Colocalization of insulin receptor and insulin receptor substrate-1 to caveolae in primary human adipocytes Cholesterol depletion blocks insulin signalling for metabolic and mitogenic control Margareta Karlsson 1 , Hans Thorn 1 , Anna Danielsson 1 , Karin G. Stenkula 1 , Anita O ¨ st 1 , Johanna Gustavsson 1 , Fredrik H. Nystrom 2 and Peter Stra ˚ lfors 1 1 Department of Cell Biology and Diabetes Research Centre, and 2 Department of Medicine and Care and Diabetes Research Centre, University of Linko ¨ ping, Sweden Caveolae are plasma membrane invaginations with several functions, one of which appears to be to organize receptor mediated signalling. Here we report that in primary human subcutaneous adipocytes the insulin receptor was localized to caveolae by electron microscopy/immunogold detection and by isolating caveolae from plasma membranes. Part of insulin receptor substrate 1 (IRS1), the immediate down- stream signal mediator, was colocalized with the insulin receptor in the plasma membrane and caveolae, as demon- strated by immunofluorescence microscopy, immunogold electron microscopy, and immunogold electron microscopy of transfected recombinant HA-IRS1. In contrast, rat epi- didymal adipocytes lacked IRS1 at the plasma membrane. Depletion of cholesterol from the cells using b-cyclodextrin blocked insulin stimulation of glucose uptake, insulin inhibition of perilipin phosphorylation in response to iso- proterenol, and insulin stimulation of protein kinase B and Map-kinases extracellular signal-related kinase (ERK)1/2 phosphorylation. Insulin-stimulated phosphorylation of the insulin receptor and IRS1 was not affected, indicating that caveolae integrity is required downstream of IRS1. In conclusion we show that insulin receptor and IRS1 are both caveolar proteins and that caveolae are required for both metabolic and mitogenic control in human adipocytes. Our results establish caveolae as foci of insulin action and stress the importance of examining human cells in addition to animal cells and cell lines. Keywords: extracellular signal-related kinase; ultrastructure; protein kinase B; b-cyclodextrin; glucose transport. Insulin exerts control over cell metabolism by binding to its cell surface receptor, which has been characterized in great detail [1–5]. The occupied receptor is autophosphorylated on tyrosine residues and can thereby tyrosine phosphorylate other cellular proteins, to transduce the insulin signal into the signal network of the cell. Chief among these proteins are the insulin receptor substrate (IRS) family of proteins [1]. When tyrosine is phosphorylated they can transmit metabolic and mitogenic signals. The further downstream events involve the generation of second messengers and phosphorylation of protein kinase B/Akt (PKB). Eventually glucose transporter GLUT4 is translocated to the plasma membrane for glucose uptake and other target proteins (e.g. perilipin [6]) are phosphorylated/dephosphorylated. Insu- lin’s ability for mitogenic signalling is transmitted via the Map-kinases extracellular signal-related kinase (ERK)1/2 for phosphorylation control of transcription factors. The precise mechanisms for insulin’s cellular control are not yet known in detail, and especially not so in human cells and tissues. Caveolae are 25–150 nm invaginations of the plasma membrane and are found in most cell types. They are particularly abundant in rat adipocytes and increase dramatically in number when 3T3-L1 fibroblasts are differentiated into fat cells and become insulin responsive [7–9]. Cholesterol and sphingolipids together with the principal structural protein caveolin are required for caveo- lae to form. Caveolae are involved in numerous cellular processes such as receptor mediated uptake, receptor mediated signalling, and vesicular trafficking, reviewed in [10]. A number of proteins involved in signal transduction have been found in caveolae [11–15]. We have shown that in rat adipocytes and in 3T3-L1 adipocytes the insulin receptor is localized to caveolae [16,17]. The immediate downstream signal-mediator IRS1, on the other hand, has been reported not to be associated with the plasma membrane in rat adipocytes [18] or in other cell types [19–21]. In rat adipocytes the disruption of caveolar integrity by cholesterol depletion using b-cyclodextrin, reversibly made the cells insulin resistant by inhibiting the insulin receptor from phosphorylating IRS1. On the other hand, insulin signalling for mitogenic control via Map-kinases ERK1 and 2 was not impaired in the rat cells [17,22]. The importance of caveolae for insulin action is reinforced by the finding that, in response to insulin stimulation, the insulin regulated glucose transporter GLUT4 is translocated to caveolae for glucose Correspondence to P. Stralfors, Department of Cell Biology and Diabetes Research Centre, University of Linko ¨ ping, SE58185 Linko ¨ ping, Sweden. Fax: + 46 13 224314, Tel.: + 46 13 224315, E-mail: peter.stralfors@ibk.liu.se Abbreviations: ERK, extracellular signal-related kinase; GLUT4, insulin-stimulated glucose transporter 4; IRS, insulin receptor sub- strate; Map, mitogen activated protein; PKB, protein kinase B; TEM, transmission electron microscopy. (Received 11 March 2004, revised 16 April 2004, accepted 21 April 2004) Eur. J. Biochem. 