Cholesterol Depletion Disrupts Caveolae and Insulin Receptor Signaling for Metabolic Control via Insulin Receptor Substrate-1, but Not for Mitogen-activated Protein Kinase Control* through

Insulin exerts its cellular control receptor binding in caveolae in plasmalemma of target cells Mag-nusson, and P. (1999) J. 1961– 1971). We now report that a progressive cholesterol depletion of 3T3-L1 adipocytes with b -cyclodextrin gradually destroyed caveolae structures and concomitantly attenuated insulin stimulation of glucose transport, in effect making cells insulin-resistant. Insulin access to or affinity for the insulin receptor on rat adipocytes was not affected as determined by 125 I-insu-lin binding. By immunoblotting of plasma membranes, total amount of insulin receptor and of caveolin remained unchanged. Receptor autophosphorylation in response to insulin was not affected by cholesterol depletion. Insulin treatment of isolated caveolae prepara-tions increased autophosphorylation of receptor before and following cholesterol depletion. Insulin-increased tyrosine phosphorylation of an immediate downstream signal transducer, insulin receptor substrate-1, and activation of the further downstream protein kinase B were inhibited. In contrast, insulin signaling to mitogenic control as determined by control of the extracellular signal-related kinases

Insulin controls target cells by binding to its cell surface receptor. The further intracellular transmission of the insulin signal involves phosphorylation of the receptor as well as other proteins, in particular the insulin receptor substrate (IRS), 1 on specific tyrosine residues. After tyrosine phosphorylation IRS is recognized by Src homology 2 domain-containing proteins for metabolic and glucose transport control, or activation of the mitogen-activated protein kinase (MAP kinase) pathway and mitogenic control (1)(2)(3)(4). In type 2 diabetes target cells of the hormone are not fully responsive, which is compensated temporarily by enhanced insulin secretion. The pathogenic mechanisms for this insulin resistance are not known, but an important common feature appears to be a reduced activation/ tyrosine phosphorylation of IRS-1 (5).
The insulin receptors are sequestered in the caveolae microdomains of the plasma membrane in adipocytes, and caveolae appear to be critical for insulin control (6). By thin-section electron microscopy, caveolae appear as omega-shaped invaginations of 50 -100 nm diameter in the plasma membrane (7). Caveolae invaginations are found in the plasma membrane of many cell types, but are particularly abundant in adipocytes where they increase in number in conjunction with the differentiation of 3T3-L1 fibroblasts to mature adipocytes (8 -10). Caveolae are involved in receptor-mediated uptake of solutes into the cytosol (11) and in transcytosis (12). A number of proteins, in addition to the insulin receptor, involved in signal transduction are localized to caveolae, which suggests that they may be involved in cellular signaling and control (reviewed in Refs. [13][14][15][16]. Caveolae are rich in cholesterol and sphingolipids. Caveolae may indeed form from cholesterol-and sphingolipid-rich rafts in the membrane in a process requiring the caveolae-specific structural protein caveolin. Caveolin is found in the plasma membrane and intracellularly, but in the plasma membrane is confined to caveolae; it is therefore used as a marker for these structures. The function of caveolae is dependent on a sufficient level of cholesterol in the plasma membrane and caveolae (12,17). We have also demonstrated a critical dependence of the insulin receptor signal transduction on cholesterol; depletion of cholesterol from the plasma membrane of rat adipocytes reversibly inhibited insulin stimulation of glucose transport and metabolic protein phosphorylation control (6). The importance of caveolae for insulin receptor signaling is further indicated by a consensus binding site for interaction with caveolin (18), and coprecipitation of the receptor with caveolin (4) indicates that the interaction may be physiological. Moreover, the insulin receptor appears to phosphorylate caveolin (19), whereas caveolin was shown to activate the isolated receptor, although the physiological relevance of this is not known (20).
Herein we examine in detail the dependence of the insulin receptor on caveolae for signal transduction: the effects of cholesterol depletion on plasma membrane and caveolae morphology, on the insulin receptor and on the downstream propagation of the insulin signal. In short, the insulin receptor is unaffected, but interaction with the immediate downstream mediator molecule IRS-1 and metabolic control is inhibited by cholesterol depletion and caveolae destruction, while insulin signaling via the MAP kinase pathway remains intact. We also discuss implications for the pathogenesis of insulin-resistant states.

