Regulation of adipogenic differentiation by LAR tyrosine phosphatase in human mesenchymal stem cells and 3T3-L1 preadipocytes.

Mesenchymal stem cells (MSCs) are multipotent adult stem cells that can differentiate into a variety of mesodermal-lineage cells. MSCs have significant potential in tissue engineering and therapeutic applications; however, the low differentiation and proliferation efficiencies of these cells in the laboratory are fundamental obstacles to their therapeutic use, mainly owing to the lack of information on the detailed signal-transduction mechanisms of differentiation into distinct lineages. With the aid of protein-tyrosine-phosphatase profiling studies, we show that the expression of leukocyte common antigen related (LAR) tyrosine phosphatase is significantly decreased during the early adipogenic stages of MSCs. Knockdown of endogenous LAR induced a dramatic increase in adipogenic differentiation, whereas its overexpression led to decreased adipogenic differentiation in both 3T3-L1 preadipocytes and MSCs. LAR reduces tyrosine phosphorylation of the insulin receptor, in turn leading to decreased phosphorylation of the adaptor protein IRS-1 and its downstream molecule Akt (also known as PKB). We propose that LAR functions as a negative regulator of adipogenesis. Furthermore, our data support the possibility that LAR controls the balance between osteoblast and adipocyte differentiation. Overall, our findings contribute to the clarification of the mechanisms underlying LAR activity in the differentiation of MSCs and suggest that LAR is a candidate target protein for the control of stem-cell differentiation.


Introduction
Adipocytes are highly specialized cells that have an important role in energy homeostasis by harboring energy reservoirs as lipid droplets (Cornelius et al., 1994;Hwang et al., 1997). However, these reservoirs are implicated in a host of major human health problems, because excessive or insufficient energy reserves result in metabolic disorders, such as obesity and lipodystrophy (Garg, 2000;Otto and Lane, 2005). Adipogenesis involves the formation of preadipocytes from mesenchymal progenitor cells and their subsequent differentiation into adipocytes (Gregoire et al., 1998;Rosen and Spiegelman, 2000). The cellular and molecular mechanisms of adipocyte differentiation are regulated via activation of several adipogenic intracellular signaling pathways.
Insulin signaling clearly has marked effects on adipogenesis. The mechanisms of insulin are mediated by a cascade of tyrosinephosphorylation events initiated by the binding of insulin to its receptor (White and Kahn, 1994;White, 1997;White and Yenush, 1998). Binding stimulates the kinase activity of the insulin receptor (IR) and the phosphorylation of IR substrates (IRSs), leading to the activation of downstream signaling molecules, including phosphoinositide 3-kinase (PI3K) and Akt (also known as PKB). Activated Akt regulates the activity of several downstream proteins involved in gluconeogenesis, lipogenesis and adipogenesis (White and Kahn, 1994;Wang and Sul, 1998;Czech and Corvera, 1999;Withers et al., 1999;Baudry et al., 2006;Rosen and MacDougald, 2006). Downstream components of the insulin signaling cascade are crucial for adipogenesis. The loss of individual IRS proteins, including the combined deletion of IRS1 and IRS2, leads to inhibition of adipogenesis (Laustsen et al., 2002;Tseng et al., 2004;Rosen and MacDougald, 2006). Moreover, inhibition of PI3K and the loss of Akt repress adipogenesis via regulation of adipogenic and anti-adipogenic transcription factors (Garofalo et al., 2003;Nakae et al., 2003;Wofrum et al., 2003;George et al., 2004;Menghini et al., 2005;Rosen and MacDougald, 2006).
Numerous growth-factor and -hormone receptors belong to the tyrosine-kinase-receptor family and undergo phosphorylation and dephosphorylation at tyrosine residues in a concerted manner in response to a stimulus in order to initiate the signaling cascade (Meng et al., 2004;Niu et al., 2007). Termination of insulin activity is also mediated by protein tyrosine phosphatases (PTPs), which dephosphorylate and inactivate the IR and, subsequently, postreceptor substrates. PTPs constitute a large family of transmembrane or intracellular enzymes that function as either positive or negative regulators of a number of signaling pathways (Goldstein, 1993;Denu et al., 1996;Tonks and Neel, 1996). Several PTPs, such as leukocyte common antigen related (LAR; also known as PTP-RF) phosphatase, PTP-a, PTP-1B and SHPTP2 (also known as Syp), are highly expressed in the major insulin-sensitive tissues, including the liver, skeletal muscle and adipose tissue. These enzymes are involved in insulin signaling, as established using a variety of

