Insulin-stimulated tyrosine phosphorylation of caveolin is specific for the differentiated adipocyte phenotype in 3T3-L1 cells.

Previous work from this laboratory has shown that insulin stimulates the tyrosine phosphorylation of caveolin in 3T3-L1 adipocytes (Mastick, C. C., Brady, M. J., and Saltiel, A. R. (1995) J. Cell Biol. 129, 1523-1531). This phosphorylation is specific for insulin and involves the activation of a tyrosine kinase downstream of the insulin receptor. We report here that tyrosine phosphorylation of caveolin is detected only in fully differentiated adipocytes, not in fibroblasts (preadipocytes), despite the fact that both cell types express caveolin-1 and active insulin receptor. Caveolin copurifies with caveolin tyrosine kinase activity in both preadipocytes and adipocytes. Accumulating evidence indicates that this kinase is the Src family kinase Fyn. Fyn is expressed in the preadipocytes and the adipocytes and is colocalized with caveolin in low density Triton-insoluble complexes in both cell types. In adipocytes, overexpression of wild type Fyn leads to increased basal phosphorylation of caveolin and hyperphosphorylation of caveolin in response to insulin. In vitro kinase assays suggest that Fyn may be activated in response to insulin through the binding of a tyrosine-phosphorylated insulin receptor substrate protein. Previous work suggested that this protein may be c-Cbl (Ribon, V., and Saltiel, A. R. (1997) Biochem. J. 324, 839-846). In 3T3-L1 adipocytes, Cbl binds to Fyn in an insulin-dependent manner, and Cbl phosphorylation is adipocyte-specific. Here we show that phosphorylated Cbl is translocated into caveolin-enriched Triton-insoluble complexes after insulin stimulation. This may lead to the cell type-specific, compartmentalized activation of Fyn and the specific phosphorylation of proteins in the caveolae in response to insulin in adipocytes.

Like receptors for many growth factors, the insulin receptor is a tyrosine kinase that undergoes activation upon insulin binding, leading to the tyrosine phosphorylation of a specific set of substrate proteins (1,2). Like other growth factors, insulin acts as a mitogen in many cell types and therefore shares a number of signaling pathways with these growth factors (3). However, in adipocytes insulin increases glucose uptake and metabolism 10 to hundreds of fold, whereas growth factors such as PDGF 1 and EGF are without effect on these processes (4 -7). This indicates that key regulators of metabolic pathways exist in adipocytes that respond uniquely to insulin signaling. Despite intensive investigation, the basis for this specificity remains largely unknown.
One proposed mechanism for specificity in signal transduction involves the spatial compartmentalization of signaling proteins (8). The specialized regions of the plasma membrane termed caveolae have been implicated in the segregation of signal initiation events (9,10). Caveolae are small invaginations of the plasma membrane with unique protein and lipid compositions (9 -12). These structures have a characteristic striated coat, which is made up largely of the caveolins (10,13). Caveolin was originally identified as a major phosphoprotein in v-Src-transformed cells (14 -16), and caveolin expression is down-regulated in cell lines transformed by a number of different oncogenes (17). In many different cell types, caveolin copurifies with or binds to signaling molecules, including glycosylphosphatidylinositol-anchored proteins, receptors, and effectors (18 -26). Caveolins have also been reported to have intrinsic GDP dissociation inhibitor (GDI) and GTPase-activating protein (GAP) activities for G-protein ␣-subunits (27,28). Despite these reports, the exact role and composition of these caveolin-containing structures remain controversial (29 -31; reviewed in Ref. 8).
Caveolae are abundant in adipocytes, covering a significant fraction of the inner surface of the plasma membrane (32). Caveolins-1 and -2 are highly expressed in adipocytes (33)(34)(35), and caveolin-1 expression increases upon adipocyte differentiation (34). Recently, we reported that caveolin is tyrosinephosphorylated in response to insulin in 3T3-L1 adipocytes (36). Consistent with an important role for caveolae in insulin action, phosphorylation of caveolin is specific for insulin and does not occur in response to other growth factors such as PDGF or EGF, suggesting that caveolae may be involved in an insulin-specific signaling pathway.
Here we show that the tyrosine phosphorylation of caveolin in response to insulin is specific for the differentiated adipocyte phenotype. Although preadipocytes express active insulin receptors, caveolin-1, and the caveolin-kinase Fyn, caveolin is not tyrosine-phosphorylated in response to insulin in these cells.
The data indicate that a key molecule(s) in the signaling pathway leading from the activated insulin receptor to caveolin phosphorylation is regulated by adipocyte differentiation.