EXPERIMENTAL PROCEDURES
Materials-Rabbit anti-insulin receptor ␤-chain polyclonal antibodies were from Transduction Laboratories, and mouse monoclonal antibodies were from Santa Cruz Biotechnology. Rabbit anti-caveolin polyclonal antibodies, and mouse anti-phosphotyrosine (PY20) monoclonal antibodies were from Transduction Laboratories, and rabbit anti-IRS-1 polyclonal antibodies were from Santa Cruz Biotechnology. Rabbit polyclonal antibodies against phospho-p44 and phospho-p42 MAP kinases (ERK1 and ERK2) were from New England Biolabs, Inc., and antiphospho-Akt1/protein kinase B␣ sheep polyclonal antibodies and rabbit polyclonal antibodies against Shc were from Upstate Biotechnology Inc. DMEM culture medium and sera were from Life Technologies, Inc. 2-Deoxy-D-[1-3 H]glucose and [ 32 P]phosphate were from Amersham Pharmacia Biotech. Insulin, ␤-cyclodextrin, and other chemicals were from Sigma-Aldrich or as indicated in the text. Harlan Sprague-Dawley rats were from B&K Universal (Stockholm, Sweden).
3T3-L1 Cell Culture and Differentiation-3T3-L1 fibroblasts were grown on 13 mm (diam.) glass coverslips in DMEM with 25 mM Dglucose, supplemented with 10% newborn calf serum, 50 UI/ml penicillin, and 50 g/ml streptomycin in 10% CO 2 /humidified atmosphere at 37°C. The medium was changed every 2-3 days. Two days after the fibroblasts reached confluence, differentiation was induced with slight modifications to the procedure in Ref. 21; cells were incubated for 2 days in DMEM containing 10% fetal bovine serum, 5 g/ml insulin, 0.25 M dexamethasone, and 0.1 mM 3-isobutyl-1-methylxanthine. The cells were then incubated for an additional 2 days in the same medium excluding dexamethasone and 3-isobutyl-1-methylxanthine. Cells were maintained for 8 -10 days in DMEM with 10% fetal bovine serum to attain maximal differentiation. More than 95% of the cells expressed the adipocyte phenotype, as determined from accumulation of triacylglycerol droplets.
Membranes were then fixed in 3% paraformaldehyde, 0.05% glutaraldehyde for 30 min at room temperature. Nonspecific binding was blocked with 1% bovine serum albumin and 0.1% gelatin before incubation with primary antibodies: mouse anti-insulin receptor ␤-chain monoclonal antibodies (20 g/ml) and rabbit anti-caveolin polyclonal antibodies (20 g/ml) for 90 min at 37°C. These were detected with gold(15 nm)-conjugated anti-mouse antibody and gold (6 nm)-conjugated anti-rabbit antibody for 15 h at 4°C. The membrane preparation was finally fixed in 2% glutaraldehyde for 10 min and 1% OsO 4 for 30 min at room temperature. After rinsing, the grids with membranes were frozen, lyophilized, and covered with 2-nm tungsten. Transmission electron microscopy was done with Jeol EX1200 TEM-Scan (Tokyo, Japan), equipped with Gatan Bioscan CCD camera; images were obtained with GATAN Digital Micrograph software.
Isolation and Incubation of Rat Adipocytes-Adipocytes were isolated by collagenase digestion from epididymal fat pads of Harlan Sprague-Dawley rats (160 -200 g) (23). Cells, at a final concentration 100 l of packed cell volume/ml, were freshly incubated in Krebs-Ringer solution (0.12 M NaCl, 4.7 mM KCl, 2.5 mM CaCl 2 , 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 ) containing 20 mM Hepes, pH 7.40, 3.5% (w/v) fatty acidfree bovine serum albumin, 100 nM phenylisopropyladenosine, 0.5 units⅐ml Ϫ1 adenosine deaminase with 2 mM D-glucose, at 37°C on a shaking water bath. 125 I-Insulin Binding to Adipocytes-Rat adipocytes were incubated in a shaking water bath with 125 I-insulin and unlabeled insulin as indicated at 37°C. Cells were rapidly separated from incubation medium by centrifugation through dinonylphthalate. 125 I in cell cake and medium was separately determined. Unspecific binding, determined on cells incubated with an excess of unlabeled insulin (1 M), was on average 0.2% of total and was subtracted. The amount of cells was determined as packed cell volume at the end of each incubation.
