Insulin stimulates mitogen-activated protein kinase by a Ras-independent pathway in 3T3-L1 adipocytes.

To characterize tissue-specific differences in insulin signaling, we compared the mechanisms of mitogen-activated protein (MAP) kinase activation by insulin in the mitogenically active 3T3-L1 fibroblasts with the metabolically active 3T3-L1 adipocytes. In both cell lines, insulin significantly increased p21ras·GTP loading (1.5-2-fold) and MAP kinase activity (5-8-fold). Inhibition of Ras farnesylation with lovastatin blocked activation of p21ras and Raf-1 kinase in both 3T3-L1 fibroblasts and 3T3-L1 adipocytes. In 3T3-L1 fibroblasts, this was accompanied by an inhibition of the stimulatory effect of insulin on MAP kinase. In contrast, in 3T3-L1 adipocytes, despite an inhibition of activation of p21ras and Raf-1 by lovastatin, insulin continued to stimulate MAP kinase activity. Fractionation of the cell lysates on the FPLC Mono-Q column revealed that lovastatin inhibited insulin stimulation of ERK2 (and, to a lesser extent, ERK1) in 3T3-L1 fibroblasts and had no effect on the insulin-stimulated ERK2 in 3T3-L1 adipocytes. These results demonstrate an important distinction between the mechanism of insulin signaling in the metabolically and mitogenically active cells. Insulin activates MAP kinase by the Ras-dependent pathway in the 3T3-L1 fibroblasts and by the Ras-independent pathway in the 3T3-L1 adipocytes.

To characterize tissue-specific differences in insulin signaling, we compared the mechanisms of mitogen-activated protein (MAP) kinase activation by insulin in the mitogenically active 3T3-L1 fibroblasts with the metabolically active 3T3-L1 adipocytes. In both cell lines, insulin significantly increased p21 ras ⅐GTP loading (1.5-2-fold) and MAP kinase activity (5-8-fold). Inhibition of Ras farnesylation with lovastatin blocked activation of p21 ras and Raf-1 kinase in both 3T3-L1 fibroblasts and 3T3-L1 adipocytes. In 3T3-L1 fibroblasts, this was accompanied by an inhibition of the stimulatory effect of insulin on MAP kinase. In contrast, in 3T3-L1 adipocytes, despite an inhibition of activation of p21 ras and Raf-1 by lovastatin, insulin continued to stimulate MAP kinase activity. Fractionation of the cell lysates on the FPLC Mono-Q column revealed that lovastatin inhibited insulin stimulation of ERK2 (and, to a lesser extent, ERK1) in 3T3-L1 fibroblasts and had no effect on the insulin-stimulated ERK2 in 3T3-L1 adipocytes. These results demonstrate an important distinction between the mechanism of insulin signaling in the metabolically and mitogenically active cells. Insulin activates MAP kinase by the Ras-dependent pathway in the 3T3-L1 fibroblasts and by the Ras-independent pathway in the 3T3-L1 adipocytes.
Insulin's interaction with its cell surface receptor triggers both metabolic and mitogenic cellular responses. Insulin binding activates the tyrosine kinase of the ␤-subunit of the insulin receptor, which immediately phosphorylates insulin receptor substrate-1 and Shc. Phosphorylated sites of insulin receptor substrate-1 and Shc bind Src homology-2 domain-containing intermediates such as phosphatidylinositol 3-kinase (PI 3-kinase) 1 and Grb-2, which, in turn, propagate insulin signaling downstream. PI 3-kinase appears to be involved in regulation of glucose transport and protein synthesis, while Grb-2 associ-ation with Sos appears to be an important step in activating the Ras-Raf-MEK-MAP kinase pathway (reviewed in Ref. 1).
Although many intermediates of the insulin signaling have been identified, numerous questions remain unanswered. In particular, most of the signaling molecules activated by insulin are also significantly stimulated by other growth factors, such as platelet-derived growth factor and EGF (2). However, in contrast to these growth factors, only insulin elicits defined metabolic responses, suggesting that additional steps must be involved in the mechanism of insulin action.
