Endocytosis of the Glucose Transporter GLUT4 Is Mediated by the GTPase Dynamin*

To study the role of the GTPase dynamin in GLUT4 intracellular recycling, we have overexpressed dynamin−1 wild type and a GTPase-negative mutant (K44A) in primary rat adipose cells. Transfection was accomplished by electroporation using an hemagglutinin (HA)-tagged GLUT4 as a reporter protein. In cells expressing HA-GLUT4 alone, insulin results in an ≈7-fold increase in cell surface anti-HA antibody binding. Studies with wortmannin indicate that the kinetics of HA-GLUT4-trafficking parallel those of the native GLUT4 and in addition, that newly synthesized HA-GLUT4 goes to the plasma membrane before being sorted into the insulin-responsive compartments. Short term (4 h) coexpression of dynamin-K44A and HA-GLUT4 increases the amount of cell surface HA-GLUT4 in both the basal and insulin-stimulated states. Under conditions of maximal expression of dynamin-K44A (24 h), most or all of the intracellular HA-GLUT4 appears to be present on the cell surface in the basal state, and insulin has no further effect. Measurements of the kinetics of HA-GLUT4 endocytosis show that dynamin-K44A blocks internalization of the glucose transporters. In contrast, expression of dynamin wild type decreases the amount of cell surface HA-GLUT4 in both the basal and insulin-stimulated states. These data demonstrate that the endocytosis of GLUT4 is largely mediated by processes which require dynamin.

In adipose cells, GLUT4 glucose transporters are constantly recycling between an intracellular compartment and the plasma membrane (1)(2)(3)(4). In the basal state, where the rate of exocytosis is relatively low, the vast majority of the GLUT4 glucose transporters reside in an as yet poorly characterized intracellular compartment (1,3,5). Stimulation of adipose cells with insulin leads to an increase in the rate of exocytosis of GLUT4-containing vesicles, resulting in a rapid shift in the steady state distribution of GLUT4 to the plasma membrane (1)(2)(3). After clearance of the hormone, the rate of GLUT4 exocytosis decreases and the steady state distribution shifts back to the intracellular compartment.
The primary focus of recent investigations has been the identification and characterization of signaling molecules (e.g. p85/ p110 phosphatidylinositol 3-kinase) and other cellular components (e.g. soluble NSF attachment protein receptors (SNAREs)) possibly involved in the regulated exocytosis of GLUT4 (6 -11). However, little is known about the mechanism of GLUT4 endocytosis. Previous reports provided indirect evidence that GLUT4 might be internalized by a mechanism involving clathrin-mediated endocytosis. Potassium depletion, known to disrupt formation of clathrin-coated vesicles (12), results in a decreased internalization of GLUT4 and mannose-6-phosphate receptors in rat adipose cells (13). In 3T3-L1 adipocytes, GLUT4 has been shown to co-purify with clathrin-coated vesicles derived from the plasma membrane after treatment of the cells with the fungal toxin brefeldin A (14). Previous morphological analysis showed association of GLUT4 with clathrin-coated pits (4), whereas little colocalization of GLUT4 with clathrin is observed in a recent study from our laboratory (5). Since no functional studies involving components of clathrin-mediated endocytosis in insulin target cells have been reported, the mechanism of GLUT4 internalization still remains unclear.