Isolation of Caveolae-enriched Membrane Fraction-Adipocytes were homogenized in 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.5 mM EGTA, 0.25 M sucrose, 25 mM NaF, 1 mM Na 2 -pyrophosphate, with protease inhibitors, 10 M leupeptin, 1 M pepstatin, 1 M aprotinin, 4 mM iodoacetate, and 50 M phenylmethylsulfonyl fluoride using a motordriven Teflon/glass homogenizer at room temperature. Subsequent procedures were carried out at 0 -4°C. A plasma membrane-containing pellet, obtained by centrifugation at 16,000 ϫ g for 20 min, was resuspended in 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, and protease inhibitors. Purified plasma membranes were isolated by sucrose density gradient centrifugation (24). Aliquots of this fraction were pelleted and resuspended in 0.5 M Na 2 CO 3 , pH 11, and protease inhibitors (25) and sonicated with a probe-type sonifier (MSE, Soniprep 150) three times for 20 s each. The homogenate was then adjusted to 45% sucrose in 12 mM Mes, pH 6.5, 75 mM NaCl, 0.25 M Na 2 CO 3 and loaded under a 5-35% discontinuous sucrose gradient in the same buffer solution and centrifuged at 39,000 rpm for 16 -20 h in a SW41 rotor (Beckman Instruments). The light-scattering band, enriched in caveolae, at the 5-35% sucrose interface was collected. Enrichment of caveolae was demonstrated by the enrichment of caveolin (7-fold) and cholesterol (3-fold) compared with the intact plasma membrane fraction; by electron microscopy and immunogold labeling of caveolin, between 50% and 70% of the isolated membranes constituted caveolae (6).
For insulin treatment of isolated caveolae, the caveolae-enriched fraction was suspended in the Krebs-Ringer Hepes-buffered cell incubation medium (above) with 0.01% (w/v) bovine serum albumin, 0.2 mM NaF, 0.2 mM Na 3 VO 4 , 1 mM MgCl 2 , and 0.1 mM ATP. Insulin or vehicle was added as indicated and the suspension sonicated with the probe sonifier five times for 3 s each time (the sonication was necessary, presumably for insulin to gain access to the interior of caveolae vesicles) on ice, and then incubated for 10 min at 37°C. Equal amounts of caveolae protein was subjected to SDS-PAGE and immunoblotting.
Additional Procedures-Total cell protein was prepared for SDS-PAGE as described (23); cells were immediately spun through dinonylphthalate and frozen. The cell cake was thawed by boiling in SDS-PAGE sample solution (26). IRS-1 tyrosine-specific phosphorylation, phosphoprotein kinase B, or phospho-ERK1/2 was thus determined by immunoblotting after SDS-PAGE of total cell protein (23). Equal amounts of cells were loaded within each experiment.
For immunoprecipitation cells were rapidly rinsed in Krebs-Ringer solution without serum albumin (with or without insulin). Cells were lysed in 50 mM Hepes, pH 7.4, 10 mM EDTA, 100 mM NaF, 10 mM sodium pyrophosphate, 2 mM Na 3 VO 4, 0.5 mM phenylmethylsulfonyl fluoride, 1 M aprotinin, 1 M pepstatin, 10 M leupeptin, 1% Nonidet P-40 with or without 100 M genistein (Calbiochem). The solubilized cell lysate was filtered through a 0.22-m pore size filter to remove fat. Equal amounts of protein were incubated with antibodies for 14 h at 4°C, when Sepharose G-Plus (Santa Cruz Biotechnology) was added for 1 h at 4°C with constant mixing (turning over). The beads were washed and boiled with SDS-PAGE sample solution and subjected to SDS-PAGE.
After SDS-PAGE (7-11% acrylamide), separated proteins were electrophoretically transferred to a polyvinylidene difluoride blotting membrane (Immobilon-P, Millipore) (27) and then incubated with indicated antibodies. Bound antibodies were detected according to the ECLϩ protocol and reagents from Amersham Pharmacia Biotech or Renaissanceϩ from PerkinElmer Life Sciences, using horseradish peroxidaseconjugated anti-IgG as secondary antibodies. Blots were quantitated by chemiluminescence imaging (Las 1000, Fuji, Japan) and Image Gauge software (Fuji).