271, 2471–2479 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04177.x uptake [9,23,24], and that some of the downstream signalling for enhanced glucose uptake may take place in caveolae [25]. The role of caveolae in insulin signalling has, however, not been examined in human cells. As animal cells in general and immortalized cell lines in particular often deviate from primary human cells, which are the cells most relevant to the study of human disease, we set out to investigate this in human adipocytes. Here we report that in primary human adipocytes the insulin receptor along with its immediate downstream signal mediator IRS1 is localized in caveolae of the plasma membrane. Destruction of caveolae interrupts insulin signal transduction downstream of IRS1 and makes these cells insulin resistant to both metabolic and mitogenic signalling. Materials and methods Subjects Samples of subcutaneous abdominal fat were obtained, after the informed consent, from patients (male or female, age 40–76 years) undergoing elective abdominal surgery at the University Hospital of Linko ¨ ping. Patients diagnosed with diabetes were excluded. The Local Ethics Committee approved the study. Materials Rabbit anti-insulin receptor b-chain and anti-caveolin polyclonal and mouse anti-phosphotyrosine (PY20) mono- clonal antibodies were from Transduction Laboratories (Lexington, KY, USA). Rabbit anti-phospho(Thr308)- PKB/Akt polyclonal antibodies were from Upstate Biotech (Charlottesville, VI, USA). Rabbit polyclonal antibodies against phospho-ERK1/2 and against phospho-p38 Map- kinase were from Cell Signalling Techn. (Beverly, MA, USA). Anti-insulin receptor b-chain monoclonal, anti-IRS1 rabbit polyclonal antibodies (sc-559, sc-560), and anti-HA rabbit polyclonal antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Colloidal gold conjugated anti-(IgG) Ig was from Aurion (Wageningen, the Netherlands). 2-Deoxy- D -[1- 3 H] glucose was from Amersham Biotech (UK). Insulin, b-cyclodextrin, and other chemicals were from Sigma-Aldrich (St Louis, MO, USA) or as indicated in the text. Isolation and incubation of adipocytes Human adipocytes were isolated by collagenase (type 1; Worthington, NJ, USA) digestion as described [26] except that the cells were allowed to floatate by gravity. At a final concentration of 100 lL packed cell volume per ml, cells were incubated in Krebs–Ringer solution (0.12 M NaCl, 4.7 m M KCl, 2.5 m M CaCl 2 ,1.2m M MgSO 4 ,1.2m M KH 2 PO 4 ) containing 20 m M Hepes, pH 7.40, 1% (w/v) fatty acid-free bovine serum albumin, 100 n M phenyliso- propyladenosine, 0.5 UÆmL )1 adenosine deaminase with 2m M glucose, at 37 °C on a shaking water bath. For analysis of b-cyclodextrin effects on insulin control of insulin receptor, IRS1, PKB, ERK1/2, and p38, cells were incubated overnight before analysis. Rat adipocytes were isolated [26] from the epididymal adipose tissue of Sprague–Dawley rats (130–160 g, B and K Universal, Sollentuna, Sweden). The animals were treated in accor- dance with Swedish Animal Care regulations. Transfection of adipocytes with myc-tagged caveolin-1 or HA-tagged IRS1 Isolated adipocytes were transfected as follows: 200 lLcells (400 lL cell volume per ml) was mixed with 200 lLof buffer (137 m M NaCl, 2.7 m M KCl, 10 m M Na 2 HPO 4 and 1.8 m M KH 2 PO 4 , pH 7.5) containing 5 lg of empty vector pcis2, pcis2-myc-caveolin-1, or pcis2-HA-IRS1 (kindly supplied by M. Quon, NIH, Bethesda, MD, USA) in an electroporation cuvette. Cells were electroporated with 6 pulses at 600 V and 25 lF using Gene pulser II (Bio-Rad). Cells from 15 cuvettes were pooled and kept at 37 °Cin 10% (v/v) CO 2 . After 1 h an equal volume of Dulbecco’s modified Eagle’s medium pH 7.5, containing 25 m M glu- cose, 50 unitsÆmL )1 penicillin, 50 lgÆmL )1 streptomycin, 200 n M phenylisopropyladenosine, 7% (w/v) bovine serum albumin, and 25 m M Hepes, was added. After 18 h of incubation cells were collected and plasma membranes prepared for electron microscopy. Electron microscopy of plasma membranes Plasma membrane sheets were prepared for electron microscopy as described previously [27]. After rinsing the adipocytes in ice-cold phosphate-buffer (10 m M Na 2 HPO 4 / NaH 2 PO 4 , 150 m M NaCl, pH 7.5) they were attached to nickel grids [28]: poly( L -lysine) and formvar coated grids were rehydrated by floating on ice-cold phosphate- buffer containing the cells. Grids with captured adipocytes were flushed with ice-cold 150 m M KCl, 1.9 m M Tris/HCl buffer, pH 7.6. Plasma membranes remaining on the grids were washed three times in 150 m M Hepes, pH 7.5, and fixed in 0.1 M sodium cacodylate buffer, containing 0.1 M sucrose, 3% (w/v) paraformaldehyde, and 0.05% (v/v) glutaraldehyde, for 30 min at room temperature. Plasma membranes were blocked for 1 h in phosphate- buffer, containing 5% (w/v) bovine serum albumin (BSA-c, Aurion, the Netherlands), 0.1% (w/v) gelatine, and 5% normal goat serum (Aurion) at 37 °C. This was followed by incubation with mouse anti-insulin receptor b-chain mono- clonal antibodies, or anti-IRS1 polyclonal antibodies, or rabbit anti-caveolin polyclonal antibodies for 1.5 h at 37 °C. Grids were rinsed with phosphate-buffer, containing 0.1% (w/v) BSA-c before incubation with secondary antibodies. Goat anti-(rabbit IgG) Ig and/or goat anti- (mouse IgG) Ig, conjugated with 15 nm or 6 nm colloidal gold, was added to plasma membranes and incubated overnight at 4 °C. After immunolabelling, the plasma membranes were rinsed and fixed in 2% (v/v) glutaraldehyde for 10 min followed by 1% (w/v) OsO 4 for 30 min in 0.1 M sodium cacodylate buffer, with 0.1 M sucrose, pH 7.5, at room temperature. Grids were rinsed with water, frozen, lyo- philized, and coated with 2 nm tungsten by magnetron sputtering directly in the freeze-dryer [29]. Transmission electron microscopy was performed with Jeol EM1230 2472 M. Karlsson et al. (Eur. J. Biochem. 