Mesenchymal stem cells (MSCs) are multipotent adult stem cells that can differentiate into a variety of mesodermal-lineage cells.
MSCs have significant potential in tissue engineering and therapeutic applications; however, the low differentiation and proliferation efficiencies of these cells in the laboratory are fundamental obstacles to their therapeutic use, mainly owing to the lack of information on the detailed signal-transduction mechanisms of differentiation into distinct lineages. With the aid of protein-tyrosine-phosphatase profiling studies, we show that the expression of leukocyte common antigen related (LAR) tyrosine phosphatase is significantly decreased during the early adipogenic stages of MSCs. Knockdown of endogenous LAR induced a dramatic increase in adipogenic differentiation, whereas its overexpression led to decreased adipogenic differentiation in both 3T3-L1 preadipocytes and MSCs. LAR reduces tyrosine phosphorylation of the insulin receptor, in turn leading to decreased phosphorylation of the adaptor protein IRS-1 and its downstream molecule Akt (also known as PKB). We propose that LAR functions as a negative regulator of adipogenesis. Furthermore, our data support the possibility that LAR controls the balance between osteoblast and adipocyte differentiation. Overall, our findings contribute to the clarification of the mechanisms underlying LAR activity in the differentiation of MSCs and suggest that LAR is a candidate target protein for the control of stem-cell differentiation.
LAR is a receptor-type PTP that has two tandem-repeat cytoplasmic phosphatase domains (D1 and D2), of which the membrane-proximal D1 domain possesses catalytic activity (Streuli et al., 1990;Nam et al., 1999). LAR is widely expressed in insulinsensitive tissues. Increased association of LAR with the IR occurs after treatment with insulin (Ahmad et al., 1995;Ahmad and Goldstein, 1997). Several studies have reported that LAR is negatively regulated during insulin signaling. Moreover, overexpression of LAR leads to suppression of insulin activity, and knockdown of LAR enhances IR phosphorylation and PI3K activity in hepatoma cells (Kulas et al., 1995;Kulas et al., 1996;Li et al., 1996). An in vivo study similarly reported that overexpression of LAR in mouse skeletal muscle suppresses IR signaling (Zabolotny et al., 2001). LAR also regulates insulin-like growth factor-1 receptor signaling in vascular smooth-muscle cells (Niu et al., 2007). These reports strongly implicate LAR in insulin-mediated adipocyte differentiation. There are no direct reports on the exact role of LAR in adipogenesis, although earlier studies have demonstrated that LAR expression is decreased during adipogenesis of human mesenchymal stem cells (MSCs) (Song et al., 2006). The present study was undertaken to determine the regulatory effects of LAR on the differentiation of preadipocytes and human MSCs into adipocytes.

Expression of LAR during the differentiation of MSCs and 3T3-L1 cells into adipocytes
To determine the roles of the PTP family during the early phase of adipogenesis, we extensively assessed the expression changes of all PTP members via reverse-transcriptase PCR (RT-PCR) analysis (data not shown). Interestingly, expression of LAR mRNA was markedly decreased during adipocyte differentiation of human MSCs (Fig. 1A). To obtain a detailed assessment of the LAR level, real-time PCR analysis was performed. LAR mRNA expression was clearly decreased at the later as well as the early stages of adipogenesis in MSCs (Fig. 1B). Next, we assessed LAR expression in 3T3-L1 preadipocyte cell lines using both RT-PCR and western blot analyses. As shown in Fig. 1C,D, LAR expression levels were significantly decreased during the differentiation of 3T3-L1 preadipocytes into adipocytes. The protein stability of LAR seemed to be higher than that of LAR mRNA, but on the whole a correlation between them was observed. Levels of adipogenic markers, such as aP2 (fatty-acid-binding protein; FABP) and PPAR-, were increased following hormonal stimulation. These results clearly show that LAR expression decreased during differentiation into adipocytes in both MSCs and preadipocyte cells.

