INTRODUCTION

In adipocytes and in skeletal muscle, insulin causes the tissue-specific glucose transporter isoform, GLUT4, to translocate to the cell surface from an intracellular storage pool (1, 2, 3, 4, 5). The amount of GLUT4 translocated to the plasma membrane in response to insulin corresponds very well to the increased rate of glucose uptake, particularly in fat cells where these two parameters can be most easily measured (4). Therefore, GLUT4 translocation appears to account for the majority if not all of the insulin-dependent increased glucose uptake required for postprandial blood glucose homeostasis. The mechanism of this process has been extensively studied, both for fundamental questions of cell biology and for its possible relevance to diabetes mellitus (for recent reviews, see Refs. 6, 7, 8, 9, 10). A major unresolved question concerning the GLUT4 translocation process concerns the identification and characterization of the membrane compartments utilized by GLUT4 as it cycles to and from the cell surface. The limited information available on this topic comes from immunoelectron microscopy studies in delipidated brown adipose cells (11) as well as in white adipocytes (12). Both studies showed that under basal conditions, most of the intracellular GLUT4 is present in vesicles and short tubules near the cell surface, and insulin administration depletes intracellular GLUT4 by redistributing it to the plasma membrane. Slot et al. (11) also showed that cellular insulin exposure led to some GLUT4 localization in early endosomes as marked by albumin uptake. These studies left open the question as to whether other proteins followed the same trafficking pathway as GLUT4.

Recent work from our lab and elsewhere has begun to address this question. Using cell surface biotinylation of insulin-treated adipocytes followed by immunoadsorption of GLUT4-containing membranes, we identified three glycoproteins (gp)¹ of molecular masses 110, 160, and 230 kilodaltons that appear to cycle to and from the cell surface together with GLUT4 (13). These proteins correspond to major vesicle constituents because they are the most prominent silver-staining bands observed upon immunoisolation of GLUT4-containing membranes. We (14, 15) and others (16) have identified gp160 as an aminopeptidase whose expression, like that of GLUT4, is restricted to fat and muscle (14) and whose distribution and trafficking in adipocytes appears identical to that of the transporter (14, 17).

We show here that gp230, also a major protein component of GLUT4-containing vesicles, is the IGF-II/Man-6-P receptor. The 10-15% of the total IGF-II/Man-6-P receptors that is found co-localized with GLUT4 comprises the entire population of the receptor that cycles to and from the cell surface in response to insulin. In other words, the previously described insulin-dependent translocation of IGF-II/Man-6-P receptor to the cell surface (18, 19) goes exclusively through GLUT4-containing compartments. Taken together with previously published data, we postulate that the glucose transporting machinery, that is, GLUT4-containing vesicles in unstimulated adipocytes, may represent a specialized compartment that accumulates a number of recycling proteins and mediates their translocation to the cell surface in an insulin-sensitive fashion.

EXPERIMENTAL PROCEDURES

In the present study, we used the monoclonal anti-GLUT4 antibody 1F8 (1) and DEAE-cellulose purified anti-IGF-II/Man-6P receptor polyclonal antibodies (a kind gift of Dr. M. Czech, University of Massachusetts Medical School, Worcester, MA).

Adipocytes were isolated from the epididymal fat pads of male Sprague-Dawley rats (200-250 g) by collagenase digestion (20) and transferred to KRP buffer (12.5 mM HEPES, 120 mM NaCl, 6 mM KCl, 1.2 mM MgSO₄, 1 mM CaCl₂, 0.6 mM Na₂HPO₄, 0.4 mM NaH₂PO₄, 2.5 mM D-glucose, 2% bovine serum albumin, pH 7.4). In biotinylation experiments, before use with cells, the buffer was preincubated with 0.4-0.5 mg/ml of sulfo-N-hydroxysuccinimide-acetate (Pierce) for 3-4 h at 37 °C and overnight at 4 °C to block free amino groups of the bovine serum albumin present in the buffer. Insulin was administered to cells (where indicated) to final concentration 10 nM. Sulfo-N-hydroxysuccinimide-biotin (Pierce) was added to cells 2 min. after insulin to final concentration of 0.5 mg/ml. Biotinylation was usually performed for 16-17 min at 37 °C, and then 1 M Tris, pH 7.4, and 0.2 M KCN were added to final concentrations 50 mM and 2 mM, respectively, for 5-15 min. For ¹²⁵I-IGF-II binding, iodinated growth factor (Amersham Corp.) was added to cell suspension at 0.05 µCi/ml alone or together with 0.5 µM of nonradioactive IGF-II (Calbiochem). As in the biotinylation experiments, insulin was added to a final concentration of 10 nM for 18-20 min. After that, cells were washed 3-4 times with HES buffer cooled to 14-16 °C (20 mM HEPES, 250 mM sucrose, 1 mM EDTA, 5 mM benzamidine, 1 mM phenylmethanesulfonyl fluoride, 1 µM pepstatin, 1 µM aprotinin, 1 µM leupeptin, pH 7.4), homogenized with a Potter-Elvehjem Teflon pestle, and subcellular fractions were prepared as described previously (21). Isolated fractions were resuspended in PBS, which contained all of the protease inhibitors listed above.

