracellular triglyceride in resting cells, whereas the cyclic stretching counteracted the action of troglitazone (Fig. 8C).

The present results provide direct evidence for a regulatory role of mechanical stress in adipocyte differentiation, mediated through the activation of the ERK/MAPK system. Controversial observations concerning the role of ERK/MAPK in adipocyte differentiation have been reported by several laboratories – the activation of the ERK/MAPK pathway has been shown to be involved in both the inhibition (Font de Mora et al., 1997; Hu et al., 1996; Kim et al., 2001; Shimba et al., 2001) and the promotion (Bost et al., 2002; Klemm et al., 2001; Machinal-Quelin et al., 2002; Prusty et al., 2002; Zhang et al., 1996) of adipocyte differentiation. Along these lines, Prusty et al. recently suggested that stimulation of the ERK/MAPK pathway might have opposing effects in the process of adipogenesis, depending on the time of activation during the differentiation process (Prusty et al., 2002). In the present study, the activated state of ERK1/2 was more prolonged during the induction period in response to the stretching condition than it was during the same period under the unstretched condition (Fig. 4). Moreover, only the inhibitory role of the ERK/MAPK pathway in the adipocyte differentiation of 3T3-L1 cells was elicited in response to cyclic stretching, suggesting that prolonged or continuous activation of the ERK/MAPK pathway acts in an inhibitory manner in the context of adipocyte differentiation.

It has already been shown that the cascade-like expression of three members of the C/EBP family (C/EBPα, C/EBPβ, C/EBPδ) and PPARγ (γ₁ and γ₂ isoforms of PPAR) plays an important role in the terminal differentiation of preadipocytes to triglyceride-laden adipocytes (Cao et al., 1991; Tanaka et al., 1997; Tontonoz et al., 1994a; Tontonoz et al., 1994b; Wu et al., 1995; Wu et al., 1999; Yeh et al., 1995). In the present study, the application of cyclic stretching during the entire induction period significantly reduced the expression of C/EBPδ, PPARγ₁ and PPARγ₂ mRNAs (Fig. 2D-F). However, when the stretching condition was applied during the late stage of the induction period only, the expression of C/EBPδ mRNA was not inhibited, although this condition did inhibit the expression of both PPARγ₂ and PPARγ₁ mRNAs (Fig. 3D). Furthermore, when the stretching condition was applied only during the early stage of the induction period (a condition under which the expression of C/EBPδ mRNA would be expected to be downregulated), no effect was observed on the terminal differentiation of 3T3-L1 cells (Fig. 3B,C). Regardless of the protocol, neither C/EPBα nor C/EPBβ mRNA showed any notable changes in response to the stretching condition (Fig. 2B,C, Fig. 3D). Adipocyte differentiation of the cells was also inhibited at a similar level by the `delayed' stretching protocol, which suggests that the downregulation of PPARγs was the most likely reason for the inhibition of adipocyte differentiation under the stretching condition. Consistent with the finding of the downregulation of PPARγs by the stretching, the adipocyte differentiation was restored by troglitazone, an activator of PPARγ during the induction period (Fig. 8). However, there seems to be an alternative possibility that the mechanical deformation of the cells might affect either generation or availability of endogenous PPARγ ligands. It should be noted that administration of troglitazone during induction period significantly augmented accumulation of triglyceride in resting cells, whereas the stretching restrained this action of troglitazone (Fig. 8C). Therefore, the stretching might counteract the stimulating effect of troglitazone on differentiation.

An earlier study using ectopic overexpression experiments in uncommitted NIH3T3 fibroblasts suggested that the expression of C/EBPβ alone could induce endogenous PPARγ mRNA and stimulate adipogenesis (Wu et al., 1995). By contrast, the ectopic overexpression of C/EBPδ alone was not found to induce PPARγ and adipocyte differentiation. Moreover, it was shown that the dual overexpression of both C/EBPδ and C/EBPβ enhanced PPARγ mRNA expression and differentiation (in this context, it should be noted that DEX was still required to achieve a fully induced level of PPARγ mRNA) (Wu et al., 1996). The synergistic role of C/EBPβ and C/EBPδ in adipocyte differentiation was also indicated using C/EBPβ^–/– and C/EPBδ^–/– mice both in vivo and in vitro (Tanaka et al., 1997). These results, taken together, suggest that the coordinated expression of C/EBPβ and C/EBPδ is a prerequisite for the efficient expression of PPARγ, leading to fully differentiated adipocytes. Therefore, the downregulation of C/EBPδ mRNA by itself as a result of the application of stretching during the early stage of induction period showed minor role in the inhibition of adipocyte differentiation.