EXPERIMENTAL PROCEDURES
Materials-All tissue culture reagents were purchased from Life Technologies, Inc. Insulin (porcine) was purchased from Sigma; IGF-I (human recombinant, receptor grade) was purchased from Mallinckrodt (Chesterfield, MO). Monoclonal and polyclonal anti-caveolin antibodies were purchased from Transduction Laboratories (Lexington, KY). * 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.
Monoclonal anti-phosphotyrosine antibodies were purchased from Upstate Biotechnology Inc. (Lake Placid, NY) and Transduction Laboratories. Monoclonal anti-insulin receptor antiserum was purchased from Oncogene Sciences (Cambridge, MA), and polyclonal anti-IRS-1 antisera was prepared as described (37). Polyclonal anti-Fyn antiserum (␣-Fyn-268 -389) was prepared as described (36) or was purchased from Santa Cruz Biotechnology (Santa Cruz, CA) (␣-Fyn-29 -48); monoclonal anti-Fyn antiserum was purchased from Transduction Laboratories. Polyclonal anti-Src antiserum was purchased from Santa Cruz; this antibody cross-reacts with Fyn and Yes. Polyclonal anti-c-Cbl antiserum was purchased from Santa Cruz. Prestained molecular weight markers (high range), protein A-agarose, and horseradish peroxidaseconjugated goat anti-rabbit and anti-mouse IgG antibodies were purchased from Life Technologies, Inc. Octylthioglucoside and Triton X-100 were purchased from Boehringer Mannheim. All other reagents were purchased from Sigma. Protein concentrations were determined using the BCA protein determination kit from Pierce.
Cell Culture-3T3-L1 fibroblast cells (CCL 92.1; American Type Culture Collection, Rockville, MD) were grown in DME (high glucose) supplemented with 10% calf serum, 2 mM L-glutamine, 50 units/ml penicillin, and 50 mg/ml streptomycin. They were differentiated into adipocytes on 150-mm culture plates as described (38). Cells were used for experiments 7-10 days after initiation of differentiation. At this point in the differentiation protocol, most of the cells (Ͼ90%) were filled with fat droplets. These cells showed large increases in glucose uptake and metabolism in response to nanomolar concentrations of insulin when serum-deprived as described below. Preadipocytes were grown to confluence and maintained in DME supplemented with calf serum in parallel with the differentiated cells.
Generation of Fyn Expressing Cell Lines-Fyn kinase constructs (a kind gift from Dr. S. A. Courtneidge, EMBL, Heidelberg, Germany) (39) were subcloned into the BamHI site of pUHD 10.3 (a kind gift from H. Bujard, ZMDH, Heidelberg, Germany) (40). 3T3-L1 cells were co-transfected using LipofectAMINE (Life Technologies, Inc.) with Fyn constructs (linearized with Pvu I) and an empty pOPI3CAT vector containing neomycin resistance (Stratagene Cloning Systems, La Jolla, CA; linearized with NotI and BstXI) at a 1:10 ratio. After 2 days, resistant cells were selected in DME supplemented with 750 g/ml G418 (Life Technologies, Inc.). Cell lines were isolated from single cells by limiting dilution and were screened for Fyn expression by Western blotting.
Preparation of Triton-insoluble Complexes-Caveolin-enriched Triton-insoluble complexes were prepared essentially as described (36). Adipocytes were incubated overnight in DME supplemented with 0.5% calf serum. Insulin or IGF-1 (100 nM) were added, and incubation was continued at 37°C for the indicated times. Cells for basal samples were treated identically, without addition of growth factors. Cells (1 or 2 150-mm plates per condition) were washed 3 times in ice-cold phosphate-buffered saline, harvested by scraping into 25 mM MES, pH 6.0, 150 mM NaCl, 1% Triton X-100, 1 mM NaVO 4 , 10 mg/ml aprotinin, 1 mM benzamidine, and 0.1 mM PMSF (MES/NaCl/Triton X-100), and homogenized (10 strokes in glass homogenizer with a Teflon pestle). To prepare low density, Triton-insoluble complexes, 2 ml of cell lysate were diluted 1:1 with 80% sucrose (40% sucrose final concentration), 25 mM MES, pH 6.0, 150 mM NaCl. A gradient was formed by overlaying the extract with 2 ml each of 35,25,15, and 5% sucrose in 25 mM MES, pH 6.0, 150 mM NaCl. The gradients were centrifuged for 20 h at 39,000 rpm in an SW40Ti rotor (Beckman Instruments, Palo Alto, CA). Tritoninsoluble complexes enriched in caveolin were collected as a flocculent band of material just below the 15-25% interface using a 19-gauge needle and a syringe (approximately 1-2 ml). This fraction was diluted 2-to 3-fold with MES/NaCl/Triton X-100, and the complexes were collected by centrifugation at 13,500 rpm in a microcentrifuge for 20 min at 4°C. Alternatively, 1-ml fractions were collected from the top of the gradients using a 19-gauge needle and syringe. Fractions were assayed directly for caveolin and Fyn and then low density, Tritoninsoluble complexes prepared from pooled fractions.