Protein A-purified 1F8 antibody (1), as well as nonspecific mouse IgG (Sigma), were each coupled to acrylic beads (Reacti-gel GF 2000, Pierce) at a concentration of 0.4 and 0.6 mg of antibody/ml of resin, respectively, according to the manufacturer's instructions. Before usage, the beads were saturated with 2% bovine serum albumin in PBS for at least 1 h and washed with PBS. The light microsomes (LM) from rat adipocytes were incubated separately with each of the specific and nonspecific antibody-coupled beads overnight at 4 °C. The beads were washed three times with PBS, 10 mM Tris, pH 7.4, and the adsorbed material was eluted with 1% Triton X-100 in PBS or Laemmli sample buffer (22) without 2-mercaptoethanol.

Lyophilized anti-IGF-II/Man-6-P receptor antibodies (purified by DEAE-cellulose chromatography) were reconstituted to the volume of the original serum with PBS. Protein fractions in 1% Triton X-100 were supplied with 10-15 µl of reconstituted antibody solution and 100 µl of 50% Protein A-Trisacryl suspension (Pierce). After overnight incubation at 4 °C, beads were washed three times with 1% Triton X-100 in PBS, 10 mM Tris, pH 7.4, and eluted with Laemmli's sample buffer (22) without mercaptoethanol.

Proteins were separated in SDS-polyacrylamide gels according to Laemmli (22) but without reducing agents and were transferred to Immobilon-P membrane (Millipore) in 25 mM Tris, 192 mM glycine. Following transfer, the membrane was blocked with 10% nonfat dry milk in PBS for 2 h at 37 °C. Proteins were visualized with specific antibodies, horseradish peroxidase-conjugated secondary antibodies (Sigma) and an enhanced chemiluminescent substrate kit (DuPont NEN). Autoradiograms were scanned on a densitometer (Molecular Dynamics) for quantitative analysis of Western blots. Biotinylated proteins were stained with 2000-fold diluted streptavidin-alkaline phosphatase conjugate obtained from Boehringer and used according to the manufacturer's instructions.

Protein content was determined with BCA kit (Pierce) according to manufacturer's instructions.

RESULTS

Fig. 1 is a representative Western blot showing the distribution of the IGF-II/Man-6-P receptor in subcellular fractions of rat adipocytes. Under basal conditions (without insulin), the major part (78%) of the receptor is localized in intracellular LM, which are enriched in Golgi and trans-Golgi network markers (21). The intermediate fraction of heavy microsomes, which is enriched in endoplasmic reticulum, and the plasma membrane (PM) have 13 and 9% of the total IGF-II/Man-6-P receptor population, respectively. The combined fractions of mitochondria and nuclei, as well as cytosol, do not contain any detectable amount of the receptor. In agreement with ¹²⁵I-IGF-II binding studies (18, 19, 23), insulin causes a translocation from LM to PM fractions of a small portion of the IGF-II/Man-6-P receptor in such a way that its amount in LM decreases to 67% and increases in PM to 20% of the total receptor content. The amount of this receptor in heavy microsomes does not change in response to insulin. The intracellular distribution of IGF-II/Man-6-P receptors in basal and insulin-treated adipocytes somewhat resembles that of GLUT4 and another marker protein of GLUT4-containing membranes, aminopeptidase, gp160 (14). However, the insulin-dependent translocation of the IGF-II/Man-6-P receptor is of a lesser magnitude than the translocation of these two proteins, which are depleted by 40-50% in intracellular membranes in response to insulin (14).