Recent studies from our laboratory have identified significant differences between the regulation of certain aspects of insulin signaling in metabolically responsive 3T3-L1 adipocytes and mitogenically responsive 3T3-L1 fibroblasts (3,4). For example, in 3T3-L1 adipocytes, PI 3-kinase and PKC exert a constitutively inhibitory influence on GTPase-activating protein (GAP), allowing insulin signaling to proceed through Sos and p21 ras . Removal of this inhibitory influence results in activation of GAP and inhibition of the p21 ras ⅐GTP loading. The inhibitory influence of PI 3-kinase and PKC on GAP activation was not observed in 3T3-L1 fibroblasts (3,4). Since activation of p21 ras is believed to be required for the activation of MAP kinase by growth factors (reviewed in Refs. 5-8), we compared the mechanism of MAP kinase activation by insulin in 3T3-L1 adipocytes and 3T3-L1 fibroblasts. We found that, while in fibroblasts, insulin activates MAP kinase exclusively by a Rasdependent pathway, in adipocytes, insulin stimulates MAP kinase predominantly via a Ras-independent pathway.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media and supplies were from Life Technologies, Inc. and Gemini Bioproducts (Calabasas, CA); radioisotopes from DuPont NEN. All standard chemicals were from Sigma; insulin was a gift from Eli Lilly and Co. Anti-Ras antibodies were from Transduction Laboratories (Lexington, KY); anti-ERK1/2 antibody and Raf-1 antibody were from Upstate Biotechnology Inc. (Lake Placid, NY); lovastatin and the farnesyl transferase inhibitor, ␣-hydroxyfarnesylphosphonic acid, were from Merck and Co. MEK was kindly provided by Dr. R. Nemenoff (University of Colorado Health Sciences Center, Denver, CO), the MEK inhibitor (PD98059) was a gift from Dr. Alan Saltiel (Parke-Davis, Ann Arbor, MI), and Rat-1 fibroblasts transfected with a dominant negative mutant of p21 ras (N17) were courtesy of Dr. Jerrold Olefsky (University of California, San Diego).
Cell Culture-3T3-L1 fibroblasts were grown to confluence in fibroblast growth media (Dulbecco's modified Eagle's medium (DMEM) containing 5.5 mM glucose with 10% fetal calf serum (FCS), 50 g/ml gentamicin, 1 mM L-glutamine). Differentiation of adipocytes was induced via the following protocol. Cells were refed with fibroblast growth media when 80% confluent. Two days later, cells were fed differentiation media (DMEM containing 25 mM glucose, 10% FCS, 50 g/ml gentamicin, 1 mM L-glutamine, 2.5 ml of 10 ϫ PBS, 0.055 g of isobutylmethylxanthene, 20 ml of deionized water, 250 l of 49 mM dexamethasone, and 2.5 mg of insulin). On day 4, cells were fed adipocyte growth media (DMEM containing 25 mM glucose with 10% FCS, 50 g/ml gentamicin, 1 mM L-glutamine, and 1 g/ml insulin). Cells were refed * This work was supported by the Veterans Affairs Research Service, The Health One Foundation: Diabetes Research and Education Fund, and The Foundation for Biomedical Education and Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Insulin-mediated p21 ras ⅐GTP Formation-Confluent 3T3-L1 fibroblasts and adipocytes were serum-and phosphate-starved for 24 h and labeled with [ 32 P]orthophosphate (0.25 mCi/ml) overnight. Cells were then incubated with insulin (100 nM) for 10 min. In some experiments, cells were preincubated with lovastatin (2 g/ml) for 24 h prior to insulin. The increment in percent p21 ras ⅐GTP was determined as described previously (10). Briefly, precleared lysates were immunoprecipitated with anti-Ras antibody (Y13-259), and the nucleotides were separated by thin layer chromatography. GTP and GDP were visualized by autoradiography and using acid molybate reagent, cut, and quantified by liquid scintillation counting.