To prepare caveolin-enriched Triton-insoluble pellets, cells were lysed as described above and then centrifuged for 5 min at 1500 rpm in a microcentrifuge (to remove nuclei). The supernatant was then centrifuged at 13,500 rpm at 4°C to prepare Triton-insoluble pellets (pellet) and post-nuclear supernatants (supernatant). Triton-insoluble pellets were washed 1 ϫ with 1 ml of MES/NaCl/Triton X-100. To detect IRS-1 phosphorylation, post-nuclear supernatants were prepared from cells solubilized with MES/NaCl/Triton X-100 supplemented with 1 mM EGTA, 10 mM Na 4 P 2 0 7 , and 100 mM NaF.
In Vitro Kinase Assays-Assays were performed essentially as described (36). Briefly, Triton-insoluble pellets or low density Tritoninsoluble complexes were resuspended in 5 l of MES/NaCl/Triton X-100 and 20 l of kinase assay buffer (22.5 mM Hepes, pH 7.5, 12.5 mM MgCl 2 , 1.25 mM MnCl 2 , 1.25 mM EGTA). The reaction was initiated with the addition of 1 mM ATP (or buffer); the samples were incubated for 10 min at room temperature, and the reaction was terminated with 4 ϫ concentrated Laemmli sample buffer. Protein tyrosine phosphorylation was determined by Western blotting. In some experiments, reactions were terminated with 0.5 ml of solubilization buffer supplemented with 5 mM EDTA, and caveolin was immunoprecipitated prior to analysis.
To determine the effects of tyrosine-phosphorylated peptides on caveolin-kinase activity, the complexes were preincubated for 1 h at 4°C with 100 M peptide or Me 2 SO (2%) prior to assay. The peptides used were pY-MT (EPQ-pY-EEIPIYL) and MT (EPQYEEIPIYL) derived from the middle T antigen, and pY-VPM derived from tyrosine 751 of the human PDGF␤ receptor. The peptides were synthesized, purified, and characterized as described previously (41). In vitro competition binding assays using glutathione S-transferase fusion proteins of SH2 domains indicate that the pY-MT peptide binds with high affinity to the SH2 domains of Fyn and Src, but not p85, whereas the pY-VPM peptide binds with high affinity to both SH2 domains of PI 3Ј-kinase but not to the SH2 domains of Fyn or Src (data not shown).
Image Processing-Autoradiographs were quantified by computerassisted video densitometry using the BioImage system (Imaging Systems, Millipore Corp., Ann Arbor, MI). Figures of autoradiographs were constructed using the images collected above and Adobe Photoshop software (Adobe Systems Incorporated, Mountain View, CA).

Tyrosine Phosphorylation in Preadipocytes and Adipocytes-
Differentiation of 3T3-L1 fibroblasts into adipocytes is accompanied by a significant increase in the responsiveness of glucose uptake and metabolism to insulin. 2 To further explore the role of caveolae in insulin action, the tyrosine phosphorylations of caveolin and the caveolin-associated proteins were compared in preadipocytes and adipocytes (Fig. 1). Cells were grown to confluency and either maintained in calf serum (pread.) or induced to differentiate (adip.) as described (see "Experimental Procedures"). 10 days after the initiation of differentiation, cells were serum-starved over night and then incubated with insulin (100 nM, 5 min). The cells were lysed in MES/NaCl/ Triton X-100; caveolin was immunoprecipitated, and tyrosine phosphorylation was determined by anti-phosphotyrosine Western blotting. As reported previously (36), in response to insulin there was a significant increase in the tyrosine phosphorylation of two forms of caveolin (22 and 24 kDa) and a 29-kDa caveolin-associated protein in adipocytes (Fig. 1A). This was not observed in the preadipocytes, although basal tyrosine phosphorylation of both proteins was detected in these cells. In this experiment, the amount of cell lysate (mg of protein) used for the preadipocyte and adipocyte samples was adjusted such that comparable amounts of caveolin were immunoprecipitated from both (Fig. 1B). However, while the adipocytes express both the 22-(␣) and 24-kDa (␤) forms of caveolin-1, the preadipocytes express only the 24-kDa form (see below). The preadipocytes and adipocytes express comparable numbers of active insulin receptors (Fig. 1C, approximately 3-fold increase upon differentiation). These receptors appear to be fully functional, as IRS-1 was highly phosphorylated in response to insulin in both cell types (Fig. 1D). The identity of the 185-kDa phosphoprotein as IRS-1 was confirmed by immunoprecipitation (data not shown).