Total and phosphoamino acid-specific phosphorylation of IRS-1 was determined after immunoprecipitation of IRS-1 from cells preincubated with [ 32 P]phosphate for 1 h (28). Phosphoamino acid identification was by electrophoresis on thin-layer silica plates at pH 1.9 (29) after partial acid hydrolysis in 5.7 M HCl at 110°C for 2 h, using authentic internal standards (Sigma) that were identified by ninhydrin staining.
Glucose transport was determined in 3T3-L1 adipocytes as uptake of 2-deoxy-D-[1-3 H]glucose (21). Cells were grown on 13-mm plastic coverslips (Thermanox coverslips, Nunc Inc., Copenhagen, Denmark) in 24well culture dishes. 2-Deoxy-D-[1-3 H]glucose was added to a final concentration of 50 M (0.3 Ci/ml), and the cells were incubated for 9 min. Glucose uptake was stopped by rinsing the coverslips in three successive solutions of ice-cold buffer. Nonspecific uptake was determined in the presence of 25 M cytochalasin B. Coverslips were transferred to scintillation vials and the cells were dissolved in 1% SDS. Radioactivity was measured after adding 5 ml of scintillant (Ready Gel, Beckman).
For determination of cholesterol content, membranes were pelleted and lipids were extracted with 2-propanol. Cholesterol was then quantitated spectrofluorometrically by measuring the amount of H 2 O 2 produced by cholesterol oxidase (30).
Protein was determined by Coomassie Blue binding (31) using bovine serum albumin as standard.

Effect of Cholesterol Depletion on Caveolae Structure in Adi-
pocyte Plasma Membrane-Here we in detail examine effects of cholesterol depletion on caveolar structure in relation to insulin signaling. We have chosen to study cultured 3T3-L1 adipocytes because their plasma membranes can be prepared intact for electron microscopic examination (6). Incubation of 3T3-L1 adipocytes with increasing concentrations of ␤-cyclodextrin in a dose-dependent manner reduced the cholesterol content of the plasma membrane (Fig. 1). Concomitantly the structural integrity of caveolae in the plasma membrane was progressively corrupted, as revealed by transmission electron microscopy of isolated plasma membranes (Fig. 2). Control membranes incubated with vehicle typically displayed an abundance of caveolar structures of 50 -100 nm diameter (Fig. 2a), identified by labeling with antibodies against the caveolar structural protein caveolin (6). Incubation with 2 mM ␤-cyclodextrin had no visible effect on caveolar morphology (data not shown), whereas 5 mM ␤-cyclodextrin was seen to clearly have an effect on the caveolae, which became flattened, less distinct, and reduced in number ( Fig. 2b; cf. Refs. 32 and 33). At 10 mM ␤-cyclodextrin, caveolae invaginations have almost vanished and may be discerned as minor irregular structures at the membrane surface (Fig. 2c); after 20 mM ␤-cyclodextrin treatment, no invaginations were seen (data not shown).
Immunogold labeling of caveolin showed that the caveolin protein remained clustered, indicating that underlying caveolar structures existed in the membrane after 50% reduction of the amount of plasma membrane cholesterol by treatment of the cells with 10 mM ␤-cyclodextrin (Fig. 2c), perhaps reflecting flattened caveolar patches or "rafts" remaining in the membrane after the extensive but partial cholesterol depletion (34).
Insulin-stimulated Glucose Uptake after Cholesterol Depletion-The cholesterol depletion concomitantly also inhibited progressively the ability of insulin to enhance glucose uptake (Fig. 3a). The extent of insulin stimulation of glucose transport was correlated to the amount of cholesterol, and hence to degree of caveolar intactness, in the plasma membrane of the cells (Fig. 3b).