Depletion of LAR promotes the differentiation of 3T3-L1 preadipocytes into mature adipocytes
To determine whether endogenous LAR influences differentiation into adipocytes, we infected 3T3-L1 preadipocytes with a retrovirus expressing LAR shRNA, scrambled insert or control vector only. Infected cells were isolated by FACS sorting and then cultured further. As shown in Fig. 2A (upper panels), cells infected with a retrovirus expressing shRNA against LAR were successfully enriched. Knockdown of endogenous LAR expression was confirmed by RT-PCR and western blot analyses (Fig. 2B). Next, retrovirally transduced 3T3-L1 preadipocytes were induced to differentiate into adipocytes for 2 days in growth medium containing MDI (see Materials and Methods), followed by further differentiation and maturation in growth medium supplemented with insulin for 6 days. In parallel, fat accumulation was visualized by staining lipid droplets with Oil red-O (Fig. 2C). Interestingly, depletion of LAR dramatically facilitated the differentiation of these cells into mature adipocytes compared with control and scrambled retrovirus-infected cells (Fig.  2D). Next, to exclude off-target effects of shRNA treatment, we attempted to rescue the effect on differentiation by re-introducing an shRNA-resistant cytoLAR or mutant LAR (D1-CS; activity-dead mutant) into the LAR-knockdown 3T3-L1 cells. Most of the reinfected cells were detected as both GFP-and RFP-positive by fluorescence microscopy ( Fig. 2A, lower panels). Knockdown of endogenous LAR and expression of shRNA-resistant cytoLAR or mutant LAR D1-CS were confirmed by RT-PCR analysis (Fig. 2E). Consistently, the knockdown of endogenous LAR mRNA levels was continuously maintained until 6 days after the adipogenic differentiation of 3T3-L1 cells (Fig. 2E). The re-introduction of wildtype cytoLAR into LAR-knockdown cells induced a full recovery of the differentiation rate. By contrast, the re-introduction of LAR D1-CS (C)Analysis of LAR expression during adipocyte differentiation of 3T3-L1 preadipocytes. LAR mRNA (upper) and protein (lower) expression levels decreased during adipocyte differentiation of 3T3-L1 preadipocytes. For RT-PCR analysis, total RNA was extracted on the indicated days of differentiation and used to synthesize cDNA for RT-PCR. Two primer sets were used to check for LAR expression, one for amplification of the mRNA region encoding the extracellular domain (Extra LAR) and another for amplification of the mRNA region encoding the cytoplasmic domain (Cyto LAR). The results obtained using the two sets of primers revealed similarly decreased LAR expression patterns during adipogenesis of 3T3-L1 cells. aP2 and PPAR- were used as positive controls for adipocyte differentiation and actin was used as a loading control. For western blot analysis, whole-cell lysates were extracted on the indicated days of differentiation, and subjected to analysis using antibodies against LAR and adipocyte-differentiation markers, such as aP2 and PPAR-. -actin was used as a loading control. (D)Densitometric analysis of LAR protein expression during adipocyte differentiation of 3T3-L1 cells (means ± s.d.; n3; *P<0.05).
showed no significant changes in the differentiation rate of the cells, indicating the crucial involvement of LAR phosphatase activity in adipogenesis (Fig. 2F,G).

Ectopic expression of LAR inhibits the differentiation of 3T3-L1 preadipocytes into adipocytes
Depletion of LAR induced a dramatic increase in the adipocyte differentiation of 3T3-L1 cells, indicating that LAR phosphatase is involved in the negative regulation of adipocyte differentiation.