271) Ó FEBS 2004 TEM-SCAN (Tokyo, Japan). No labelling was observed in the absence of primary antibody, nor was any cross-reactivity detected between secondary and primary antibodies. Immunofluorescence microscopy of plasma membrane sheets Plasma membrane sheets were prepared as described [17, 30] and fixed in 3% (w/v) paraformaldehyde for 30 min at room temperature. After blocking in bovine serum albumin and gelatine membranes were incubated with rabbit anti- IRS1, caveolin, or insulin receptor antibodies. Primary antibodies were detected with fluorescent secondary anti- bodies (Alexa fluor 488 or 594, from Molecular Probes, USA) by fluorescence microscopy (Nikon, D-Eclipse CI confocal microscope, Tokyo, Japan). No labelling was observed in the absence of the primary antibody, nor was any cross-reactivity detected between secondary and pri- mary antibodies. Isolation of caveolae-enriched membrane fraction A caveolae-enriched membrane fraction was isolated as described [22]. Adipocytes were homogenized in 10 m M Tris/HCl, pH 7.4, 1 m M EDTA, 0.5 m M EGTA, 0.25 M sucrose, 25 m M NaF, 1 m M pyrophosphate with protease inhibitors (10 l M leupeptin, 1 l M pepstatin, 1 l M apro- tinin, 4 m M iodoacetate and 50 m M phenylmethanesulfo- nyl fluoride) using a motor-driven Teflon/glass homogenizer at room temperature. Subsequent proce- dures were carried out at 0–4 °C. A plasma membrane containing pellet, obtained by centrifugation at 16 000 g for 20 min, was resuspended in 10 m M Tris/HCl, pH 7.4, 1m M EDTA and protease inhibitors. Purified plasma membranes were isolated by sucrose gradient centrifuga- tion [31,32], referred to as the plasma membrane fraction. Aliquots of plasma membrane fraction were pelleted and resuspended in 0.5 M Na 2 CO 3 , pH 11, and protease inhibitors [33], and sonicated with a probe-type sonifier (MSE, Soniprep 150) 3 · 20 s. The homogenate was then adjusted to 45% sucrose in 12 m M Mes, pH 6.5, 75 m M NaCl, 0.25 M Na 2 CO 3 and loaded under a 5–35% discontinuous sucrose gradient in the same buffer and centrifuged at 39 000 r.p.m. for 16–20 h in a SW41 rotor (Beckman Instruments, Fullerton, CA, USA). The light- scattering band at the 5–35% sucrose interface, enriched in caveolae, was collected and referred to as the caveolae fraction. A cytosolic fraction was obtained as the supernatant from centrifuging the 16 000 g supernatant at 150 000 g 75 min and discarding the pellet. One third of the plasma membrane from rat adipocytes constitutes caveolae membrane [27], which thus is a minimum figure for human adipocyte plasma membranes that have a higher density of caveolae. To examine the purity of our caveolae preparation we isolated nuclear and mitochond- rial membranes, which are the chief contaminants of adipocyte plasma membrane [31], and identified seven proteins by peptide sequence analysis using collision induced dissociation and electrospray ionization mass spectrometry. The same analysis was performed in parallel on the caveolae fraction and no nuclear/ mitochondrial contamination was detected (N. Aboulaich, J. Vainonen, P. Stra ˚ lfors, and A. Vener, unpublished data). SDS/PAGE and immunoblotting Proteins were separated by SDS/PAGE, transferred to a poly(vinylidene difluoride) blotting membrane (Immobi- lone-P, Millipore, Bedford, MA, USA), and incubated with indicated primary antibodies. Bound antibodies were detected using Renaissence+ (New England Nuclear) with horseradish peroxidase-conjugated anti-IgG as secondary antibody. Blots were evaluated by chemiluminescence imaging (Las1000, Fuji, Japan). By two-dimensional elec- trofocusing (pH 3–10) SDS/PAGE [34] analysis and immu- noblotting against phosphotyrosine and IRS1, we found that > 95% of the tyrosine phosphorylated 180 kDa band represents IRS1. Cholesterol and protein determination Cholesterol was quantitated spectrofluorometrically [35] by measuring the amount of H 2 O 2 produced by cholesterol oxidase (Rhodococus erythropolis, from Boehringer Mann- heim, Mannheim, Germany) on a Fluoroscan spectrofluo- rometer (Labsystems, Finland). Protein was determined with the Micro BCA kit from Pierce. Bovine serum albumin wasusedasstandard. Determination of glucose transport Glucose transport was determined as uptake of 2-deoxy- D -[1- 3 H]glucose [36] after transfer of cells to medium without glucose. Cells were treated with or without insulin for 15 min when 2-deoxy- D -[1- 3 H]glucose was added to a final concentration of 50 l M (10 lCiÆmL )1 )andthecells were incubated for another 30 min; uptake was linear for at least 30 min. Glucose uptake was stopped by centrifuging through dinonylphtalate and freezing in an aluminium block kept at )20 °C. Non-specific uptake was determined in the presence of 25 l M cytochalasin B by terminating the incubation immediately after addition of the 2-deoxy- D -[1- 3 H]glucose. Tubes were cut frozen through the oil-layer, the cell cakes were dissolved in 0.4 mL 1% (w/v) SDS, and radioactivity was measured in 3 mL of scintillant (QuickSafe A, Zinsser Analytic Ltd). Perilipin protein phosphorylation analysis For analysis of protein phosphorylation, adipocytes (50 lL packed cell volume per ml) were prelabelled with [ 32 P]phos- phate for 1 h [37]. Total cell protein was prepared for SDS/ PAGE as described [26]. 