Deregulated MAPK Activity Prevents Adipocyte Differentiation of Fibroblasts Lacking the Retinoblastoma Protein*

A functional retinoblastoma protein (pRB) is required for adipose conversion of preadipocyte cell lines and primary mouse embryo fibroblasts (MEFs) in response to treatment with standard adipogenic inducers. Inter-estingly, lack of functional pRB in MEFs was recently linked to elevated Ras activity. Ras-dependent signaling plays a significant, although incompletely understood, role in adipocyte differentiation, because activated Ras has been reported to either promote or inhibit adipogenesis depending on the cellular context. In various cell types activation of Ras leads to activation of the mitogen-activated protein kinases (MAPKs), extracellular signal-regulated kinase 1/2 (ERK1/2), and protein kinase B (PKB)/Akt, which exert opposing effects on adipogenesis, with ERK1/2 inhibiting and PKB/Akt promoting terminal differentiation. Here we report that the levels of activated ERK1/2 and PKB/Akt are significantly increased in pRB-deficient MEFs both before and after the addition of adipogenic inducers. Consistently, we detected higher levels of activated Ras in MEFs lacking pRB. Suppression of ERK1/2 activation by the MEK inhibitor UO126 restored the ability of pRB-deficient MEFs to undergo adipocyte differentiation, as mani-fested by expression of adipocyte marker genes and lipid accumulation. Furthermore and

The conversion of fibroblast-like precursor cells to fully differentiated adipocytes is a process controlled by a complex network of signaling pathways and is tightly regulated by numerous transcription factors, including members of the peroxisome proliferator-activated receptor (PPAR) 1 and CCAAT/ enhancer-binding protein (C/EBP) families, and the adipocyte determination and differentiation-dependent factor-1/sterol regulatory element-binding protein-1 (for review, see Refs. [1][2][3]. Ras-dependent signaling has been shown to have a significant impact on adipogenesis, although the reported results are difficult to reconcile (4 -7). In a series of studies, it was shown that expression of oncogenic Ras in 3T3-L1 preadipocytes induces adipocyte differentiation, whereas expression of dominant-negative Ras abrogates adipose conversion of these cells (4,5). In contrast, it was reported that expression of an activated Ras mutant abolished differentiation of both C3H10T1/2 and certain subclones of 3T3-L1 cells (6,7). Terminal adipocyte differentiation is promoted by or dependent on the activity of a number of kinases, most notably kinases of the phosphatidylinositol 3-kinase-protein kinase B (PKB/Akt) pathway (8,9) and kinases leading to activation of p38 mitogen-activated protein kinases (MAPKs) (10,11). In contrast, sustained activation of the ERK1/2 MAPKs has been demonstrated to inhibit adipogenesis (12), at least in part, by mediating an inhibitory phosphorylation of PPAR␥, a crucial regulator of terminal adipose conversion (13,14).
The retinoblastoma protein (pRB), encoded by the retinoblastoma tumor suppressor gene (Rb), is a key regulator of proliferation, development, and the differentiation of multiple cell types, including skeletal muscle and fat cells (reviewed in Ref. 15). The role of pRB in development and differentiation relates to the ability of pRB to repress E2F transcription factors and/or enhance the activity of transcription factors promoting differentiation. Examples of such pRB targets are MyoD and C/EBP␤ in myogenesis and adipogenesis, respectively (16 -19). A recent report indicated an alternative function of pRB in cellular differentiation by showing that the absence or the inactivation of pRB led to an increased activity of Ras (20). Expression of dominant-negative Ras in pRB-deficient fibroblasts partially restored the transactivation capacity of MyoD but did not restore repression of E2F. However, the downstream effectors of Ras-dependent repression of MyoD activity were not identified (20). Although it was not reported whether dominant-negative Ras normalized the defective myocyte differentiation of RbϪ/Ϫ cells, these data indicate that deregulated Ras signaling may be an important event responsible for the defects in differentiation associated with the absence of a functional pRB, at least in the case of skeletal muscle. Recent work in Caenorhabditis elegans further underscores the important link between pRB and Ras signaling in regulating embryonic asymmetry and vulval development (21,22).