Many signaling pathways stimulated by insulin are shared by IGF-I. Although adipocytes express predominantly insulin receptors (43), preadipocytes express similar levels of both IGF-1 and insulin receptors (Fig. 2). To maximally stimulate the insulin/IGF-I pathway in the preadipocytes, we used high concentrations (100 nM) of IGF-1 to stimulate both the IGF-1 receptors and insulin receptors in these cells. The major tyrosine-phosphorylated proteins detected in the post-nuclear supernatants after insulin stimulation were the ␤-subunit of the insulin receptor (95 kDa) and the ␤-subunit of the IGF-I receptor (105 kDa) ( Fig. 2A). The identity of the 185-kDa tyrosine-phosphorylated protein as IRS-1 was verified by immunoprecipitation (Fig. 2B). Although there was a rapid phosphorylation of IRS-1 in these cells, there was no tyrosine phosphorylation of caveolin in response to IGF-I (Fig. 2C), despite immunoprecipitation of a significant amount of caveolin from these cells (Fig. 2D). Together with Fig. 1, these data indicate that the block in caveolin phosphorylation in the preadipocytes occurs downstream of the insulin receptor itself.
Colocalization of Caveolin and Fyn in Low Density Tritoninsoluble Complexes-It has previously been reported that caveolin protein expression increases 20-fold upon differentiation of 3T3-L1 fibroblasts into adipocytes (34). We found that while there is an increase (approximately 3-fold) in total caveolin expression upon differentiation, there was a significant amount of the 24-kDa or ␣-isoform of caveolin-1 expressed in the fibroblasts (Fig. 3A, lower panel). Although there was little increase in the expression of caveolin-1␣ (less than 2-fold), a large increase was detected in the expression of the 22-kDa or ␤-isoform of caveolin-1 (approximately 11-fold) upon differentiation (the caveolin migrated as a doublet after differentiation). The differences between our results and those of Scherer et al. (36) may be due to differences in antibodies or variations in the cell lines. We have previously reported that different caveolin-1 antibody preparations recognize either one or both of the caveolin-1 isoforms. Scherer et al. (36) detected only a single band of caveolin, with an apparent molecular mass of approximately 22 kDa, the molecular mass of caveolin-1␤, which is consistent with our results. Both isoforms of caveolin-1 are phosphorylated in the adipocytes in response to insulin treatment ( Fig. 1 and Ref. 36). In addition, it has been shown that caveolin-1␣ is a substrate for v-Src in vitro and in vivo (44). Therefore, the caveolin expressed in the preadipocytes should serve as a substrate for tyrosine phosphorylation, if the caveolin and the caveolin-kinase are colocalized in caveolae, and the kinase is activated in response to insulin.
The caveolin in both the preadipocytes and adipocytes was enriched in low density Triton-insoluble complexes (Fig. 3, C  and D). However, the efficiency with which the caveolin was recovered in these fractions increased upon differentiation. There was a significant increase in the total amount of protein isolated in the caveolar fractions after differentiation, although it remained only a small fraction of the total protein in the cell lysates (data not shown). Although the mechanism is still unclear, the increase in the recovery of caveolin in the low density complexes in differentiated cells may result from the increased expression of caveolin-1␤ (perhaps decreasing the Triton solubility of the complexes) or from an increase in the abundance of a limiting component that confers low density or Triton insolubility on the caveolin-containing complexes (i.e. cholesterol or sphingolipids). The failure to detect caveolin phosphorylation in the preadipocytes was not simply due to the reduced efficiency of caveolin localization in caveolae. There was no increase in the tyrosine phosphorylation of caveolin found in the low density Triton-insoluble complexes in the preadipocytes in response to insulin (Fig. 3B), although basal phosphorylation of caveolin and the 29-kDa caveolin-associated protein could be detected in these fractions with long exposures (data not shown). In contrast, there was a clear increase in the phosphorylation of the caveolins and the 29-kDa caveolin-associated protein in response to insulin in the same fractions from the adipocytes.