Journal of Cell Science 122 (22)
To clarify the role of LAR in adipogenesis, we infected 3T3-L1 cells with the cytoplasmic domain of LAR using a retroviral system (cytoLAR-IRES-GFP). LAR D1-CS and control vector were used as negative controls. Infected cells were isolated using a FACS sorter. Most of the cells were GFP-positive by fluorescence microscopy (Fig. 3A). Overexpression of cytoplasmic and mutant LAR protein was verified by western blot analysis using anti-FLAG and anti-LAR antibodies (Fig. 3B). Expression of LAR protein was continuously detected during the late phases of differentiation (Fig. 3E). Infected cells were induced to differentiate into mature adipocytes, and lipid accumulation was assessed by Oil-red-O staining 10 days after culturing with differentiation medium (Fig. 3C). Overexpression of wild-type LAR resulted in a lower degree of lipid accumulation compared with control and LAR-mutant protein (Fig. 3C,D). This antiadipogenic effect of LAR was also demonstrated in the adipogenic differentiation induced by treatment with insulin only (data not shown). Consistently, expression levels of adipogenic markers, such as aP2 and PPAR-, were reduced in mature adipocytes upon LAR overexpression (Fig. 3E).

Effects of LAR on adipocyte differentiation of human MSCs
LAR and specific shRNA against LAR were stably expressed in human MSCs via retroviral infection to establish the importance of LAR during the adipogenesis of human MSCs (Fig. 4A). LAR Fig. 2. LAR shRNA promotes 3T3-L1 adipogenic differentiation. (A)Direct monitoring of fluorescence protein expression using fluorescence microscopy after enrichment using a FACS sorter. 3T3-L1 cells were infected with retrovirus containing the vector alone (pSIREN-RetroQ-DsRed), scrambled insert or LAR shRNA (upper). In addition, retroviruses containing the vector alone (pRetroX-IRES-ZsGreen1), cytoLAR or LAR D1-CS mutant were used to re-infect LARknockdown cells (lower). Infected cells were then selected with a FACS sorter. (B)Knockdown of LAR was confirmed by western blot (upper) and RT-PCR (lower) analyses. As shown, shLAR-II provided more efficient LAR knockdown than the other constructs tested. Thus, this construct was used for the following phenotype experiments. (C)LAR-depleted cells (using the shLAR-II construct) were induced to differentiate into adipocytes by treatment with MDI medium. Six days later, the cells were stained with Oil red-O to visualize the degree of lipid accumulation. (D)Quantification of Oil-red-O staining. Data represent the mean percentage levels ± s.d. compared with control vector (n3; *P<0.05). (E)Analysis of LAR mRNA levels during adipogenic differentiation of LARknockdown cells after re-introduction of the LAR-shRNA-resistant cytoLAR or LAR D1-CS mutant. Total RNA was extracted on day 0 or 6 for RT-PCR. Human LAR primer set (see Materials and Methods) was used for RT-PCR. actin was used as a loading control. (F)The rescue experiment was performed by investigating the effect on differentiation of the re-introduction of an shRNAresistant cytoLAR or LAR D1-CS mutant into LAR-knockdown 3T3-L1 cells. Cells were differentiated into adipocytes by the addition of MDI medium. Six days later, cells were stained with Oil red-O. (G)Quantification of Oil-red-O staining. Data represent the mean percentage levels ± s.d. compared with control vector (n3; *P<0.05). expression and knockdown were confirmed by western blot and RT-PCR analyses, respectively (Fig. 4B). Analogous to the data obtained with 3T3-L1 preadipocytes, knockdown of LAR promoted adipocyte differentiation of human MSCs (Fig. 4C,D), whereas LAR overexpression suppressed adipogenic differentiation (Fig. 4E,F).