Simian virus 40 TAg has been shown to inhibit 3T3-L1 preadipocyte differentiation in a manner partly dependent on the ability of TAg to bind and inactivate pRB (23). Similarly, RbϪ/Ϫ MEFs do not undergo significant adipocyte differentiation in response to treatment with standard adipogenic inducers (19), but the block in differentiation caused by pRB deficiency can be bypassed by administration of the potent PPAR␥ agonist BRL49653 (24). Here we show that the activities of two targets of Ras signaling, the ERK1/2 MAPKs and PKB/Akt, are up-regulated in RbϪ/Ϫ MEFs compared with wild-type MEFs. This is most likely due to a pRB-dependent difference in Ras signaling, as we observed an increased amount of active Ras in RbϪ/Ϫ cells. We show that BRL49653-induced differentiation is accompanied by down-regulation of ERK1/2 activities. In contrast, treatment of RbϪ/Ϫ MEFs with standard adipogenic inducers even led to an elevation of the levels of activated ERK1/2 during the first 6 days of treatment. We hypothesized that the up-regulation of ERK1/2 activities might prevent adipocyte differentiation of pRB-deficient MEFs and, furthermore, that blocking ERK1/2 activation would allow adipogenesis even in the absence of exogenously added insulin-like growth factor-1/insulin due to the elevated levels of active PKB/Akt. In keeping with these hypotheses, we demonstrate that suppression of ERK1/2 activation by UO126 to approximately the level of activation observed in wild-type MEFs indeed restored adipogenesis of RbϪ/Ϫ MEFs in response to a standard adipogenic treatment and, furthermore, that differentiation of the pRBdeficient MEFs proceeded independently of the addition of insulin.
Activated Ras Interaction Assay-The activated Ras interaction assay (ARIA) allows a measure of the amount of active (GTP-bound) Ras present in whole cell extracts and was performed essentially as described (25), with minor modifications. Briefly, glutathione S-transferase-Raf-1 (amino acids 1-149) was expressed in Escherichia coli and purified on glutathione-Sepharose beads. Approximately 20 g of fusion protein were used per assay. The fusion protein was washed twice in magnesium lysis buffer (MLB) (25 mM HEPES, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 10% glycerol, 1 mM EDTA, 10 mM MgCl 2 , 0.5 mM phenylmethylsulfonyl fluoride, and Com-plete TM (Roche Molecular Biochemicals)). One 15-cm dish of confluent cells was lysed in 1.5 ml of MLB on ice for 10 min, after which the cells were scraped off, vortexed, and incubated on ice for another 10 min. After centrifugation (14000 rpm, 15 min, 4°C) the protein concentration of the supernatant was determined according to the Bradford procedure to equalize input. 3.8 mg of total protein were used per assay. The cell lysate was incubated with the fusion protein at 4°C for 60 min with rotation. After the incubation, the beads were washed 3 times with 1 ml of MLB, boiled in SDS sample buffer, run on 12.5% SDS-polyacrylamide gels, and blotted. Interacting Ras was visualized with chemiluminescence using a pan anti-Ras mouse monoclonal antibody (Quality Biotech, clone LA045). As a control for equal input, 0.5% of the cell lysate from confluent Rbϩ/ϩ and RbϪ/Ϫ MEFs used for the ARIA experiments was loaded on a gel, and the amount of total Ras was determined by Western blotting using the anti-Ras antibody described above. Therefore, the input represents the combined amount of GTPand GDP-bound Ras, whereas the ARIA method provides a measure of the amount of Ras bound to GTP.
Reverse Transcription-PCR-RNA purification, reverse transcription, and multiplex reverse transcription-PCR were performed as described (24). Primer sets have been described (24,26). The number of cycles used for the different primer sets is described in the legend to Fig. 5.