Our previous work (36) indicated that phosphorylation of caveolin occurs through the activation of a tyrosine kinase downstream of the insulin receptor, which is a resident of the caveolae. Accumulating evidence indicates that this kinase is the Src family kinase Fyn (36; see below). The expression and subcellular distribution of Fyn were compared in preadipocytes and adipocytes (Fig. 3). Fyn is easily detected in the preadipocytes by Western blotting (Fig. 3A, upper panel). The levels of Fyn kinase per mg of protein (but not per cell) decreased upon differentiation. Fyn was highly enriched in low density caveolin-enriched Triton-insoluble complexes from both preadipocytes and adipocytes (Fig. 3, C and D, fractions 4 and 5).
Copurification of Caveolin and Caveolin-Tyrosine Kinase Activity-We have previously observed in in vitro kinase assays that caveolin copurifies with a caveolin-tyrosine kinase in Triton-insoluble complexes (36). Isolation of these complexes results in a significant activation of the caveolin kinase activity in these fractions, as incubation of complexes from either basal or insulin-stimulated cells with ATP resulted in a significant and comparable increase in the level of phosphorylation of caveolin and associated proteins. The small apparent effect of insulin on the activity of the caveolin kinase in vitro is due to the elevated (near maximal) activity of the kinase in the complexes prepared from basal cells. No detectable insulin receptor kinase activity was detected in these fractions (36). We took advantage of this activation of the caveolin kinase activity to determine if caveolin was colocalized with functional caveolin tyrosine kinase in both preadipocytes and adipocytes. In vitro kinase assays were performed on either Triton-insoluble pellets (TIP) or highly purified low density Triton-insoluble complexes from both cell types (Fig. 4). Intact complexes were incubated with or without 1 mM ATP in kinase assay buffer for 10 min, and kinase activity was measured as an increase in tyrosine phosphorylation by Western blotting. Prior to analysis, the Triton-insoluble pellet samples were solubilized with octylthioglucoside, and caveolin-associated proteins were isolated by immunoprecipitation. The entire low density, Tritoninsoluble complex samples were analyzed. Incubation of either of the complexes with ATP resulted in a significant increase in the tyrosine phosphorylation of caveolin and the 29-kDa caveolin-associated protein in samples from both preadipocytes and adipocytes. As previously observed (36), an additional 27-kDa caveolin-associated protein was also tyrosine-phosphorylated in vitro, although this protein was not phosphorylated in response to insulin in the intact cells. These data indicate that the caveolin kinase is expressed and properly localized in complexes with caveolin in the preadipocytes; however, it is not activated in response to insulin in the intact cells.
Tyrosine Phosphorylation in Cell Lines Overexpressing Fyn-To explore in more detail whether Fyn is the insulinregulatable caveolin-tyrosine kinase, 3T3-L1 cells were stably transfected with plasmids encoding either kinase-inactive Fyn or wild type Fyn kinase. We were unable to differentiate any cell lines that expressed the kinase negative Fyn construct, so the effect of this construct on caveolin phosphorylation could not be determined. However, a cell line expressing high levels of the wild type Fyn kinase was isolated and analyzed (Fig. 5). This cell line differentiated well, as determined by accumulation of fat droplets, with kinetics that were faster than the parental cells (data not shown). Unexpectedly, the Fyn-overexpressing cell line grew significantly more slowly than the parental cells. Fyn levels were elevated in both the post-nuclear supernatants and the caveolin-enriched Triton-insoluble pellets in the overexpressing cell line (Fig. 5, A and B). In cells overexpressing Fyn, there was a significant increase in the basal level of phosphorylation in the caveolin-enriched, low

FIG. 4. Caveolin tyrosine kinase activity in caveolin-enriched Triton-insoluble complexes from preadipocytes and adipocytes.
Triton-insoluble pellets (TIP) or low density Triton-insoluble complexes prepared from preadipocytes (pread.) or adipocytes (adip.) were incubated in the absence (Ϫ) or presence (ϩ) of 1 mM ATP in kinase buffer, and kinase activity was measured by anti-phosphotyrosine (␣-ptyr) Western blotting. Left, caveolin and caveolin-associated proteins were immunoprecipitated from the less highly purified Triton-insoluble pellets after kinase reaction. Right, low density Triton-insoluble complexes after kinase reaction. density, Triton-insoluble complexes, suggesting that the high level of expression of Fyn was sufficient to cause activation of the kinase in this fraction (Fig. 5C). In contrast, there was very little increase in basal phosphorylation seen in the Tritoninsoluble pellet and post-nuclear supernatant fractions from these cells (Fig. 6B).