Effect of Cholesterol Depletion on Insulin Receptors in the Plasma Membrane-By immunogold labeling and electron microscopy examination, it appeared that the insulin receptors remained associated with the caveolin clusters and hence with the underlying caveolar remains or rafts after the depletion of cholesterol (Fig. 2). This was affirmed by immunoprecipitation of the insulin receptor, which coprecipitated caveolin to the same extent with or without prior cholesterol depletion (data not shown). We used competitive 125 I-insulin binding to isolated rat adipocytes to determine the effect of cholesterol depletion on the accessibility of insulin receptors at the cell surface (Fig. 4). Cholesterol depletion (with 10 mM ␤-cyclodextrin, which reduced plasma membrane cholesterol concentration by ϳ60% (Ref. 6)) did not affect the time taken for steady state binding of insulin (Fig. 4a). We subsequently incubated the cells with 125 I-insulin for 30 min when near-maximal binding was obtained. Specific insulin binding was not affected by prior cholesterol depletion (Fig. 4b). We used nonlinear regression to fit a one-site binding equation (GraphPad Software, Inc.) to the data in Fig. 4b, to determine the effect on number of insulin receptors and on the dissociation constant for insulin binding. Neither the number of receptors nor the binding affinity was significantly affected by the cholesterol depletion.
After subcellular fractionation of rat adipocytes and isolation of the plasma membranes, SDS-PAGE and immunoblotting with antibodies against the insulin receptor indicated that the total amount of receptor protein in the plasma membrane fraction was not affected by the 10 mM ␤-cyclodextrin treatment (Fig. 5a). This agrees well with the electron microscopy and 125 I-insulin binding results. Similarly, the amount of caveolin in the plasma membrane fraction was not affected by treatment with up to 10 mM ␤-cyclodextrin (Fig. 5b), also in agreement with the electron microscopic findings.
Effect of Cholesterol Depletion on Insulin Receptor Function-The ␤-cyclodextrin depletion of cholesterol from the isolated rat adipocytes did not affect the ability of subsequent insulin treatment to stimulate tyrosine-specific receptor autophosphorylation (Fig. 6a). Nor did 10 mM ␤-cyclodextrin extraction of cholesterol from the isolated caveolae preparation affect the subsequent ability of insulin to stimulate receptor autophosphorylation in the caveolae fraction (Fig. 6b).
Effect of Cholesterol Depletion on Downstream IRS-1 and Protein Kinase B-We concluded that cholesterol depletion does not adversely affect the number of insulin receptors at the cell surface, the insulin receptor's interaction with or activation by insulin, or its ability to autophosphorylate. We therefore next examined the ability of the receptor to tyrosine phosphorylate an immediate downstream target for insulin's metabolic control, IRS-1. After cholesterol depletion of rat adipocytes, the tyrosine-specific phosphorylation of IRS-1 in response to insulin treatment was severely reduced (Fig. 7). The depletion of cholesterol inhibited the 1 nM insulin-stimulated increase in IRS-1 phosphorylation by 71 Ϯ 6% (mean Ϯ S.E., n ϭ 10). Cholesterol or intact caveolae are apparently required for signal transfer from the insulin receptor to IRS-1. As an expected consequence of the inhibition of IRS-1 phosphorylation, cholesterol depletion also completely blocked insulin stimulation to phosphorylation/activation of the further downstream protein kinase B (Akt) (Fig. 8).
It has been suggested that serine phosphorylation of IRS-1 acts to inhibit the insulin receptor (35,36), and serine/threonine phosphorylation of IRS-1 has been found to inhibit IRS-1 binding to the insulin receptor and hence IRS-1 tyrosine phosphorylation (37,38). Similarly, an unknown serine kinase activity against IRS-1 was detected in insulin-resistant cells (39), and it has been suggested that caveolin acts as a general serine/threonine protein kinase inhibitor (40). Cholesterol depletion had, however, no effect on the total phosphorylation of IRS-1 or on the distribution of phosphorylation between serine, threonine and tyrosine (Fig. 9). An enhanced degradation and lowered levels of IRS-1 in states of insulin resistance have been reported (41,42). We found no effect of cholesterol depletion on the total cellular levels of IRS-1; the amount of IRS-1 protein in cholesterol-depleted rat adipocytes determined by immunoblotting after immunoprecipitation of IRS-1 and SDS-PAGE was not significantly different from control cells (107 Ϯ 6%, mean Ϯ S.E., n ϭ 3). To directly determine the effect of cholesterol depletion on IRS-1 binding to the insulin receptor, we immunoprecipitated the insulin receptor and looked for IRS-1. The effect of insulin to increase the amount of IRS-1 and tyrosinephosphorylated IRS-1 that coimmunoprecipitated with the in- sulin receptor was much reduced in cholesterol-depleted cells (Fig. 10), which suggests that IRS-1 binding to the insulin receptor is inhibited in cholesterol/caveolae-depleted cells.