RESULTS
MEFs lacking pRB are blocked in their ability to undergo adipocyte differentiation in response to treatment with standard adipogenic inducers (19) but undergo efficient adipocyte differentiation when treated with the potent PPAR␥ agonist BRL49653 (24). Recently, it was reported that the level of activated Ras was elevated in RbϪ/Ϫ compared with wild-type cells (20). It is well established that Ras-dependent signaling generally activates the ERK1/2 MAPKs and that activation of Ras in certain cellular settings also leads to activation of the phosphatidylinositol 3-kinase-PKB/Akt pathway (for review, see Ref. 27). Because both the ERK1/2 kinases and the phosphatidylinositol 3-kinase-PKB/Akt pathway have been shown to play pivotal roles in the control of adipogenesis (9,12,13), we decided to determine whether the activities of these kinases were deregulated in RbϪ/Ϫ cells. Initially, we performed Western blot analyses using antibodies recognizing either ERK1/2, irrespective of phosphorylation status, total ERK1/2, or antibodies specifically recognizing the phosphorylated (and active) forms. To allow comparison between the different blots, we applied a day 0 sample from the opposite genotype on each gel. Such analyses revealed that the levels of activated ERK1/2 were considerably higher in RbϪ/Ϫ than in wild-type cells, whereas the levels of total ERK1/2 were comparable (Fig. 1A). In wild-type MEFs, treatment with the standard adipogenic inducers methylisobutylxanthine, dexamethasone, and insulin (MDI treatment) did not significantly change the levels of activated ERK1/2 (Fig. 1A, panel labeled Me 2 SO for Rbϩ/ϩ cells). In contrast, MDI treatment of the pRB-deficient MEFs resulted in a pronounced increase in the levels of activated ERK1/2 up to day 6 ( Fig. 1A, panel labeled Me 2 SO for RbϪ/Ϫ cells). To validate the use of Western blotting using a phosphospecific anti-ERK1/2 antibody, we also performed an immunoprecipitation-coupled kinase assay on whole cell extracts from confluent cells. The immunoprecipitation-coupled kinase assay with Elk-1 as the ERK1/2 substrate confirmed the increased activity of ERK1/2 in RbϪ/Ϫ MEFs (Fig. 1B). This assay revealed an ϳ3-fold higher activity of ERK1/2 in pRB-deficient compared with wild-type MEFs, consistent with the results in Fig. 1A. Therefore, we conclude that Western blotting using the phospho-specific anti-ERK1/2 antibody fairly accurately determines the ERK1/2 activity in the MEFs. BRL49653-induced adipocyte differentiation of pRB-deficient MEFs was accompanied by a marked reduction in the level of activated ERK1/2 after day 2 (Fig. 1A). This is likely to be an indirect effect of the efficient BRL49653-driven differentiation (see below), because BRL49653 had no acute effect on the activities of ERK1/2 in either undifferentiated or differentiated cells (data not shown).
Similarly, significantly more active PKB/Akt was present at day 0 in RbϪ/Ϫ MEFs compared with wild-type MEFs as determined by Western blotting using a phospho-specific anti-PKB/Akt antibody ( Fig. 2A). The difference in PKB/Akt activity between Rbϩ/ϩ and RbϪ/Ϫ MEFs was, however, not as prominent as the difference in ERK1/2 activity described above. The higher abundance of active PKB/Akt in RbϪ/Ϫ compared with Rbϩ/ϩ MEFs was detected only in the earlier stages of the differentiation, i.e. at days 0, 2, and 4. Consistently, an immunoprecipitation-coupled kinase assay with GSK3␣/␤ cross-tide as a substrate demonstrated a 1.5-2-fold higher PKB/Akt activity in day 0 RbϪ/Ϫ compared with wild-type MEFs, thereby validating the use of Western blotting using the phospho-specific anti-PKB/Akt antibody to determine PKB/Akt activity in the MEFs (Fig. 2C). To determine the activity of PKB/Akt in the very early stages of differentiation, we also measured PKB/ Akt activity 1 and 4 h after the addition of adipogenic hormones. Also, at these time points the activity of PKB/Akt was higher in RbϪ/Ϫ cells compared with wild-type cells (Fig. 2B, panel labeled Me 2 SO). The presence of active PKB/Akt appeared to decrease at later time points during differentiation independent of Rb status ( Fig. 2A). Activation of p38 MAPKs has been shown to be required for terminal differentiation of 3T3-L1 cells (10). However, we observed no differences between wild-type and RbϪ/Ϫ MEFs in the activation profiles of p38 MAPKs during differentiation, as determined by Western blotting using antibodies specifically recognizing phosphorylated (active) p38 MAPKs (data not shown).
To investigate whether the higher levels of active ERK1/2 and PKB/Akt in RbϪ/Ϫ MEFs correlated with the presence of more active Ras, we measured the level of activated Ras in confluent MEFs by the ARIA method (25). This assay measures the amount of Ras in cell lysates that is able to bind recombinant glutathione S-transferase-Raf-1. This is an indirect measure of Ras activity, as only GTP-bound (active) Ras binds Raf-1 with high affinity. As shown in Fig. 3, significantly more active Ras was present in day 0 RbϪ/Ϫ relative to Rbϩ/ϩ MEFs. This is consistent with previous results (20) and suggests that the deregulated activities of ERK1/2 and PKB/Akt in pRB-deficient MEFs are due to increased Ras signaling.