LAR affects insulin-mediated adipogenic signal transduction
To clarify the mechanism underlying LAR-induced regulation of adipogenesis, we examined the phosphorylation states of IR, IRS-1, and its downstream target molecule Akt following treatment with insulin or MDI. Consistent with a previous report , tyrosine phosphorylation of IR in 3T3-L1 preadipocytes was rapidly increased in response to insulin or MDI. Notably, overexpression of LAR led to the inhibition of insulin-and MDI-induced tyrosine phosphorylation of IR and IRS-1 (Fig. 5A,B). Consistently, this decrease in tyrosine phosphorylation resulted in reduced phospho-Akt levels in cells overexpressing LAR, compared with control cells (Fig. 5A,B). A marginal effect of activity-dead mutant LAR on the phosphorylation of IR, IRS-1 and Akt was detected, indicating that the phosphatase activity of LAR is directly involved in the regulation of adipogenesis. Next, we checked the phosphorylation states of IR, IRS-1 and its downstream target molecule Akt after knockdown of LAR using an shRNA construct. LAR knockdown induced an increase in the phosphorylation of IR and IRS-1, resulting in augmented phospho-Akt levels (Fig. 5C,D). These results strongly suggest that LAR functions as a negative regulator of adipogenesis by controlling the phosphorylation level of IR, IRS-1 and, consequently, Akt.

Effects of LAR on osteoblast differentiation
To establish whether LAR is involved in regulating osteoblastogenesis, we first examined the change in the expression level of LAR during osteoblastogenesis of MC3T3-E1 cells. As expected, the mRNA level of LAR was continuously increased during the differentiation of MC3T3-E1 cells into osteoblasts (Fig.  6A). LAR shRNA or cytoplasmic LAR was stably introduced into MC3T3-El preosteoblast cells using a retroviral system. Knockdown and overexpression of LAR were confirmed by fluorescence microscopy, western blot and RT-PCR analyses (Fig. 6B,C). Cells were differentiated into osteoblasts by culturing them in osteogenic induction medium for 21 days. Cells were then stained with Alizarin-Red-S solution to visually detect mineralization. Interestingly, cells depleted of LAR displayed a significantly lower degree of mineralization than control cells (Fig. 6D,E). By contrast, the degree of mineralization was significantly increased in cells overexpressing LAR, compared with control cells. LAR D1-CS  displayed a mineralization rate similar to the control cells (Fig.  6F,G). These results support the possibility that LAR is involved in osteoblastogenesis as well as adipogenesis.

Discussion
Tyrosine phosphorylation is one of the fundamental mechanisms underlying the numerous important aspects of eukaryote physiology, including cell growth and differentiation. Adipocyte differentiation is regulated by tyrosine phosphorylation balanced by the antagonistic actions of protein tyrosine kinases (PTKs) and PTPs (Hunter, 1987;Hunter, 1998; Blume-Jensen and Hunter, 2001).