32 P-incorporation in perilipin was evaluated by radioimaging (Bas1000, Fuji). Results Localization of the insulin receptor in caveolae of human adipocytes We analyzed isolated adipocytes from human abdominal subcutaneous adipose tissue. As seen by transmission Ó FEBS 2004 Insulin receptor and IRS1 in caveolae (Eur. J. Biochem. 271) 2473 electron microscopy, the inside of the plasma membrane of a human fat cell was covered with large numbers of caveolae. These were seen as bulbs protruding from the plasma membrane, either singularly (Fig. 1A) or in clusters of varying size (Figs 1B and 2). Caveolae clusters were frequent in the plasma membrane of these cells, in contrast to rat adipocytes [27]. The caveolae structures were labelled with antibodies against the caveolae specific protein caveolin (6 nm gold particles) and against the insulin receptor (15 nm gold particles). Labelling of the insulin receptor was largely restricted to the caveolae structures: as single caveolae (Fig. 1A) or in clusters of caveolae (Fig. 1B). A close up view of a single caveola showed that only the neck attaching the caveolar bulb to the plasma membrane was significantly labelled by antibodies against caveolin (Fig. 2). The specific localization of caveolin in the necks of caveolae has been described for rat adipocytes [27] and human fibroblasts [27]. We verified this finding here by transfecting human adipocytes with myc-tagged caveolin-1 and examination with gold-conjugated antibodies against the myc tag. Only the necks and not the bulbs of caveolae were labelled (not shown). We also isolated a caveolae fraction, without using detergent [17], from purified human adipocyte plasma membranes and examined the presence of insulin receptor by immunoblotting after gel electrophoresis. The insulin receptor was present and enriched in this caveolae fraction, compared to the total plasma membrane fraction (Fig. 3A). The chief structural constituents of caveolae, caveolin (Fig. 3C), and cholesterol (0.59 ± 0.05 nmol cholesterol Fig. 1. Transmission electron microscopy of the inside of human adi- pocyte plasma membrane. Plasma membrane sheets were immunogold labelled against the insulin receptor (15 nm gold particles, black arrowheads) and caveolin (6 nm gold particles, white arrowheads). (A) Membrane area with single caveolae, (B) a cluster of caveolae. Images contrast inverted. Fig. 2. Stereo image of a human adipocyte caveola by transmission electron microscopy. Plasma membrane sheets were immunogold labelled against the insulin receptor (15 nm gold particle, black arrowheads) and caveolin (6 nm gold particles, white arrowheads). Image contrast inverted. Fig. 3. Insulin receptor in plasma membrane and isolated caveolae fractions. Human adipocytes were incubated with or without 1 n M insulin for 20 min, as indicated; plasma membrane (pm) and caveolae (cav) fractions were prepared. Aliquots of 3 lg protein was subjected to SDS/PAGE and immunoblotting with antibodies against (A) the insulin receptor b-subunit, (B) phospho-tyrosine and (C) caveolin. 2474 M. Karlsson et al. (Eur. J. Biochem. 271) Ó FEBS 2004 per lgprotein(mean±SE,n ¼ 6) in the caveolae fraction and 0.31 ± 0.02 in the plasma membrane fraction), were also enriched. The enrichment was compatible with caveo- lae constituting between a third and half of the plasma membrane; in human adipocytes the numerous caveolae clusters increase the caveolae membrane more than in rat adipocytes, which have a third of their plasma membrane being caveolae [27]. Insulin receptor signalling in caveolae We next examined the functional status of caveolae- localized receptors. After incubating cells with a submax- imal concentration of insulin the insulin-induced increase in receptor autophosphorylation was comparable in the iso- lated caveolae fraction and in the total plasma membrane fraction (Fig. 3B). We have previously shown that reduction of the amount of cholesterol in the plasma membrane with b-cyclodextrin destroys the structural integrity of caveolae [27]. Depletion of cholesterol made the human adipocytes insulin resistant: insulin stimulation of glucose uptake was impaired by prior cholesterol depletion, but without any effect on the basal glucose uptake (Fig. 4). The protein perilipin is believed to be involved in the hormonal regulation of lipolysis [6]. Insulin’s ability to counteract catecholamine-stimulated phosphorylation of perilipin was also inhibited by the cholesterol depletion (Fig. 5). In line with these findings, insulin signalling to increased phosphorylation of protein kinase B (Fig. 6A) was curtailed. The insulin-stimulated phosphorylation of Map- kinases ERK1 and 2 (Fig. 6B) was also attenuated by b-cyclodextrin. In contrast, insulin-stimulated phosphory- lation (Fig. 6C) of the insulin receptor was not affected by this level of prior cholesterol depletion, nor was