The dynamins belong to a family of 100-kDa GTPases that mediate the initial stages of endocytosis (15)(16)(17)(18). To date, three mammalian dynamin genes (referred as dynaminϪ1 to dy-naminϪ3) have been described (15, 19 -21). DynaminϪ1 is found in neurons, dynaminϪ2 is expressed ubiquitously, and dynaminϪ3 is enriched in testis (15, 19 -21). Although the distinct functions of these different dynamin proteins are not fully understood, considerable evidence now indicates that dy-naminϪ1 participates in clathrin-mediated endocytosis. Dy-naminϪ1 colocalizes with clathrin in intact cells on the light and electron microscopy levels (22,23), and binds ␣-adaptin, a component of clathrin-coated pits, in vitro (24). Furthermore, transfection of cultured mammalian cells with dominant-negative dynaminϪ1 mutants results in the accumulation of clathrin-coated pits at the plasma membrane (18,22), whereas internalization of transferrin receptors, epidermal growth factor receptors, and ␤2-adrenergic receptors is inhibited (17,18,22,25). Even though a role of dynaminϪ2 in endocytosis remains unclear, a recent report suggests that dynaminϪ2 might also be localized to coated pits on the plasma membrane (26). Thus, considering the high degree of amino acid sequence homology between dynaminϪ1 and dynaminϪ2 and the ability of dynaminϪ1 mutants to inhibit receptor-mediated endocytosis even in nonneuronal cells, both isoforms might act as functional homologues in endocytosis in nonneuronal cells (17,18,22,25). However, the finding that dynaminϪ2 localizes to vesicles in the Golgi complex (27)(28)(29) implies additional functions of this isoform as well. To characterize the mechanism of GLUT4 endocytosis, we have overexpressed a dominant-negative mutant of dynaminϪ1 in isolated rat adipose cells. The effects of dynaminϪ1 on GLUT4-trafficking in vivo were monitored by utilizing a co-transfected recombinant GLUT4 containing an HA 1 epitope tag in the first exofacial loop (30).

EXPERIMENTAL PROCEDURES
Plasmid Constructs-All constructs were generated in the pCIS2 mammalian expression vector (a generous gift from Dr. C. Gorman). cDNAs for HA epitope-tagged dominant-negative K44A dynaminϪ1 and HA-tagged wild-type dynaminϪ1 (a generous gift from Drs. H. Damke and S. L. Schmid) were subcloned into the expression vector. Construction of the HA-tagged GLUT4 has been described previously (30). For transfection experiments, the plasmids were purified in mg quantities using a maxiprep kit (Qiagen).
Cell Culture and Transfection of Rat Adipose Cells-Preparation of isolated rat epididymal adipose cells from male rats (CD strain, Charles River Breeding Laboratories, Inc.) was performed as described previously (31). Isolated cells were washed twice with Dulbecco's modified Eagle's medium containing 25 mM glucose, 25 mM HEPES, 4 mM Lglutamine, 200 nM (Ϫ)-N 6 -(2-phenylisopropyl)-adenosine, and 75 g/ml gentamycin, and resuspended to a cytocrit of 40% (Ϸ5-6 ϫ 10 6 cells/ml). 200 l of the cell suspension were added to 200 l of Dulbecco's modified Eagle's medium containing 100 g of carrier DNA (sheared herring sperm DNA; Boehringer Mannheim) and expression plasmids as indicated. The total concentration of plasmid DNA in each cuvette was adjusted to 5 g/cuvette with empty pCIS2. Electroporation was carried out in 0.4-cm gap-width cuvettes (Bio-Rad) using a T810 square wave pulse generator (BTX). After applying three pulses (12 ms, 200 V), the cells were washed once in Dulbecco's modified Eagle's medium, pooled in groups of 4 -10 cuvettes, and cultured at 37°C, 5% CO 2 in Dulbecco's modified Eagle's medium containing 3.5% bovine serum albumin.
Cell Surface Antibody Binding Assay-Rat adipose cells were harvested 3.5 or 20 -24 h post-transfection and washed in Krebs-Ringer bicarbonate HEPES buffer, pH 7.4, 200 nM adenosine (KRBH buffer) containing 5% bovine serum albumin. Samples corresponding to the cells from one cuvette were distributed into 1.5-ml microcentrifuge tubes. After stimulation with 67 nM (1 ϫ 10 4 microunits/ml) insulin for 30 min at 37°C, subcellular trafficking of GLUT4 was stopped by the addition of 2 mM KCN (32). All of the following steps were performed at room temperature. A monoclonal anti-HA antibody (HA.11, Berkeley Antibody Co.) was added at a dilution of 1:1000, and the cells were incubated for 1 h. Excess antibody was removed by washing the cells three times with KRBH, 5% bovine serum albumin. Then 0.1 Ci of 125 I-sheep anti-mouse antibody (Amersham Pharmacia Biotech) was added to each reaction, and the cells were incubated for 1 h. Finally, the cells were spun through dinonylphtalate oil to remove the unbound antibody (31), and the cell surface-associated radioactivity was counted in a ␥-counter. The resulting counts were normalized to the lipid weight of the samples (31). Unless stated otherwise, the values obtained for pCIS-transfected cells were subtracted from all other values to correct for nonspecific antibody binding. Antibody binding assays were performed in duplicate or quadruplicate.