Overexpression of Fyn led to increased basal tyrosine phosphorylation of four proteins in the low density, Triton-insoluble complexes with apparent molecular masses of 22, 24, 27, and 29 kDa. Hyperphosphorylation of these proteins was observed in response to insulin (Fig. 5C). These four proteins were the only proteins in the low density, Triton-insoluble fractions to show an increase in phosphorylation after insulin stimulation. Similar levels of caveolin were isolated in the fractions from the two cells types (Fig. 5D).
The identities of the Fyn substrate proteins in the Tritoninsoluble fractions as the caveolins and the caveolin-associated proteins were verified by immunoprecipitation (Fig. 6). Tritoninsoluble pellets from basal or insulin-stimulated cells were solubilized with octylthioglucoside, and caveolin-associated proteins were isolated by immunoprecipitation. There was an increase in the basal phosphorylation of the caveolins (22 and 24 kDa) and the 29-kDa caveolin-associated protein and hyperphosphorylation of these three proteins after insulin stimulation (Fig. 6A, upper panel). Interestingly, overexpression of Fyn led to the constitutive phosphorylation of the 27-kDa caveolinassociated protein. The phosphorylation of this protein had previously only been detected in vitro. Surprisingly, there was no change in the phosphorylation of caveolin, or the caveolinassociated proteins in the preadipocytes (Fig. 6C, upper panel), despite a significant increase in the amount of Fyn localized to the Triton-insoluble pellets in these cells (Fig. 6D), indicating that overexpression of Fyn is not sufficient to activate this kinase in the preadipocytes. Similar amounts of caveolin were immunoprecipitated from both preadipocytes and adipocytes from both cell lines (Fig. 6, A and C, lower panel). Taken together, these data strongly suggest that the block in caveolin phosphorylation in the preadipocytes is in the signaling pathway leading to the activation of the caveolin kinase Fyn, rather than in the localization or expression of the kinase itself.
Potential Mechanism for Fyn Activation in Adipocytes-How is the Fyn that is resident in caveolae activated in response to insulin? Src family kinases share two tyrosines that can be phosphorylated in vivo (45). The Src family kinases are normally phosphorylated at the C-terminal regulatory site (tyrosine 527 in Src) by Csk. This phosphorylation inhibits the activity of the kinases, by inducing the binding of this residue to the SH2 domain in the N termini of these kinases in an intramolecular interaction. This allows for an intramolecular SH3 domain interaction within the kinases, folding the kinases into inactive conformations (46,47). There is an additional autophosphorylation site that is required for full activation of these kinases. The Src family kinases can be activated either through the dephosphorylation of the tyrosine 527 residue (48) or through the displacement of the phosphorylated C-terminal tail by another tyrosine-phosphorylated protein binding to the SH2 and SH3 domains (49,50).
Fyn was constitutively associated with the caveolinenriched, low density, Triton-insoluble complexes in 3T3-L1 adipocytes (Fig. 7A, ␣-Fyn-(29 -48) and ␣-Fyn- (268 -389)). The Fyn in these complexes exactly comigrated with the 60-kDa phosphoprotein found in these complexes (Fig. 7A, ␣-ptyr), and immunoprecipitation with an anti-Fyn antibody completely immunodepleted the Fyn from the supernatants of octylthioglucoside-solubilized complexes, indicating that Fyn is the predominant Src family kinase in these fractions (36). There is no detectable change in the phosphorylation of Fyn in the low density, Triton-insoluble complexes in response to insulin (Fig.  7B). Consistent with this, it has been shown that CD45 is largely excluded from the Triton-insoluble fractions in lymphocytes (51). Fyn colocalizes with caveolin in Triton-insoluble complexes from both preadipocytes and adipocytes (Fig. 7C, ␣-Fyn and ␣-Src), and insulin has no effect on the localization or phosphorylation of Fyn in either cell type. Interestingly, however, differentiation leads to a significant increase in the constitutive tyrosine phosphorylation of Fyn in the Tritoninsoluble pellets (Fig. 7C, ␣-ptyr).