Lack of Effect of Cholesterol Depletion on MAP Kinase Pathway-IRS-1 is believed to transmit insulin signaling for metabolic as well as mitogenic control. However, there is a redundancy among insulin receptor substrates and mitogenic control by insulin has in some cell-types been described via, e.g., the protein Shc. It has also been shown that cholesterol depletion of fibroblasts activates the MAP kinase pathway and epidermal growth factor causes its hyperactivation (43). We therefore examined the effect of cholesterol depletion of rat adipocytes on the extracellular signal-related kinases (ERK1 and ERK2) of the MAP kinase pathway. Extensive cholesterol depletion (60% reduction) did not by itself have an effect on ERK1/2 phosphorylation (Fig. 11). Moreover, insulin enhanced the phosphorylation/activation of ERK1/2 to the same extent whether cholesterol was depleted or not (Fig. 11). It was verified in the same experiments that insulin-stimulated tyrosine-specific phosphorylation of IRS-1 was inhibited.
In contrast to ERK1/2, the cholesterol depletion strongly enhanced the state of Shc 52-kDa protein tyrosine phosphorylation (Fig. 12). Inclusion of the protein tyrosine kinase inhibitor genistein in the lysis buffer and during immunoprecipitation did not affect the increase in Shc phosphorylation, making a postlysis artifact unlikely. Insulin on the other hand had no effect on Shc phosphorylation, neither before nor after cholesterol depletion, even at a supraphysiological insulin concentration (Fig. 12), in accordance with recent findings in rat adipocytes (44). DISCUSSION ␤-Cyclodextrin is a widely used tool for control of cellular levels of cholesterol through its ability to extract cholesterol from the plasma membrane of intact cells. Without itself incorporating into the membrane, ␤-cyclodextrin selectively extracts cholesterol from the surface of the cells (45,46). We have shown previously that the inhibition of insulin signaling by cholesterol extraction using ␤-cyclodextrin was reproduced by cholesterol oxidation using cholesterol oxidase (6). Moreover, cholesterolloaded ␤-cyclodextrin did not inhibit insulin signaling (6), attesting against cholesterol-unrelated effects on insulin signaling by the ␤-cyclodextrin treatment. We now demonstrate that a reduction of adipocyte plasma membrane cholesterol concentration with ␤-cyclodextrin in a dose-dependent manner affected the structural integrity of caveolae in the plasma membrane. Eventually, the caveolar structure was lost by cholesterol depletion as determined by transmission electron microscopy of plasma membranes. Likewise, insulin signaling to enhanced glucose uptake was progressively inhibited in an apparently linear manner by cholesterol depletion. Hence, there was a correlation between the level of plasma membrane cholesterol, caveolae structural integrity, and insulin stimulation of glucose transport. Earlier studies have indicated that experimental modulation of membrane physical properties affects insulin signaling (47)(48)(49), but caveolae or cholesterol levels were not examined.
Cholesterol depletion and loss of the caveolae structures in the plasma membrane did not reduce the amount of insulin receptors in the membrane, or their affinity for insulin, as determined by 125 I-insulin binding to intact adipocytes, by immunogold-electron microscopy, or by immunoblotting for the receptor in isolated plasma membranes after SDS-PAGE. Nor was insulin receptor interaction with caveolin affected as shown by coimmunoprecipitation and by electron microscopy. Insulin-stimulated autophosphorylation of the insulin receptor was not affected by cholesterol depletion either of intact cells or of isolated caveolae. This is compatible with the demonstrated insulin-responsiveness of the detergent-solubilized receptor (50). In contrast, cholesterol has been found to be necessary for ligand binding (51) or receptor activity (52), with GalaninR2 and the nicotinic acetylcholine receptor, respectively.
Hence, intact caveolae are necessary not for the insulin receptor to bind insulin or to signal per se, but for the further downstream propagation of the signal. Activation of the immediate downstream protein IRS-1 was clearly abrogated, since after cholesterol depletion its insulin-stimulated tyrosine phosphorylation was inhibited. As a consequence, the further downstream propagation of the signal to protein kinase B phosphorylation was also blocked. There is a redundancy of IRS proteins in mediating metabolic control by insulin; IRS-2, -3, and -4 have also to varying degrees been implicated, although IRS-4 appears to have a very restricted tissue expression. The fact that cholesterol depletion inhibits the end effects of metabolic signal transduction suggests that insulin receptor interaction with all involved IRS proteins was inhibited.