RESULTS
Expression of HA-GLUT4 -To increase the insulin response of rat adipose cells transfected with epitope-tagged GLUT4 above that observed with the original technique (30), we tested the experimental procedure as follows. Rat adipose cells were transfected with various amounts of HA-GLUT4 expression plasmid and analyzed at different time points for basal and insulin-stimulated cell surface HA-GLUT4 using the anti-HA antibody binding assay. The tagged glucose transporters become detectable at the cell surface in response to insulin as early as 2 h post-transfection. Synthesis of HA-GLUT4 continues until about 16 h post-transfection, at which time its level stays relatively constant and maximal to 24 h post-transfection (data not shown). As judged by immunohistochemistry using the same monoclonal anti-HA antibody and nonpermeabilized, insulin-stimulated cells, the electroporation procedure yields Ϸ10% HA-positive cells (Ref. 33 and data not shown).
As shown in Fig. 1, after culturing the transfected rat adipose cells for 3.5 h, acute (30 min) insulin typically resulted in a 7-10-fold increase in cell surface anti-HA antibody binding after correction for transfection with the empty vector alone (Fig. 1A). Prolonged expression of the tagged GLUT4 (24 h) resulted in higher amounts of HA-GLUT4 at the plasma membrane in both the basal and insulin-stimulated states (Fig. 1B).
Compared with 3.5 h of expression, the amount of insulinstimulated cell surface HA-GLUT4 with 24 h of expression was increased about 4 -5-fold. However, because the -fold increase in basal cell surface HA-GLUT4 from 3.5 to 24 h of expression was greater than that observed in the insulin-stimulated state, insulin stimulation resulted only in a 2-3-fold increase in cell surface glucose transporters. As a compromise between signal strength and insulin response, we selected transfection using 0.5 g of plasmid/cuvette and 3.5 h of protein expression as our standard conditions.
To determine the extent of overexpression of GLUT4, total membrane fractions from adipose cells transfected with 0.1 to 5 g of HA-GLUT4 plasmid/cuvette and cultured for 24 h were analyzed by Western blotting (data not shown). Both endogenous and recombinant GLUT4 were detected using the same polyclonal antibody against the C terminus of GLUT4. Quantitation of the blots shows that the amount of GLUT4 was increased by 1.1-2-fold compared with pCIS-transfected cells. Thus, taking into account that only Ϸ10% of the cells are transfected (33), overexpression of GLUT4 in transfected cells ranged from Ͻ2to 10-fold of endogenous GLUT4 levels, depending on the amount of expression plasmid used.
Coexpression of Dynamins-To study the effects of dynamin overexpression on the subcellular distribution of epitopetagged GLUT4, rat adipose cells were co-transfected with HA-GLUT4 and various concentrations of expression plasmids for wild-type and mutant dynamin. Fig. 2 illustrates the data when protein expression is carried out for 3.5 and 20 h. cell surface level of HA-GLUT4 was increased as much as 2.6 Ϯ 0.1-fold (mean Ϯ S.D.) compared with cells transfected with HA-GLUT4 alone ( Fig. 2A). Likewise, a concomitant increase in cell surface HA-GLUT4 in the insulin-stimulated state was observed. (Fig. 2A). In contrast, expression of wild-type dynamin decreased the amount of cell surface HA-GLUT4 in both the basal and insulin-stimulated states to 51 Ϯ 4 and 58 Ϯ 13% that of the controls, respectively. After 20 h of expression of the dynamin mutant, the basal level of cell surface HA-GLUT4 equaled that for the insulin-stimulated state (Fig. 2B). In addition, the absolute amount of cell surface HA-GLUT4 was increased by about 30 -40% (37 Ϯ 1% and 32 Ϯ 12% in the absence and presence of insulin, respectively) compared with the insulin-stimulated control. Prolonged overexpression (20 h) of the wild-type dynamin decreased the amount of cell surface HA-GLUT4 in the insulin-stimulated state compared with both the control cells and the cells transfected with mutant dynamin and decreased the basal level compared with cells transfected with mutant dynamin (Fig. 2B), as shown for 3.5 h of expression ( Fig. 2A).