Statistics-All statistics were performed by Student's t test, with p value Ͻ 0.05 considered significant. Results are expressed as mean Ϯ S.E. in comparison to control groups.

RESULTS
To investigate the role of p21 ras in mediating insulin activation of MAP kinase, we initially employed the hydroxymethylglutaryl-CoA-reductase inhibitor lovastatin to block the activation of p21 ras by insulin in 3T3-L1 fibroblasts and 3T3-L1 adipocytes. Lovastatin blocks farnesylation of p21 ras , thus decreasing the pool of the intracellular Ras available for subsequent activation by growth factors, including insulin (9). Lovastatin (2 g/ml for 18 h) significantly inhibited the ability of insulin to stimulate p21 ras ⅐GTP loading in both 3T3-L1 fibroblasts and 3T3-L1 adipocytes (Fig. 1). However, insulin-stimulated MAP kinase activity was inhibited by lovastatin in the 3T3-L1 fibroblasts (by 85%), but not in 3T3-L1 adipocytes (Fig.  2). Despite complete inhibition of Ras stimulation, insulin continued to activate MAP kinase (5-8-fold) in 3T3-L1 adipocytes.
In the next series of experiments, activation of p21 ras was inhibited by 1 M ␣-hydroxyfarnesylphosphonic acid, a specific inhibitor of farnesyltransferase. This enzyme catalyzes farnesylation of p21 ras , which is necessary for its subsequent activation. The presence of the farnesyltransferase inhibitor completely blocked the ability of insulin to stimulate MAP kinase in 3T3-L1 fibroblasts, but had no effect in 3T3-L1 adipocytes (Fig. 3).
We have recently observed that, in 3T3-L1 adipocytes, deple-tion of PKC with prolonged exposure to the phorbol ester, TPA, results in activation of GAP and inhibition of the insulininduced p21 ras ⅐GTP loading (4). Therefore, we used this paradigm to further examine the role of p21 ras ⅐GTP in mediating insulin's effect on MAP kinase in 3T3-L1 adipocytes. Cells treated with TPA (100 nM for 18 h) were unable to increase p21 ras ⅐GTP formation in response to insulin (4), but responded fully to activation of MAP kinase by insulin (Fig. 4), suggesting that the latter is activated by a Ras-independent pathway. Assuming a linear progression of the insulin signal from Ras to Raf to MEK to ERK1/ERK2, we compared the involvement of Raf and MEK kinases in the activation of MAP kinase in 3T3-L1 fibroblasts and adipocytes. Activation of Raf-1 kinase activity by insulin was clearly Ras-dependent in both cell lines, as lovastatin inhibited insulin-stimulated Raf-1 activity in fibroblasts and adipocytes (Fig. 5). Similarly in both cell lines, the MEK inhibitor, PD98059, was equipotent in inhibiting the activation of MAP kinase by insulin in both cell lines (Fig. 6), implying the importance of MEK as the upstream activator of MAP kinase in both fibroblasts and adipocytes.
In order to confirm that in fibroblasts insulin stimulates MAP kinase activity via the Ras-dependent pathway, we examined the effect of insulin on MAP kinase in the Rat-1 fibroblasts transfected with the dominant negative mutant (N17) of Ras. As shown in Fig. 7, insulin failed to stimulate MAP kinase activity in these cells.
In the experiments described above, MAP kinase activity reflected the total cellular kinase activity assayed using the EGF receptor peptide as a substrate. Conceivably, differential expression of various members of the MAP kinase family (11)(12)(13)(14) may explain the differences in response to insulin between the adipocytes and fibroblasts. To examine this possibility, we fractionated cell lysates from 3T3-L1 fibroblasts and 3T3-L1 adipocytes on FPLC Mono-Q column and measured MAP kinase activity in each fraction. Two major peaks of MAP kinase activity were detected in both cell types. The first peak corresponded to ERK2 and the second smaller peak to ERK1. Insu- lin activated ERK2 and ERK1 in both cell types. Inhibition of p21 ras activation with lovastatin resulted in a decreased ability of insulin to stimulate ERK2 in fibroblasts (Fig. 8A), but had no effect on the activation of ERK2 by insulin in adipocytes (Fig.  8B).