Journal of Cell Science 122 (22)
Among the ~100 PTPs, several play important roles in insulin signaling, including LAR. Insulin signaling promotes adipogenesis (Smith et al., 1988;Rosen and MacDougald, 2006); however, limited information is available on the relationship between LAR and adipogenesis. In the present study, the effects of LAR on adipocyte differentiation and insulin signal transduction during adipogenesis in 3T3-L1 preadipocytes and human MSCs were investigated.
LAR mRNA and protein levels were markedly suppressed during adipocyte differentiation, suggesting that LAR might be involved in adipogenesis. The data are consistent with the previous finding that LAR expression is decreased during differentiation into adipocytes and increased during dedifferentiation (Zhou et al., 2005;Song et al., 2006). However, there is no direct evidence for the mechanism of LAR action in adipogenesis, either in preadipocytes or MSCs. In our experiments, ectopic expression of the cytoplasmic region of LAR in 3T3-L1 preadipocytes was associated with significant inhibition of adipogenesis (Fig. 3C,D), implying that LAR is crucial for the adipocyte differentiation process. In addition, the overexpression of LAR, but not of the LAR D1-CS mutant, was associated with a marked decrease in PPAR- and aP2 expression (Fig. 3E), indicating that the phosphatase functions via the downregulation of these key adipogenic factors. The finding that an increase in LAR levels resulted in the inhibition of the adipogenic differentiation of human MSCs (Fig. 4C) suggests that LAR blocks adipogenesis in both committed preadipocytes and multi-lineage stem cells. In addition, silencing of LAR in both 3T3-L1 cells and MSCs dramatically induced adipogenic differentiation. These results indicate that LAR has an anti-adipogenic effect in both stem cells and committed cells.
LAR functions as a negative regulator of insulin signaling in cellculture and in animal model studies (Kulas et al., 1995;Kulas et al., 1996;Li et al., 1996;Zabolotny et al., 2001). Because insulin plays an important role in adipogenesis, we propose that LAR suppresses adipogenesis by modulating insulin signaling through its tyrosine-phosphatase activity. Consistent with previous observations (Kulas et al., 1995;Kulas et al., 1996;Li et al., 1996), LAR markedly reduced IR tyrosine phosphorylation (possibly in a direct manner) following insulin or MDI treatment, which in turn led to decreased phosphorylation of the adaptor protein IRS-1 and its downstream signaling molecule Akt (Fig. 5A,B). Previous investigations reported that interference with the IRS-1-and Aktassociated signaling pathways represses adipogenesis (Laustsen et al., 2002;Garofalo et al., 2003;George et al., 2004;Tseng et al., 2004;Baudry et al., 2006). Our data clearly show that the reduced adipogenic differentiation of 3T3-L1 and human MSCs by LAR occurs via decreased activation of IRS-I and Akt. Whereas cytoplasmic LAR was previously shown to have no effect on phosphorylation of components of the insulin signaling pathway in CHO cells , the protein clearly affected the insulin signaling pathway via dephosphorylation of IR under our experimental conditions. Furthermore, the marginal effect of the inactive LAR mutant strongly supports the importance of phosphatase activity in controlling adipogenesis.
The relationship between bone and fat has been established at both the developmental and physiological level. During development, mesenchymal precursor cells differentiate into adipocytes, osteoblasts and other cell types in vitro, and a reciprocal relationship exists such that increased osteoblast differentiation is associated with decreased adipocyte differentiation and vice versa (Nuttall and Gimble, 2000; Ichida analyses. (D)MC3T3-E1 LAR-knockdown cells (using the shLAR-II construct) were induced to undergo osteoblastic differentiation for 21 days, according to standard procedures. Cells were stained with Alizarin-Red-S to visualize the bone matrix. (E)Quantification of stained cells was performed. Data represent the mean percentage levels ± s.d. compared with control vector (n3; *P<0.05). (F)MC3T3-E1 cells overexpressing cytoLAR or LAR D1-CS mutant were induced to differentiate into osteoblasts for 21 days, according to standard procedures. Cells were stained with Alizarin-Red-S to visualize the bone matrix. (G)Quantification of stained cells was performed using dye extraction buffer. Data represent the mean percentage levels ± s.d. compared with control vector (n3; *P<0.05). et al., 2004). Interestingly, in contrast to adipogenic differentiation, osteoblast differentiation was augmented upon LAR overexpression and reduced following LAR depletion. This reciprocal effect of LAR on adipogenesis and osteoblastogenesis is consistent with previous findings that GILZ (glucocorticoidinduced leucine zipper) induces osteoblastogenesis and inhibits adipogenesis in MSCs, and that TAZ (transcriptional co-activator with PDZ-binding motif) controls the balance of osteoblast and adipocyte differentiation (Hong et al., 2005;Zhang et al., 2008). In conjunction with earlier data, our results suggest that inhibition of adipogenic differentiation by LAR occurs through the enhancement of osteogenic differentiation. Thus, LAR might act as a modulator of the balance between osteoblast and adipocyte differentiation in MSCs. However, the reason for the marginal effect of LAR on osteoblast differentiation, compared with that on adipocyte differentiation, remains to be defined. Thus, extensive studies to establish the mechanism underlying LAR involvement in osteoblastogenesis are warranted. In addition, LAR knockout (KO) mice were smaller in size than wild-type mice, were insulinresistant and exhibited subtle alterations in neurogenesis, although they have been reported to be, in general, normal and to possess no adipose-tissue or bone morphogenesis defects (Ren et al., 1998). Therefore, to elucidate the detailed functions of LAR in adipogenesis and osteoblastogenesis at an organism level, it is necessary to conduct more detailed investigations of LAR KO mice.