It has been shown that ligands for the Src SH2 domain, such as tyrosine-phosphorylated peptides derived from the PDGF receptor or middle T antigen, can activate Src enzymatic activity (49,52). Therefore, we investigated whether the caveolin kinase in the Triton-insoluble complexes could be stimulated with tyrosine-phosphorylated peptides that bind to the SH2 domains of Src family kinases. Intact Triton-insoluble pellets (Fig. 7D) or low density Triton-insoluble complexes (Fig. 7E) were preincubated with peptides and further incubated with or without 1 mM ATP. Kinase activity was measured as an increase in tyrosine phosphorylation by anti-phosphotyrosine Western blotting. Incubation with a tyrosine-phosphorylated peptide derived from middle T antigen (pY-MT) increased the autophosphorylation of Fyn (60 kDa) 2-3-fold. The phosphorylations of a number of other proteins were significantly increased as well (Fig. 7, D and E, filled arrowheads), including caveolin and the 29-kDa caveolin-associated protein (Fig. 7E, and data not shown). This peptide binds with high affinity to the SH2 domains of Src and Fyn but not to SH2 domains from other proteins such as SHP-2 or PI 3Ј-kinase (41). No stimulation of the kinase was observed with non-phosphorylated peptide or with a phosphorylated peptide (pY-VPM) that binds to the SH2 domain of PI 3Ј-kinase, but not Src or Fyn. Interest- ingly, the pY-MT peptide inhibited the phosphorylation of a 27-kDa protein in the Triton-insoluble pellet (Fig. 7D, open  arrowhead).
The in vitro kinase data suggested that insulin might activate the caveolin kinase through the tyrosine phosphorylation of a specific insulin receptor substrate protein that binds to the SH2 domain in Fyn. Previous work suggested that the insulinstimulated tyrosine phosphorylation of Cbl may be involved in the activation of Fyn in adipocytes (53). Cbl is rapidly and specifically phosphorylated on tyrosine in response to insulin in adipocytes. Cbl binds to Fyn in adipocytes, and this association is significantly increased after insulin stimulation. In addition, tyrosine-phosphorylated Cbl from adipocyte lysates binds specifically to a fusion protein containing the SH2 domain of Fyn. To further explore the possible relationship between Cbl phosphorylation and activation of the caveolin kinase, the effect of insulin on the association of Cbl with caveolin-enriched fractions was analyzed (Fig. 8). Cells were stimulated with insulin for increasing amounts of time, and the cells were fractionated into Triton-insoluble pellets or post-nuclear supernatants as described under "Experimental Procedures." Cbl was detected in the caveolin-enriched Triton-insoluble pellets by anti-Cbl Western blotting. Consistent with a role for Cbl in the activation of Fyn, Cbl was translocated into the Triton-insoluble fractions after insulin stimulation (Fig. 8A). Cbl association rapidly increased after insulin stimulation, reaching a maximum of 4-fold within 1-2 min, and levels declined thereafter. Cbl was immunoprecipitated either from octylthioglucosidesolubilized complexes or from the post-nuclear supernatants and analyzed by anti-phosphotyrosine Western blotting (Fig.  7B). The Cbl translocated into the Triton-insoluble pellets was tyrosine-phosphorylated, and the kinetics of the association of tyrosine-phosphorylated Cbl with the Triton-insoluble pellets was similar to the kinetics of phosphorylation of Cbl in the cell lysates and to the phosphorylation of caveolin in the Tritoninsoluble pellets (Fig. 8C). Very little of the tyrosine-phosphorylated insulin receptor and no tyrosine-phosphorylated IRS-1 were detected in the Triton-insoluble pellets by antiphosphotyrosine Western blotting (Fig. 8D). DISCUSSION The ability of insulin to stimulate glucose uptake and storage of glucose as glycogen and lipid is significantly increased after adipocyte differentiation. 2 Differentiation leads to an increase in the expression of the enzymes involved in the storage of glucose as lipid and glycogen. However, the increase in the responsiveness of these pathways to insulin cannot be fully accounted for by the combined increases in insulin-stimulated glucose uptake and the increased levels of expression of these enzymes after differentiation. 3 This indicates that the metabolic signaling pathways stimulated by insulin may be upregulated by differentiation as well.
Many of the signaling pathways stimulated by insulin are shared by other growth factors, such as EGF and PDGF. However, insulin stimulates glucose uptake and metabolism in adipocytes, whereas EGF and PDGF have little effect on these processes. Interestingly, caveolin-1 is specifically tyrosinephosphorylated in response to insulin. Despite the presence of receptors for other growth factors such as PDGF or EGF on adipocytes, caveolin phosphorylation does not occur in response to these other ligands (36). We demonstrate here that the tyrosine phosphorylation of caveolin is also differentiationdependent in 3T3-L1 cells (Figs. 1 and 2). Despite comparable levels of expression of active insulin receptors and caveolin in both preadipocytes and adipocytes, stimulation of the insulin receptor is uncoupled from the phosphorylation of caveolin in the preadipocytes. This suggests that caveolin is not a direct substrate of the insulin receptor and that other proteins are required for this phosphorylation.