Finally, to verify the effect of insulin on ERK1/2, we determined MAP kinase activity in the ERK1/2 immunoprecipitates of control and lovastatin-treated cells. Insulin's effect on MAP kinase was completely blocked in 3T3-L1 fibroblasts, but remained unaffected in the adipocytes (Fig. 9).

DISCUSSION
It is well known that insulin stimulates both p21 ras ⅐GTP loading and MAP kinase in most cells expressing insulin receptors (15,16). It has been assumed that these two events are obligatorily coupled in these cells. The main point of the present study is that, at least in 3T3-L1 adipocytes, these events are parallel but independent of each other.
Insulin stimulates both p21 ras and MAP kinase in 3T3-L1 adipocytes, but inhibition of p21 ras by lovastatin, farnesyltransferase inhibitor, or PKC depletion does not affect insulin's ability to stimulate MAP kinase in these cells (Figs. 2-4). This is in direct contrast to 3T3-L1 fibroblasts, where activation of p21 ras is an obligatory step for MAP kinase activation (Figs. 2-4). Despite differential regulation, the Mono-Q pattern of the MAP kinases is similar in both cell lines (Fig. 8), indicating that the same MAP kinase family members are differentially regulated in 3T3-L1 fibroblasts and 3T3-L1 adipocytes: one via a Ras-dependent and the other via a Ras-independent pathway.
The signaling molecules within the Ras-independent pathway are unknown. ERK1/2 is believed to be activated by its own kinase, MEK1/2 (5-8), and MEK1/2 activation appears to be essential for insulin signaling in both adipocytes and fibroblasts, since the MEK inhibitor blocked insulin-stimulated MAP kinase activation in both cell lines (Fig. 6). MEK itself is a dual specificity kinase, which is activated by the upstream serine/threonine kinases Raf and MEKK. The present data suggest that activation of MAP kinase by MEK in 3T3-L1 adipocytes may occur in the absence of Ras and Raf activation, suggesting that MEK is activated by another MEK kinase. Haystead et al. (29) have recently identified a novel insulinresponsive MEK kinase (I-MEKK) in Wistar rat adipocytes. This kinase was distinct from Raf and showed rapid phasic kinetics in response to insulin but not phorbol ester. They have proposed that in adipocytes, I-MEKK may represent a divergence point between the insulin-and PKC-mediated signal transduction pathways. It remains unknown whether or not this new insulin-responsive MEK kinase is Ras-dependent.
Although 3T3-L1 adipocytes are derived from 3T3-L1 fibroblasts as a result of differentiation, there are numerous distinctions between the two lines. The major difference, however, is that 3T3-L1 adipocytes are terminally differentiated, metabolically active, and non-dividing cells, whereas the 3T3-L1 fibroblasts are mitogenically active, growing, and propagating cells. Conceivably, with differentiation, the influence of insulin shifts to a different signaling cascade. It would be of great importance to investigate whether or not insulin activates MAP kinase via a Ras-independent pathway in other fully differentiated and metabolically active cells. Until this question is answered, one should exercise a great deal of caution in extrapolating results obtained in non-metabolically responsive cells (and/or non fully-differentiated cells) to metabolically active insulin target cells. Cell lysates were fractionated as described in the text. Each fraction was assayed for phosphotransferase activity using MBP as the substrate. A representative experiment is demonstrated. The first peak was identified as ERK2 and the second one as ERK1 by immunoblotting (not shown). Lovastatin significantly inhibited ERK1/2 (predominantly ERK2) activity in fibroblasts but not in adipocytes.