Insulin-resistant states and reduced insulin-stimulated tyrosine phosphorylation of IRS-1 have been associated with reduced levels of IRS-1 (41,42) or with serine-phosphorylation of IRS-1 (37,38). We have ruled out these possibilities for the effect on IRS-1 of caveolae destruction/cholesterol depletion. A pleckstrin homology and phosphotyrosine-binding domain of IRS-1 have been shown to direct binding to the insulin receptor (53,54). We assessed if IRS-1 binding to the insulin receptor was affected by cholesterol/caveolae depletion. The reduced amount of IRS-1 that coprecipitated with anti-insulin receptor antibodies suggests that cholesterol/caveolae are required for IRS-1 binding to the receptor. The exact function of caveolae/ cholesterol in insulin receptor IRS-1 interaction in intact cells will require further investigation.
Interestingly, our findings demonstrate that, although intact caveolae/caveolar cholesterol was necessary for metabolic control, it was not required for insulin's mitogenic control via ERK1 and ERK2 and the MAP kinase pathway. Shc has been suggested to dominate over IRS-1 in mediating insulin's control via the MAP kinase pathway in some cell types (55)(56)(57)(58), but apparently not in, e.g., skeletal muscle (59). Our findings demonstrate that, although Shc is present, insulin does not affect Shc in isolated adipocytes and Shc does not control ERK and the MAP kinase pathway in adipocytes. Lack of insulin effect on Shc phosphorylation in adipocytes was recently described (44). The differential effects of caveolae destruction on insulin IRS/metabolic and ERK/mitogenic control indicate that caveolae may have a role in insulin signal sorting for metabolic FIG. 11. Effects of cholesterol depletion on insulin stimulation of ERK1 and ERK2 phosphorylation. Isolated rat adipocytes were incubated without or with 10 mM ␤-cyclodextrin for 50 min when the cells were incubated with the indicated concentration of insulin for another 10 min. After SDS-PAGE, the state of ERK1/2 phosphorylation was determined by immunoblotting with antibodies against the phosphorylated form of the two proteins. The experiment was repeated five times with similar results.
FIG. 12. Effects of cholesterol depletion on Shc tyrosine phosphorylation. Isolated rat adipocytes were incubated without or with 10 mM ␤-cyclodextrin for 50 min, when the cells were incubated with the indicated concentration of insulin for another 10 min. Shc was immunoprecipitated, and after SDS-PAGE the state of Shc tyrosine phosphorylation was determined by immunoblotting with antibodies against phosphotyrosine. The identity of Shc 52-kDa protein was determined by immunoblotting with antibodies against Shc (no phosphotyrosine corresponding to Shc 46-kDa protein was detected). a, immunoblot against phosphotyrosine. Indicated is the molecular mass of reference protein (Rainbow, Amersham Pharmacia Biotech). b, quantitation of three separate experiments. Incubation with 1 nM insulin as indicated. Each experiment was normalized by setting the control without insulin to 100%. Figure  versus mitogenic control. Hence, in cell types with no or little caveolae insulin signaling may be mainly via the MAP kinase pathway, at the expense of IRS-1-mediated metabolic signaling. It remains to identify the protein(s) which transmits insulin receptor activation of the MAP kinase pathway and is immune to caveolae destruction (e.g. the insulin receptor substrates Gab1 (Ref. 60 The dependence of the insulin receptor on intact caveolae/ cholesterol for transmission of its metabolic signals stands in contrast to other tyrosine kinase receptors, which are inhibited by cholesterol/caveolae/caveolin: Cholesterol depletion (using cyclodextrin) or caveolin-1 depletion (using caveolin-1 antisense expression) therefore constitutively activated the p42/44 MAP kinase cascade (43,64), and epidermal growth factor caused hyperactivation of ERK in cholesterol-depleted cells (43). The adipocyte p42/44 ERK1 and ERK2 were, however, not phosphorylated/activated by cholesterol depletion. Cholesterol depletion of adipocytes did not significantly reduce the amount of caveolin in the plasma membrane (herein), nor the amount of caveolin associated with the insulin receptor (herein), whereas depletion or oxidation of cholesterol was found to reduce plasma membrane-localized caveolin in fibroblasts (43,65) and in MDCK cells (66). It can be argued that, in our experiments, the adipocyte plasma membrane concentration of cholesterol and caveolae were reduced, but not caveolin, and hence we do not see constitutive activation or hyperactivation of ERK in cholesterol-depleted adipocytes. However, this does not explain the reduced ability of active insulin receptors to tyrosine phosphorylate IRS-1. Moreover, caveolin has been shown to activate the isolated insulin receptor (20). The strong increase in Shc phosphorylation after cholesterol depletion may be explained by activation of a Shc kinase other than the insulin receptor (an interpretation supported by the lack of insulin effect on Shc phosphorylation in adipocytes) or by cholesterol depletion making Shc accessible as a substrate for kinases.