Cell culture
The preadipocyte cell line 3T3-L1, derived from mouse embryo fibroblasts, was cultured in growth medium [high glucose Dulbecco's modified Eagle's medium (DMEM) containing 1% antibiotic-antimycotic solution and 10% bovine calf serum (BCS); Gibco-Invitrogen] at 37°C in a humidified atmosphere with 5% CO 2 . MC3T3-E1 preosteoblasts were purchased from ATCC and cultured in maintenance medium [a-mineral essential medium (MEM) with 10% fetal bovine serum (FBS) and 1% antibiotic-antimycotic solution). Human MSCs were purchased from Cambrex Bio Science and maintained in MSC growth medium (MSCGM; Cambrex Bio Science) at 37°C in a humidified atmosphere with 5% CO 2 .

Differentiation into adipocytes
3T3-L1 cells were induced to differentiate into mature adipocytes, as described previously (Kim et al., 2008;Jung et al., 2009). Briefly, at 2 days after confluence (day 0), cells were placed in differentiation medium composed of DMEM, 10% FBS and MDI, a differentiation cocktail of 0.5 mM isobutylmethylxanthine (IBMX), 1 M dexamethasone and 10 g/ml insulin (Sigma), and then switched to maintenance medium composed of DMEM, 10% FBS and 10 g/ml insulin. The medium was replenished every other day.
Adipocyte differentiation of human MSCs was induced as recommended by the manufacturer (Cambrex). In brief, after reaching confluence, the medium was switched to adipogenic induction medium (MSCGM plus MDI and 200 M indomethacin; Cambrex) and the cells were cultured for 3 more days. The medium was switched to adipogenic maintenance medium (MSCGM plus 10 g/ml insulin; Cambrex) for 1-2 days. Three complete cycles of induction-maintenance medium were required to stimulate optimal adipogenic differentiation.

Differentiation into osteoblasts
To induce osteoblastogenesis, confluent MC3T3-E1 cells were switched to maintenance medium (a-MEM with 10% FBS) supplemented with 50 g/ml ascorbic acid and 10 mM -glycerophosphate; the medium was changed every 2-3 days for 21 days.
Osteoblast differentiation of human MSCs into mature osteoblasts was induced as described previously (McBeath et al., 2004). Confluent MSCs were switched to osteogenic differentiation medium (50 M ascorbic acid, 10 mM -glycerophosphate and 100 nM dexamethasone; Sigma) and the medium was changed every 2-3 days.