To visualize the recyclable population of IGF-II/Man-6-P receptors, we used an approach based on biotinylation of the cell surface proteins of rat adipocytes (13). With this technique, recyclable proteins are tagged with biotin groups on the adipocyte surface using a cell impermeable, amino group-specific reagent, sulfo-N-hydroxysuccinimide-biotin, and they are then recovered in the intracellular fraction of light microsomes by specific immunoadsorption. The specific biotinylation of an individual recyclable protein with accessible aminogroups should be proportional to its residence time on the cell surface. Indeed, immunoprecipitation with specific antibodies demonstrates that the specific biotinylation of the IGF-II/Man-6-P receptor present in LM fraction is 5-6-fold higher in insulin-treated than in control cells (Fig. 2). A certain level of basal biotinylation most probably reveals the relatively slow, continuous recycling of the receptor in the absence of insulin, which has also been shown for GLUT4 (5, 24, 25) and gp160 (13).

In our previous studies of the recycling of GLUT4-containing vesicles in rat adipocytes, we noticed that one of their main constituents identified by silver staining, a glycoprotein designated gp230, demonstrated electrophoretic mobility and insulin-dependent behavior close or identical to that of the IGF-II/Man-6-P receptor (13). Fig. 3 (left panel) shows by Western blotting that the IGF-II/Man-6-P receptor is indeed a component protein of GLUT4-containing vesicles, because it is specifically immunoadsorbed with the anti-GLUT4 monoclonal antibody, 1F8, and cannot be detected in the material bound to nonspecific mouse IgG. Fig. 3 (right panel) demonstrates that the IGF-II/Man-6-P receptor recycles in adipocytes in an insulin-dependent fashion as a constituent of GLUT4-containing vesicles, because its biotinylation there, as in total LM fraction, is markedly stimulated by insulin. We have shown earlier that biotinylation of gp230 is abolished by 2 mM KCN (13), and thus this reaction is most probably a consequence of the ATP-dependent translocation of vesicles and their fusion with plasma membrane.

Next, we isolated proteins of GLUT4-containing vesicles by eluting the material immunoadsorbed on 1F8 beads with 1% Triton and then immunoprecipitated the IGF-II/Man-6-P receptor from this preparation. It is shown in Fig. 4 that GLUT4-containing vesicles have several components whose biotinylation is markedly enhanced by insulin (in agreement with the previously published data, see Ref. 13). The most prominent such proteins are gp230 (now identified as IGF-II/Man-6-P receptor), gp160, a new member of the aminopeptidase family (15, 16), and an unidentified glycoprotein, gp110, which in this experiment, migrates as a doublet. Anti-IGF-II/Man-6-P receptor antibodies specifically recognize and immunoprecipitate only the gp230 band from Triton X-100 eluate, and this protein can be quantitatively recovered in the SDS-eluate of the protein A beads used for immunoprecipitation. This result confirms the data from the previous figure that the IGF-II/Man-6-P receptor is a component protein of GLUT4-containing vesicles.

Unlike GLUT4 and gp160, which can be found only in specific vesicles that are quite different from other LM in their physicochemical parameters (26), only a small part of the total population of the IGF-II/Man-6-P receptor present in microsomal fraction is brought down with 1F8 antibody. Under conditions where 90% of the GLUT4 is immunoadsorbed by the beads, 85-90% of IGF-II/Man-6-P receptor remains in the supernatant (Fig. 5). Note that 33% of 1F8-eluate (lanes 5 and 6, panel A) gives a signal for the IGF-II/Man-6-P receptor close to that produced by 3.5% of original LM and the supernatant of immunoadsorption. The fact that the latter two produce a signal of essentially the same intensity once again demonstrates that not more than 10-15% of the total population of IGF-II/Man-6-P receptor is removed by 1F8 beads.

The distribution of the biotinylated receptor between 1F8 supernatant and eluate is completely different, in comparison with that of the total receptor protein, because almost all biotinylated receptor is brought down by the beads. In the experiment illustrated by Fig. 6A, we took 50 µl of 1F8 beads (which contained 20 µg of immobilized 1F8) for immunoadsorbtion of 0.5 mg of LM. Under these conditions, about 75% of GLUT4 is immunoadsorbed (not shown). The small amount of biotinylated IGF-II/Man-6-P receptor found in 1F8 supernatant by immunoprecipitation with anti-receptor antibodies (Fig. 6A, lane 4) most probably originated from the 25% of the GLUT4 vesicles, which were not bound to 1F8 beads. When we took 150 µl of 1F8 beads (60 µg of immobilized 1F8) for the same amount of LM, we were able to immunoadsorb over 90% of GLUT4-containing vesicles (Fig. 5) and also over 90% of the biotinylated receptor (Fig. 6B).