the total tyrosine phosphorylation of IRS1 affected (Fig. 6D). A more detailed examination of the effect of cholesterol depletion, at different concentrations of insulin, still failed to Fig. 4. Effect of cholesterol depletion on insulin control of glucose up- take. Isolated human adipocytes were incubated with (closed symbols) or without (open symbols) 10 m M b-cyclodextrin for 50 min and were then incubated with the indicated concentration of insulin for 15 min. 2-Deoxyglucose uptake was then determined. Because the maximal response to insulin was variable between individuals (131, 53, and 58 nmol 2-deoxyglucoseÆmin )1 ÆmL )1 packed cells), glucose uptake was normalized to the maximal uptake for each experiment and expressed aspercentofmax.Mean±SE,n ¼ 3 separate experiments of three determinations each on cells from three subjects. Fig. 5. Effect of cholesterol depletion on insulin control of perilipin phosphorylation. Isolated human adipocytes were incubated with or without 10 m M b-cyclodextrin (bCD)asindicatedfor50minandthen they were incubated with 20 n M isoproterenol (iso) or 20 n M isopro- terenol and 1 n M insulin (iso + ins), or with vehicle (contr) for 15 min. Shown is the 32 P-phosphorylation of perilipin after SDS/PAGE of whole-cell lysates. The indicated doublet represents perilipin; the phosphorylation is typically shifted from the lower molecular mass band to an enhanced phosphorylation of the higher molecular mass band in response to cyclic AMP elevation [6]. Fig. 6. Effect of cholesterol depletion on insulin signal mediators. Human adipocytes were treated with or without 7 m M b-cyclodextrin for 50 min, and then incubated with or without 10 n M insulin for 20 min, as indicated. Whole-cell lysate aliquots were subjected to SDS/ PAGE and immunoblotting with antibodies against (A) phospho- PKB; (B) phospho-ERK1/2; (C) phospho-tyrosine (insulin receptor b-subunit); (D) phospho-tyrosine (IRS1). Ó FEBS 2004 Insulin receptor and IRS1 in caveolae (Eur. J. Biochem. 271) 2475 reveal any significant effects on insulin receptor (Fig. 7A) or IRS1 (Fig. 7B) phosphorylation at any concentration of the hormone. However, as we determined total tyrosine-phos- phorylation of IRS1, we cannot rule out effects on specific phosphorylation sites on the protein. Localization of IRS1 in the plasma membrane and caveolae of human adipocytes As IRS1 phosphorylation by the insulin receptor was not dependent on plasma membrane cholesterol/caveolae integ- rity, as has been found to be the case in rat adipocytes [22], we examined if IRS1 was localized in the plasma membrane and caveolae in human adipocytes. Immunofluorescence microscopy of plasma membrane sheets, from cells not exposed to insulin, using anti-IRS1 Igs revealed that IRS1 was localized in the plasma membrane (Fig. 8B). Antibodies against different epitopes (N- and C-terminal) of the protein similarly detected IRS1 in the plasma membrane. In contrast, under identical conditions, IRS1 was repeatedly not detected by immunofluorescence microscopy in the plasma membrane of rat epididymal adipocytes (Fig. 8E). The punctuate pattern of IRS1 binding in the plasma membrane (Fig. 8B) suggested that part of IRS1 may be associated with caveolae and hence the insulin receptor. To examine the possibility of caveolar localization of IRS1 we analyzed plasma membrane sheets for colocalization with caveolin (Fig. 8A–C) or the insulin receptor (Fig. 9A–C). The merged immunofluorescence images for IRS1 and caveolin or IRS1 and the insulin receptor indicates colocalization (as demonstrated by the yellow colour) of IRS1 with both caveolin and the insulin receptor. To further examine a caveolar localization of IRS1 we next examined plasma membrane sheets by electron micro- scopy after immunogold labelling with antibody against IRS1 (Fig. 10A,B). IRS1 was detected in the caveolae structures as well as outside: 63% of IRS1 was found in caveolae (293 of 567 identified gold particles, from 10 different cells, were found in or directly associated with caveolae structures, which constituted 30% of the plasma membrane sheets examined). Similar results were obtained with antibodies against the N- and against the C-terminal part of IRS1. IRS1 remained bound to the plasma membrane after depletion of cholesterol with b-cyclodextrin (not shown). To dispel any uncertainty of antibody cross-reactivity and detection of other plasma membrane and caveolar proteins, we transfected human adipocytes with the human IRS1 with an HA-tag (hemagglutinin) and used immunogold- antibodies against the HA-tag to examine IRS1 localization in the human adipocyte plasma membrane. Figure 10C shows that the HA-tag of HA-IRS1 was detected by electron microscopy in the plasma membrane and in caveolae, similarly to the findings with antibody against the wild-type IRS1. Discussion Our findings demonstrate that caveolae are central to insulin action in human adipocytes, similarly to the previously described situation in rat and 3T3-L1 adipo- cytes [15,17,22–24]: the insulin receptor in human adipo- cytes is located in caveolae, is autophosphorylated in caveolae, and is dependent on caveolae for its cellular control. Intriguingly, in human adipocytes the immediate downstream signal mediator IRS1 was, under basal noninsulin-stimulated conditions, colocalized with the insulin receptor