A pronounced threshold effect with caveolae structures vanishing after more than 30% reduction of the normal cellular cholesterol concentration was described in MDCK cells (66). In MA104 epithelial cells, the number of caveolae was reduced to 10% of normal after lowering total cellular cholesterol by 55% through inhibition of endogenous cholesterol synthesis (67). In the 3T3-L1 adipocytes, we found that the stucture of caveolae was not affected after 20% reduction, was partially destroyed after 40%, and was almost completely destroyed after 50% reduction of the plasma membrane cholesterol. In MDCK cells, caveolin expression was correlated to the cellular cholesterol concentration (66), but was in rat or 3T3-L1 adipocytes apparently not affected by a 50% reduction of the plasma membrane concentration of cholesterol. This discrepancy may be due to the different cell types and/or to the acute effects studied herein and the long term (overnight to several days) effects of cholesterol reduction examined in the MDCK cells. Ilangumaran and Hoessli (68) found that methyl-␤-cyclodextrin treatment of lymphocytes caused a release of apparently raft-associated proteins and lipids, but not caveolin, from the cells. The adipocytes, however, remained intact after ␤-cyclodextrin extraction of up to 50% (3T3-L1 adipocytes (herein)) or 60% (rat adipocytes (Ref. 6)) of the plasma membrane cholesterol, when the cells were able to regain insulin control upon cholesterol replenishment (6). The insulin receptor, moreover, was quantitatively retained in the plasma membrane after cholesterol depletion. However, extraction of more than 60% of the plasma membrane cholesterol caused the adipocytes to easily rupture (data not shown). Apparently the cholesterol extraction made the adipocyte membranes brittle.
A uniform finding in insulin resistance has been that IRS-1 tyrosine phosphorylation in response to insulin is decreased in adipocytes and skeletal muscle in obesity and type 2 diabetes (5). Indeed, our findings demonstrate that cholesterol depletion induces insulin resistance with decreased insulin-stimulated IRS-1 tyrosine phosphorylation. It is also interesting to note that in skeletal muscle of type 2 diabetic patients insulin stimulation to phosphorylation of MAP kinase was normal, while insulin-stimulated tyrosine phosphorylation of IRS-1 was severely impaired (69,70). Similarly, in the vasculature of insulin-resistant obese Zucker fa/fa rats, insulin-stimulated tyrosine phosphorylation of IRS-1/2 were reduced, while the phosphorylation of ERK-1/2 were equal to that in normal rats (71). Insulin-resistant state induced by ␤ 3 -adrenergic stimulation blocked insulin-stimulated phosphorylation of IRS1/2 without affecting MAP kinase (72). Severely insulin-resistant patients with pseudoacromegaly had impaired insulin activation of IRS-1-associated phosphatidylinositol 3-kinase, with intact activation/phosphorylation of MAP kinase (73). Furthermore, glucocorticoid-induced insulin resistance blocked protein synthesis-stimulation by insulin without affecting the MAP kinase pathway (74). These findings are similar to the insulin resistance incurred on adipocytes by cholesterol depletion.
In conclusion, our findings point to the caveola as forming a hub, which is critical for insulin's cellular control, apparently with the potential to sort for metabolic versus mitogenic control. Our findings, moreover, suggest that caveolae dysfunction can be involved in the pathogenesis of insulin resistance.