We have previously shown (36) that caveolin copurifies with a caveolin-tyrosine kinase in low density Triton-insoluble complexes from 3T3-L1 adipocytes from both basal and insulinstimulated cells. Although there was no detectable insulin receptor tyrosine kinase activity in these complexes, the Src family kinase Fyn was highly enriched in the caveolar fractions from 3T3-L1 adipocytes (36). Fyn copurifies with caveolin in low density Triton-insoluble complexes from both preadipocytes and adipocytes (Figs. 3 and 4). Moreover, overexpression of wild type Fyn caused an increase in basal caveolin phosphorylation, as well as hyperphosphorylation of caveolin after insulin stimulation in the adipocytes (Figs. 5 and 6). However, there was no effect of Fyn overexpression on caveolin phosphorylation in the preadipocytes.
Preadipocytes and adipocytes both express significant levels of the insulin receptor, caveolin, and the caveolin kinase Fyn. However, only in the adipocytes does activation of the insulin receptor lead to the phosphorylation of caveolin. The data imply the presence of a mediator required for the segregated activation of Fyn in caveolae, the activity of which is critically dependent on adipocyte differentiation. There was no change in the association or tyrosine phosphorylation of Fyn in the caveolin-enriched fractions in response insulin in either adipocytes or preadipocytes (Fig. 7), indicating that direct modifications of Fyn phosphorylation are not involved in the activation of this kinase in response to insulin. The in vitro kinase assays suggest that the mediator for Fyn activation in adipocytes is an insulin-stimulated tyrosine-phosphorylated protein, which activates Fyn through binding to the SH2 and SH3 domains of the kinase. While the identity of this specific insulin-receptor substrate protein has not been definitively shown, evidence suggests that this molecule may be c-Cbl.
The unique conditions under which caveolin phosphorylation is observed suggest that the insulin receptor substrate responsible for the activation of caveolar Fyn would have several unique properties as follows: 1) phosphorylation showing specificity for insulin in adipocytes; 2) association with Fyn (through an SH2 domain interaction) in response to tyrosine phosphorylation; 3) translocation of the phosphorylated protein into caveolae in response to insulin; and 4) phosphorylation in 3T3-L1 adipocytes but not in the preadipocytes. Although the well characterized insulin receptor substrates IRS-1/-2 and Shc emerged as potential candidates (54,55), none fulfilled all of these requirements. In contrast, the proto-oncogene product c-Cbl shares many properties with this presumed substrate protein. Insulin stimulates the phosphorylation of c-Cbl in adipocytes, and this phosphorylation shows specificity for insulin in this cell type (53). As has been observed in lymphocytes, Cbl constitutively binds to Fyn in unstimulated adipocytes through an SH3 domain-mediated interaction, and insulin-stimulated tyrosine phosphorylation of Cbl increases this association. The tyrosine-phosphorylated Cbl in adipocytes binds specifically to fusion proteins containing the SH2 domain of Fyn but not other insulin-stimulated proteins such as SHP-2 or PI 3Ј-kinase. Cbl is translocated into caveolin-enriched Triton-insoluble complexes after insulin stimulation, and the Cbl in these complexes is tyrosine-phosphorylated (Fig. 8). Most interestingly, unlike the other known substrates of the insulin receptor such as IRS-1/-2 and Shc, Cbl phosphorylation is specific for the differentiated adipocyte phenotype (53).
The basis for the cell type specificity of Cbl phosphorylation is currently unknown. Cbl is expressed at comparable levels in both the preadipocytes and adipocytes. Unlike IRS-1/-2 and Shc, however, Cbl does not directly interact with the insulin receptor (53). We hypothesize that tyrosine phosphorylation of Cbl may require a specific adapter protein (56), which allows for the interaction of Cbl with the insulin receptor (53), and that it is the regulation of the expression of this protein that accounts for the coordinate regulation of the phosphorylations of Cbl and caveolin in response to insulin in adipocytes. Although the model linking phosphorylation of Cbl to the phosphorylation of caveolin through the activation of Fyn is compelling, the exact relationship between these three proteins is likely to be complex (56). For example, Cbl is likely to undergo processive phosphorylation after the activation of Fyn (42,57). However, the insulin-dependent association of Cbl and Fyn in caveolae leads to the intriguing possibility that the Fyn/Cbl pathway may have a unique function in adipocytes (phosphorylation of caveolin) due to the localization of these proteins to specialized domains of the plasma membrane, the caveolae.
The fact that caveolin expression and phosphorylation are regulated during adipocyte differentiation suggests an important role for caveolae in the differentiated adipocyte phenotype. An understanding of this role will lead to further insights into the general function of caveolae in cells and may lead to insights into the mechanisms of insulin-specific signal transduction.