So far, we have demonstrated that Ras activity and the activities of two downstream targets of Ras signaling, ERK1/2 and PKB/Akt, were increased in RbϪ/Ϫ MEFs. Next, we were interested in analyzing if these deregulated downstream kinases were responsible for the different adipogenic capacities of wild-type and RbϪ/Ϫ MEFs. We speculated that the deregulated ERK1/2 activity might enforce an inhibitory effect on adipose conversion of the pRB-deficient cells. Therefore, we analyzed the effect of suppressing ERK1/2 activation by the addition of the MEK inhibitor UO126. Initially, we performed a titration with UO126 to determine which concentration would suppress the activities of ERK1/2 in RbϪ/Ϫ MEFs to approximately the levels observed in Rbϩ/ϩ cells. This concentration of UO126 was found to be 10 M (Fig. 1A, compare panel Me 2 SO for Rbϩ/ϩ cells with panel UO126 for RbϪ/Ϫ cells; data not shown). It is important to notice that this concentration of UO126 was sufficient to reduce the levels of activated ERK1/2 to those observed in the wild-type cells without causing complete ablation of ERK1/2 activity. Moreover, UO126 did not suppress the increased activity of PKB/Akt in the pRB-defi-  ERK2 (lower band). B, active ERK1/2 were immunoprecipitated from day 0 lysates prepared from wild-type and RbϪ/Ϫ MEFs, and the ERK1/2 activity in the immunoprecipitates was determined in an in vitro kinase assay with Elk-1 as a substrate. The amounts of lysate input were 100, 50, and 25 g.

FIG. 2. Activation profile of PKB/Akt in differentiating wildtype and Rb؊/؊ MEFs.
A, cells were differentiated according to the standard MDI treatment. Western blot analyses of active and total PKB/Akt were performed as described in the legend to Fig. 1A. P-PKB/ Akt indicates the activated form of PKB/Akt. B, determination of active and total PKB/Akt in wild type and RbϪ/Ϫ MEFs early after the addition of adipogenic inducers as described in the legend to Fig. 1A. C, PKB/Akt was immunoprecipitated from day 0 lysates prepared from Rbϩ/ϩ and RbϪ/Ϫ MEFs, and the activity of PKB/Akt in the immunoprecipitates was analyzed in an in vitro kinase assay using GSK3␣/␤ cross-tide as a substrate. The amounts of lysate input were 200, 100, and 75 g. cient cells in the early stages of differentiation (Fig. 2B, panel  labeled UO126). We assessed adipose conversion by oil red O staining of lipids and by multiplex reverse transcription-PCR to measure the expression of selected adipocyte marker genes. Consistent with previous reports (19,24), treatment with standard adipogenic inducers did not promote significant differentiation of RbϪ/Ϫ cells, whereas significant differentiation of wild-type cells was observed (Fig. 4, panel labeled Me 2 SO, and Fig. 5, compare lanes 1 and 6). Interestingly, the addition of UO126 bypassed the block in differentiation of RbϪ/Ϫ MEFs and induced lipid accumulation in a significant fraction of these cells, as demonstrated by oil red O staining of whole dishes (Fig. 4). In agreement with the accumulation of lipid, expression of PPAR␥, adipocyte lipid-binding protein (ALBP)/aP2 and glycerol-3-phosphate dehydrogenase (GPDH) mRNAs was induced in RbϪ/Ϫ MEFs treated with UO126 (Fig. 5, compare  lanes 6 and 7). UO126 did not significantly modulate the differentiation of wild-type cells (Figs. 4 and 5). Moreover, as predicted from the elevated level of activated PKB/Akt, UO126 promoted differentiation of RbϪ/Ϫ cells in the absence of exogenously added insulin (Figs. 4 and 5, compare lanes 7 and 10). The UO126-induced differentiation of RbϪ/Ϫ MEFs was patchier than that observed upon differentiation of wild-type MEFs treated with standard inducers, but the level of induction of adipocyte marker genes was at least as prominent (Figs. 4 and 5, compare lanes 1 and 7). Under these conditions, as mentioned above, the activities of ERK1/2 were similar in Rbϩ/ϩ and RbϪ/Ϫ MEFs (see Fig. 1A). Therefore, suppression of ERK1/2 activities in RbϪ/Ϫ MEFs by UO126 to the levels observed in wild-type MEFs treated with standard inducers restored the adipogenic potential of the RbϪ/Ϫ MEFs. The reason for the patchy differentiation of the pRB-deficient MEFs treated with UO126 is not known. However, as shown in Figs. 4 and 5, the conversion induced by UO126 is not as efficient as that observed upon treatment with BRL49653. In agreement with results obtained using normal preadipocyte cell lines (8), inhibition of phosphatidylinositol 3-kinase by LY294002 attenuated the MDI-induced differentiation of wild-type MEFs and completely blocked the low level of differentiation of similarly treated RbϪ/Ϫ MEFs (Fig. 4). If the activity of PKB/Akt in RbϪ/Ϫ MEFs contributed to the differentiation induced by the addition of UO126, we would expect that simultaneous treatment with LY294002 would inhibit this differentiation. Accordingly, treatment of RbϪ/Ϫ MEFs with MDI in combination with both UO126 and LY294002 blocked the adipogenic effect of UO126 (Fig. 4), substantiating the importance of phosphatidylinositol 3-kinase activity also in adipocyte differentiation of MEFs.