The stretch-induced blockade of adipocyte differentiation was reversed by the inhibition of the ERK/MAPK pathway (Fig. 7), with concomitant restoration of the expression of PPARγ₂ at the mRNA (Fig. 5C) and protein (Fig. 6B,C) levels. By contrast, the expression of PPARγ₁ mRNA, which was reduced in response to the stretching condition, was not restored by the inhibition of the ERK/MAPK pathway (Fig. 5D). Therefore, the reduced expression of PPARγ₂ was responsible for the inhibition of the differentiation of 3T3-L1 cells in response to cyclic stretching, whereas the reduction in expression of PPARγ₁ mRNAs played a minor role in the inhibition of cell differentiation. In addition, it has been reported that PPARγ activity is negatively regulated by phosphorylation with ERK (Hu et al., 1996; Camp and Tafuri, 1997; Adams et al., 1997). It is possible that, once PPARγ is produced during the induction period of 3T3-L1 cells, its transcriptional activity is inhibited by phosphorylation with the stretch-induced MAPK. Thus, the inhibitory phosphorylation of PPARγ by the MEK-ERK signalling pathway might also be responsible for the stretch-induced inhibition of the differentiation of 3T3-L1 cells.

A range of mechanosensor molecules, thought to transduce mechanical forces into intracellular signals, have been proposed [e.g. mechanically gated ion channels (Hamill and Martinac, 2001), membrane-integrated growth factor receptors (Li and Xu, 2000) and integrins (Geiger and Bershadsky, 2002)]. At present, it remains unclear how 3T3-L1 cells receive the mechanical force that leads to the activation of ERK (followed by the attenuation of adipocyte differentiation); in other words, the means by which a negative signal is propagated for the expression of adipogenic transcription factors remains unknown. In this regard, it has been shown that ERK are activated by integrin-mediated cell adhesion to the extracellular matrix (ECM) in rat cardiac fibroblasts (MacKenna et al., 1998). During the process of differentiation to adipocytes, drastic changes are known to take place in cell morphology, cytoskeletal components and the level and type of ECM components secreted. In turn, all of these factors are recognized to exert significant influence on adipocyte differentiation (Gregore et al., 1998). Significant morphological changes were commonly observed in our experiments, even during the induction period (Fig. 1Aa,b) (i.e. in the early phase of adipocyte differentiation of 3T3-L1 cells). It is possible that, in the present cases, the appropriate cytoskeletal rearrangement was disturbed; thus, the correct arrangement has been suggested to be a prerequisite for terminal differentiation (Lieber and Evans, 1996; Gregore et al., 1998; Kawaguchi et al., 2003). This conclusion is plausible, because exposure to cyclic stretching during the induction period induced in these cells the tendency to orient in a unidirectional manner (Fig. 1Ad). Interestingly, the stretch-induced unidirectionally oriented response of the cells disappeared with the administration of PD98,059 (Fig. 7Ab,d), suggesting a link between this orienting response and the inhibition of adipocyte differentiation through the activation of the ERK/MAPK pathway. In addition, Spiegelman and Ginty indicated that fibronectin matrices decrease lipogenic gene expression and lead to decreased triglyceride accumulation (Spiegelman and Ginty, 1983). In this context, the interaction between the cell matrix and nuclear events has been suggested to play an important role in the cell differentiation (Gregore et al., 1998). Collectively, physical deformation caused by the stretching of membrane components and/or cytoskeletal components (both of which interact directly or indirectly with the ECM) might be involved in the inhibitory mechanism of adipocyte differentiation in response to cyclic stretching.

Finally, the physiological significance of the present study should be briefly addressed. The mechanosensitivity of preadipocytes suggest that gymnastic exercise and/or massage act directly on preadipocytes as a mechanical stimulus; such stimulus might activate the ERK/MAPK system, which in turn could lead to the prevention of adipocyte differentiation and renewal. This line of reasoning further implies that local massage (vibration or passive stretching applied to the body) might ameliorate obesity-associated physical conditions in terms of the prevention of adipocyte differentiation.

We thank Sankyo for providing us with troglitazone. This work was partly supported by Grants-in-Aid for Scientific Research C from the Ministry of Education, Science, Sports and Culture of Japan, a grant from the Shizuoka Research Institute from 1999 to 2000, and Goto Research Grant from University of Shizuoka in 2003.