An independent proof for the hypothesis that the IGF-II/Man-6-P receptor is recruited to the plasma membrane from an intracellular pool via GLUT4-containing vesicles comes from the analysis of ligand internalization. We added ¹²⁵I-labeled IGF-II to adipocytes in the absence and in the presence of insulin and monitored its distribution in these cells. We observed that insulin considerably stimulated binding of ¹²⁵I-IGF-II to the plasma membrane and also its internalization, as measured by recovery in the intracellular LM fraction. Both processes are mediated through interaction with specific receptors, because unlabeled IGF-II blocks essentially all binding and internalization of the radioactive ligand (Fig. 7). These data are in a good agreement with previously published results (18, 19, 27, 28, 29).

LM fractions obtained in the previous experiment (Fig. 7) were immunoadsorbed with 1F8 beads taken in an amount sufficient to isolate all GLUT4-containing vesicles. In a parallel control experiment, we used the same volume of nonspecific IgG beads. Under these conditions, 60 ( insulin) to 70% (+ insulin) of the total ¹²⁵I-IGF-II present in LM was recovered in 1F8 adsorbed material, i.e. in GLUT4-containing vesicles. Considering that the latter comprise not more than 2-3% of LM (3, 26, 30), the specific ¹²⁵I-IGF-II content in GLUT4-containing vesicles is about 100-fold higher than in the rest of LM fraction. Nonspecific IgG, as expected, did not bind any radioactivity, all of which was recovered in supernatant (Fig. 8). Routinely, we immunoadsorb not more than 70% of total radioactive IGF-II in LM fraction with 1F8 beads. We think that this is due to the rapid sorting and exclusion of ¹²⁵I-IGF-II from GLUT4-containing vesicles and subsequent movement to lysosomes for digestion (see below). However, we cannot exclude the possibility of an artificial ligand exchange between a small population of receptors associated with GLUT4 vesicles and a 10 times larger receptor population, which is not present in the vesicles. If some of GLUT4-containing vesicles and receptor-containing endosomes are broken or damaged during homogenization, such exchange may occur.

Internalized IGF-II is degraded in adipocytes (27). Interestingly enough, a major protein of GLUT4-containing vesicles, gp160, has strong aminopeptidase N activity (15, 16), which may be responsible for cleavage of this factor in the lumen of the vesicle. However, in vitro, neither gp160 nor any other component protein of GLUT4 vesicles is capable of hydrolyzing of IGF-II (not shown), although its N terminus looks like it could serve as substrate for neutral aminopeptidases (31) and, indeed, can be cleaved by some of these enzymes.²

DISCUSSION

As noted in the introduction, the regulation of glucose homeostasis by insulin is of considerable physiological importance and has been the focus of numerous studies. The mechanism of this regulation is largely or entirely the result of the hormone-dependent movement to and fusion with the plasma membrane of the GLUT4-containing intracellular membrane compartment. GLUT4 translocation is specific to fat and muscle and cannot be reconstituted simply by expressing GLUT4 protein in a variety of mainly fibroblastic cells that possess insulin receptors and other signal transduction pathway components (reviewed in Refs. 6 and 9). Thus, insulin-dependent GLUT4 translocation requires the tissue-specific expression of additional, as yet unidentified, proteins, and an ongoing effort of our laboratory is to identify components of the GLUT4-containing membranes that may be required for their translocation. Here, we show that there are two populations of the IGF-II/Man-6-P receptor in rat adipocytes: one comprising 10-15% of the total that completely co-localizes with GLUT4 in the basal state and that translocates to the cell surface and recycles together with GLUT4 and a second population, which does not significantly exchange with the translocating pool over the time course of our experiments (15-30 min).

It has been shown by ligand binding studies that insulin causes translocation of IGF-II/Man-6-P receptors from an intracellular storage pool to the cell surface without increasing their affinity for IGF-II (18, 19, 32). Also, insulin stimulates internalization and degradation of IGF-II, whereas IGF-II itself does not have this effect (27). Similarly, it has been shown that IGF-II/Man-6-P receptors are internalized and recycle in the absence of IGF-II (28). Upon fractionation in sucrose gradients and by agarose gel electrophoresis, pools of glucose transporters and IGF-II/Man-6-P receptors were shown to partially overlap (30). In addition, major histocompatibility complex class I-derived peptides inhibit internalization of IGF-II/Man-6-P receptors and GLUT4 to the same degree and within the same time frame (33). Our data are completely consistent with this prior work. Also, Tanner and Lienhard (34) have shown earlier that anti-glucose transporter 1 (GLUT1) antibodies immunoadsorb 70-95% of IGF-II/Man-6-P receptor and vice versa. Besides the difference in antibodies used and, therefore, in the membrane fractions obtained, these studies were performed in 3T3-L1 adipocytes where the GLUT1:GLUT4 ratio is 3:1 (35) as opposed to rat fat where it is 1:10-20 (2, 3, 4). Thus, the 3T3-L1 cells are not fully differentiated, mature adipocytes, and they may lack the latter's specialized cycling vesicles that are well segregated from other intracellular vesicular compartments.