DISCUSSION
Ras is a multifunctional protein affecting several signaling pathways, and this signaling influences processes such as proliferation, apoptosis, and differentiation (for review, see Ref. 28). Conflicting results have been reported on the function and importance of Ras in adipocyte differentiation (4 -7). However, it has been demonstrated that two downstream targets of Ras signaling, ERK1/2 and PKB/Akt, exert opposing effects on adipogenesis, with ERK1/2 inhibiting and PKB/Akt promoting differentiation (9,(12)(13)(14). In this study we describe a novel mechanism through which pRB promotes adipogenesis, which relates to the ability of pRB to suppress the activities of ERK1/2, possibly by suppressing Ras signaling. This is a surprising finding, as it suggests that the role of pRB in promoting adipocyte differentiation apparently is independent of its binding to E2F and/or adipogenic transcription factors. It is not clear at present how pRB regulates Ras signaling, but several ways in which this deregulated signaling might inhibit adipose conversion are discussed below.
pRB was reported to promote C/EBP␤-mediated transactivation via a direct interaction between the two proteins, with pRB acting as a chaperone-like factor to promote binding of C/EBP␤ to its response elements (19). According to this model, the defective differentiation of RbϪ/Ϫ cells in response to standard adipogenic inducers was suggested to be caused by a reduced C/EBP␤-dependent activation of target genes due to the ab-  sence of the pRB-C/EBP␤ interaction. Therefore, it could be hypothesized that the up-regulated activities of ERK1/2 in RbϪ/Ϫ MEFs also reduced C/EBP␤ activity. However, we find this unlikely because ERK1/2 activity in several other mouse cell lines potentiates C/EBP␤-dependent transactivation (29 -31). Moreover, we have observed little or no effect of coexpression of pRB on several different C/EBP-responsive promoters in pRB-deficient cells, including RbϪ/Ϫ MEFs (Ref. 24 and data not shown).
PPAR␥ is negatively regulated by phosphorylation of a consensus MAPK site (13,14,(32)(33)(34). The effect of this phosphorylation is a reduced ligand-independent and ligand-dependent transactivation, which at least in part relate to a reduced affinity of the phosphorylated receptor for ligands and the steroid receptor coactivator-1 and/or an increased affinity for nuclear receptor corepressors (35,36). Therefore, it is conceivable that part of the defective adipogenesis in RbϪ/Ϫ MEFs relates to a reduced activity of PPAR␥ early in the differentiation program due to ERK1/2-mediated phosphorylation of PPAR␥. However, we have failed to detect the presence of elevated levels of phosphorylated PPAR␥ in RbϪ/Ϫ compared with wild-type MEFs (data not shown). In addition, we have previously reported that coexpression of pRB does not enhance PPAR␥-dependent transactivation of a PPAR-responsive promoter in a pRB-deficient human cancer cell line, and similar results have been obtained in RbϪ/Ϫ MEFs (Ref. 24 and data not shown). At least two additional ways by which increased Ras signaling can inhibit adipocyte differentiation are possible. The first relates to the ability of Ras to stabilize the E2F-1 mRNA, which leads to an increased level of E2F-1 protein (37). E2F-1 is in general an inhibitor of differentiation, and this has been demonstrated also for adipose conversion of 3T3-L1 cells (38). The second possibility relates to the effect of Ras signaling and, in particular, the activation of MAPKs on stabilization of the Myc protein (39,40). Myc is known to be a powerful inhibitor of adipose conversion (41), and it is therefore possible that accumulation of Myc prevents adipocyte differentiation of RbϪ/Ϫ MEFs in response to standard adipogenic inducers. More work is required to pinpoint the mechanism(s) through which deregulated Ras signaling and MAPK activity inhibits adipose conversion.
In summary, we provide evidence that ERK1/2 activity is deregulated in RbϪ/Ϫ MEFs and that suppression of this activity restores the adipogenic potential of RbϪ/Ϫ MEFs. Although the precise target(s) of ERK1/2 in adipocyte differentiation remains to be determined, our findings suggest that an important function of pRB in promoting adipogenesis is to suppress activation of the ERK1/2 MAPKs, presumably by inhibiting Ras signaling. Therefore, our results suggest that pRB promotes adipocyte differentiation not by inhibiting E2Fs and/or binding to adipogenic transcription factors but by attenuating signaling pathways leading to the activation of ERK1/2.