in caveolae. In type 2 diabetes the tissues respond poorly to insulin, exhibiting insulin resistance that can be overcome by increasing concentrations of circula- ting insulin. In the majority of cases it is still not known what defect in the insulin target cell causes insulin resistance. Our finding that insulin’s cellular control is dependent on caveolae/cholesterol in human adipocytes indicates that caveolae has to be taken into account when trying to understand normal insulin signal transduction as well as the dysfunction that causes insulin resistance in different human conditions. Fig. 7. Effect of cholesterol depletion on dose–response effect of insulin on tyrosine phosphorylation of insulin receptor and IRS1. Human adi- pocytes were treated with (closed symbols) or without (open symbols) 7m M b-cyclodextrin for 50 min, and incubated with indicated con- centration of insulin for 20 min. Whole-cell lysate aliquots were sub- jected to SDS/PAGE and immunoblotting with antibodies against (A) phospho-tyrosine (insulin receptor b-subunit) or (B) phospho-tyrosine (IRS1). It was verified that in each experiment insulin-stimulated phosphorylation of PKB was blocked. Insulin-stimulated phosphory- lation was obtained by setting the value with no insulin to 0% and at 100 n M insulin to 100% effect. Dose–response curves were fitted to experimental data using the sigmoid dose–response algorithm in GraphPad Prism 3 software. Mean ± SE, n ¼ 5subjects. 2476 M. Karlsson et al. (Eur. J. Biochem. 271) Ó FEBS 2004 We found that in human adipocytes IRS1 is associated with the plasma membrane and the caveolae, and hence in close proximity to the insulin receptor. This finding was demonstrated by immunofluorescence microscopy and immunogold electron microscopy, using antibodies against the N- or C-terminal part of the protein, thus reducing the possibility of antibodies crossreacting with and detecting an unrelated plasma membrane/caveolae protein. Moreover, transfected cells expressed HA-IRS1 in the caveolae and the plasma membrane as analyzed with antibodies against the HA-tag. Plasma membrane sheets on grids were extensively washed before incubation with the antibodies, thus reducing the possibility of unspecific binding of IRS1 to the plasma membrane. Also, after the same treatment no IRS1 was associated with the plasma membrane in rat epididymal adipocytes. In rat adipocytes (herein and [18]) and other cells [19–21] IRS1 has been shown not to be associated with the plasma membrane, but has been found in the cytosol or bound to intracellular membranes and the cytoskeleton, pointing to the importance of exercising great caution when extrapolating to the situation in human beings from findings in animal cells and cell lines commonly used. A consequence of IRS1 localization in the plasma membrane and caveolae appears to be that depletion of cholesterol and destruction of caveolae does not block insulin interaction with and stimulation of IRS1 tyrosine phosphorylation as happens in rat adipocytes [22], but instead blocks the further downstream phosphorylation of PKB and the target effects on glucose uptake and perilipin phosphorylation. It can be hypothesized that the occupied Fig. 8. IRS1 colocalization with caveolin in the plasma membrane. Plasma membrane sheets attached to glass cover-slips were incubated with antibodies against caveolin and against the C-terminal part of IRS1. (A–C) Human adipocyte examined by immunofluorescence microscopy using red fluorescent antibodies against caveolin (A) and green fluorescent antibodies against IRS1 (B); merged image (C). (D–F) Rat adipocyte examined by immunofluorescence microscopy using green fluorescent antibodies against caveolin (D) and red fluorescent antibodies against IRS1 (E); merged image (F). Fig. 9. IRS1 colocalization with the insulin receptor in the plasma membrane. Plasma membrane sheets attached to glass coverslips were incubated with antibodies against the insulin receptor and IRS1, and examined by immunofluorescence microscopy using green fluorescent antibodies for the insulin receptor (A) and red fluorescent antibodies for IRS1 (B); merged image (C). Ó FEBS 2004 Insulin receptor and IRS1 in caveolae (Eur. J. Biochem. 271) 2477 insulin receptor and, in human adipocytes, tyrosine-phos- phorylated IRS1 have to be internalized via caveolae in order to interact with and activate downstream signal mediators. Our finding then poses the question of IRS2 localization in these cells. Differential localization of IRS1 and IRS2 could explain why IRS1 is a preferred mediator of insulin signalling under normal conditions. Receptor inter- nalization by way of caveolae was recently reported for the kinin B(2) receptor [38]. Intriguingly, it has been suggested that a larger cyclic AMP response to stimulation of the b1-compared to the b2-adrenergic receptor in rat cardio- myocytes depends on colocalization of the b1-receptor, but not the b2-receptor, with the downstream mediator adenylyl cyclase in caveolae [39]. It will be important to examine the mechanism for IRS1 binding to caveolae in the human fat cells and whether this binding is subject to control, by, e.g. insulin. Human IRS1 does not contain the consensus sequence for binding to caveolin, as has been described for a number of signalling proteins including the insulin receptor [40]. A further property that distinguishes primary human from rat adipocytes was that insulin stimulation to increased phosphorylation of Map-kinases ERK1/2 was blocked by cholesterol depletion/caveolae destruction in the human cells. This is contrary to the situation in rat adipocytes, where only metabolic and not mitogenic control by insulin was dependent on intact caveolae [22]. Apparently this difference reflects different properties of insulin receptor signalling in rat and human tissues. This again stresses the importance of examining human cells and tissues. b-Cyclodextrin is widely used to control cellular levels of cholesterol through its ability to mildly extract cholesterol from the plasma membrane of intact cells without itself incorporating into the membrane. The effects of b-cyclodextrin treatment were critically dependent on its concentration. The effective concentration had to be experimentally tried out for each batch of the compound, which varied between manufacturers and between different lots from the same manufacturer. In conclusion, we report that in human adipocytes caveolae contain the insulin receptor and its immediate downstream signal mediator IRS1, and that cholesterol depletion and caveolae destruction make these cells insulin resistant, downstream of IRS1, for both metabolic and mitogenic control. Acknowledgements We thank Kurt Borch, Preben Kjolhede, and colleagues at the departments of Surgery and Obstetrics at the University Hospital in Linko ¨ ping for providing us with adipose tissue. Financial support was obtained from Lions Foundation, Swedish Society for Medical Research, Swedish National Board for Laboratory Animals, O ¨ stergo ¨ t- land County Council’s Medical Research Funds, Swedish Foundation for Strategic Research (National Network for Cardiovascular Research), Swedish Diabetes Association and the Swedish Research Council. References 1. Yenush, L. & White, M.F. (1997) The IRS-signalling system during insulin and cytokine action. Bioessays 19, 491–500. 2. Olefsky, J.M. (1990) The insulin receptor a multifunctional pro- tein. Diabetes 39, 1009–1016. 3. White, M.F. & Kahn, C.R. (1994) The insulin signaling system. J. Biol. Chem. 269, 1–4. 4. Cohen, P. (1999) Identification of a protein kinase cascade of major importance in insulin signal transduction. Philos. Trans. Royal Soc. 354, 485–495. 5. Nystro ¨ m, F.H. & Quon, M.J. (1999) Insulin signalling: metabolic pathways and mechanisms for specificity. Cell. Sign. 11, 563–574. Fig. 10. IRS1 localization in the plasma membrane and caveolae by electron microscopy. Plasma membrane sheets attached to grids were incubated with antibodies against IRS1 N-terminus (A,B). Human adipocytes were transfected with the recombinant gene for HA-IRS1. Plasma membrane sheets attached to grids were then incubated with antibodies against the HA-tag (C). After incubation with colloidal gold-derivatized secondary antibodies (15 nm gold particles, arrow- heads) membranes were examined by transmission electron micros- copy. Images contrast inverted. 2478 M. Karlsson et al. (Eur. J. Biochem. 271) Ó FEBS 2004 6. Clifford, G.M., Londos, C., Kraemer, F.B., Vernon, R.G. & Yeaman, S.J. (2000) Translocation of hormone-sensitive lipase and perilipin upon lipolytic stimulation of rat adipocytes. J. Biol. Chem. 275, 5011–5015. 7. Fan, J.Y., Carpentier, J.L., Obberghen, E.V., Grunfeld, C., Gorden, P. & Orci, L. (1983) Morphological changes of the 3T3- L1 fibroblast plasma membrane upon differentiation to the adi- pocyte form. J. Cell Sci. 61, 219–230. 8. Kandror, K.V., Stephens, J.M. & Pilch, P.F. (1995) Expression and compartmentalization of caveolin in adipose cells: coordinate regulation with and strucural segregation from GLUT4. J. Cell Biol. 129, 999–1006. 9. Scherer, P.E., Lisanti, M.P., Baldini, G., Sargiacomo, M., Corley Mastick, C. & Lodish, H.F. (1994) Induction of caveolin during adipogenesis and association of GLUT4 with caveolin-rich vesi- cles. J. Cell Biol. 127, 1233–1243. 10. Razani, B., Woodman, S.E. & Lisanti, M.P. (2002) Caveolae: from cell biology to animal physiology. Pharmacol. Rev. 54, 431–467. 11. Smart, E.J., Graf, G.A., McNiven, M.A., Sessa, W.C., Engelman, J.A., Scherer, P.E., Okamoto, T. & Lisanti, M.P. (1999) Caveolins, liquid-ordered domains and signal transduction. Mol. Cell. Biol. 19, 7289–7304. 12. Anderson, R.G.W. (1998) The caveolae membrane system. Annu. Rev. Biochem. 67, 199–225. 13. Kurzchalia, T.V. & Parton, R.G. (1999) Membrane microdo- mains and cavolae. Curr. Opin. Cell. Biol. 11, 424–431. 14. Simons, K & Ikonen, E. (1997) Functional rafts in cell mem- branes. Nature 387, 569–572. 15. Cohen, A.W., Combs, T.P., Scherer, P.E. & Lisanti, M.P. (2003) Role of caveolin and caveolae in insulin signaling and diabetes. Am. J. Physiol. 285, E1151–E1160. 16. Stra ˚ lfors, P. (1997) Insulin second messengers. Bioessays 19, 327–335. 17. Gustavsson, J., Parpal, S., Karlsson, M., Ramsing, C., Thorn, H., Borg, M., Lindroth, M., Peterson, K.H., Magnusson, K.E. & Stra ˚ lfors, P. (1999) Localisation of the insulin receptor in caveolae of adipocyte plasma membrane. FASEB J. 13, 1961–1971. 18.Anai,M.,Ono,H.,Funaki,M.,Fukushima,Y.,Inukai,K., Ogihara, T., Sakoda, H., Onishi, Y., Yazaki, Y., Kikuchi, M., Oka, Y. & Asano, T. (1998) Different subcellular distribution and regulation of expression of insulin receptor substrate (IRS) -3 from those of IRS-1 and IRS-2. J. Biol. Chem. 273, 29686–29692. 