So what do our data tell us in a more general sense about GLUT4-containing vesicles? Prior discoveries concerning the composition of these vesicles revealed two classes of membrane protein: a tissue-specific class consisting of GLUT4 and gp160 aminopeptidase, the physiological role of the latter being unclear (14, 15, 16, 17), and a class of proteins involved in several aspects of membrane trafficking. The latter include Scamps (36, 37), a marker for cell-surface-endosome recycling (38), VAMP/cellubrevin (39, 40) involved in formation of membrane fusion complexes at appropriate target sites, small GTP-binding protein(s) such as Rab4 (41) and its possible regulator, GDI-2 (42) whose exact physiological role is also uncertain, and lastly, phosphatidylinositol-4-kinase (43), also of unknown physiological function with respect to GLUT4 trafficking. We can now add a third class of protein, the IGF-II/Man-6-P receptor, whose physiological role may be to insure the clearance of certain proteins from the circulation in the postprandial, insulin-responsive state (44). This latter notion is reinforced by the fact that a population of transferrin receptors (~50%), although not very abundant in adipocytes, also traffic to and from the cell surface in response to insulin and do so coordinately with GLUT4 and the IGF-II/Man-6-P receptor.² The insulin-dependent translocation of the transferrin receptor to the adipocyte cell surface has been previously documented (45). Thus, four proteins likely (the aminopeptidase gp160 and the IGF-II/Man-6-P receptor) or certain (GLUT4 and transferrin receptor) to be involved in cellular nutrition are coordinately translocated to the cell surface in response to insulin.

A second point our data address is the nature and number of intracellular structures that may be involved in GLUT4 trafficking. The fact that GLUT4 traffics together with IGF-II/Man-6-P and transferrin receptors, whereas IGF-II and iron need to be sorted from their respective receptors in an acidifying compartment, suggests that GLUT4 must also pass through such a sorting endosome. Indeed, it has been shown that endocytosed IGF-II is degraded in adipocytes to trichloroacetic acid-soluble products (27), whereas the corresponding receptor along with GLUT4 recycle to the cell surface. On the one hand, this might have been expected on the basis of electron micrographs of brown fat showing the increased presence of GLUT4 in endosomal compartments after cellular insulin exposure (11). On the other hand, electron micrographic studies of white fat (12) show many more discreet vesicles than brown fat, and the physiological rationale for GLUT4 to pass through an acidifying endosome is not necessarily obvious. In any case, our data and interpretation are consistent with kinetic data that demand the existence of more than one intracellular GLUT4-containing compartment in adipocytes (46, 47).

We propose a model describing the turnover of GLUT4 and the IGF-II/Man-6-P receptor in adipocytes (Fig. 9), which is partly based on models of transferrin receptor trafficking in Hep2 cells (48) as well as on synaptic vesicle cycling (49). We suggest that under basal conditions, the bulk of GLUT4, together with the recyclable population of the IGF-II/Man-6-P receptor and some other proteins (not shown), is compartmentalized in ``ready-to-go'' vesicles (position 1). After insulin administration, these vesicles fuse with the plasma membrane and expose component proteins outside the cell where they perform their biological functions, such as glucose transport (GLUT4), cleavage of peptide substrates (gp160), binding of extracellular ligands (receptors for IGF-II/Man-6-P and transferrin), and so on. These proteins are subsequently retracted from the cell surface via a yet unidentified mechanism, possibly clathrin-mediated (50, 51). Anyway, clathrin coats (if any) should be removed immediately after endocytosis (position 2), and at this point, a substantial proportion of GLUT4 is associated with early or sorting endosomes (11). GLUT4 and the IGF-II/Man-6-P receptor are then sorted from IGF-II ligand and regeneration of insulin-responsive (see also Fig. 6 in Ref. 47) GLUT4-containing vesicles occurs. IGF-II is directed to lysosomes via late endosomes, and the bulk of adipocyte IGF-II/Man-6-P receptor targets phosphomannosylated proteins to lysosomes from the trans-Golgi network via this same pathway, thus providing lysosomes with their characteristic enzymes (52, 53, 54).