Construction of retroviral vectors and transduction
To construct 3T3-L1 cells and MSCs stably expressing the FLAG-tagged cytoplasmic domain of human LAR (spanning residues 1316-1897), a retrovirus-mediated infection system was used. For expression of cytoplasmic LAR, DNA encoding FLAG-tagged cytoplasmic LAR was inserted into the multi-cloning site of the pRetroX-IRES-ZsGreen1 vector (Clontech). Retroviruses were subsequently produced by transiently co-transfecting GP2-293 cells with a retroviral vector and VSV-G plasmid using Lipofectamine 2000 (Gibco-Invitrogen). At 48 hours after transfection, media containing retroviruses were collected, filtered with 0.45-m filters and used to infect cells in the presence of polybrene (8 g/ml). Infected cells were selected using a FACSAria cell sorter (BD Bioscience) and further maintained in growth medium. Ectopic expression of FLAG-tagged cytoplasmic LAR was confirmed by western blot analysis. As a negative control, a cytoplasmic LAR mutant (LAR D1-CS; inactive mutant; catalytic Cys1522 replaced with Ser) was constructed (Streuli et al., 1990;Nam et al., 1999).
shRNA-resistant cytoLAR and LAR D1-CS mutant were constructed by sitedirected mutagenesis of the region that shLAR-II binds in order to degrade LAR mRNA. Therefore, these constructs possess an altered mRNA sequence in the region corresponding to the shRNA sequence. The constructs were obtained by standard methods using the primers 5Ј-GGCCTATATAGCCACACAGGGGCCTC-3Ј and 5Ј-TTCTGCTGTCTATAACCATCCAGGA-3Ј. Underlined regions indicate the mutated sites (mutations did not induce a change in amino acid residues).

Immunoblot analysis
Cells were washed three times with ice-cold phosphate buffered saline (PBS) containing 1 mM sodium orthovanadate and harvested in ice-cold RIPA or NP-40 lysis buffer containing a protease-inhibitor and phosphatase-inhibitor cocktail (Roche). Protein concentrations were measured with the BCA protein assay kit (Pierce). SDS-PAGE, western blot and densitometric analyses were performed using standard protocols. The anti-LAR antibody is described in a previous report (Niu et al., 2007). Anti-aP2, anti-PPAR-, anti-phospho-Akt (cat. no. #9271S; antibody to phosphorylated Ser473), anti-Akt, anti-IR and anti-IRS-1 antibodies were purchased from Cell Signaling. The anti-phosphotyrosine antibody (4G10) was obtained from Upstate, whereas anti-FLAG and anti--actin antibodies were from Sigma. The secondary antibodies were purchased from Abcam, and membranes were visualized using the SuperSignal West Pico Chemiluminescent Substrate kit (Pierce).

Immunoprecipitation
For immunoprecipitation, anti-IR or anti-IRS-1 antibodies (1:50-1:100 dilution) were added to lysates containing equal amounts of protein (200-300 g). The mixture was incubated overnight at 4°C. Protein A/G plus agarose beads (Calbiochem) were added, followed by agitation for 1 hour at 4°C. Immunoprecipitates were recovered by centrifugation at 2500 g, washed five times with NP-40 lysis buffer and analyzed as described above.

Oil-red-O staining
Lipid droplets in differentiating or mature adipocytes were stained using the Oil-red-O method, as described previously (Kim et al., 2008;Jung et al., 2009). In brief, cultured cells were washed twice with PBS, fixed for 30 minutes with 10% formalin and washed twice with distilled water prior to staining. Lipid droplets within the cell were stained for 30 minutes using 0.3% filtered Oil-red-O solution in 60% isopropanol (Sigma). The cells were then washed twice with distilled water and micrographs were obtained. Oil-red-O staining was then quantified, as described previously (Ramirez-Zacarias et al., 1992). After elution from fixed cells with absolute isopropanol, the extracted dye was measured with a GeneQuant 1300 spectrophotometer (GE Healthcare) at 510 nm.

Alizarin-Red-S staining
Alizarin-Red-S staining and mineral-content quantitation were performed following earlier procedures with minor modifications (Stanford et al., 1995). Briefly, the cells were washed with calcium and phosphate-free saline solution, and fixed with 10% formalin for 30 minutes. After washing with distilled water, the mineral content was stained with 40 mM Alizarin-Red-S solution (pH 4.2, Sigma) at room temperature for 10 minutes. Cells were rinsed five times with distilled water and micrographs were obtained. Alizarin-Red-S was additionally quantified after elution from fixed cells with 10% (w/v) cetylpyridinium chloride (Sigma) and the absorbance was measured at 570 nm using a GeneQuint1300 spectrophotometer (GE Healthcare).

Statistical analysis
All quantitative data were analyzed using an independent Student's t-test and considered as significant at P<0.05.