19.Inoue,G.,Cheatham,B.,Emkey,R.&Kahn,C.R.(1998) Dynamics of insulin signaling in 3T3-L1 adipocytes. Differential compartmentalization and trafficking of insulin receptor substrate (IRS)-1 and IRS-2. J. Biol. Chem. 273, 11548–11555. 20. Clark, S.F., Martin, S., Carozzi, A.J., Hill, M.M. & James, D.E. (1998) Intracellular localization of phosphatidylinositide 3-kinase and insulin receptor substrate-1 in adipocytes: Potential involve- ment of a membrane skeleton. J. Cell Biol. 140, 1211–1225. 21. Kriaucinas, K.M., Meyers, M.G. & Kahn, C.R. (2000) Cellular compartmentalization in insulin action: altered signaling by a lipid-modified IRS-1. Mol. Cell. Biol. 20, 6849–6859. 22. Parpal, S., Karlsson, M., Thorn, H. & Stra ˚ lfors, P. (2001) Cho- lesterol depletion disrupts caveolae and insulin receptor signaling for metabolic control via IRS-1, but not for MAP-kinase control. J. Biol. Chem. 276, 9670–9678. 23. Gustavsson, J., Parpal, S. & Stra ˚ lfors, P. (1996) Insulin-stimulated glucose uptake involves the transition of glucose transporters to a caveolae-rich fraction within the plasma membrane: implications for type II diabetes. Mol. Med. 2, 367–372. 24. Karlsson, M., Thorn, H., Parpal, S., Stra ˚ lfors, P. & Gustavsson, J. (2002) Insulin induces translocation of glucose transporter GLUT4 to plasma membrane caveolae in adipocytes. FASEB J. 16, 249–251. 25. Baumann, C.A., Ribon, V., Kanzaki, M., Thurmond, D.C., Mora, S., Shigematsu, S., Bickel, P.E., Pessin, J.E. & Saltiel, A.R. (2000) CAP defines a second signalling pathway required for insulin-stimulated glucose transport. Nature 407, 202–207. 26. Stra ˚ lfors, P. & Honnor, R.C. (1989) Insulin-induced dephospho- rylation of hormone-sensitive lipase. Correlation with lipolysis and cAMP-dependent protein kinase activity. Eur. J. Biochem. 182, 379–385. 27. Thorn, H., Stenkula, K.G., Karlsson, M., O ¨ rtegren, U., Nystrom, F.H., Gustavsson, J. & Stra ˚ lfors, P. (2003) Cell surface orifices of caveolae and localization of caveolin to the necks of caveolae in adipocytes. Mol. Biol. Cell. 14, 3967–3976. 28. Voldstedlund, M., Tranum-Jensen, J. & Vinten, J. (1993) Quan- tification of Na+/K+-ATPase and glucose transporter isoforms in rat adipocyte plasma membrane by immunogold labeling. J. Membr. Biol. 136, 63–73. 29. Lindroth, M., Fredriksson, B A. & Bell, P.B. (1991) Cryosput- tering – a combined freeze-drying and sputtering method for high-resolution electron microscopy. J. Microsc. 161, 229–239. 30. Robinson, L.J., Pang, S., Harris, D.S., Heuser, J. & James, D.E. (1992) Translocation of the glucose transporter (GLUT4) to the cell surface in permeabilized 3T3-L1 adipocytes: effects of ATP, insulin, and GTPcS and localization of GLUT4 to clathrin lattices. J. Cell Biol. 117, 1181–1196. 31. McKeel, D.W. & Jarett, L. (1970) Preparation and characteriza- tion of a plasma membrane fraction from isolated fat cells. J. Cell Biol. 44, 417–432. 32. Oka, Y. & Czech, M.P. (1984) Photoaffinity labeling of insulin- sensitive hexose transporters in intact rat adipocytes. J. Biol. Chem. 259, 8125–8133. 33. Song, K.S., Li, S., Okamoto, T., Quilliam, L.A., Sargiacomo, M. & Lisanti, M.P. (1996) Co–purification and direct interaction of ras with caveolin, an integral membrane protein of caveolae microdomains. Detergent-free purification of caveolae mem- branes. J. Biol. Chem. 271, 9690–9697. 34. Rabilloud, T., Blisnick, T., Heller, M., Luche, S., Aebersold, R., Lunardi, J. & Braun-Breton, C. (1999) Analysis of membrane proteins by two-dimensional electrophoresis: Comparison of the proteins extracted from normal or Plasmodium falciparum- infected erythrocyte ghosts. Electrophoresis 20, 3603–3610. 35. Heider, J.G. & Boyett, R.L. (1978) The picomole determination of free and total cholesterol in cells in culture. J. Lipid Res. 19, 515–518. 36. Frost, S.C., Kohanski, R.A. & Lane, M.D. (1987) Effect of phenylarsine oxide on insulin-dependent protein phosphorylation and glucose transport in 3T3-L1 adipocytes. J. Biol. Chem. 262, 9872–9876. 37. Alemany, S., Mato, J.M. & Stra ˚ lfors, P. (1987) Phospho- dephospho-control by insulin is mimicked by a phospho-oligo- saccharide in adipocytes. Nature 330, 77–79. 38. Marceau, F., Sabourin, T., Houle, S., Fortin, J.P., Petitclerc, E., Molinaro, G. & Adam, A. (2002) Kinin receptors: functional as- pects. Int. Immunopharmacol. 2, 1729–1739. 39. Ostrom, R.S., Gregorian, C., Drenan, R.M., Xiang, Y., Regan, J.W. & Insel, P.A. (2001) Receptor number and caveolar co- localization determined receptor coupling efficiency to adenylyl cyclase. J. Biol. Chem. 276, 42063–42069. 40. Couet, J., Li, S., Okamoto, T., Ikezu, T. & Lisanti, M.P. (1997) Identification of peptide and protein ligands for the caveolin-scaffolding domain. Implications for the interaction of caveolin with caveolae-associated proteins. J. Biol. Chem. 272, 6525–6533. Ó FEBS 2004 Insulin receptor and IRS1 in caveolae (Eur. J. Biochem. 271) 2479 . Colocalization of insulin receptor and insulin receptor substrate-1 to caveolae in primary human adipocytes Cholesterol depletion blocks insulin signalling. the insulin receptor from phosphorylating IRS1. On the other hand, insulin signalling for mitogenic control via Map-kinases ERK1 and 2 was not impaired in