Time Course of HA-GLUT4 Internalization-To verify that expression of the dynamins affects the endocytosis of GLUT4 in rat adipose cells, we analyzed the redistribution of cell surface HA-GLUT4 in the presence of the fungal metabolite wortmannin (7,34,35). Acting as an inhibitor of the lipid kinase phosphatidylinositol 3-kinase (36), wortmannin blocked the insulinstimulated translocation of GLUT4 from its basal compartment to the plasma membrane (7,35,37). When added simultaneously with insulin, wortmannin (100 nM) also blocked translocation of epitope-tagged HA-GLUT4 to the cell surface in transfected cells (data not shown). Previously, Holman and co-workers (38) and a study from our laboratory (39) showed that this inhibition of GLUT4 translocation does not affect the early steps of GLUT4 endocytosis. Thus, the addition of wortmannin to insulin-stimulated cells inhibited further translocation of GLUT4 from its intracellular compartment to the plasma membrane, thereby allowing measurements of the kinetics of GLUT4 endocytosis. Fig. 3 demonstrates the results of such wortmannin experiments.
In insulin-stimulated rat adipose cells transfected with HA-GLUT4 alone, the amount of recombinant glucose transporters present on the cell surface increased over time (Fig. 3A). A similar time course of expression of luciferase activity was observed when cells were transfected with a pCIS2-luciferase construct (40) under the same conditions (data not shown). Fig.  4 illustrates that the addition of the protein synthesis inhibitor cycloheximide together with insulin prevents the increase of cell surface HA-GLUT4 with a lag time typical of the action of this reagent. A similar inhibitory effect of cycloheximide on cell surface HA-GLUT4 was observed in basal cells (data not shown). Thus, the increase of cell surface HA-GLUT4 with time apparently reflects the ongoing synthesis of recombinant glucose transporters during the assay. Nonetheless, addition of wortmannin to insulin-stimulated cells led to a decrease in cell surface HA-GLUT4 with a time course similar to that observed with native GLUT4 as previously reported (Fig. 3B) (1). Insulin-stimulated cells transfected with wild-type dynamin showed a time course of HA-GLUT4 clearance from the plasma membrane after wortmannin treatment that was similar to that observed in cells transfected with HA-GLUT4 alone (Fig.  3D). However, cell surface HA-GLUT4 levels in dynamin-K44A cells were not decreased by the addition of wortmannin after insulin but continued to increase in the presence of the inhibitor (Fig. 3C). Fig. 3E shows the time course data corrected for the synthesis of HA-GLUT4 during the assay. The addition of wortmannin to insulin-stimulated cells expressing dynamin-K44A did not change the cell surface level of the tagged glucose transporters; thus, all of the HA-GLUT4 appears to remain on the cell surface.
Targeting of Newly Synthesized HA-GLUT4 -Evidently the protein synthesis-associated increase in cell surface glucose transporters in rat adipose cells utilizes a wortmannin-insensitive trafficking pathway. As shown in Fig. 3C, newly synthesized HA-GLUT4 still appeared on the cell surface in the presence of mutant dynamin and wortmannin. Under these conditions, the endocytosis of GLUT4 was inhibited by the dynamin mutant, leading to an accumulation of glucose transporters in the plasma membrane (Fig. 3E). Likewise, the translocation of GLUT4 from the intracellular pool to the plasma membrane was inhibited by wortmannin. To further investigate the site at which the newly synthesized glucose transporters enter their recycling compartments, we studied the effects of wortmannin on the cell surface level of HA-GLUT4 under basal conditions. To increase the antibody binding signal, the incubation time with wortmannin was extended to 2 h. The results are illustrated in Fig. 5. In HA-GLUT4-transfected basal rat adipose cells, the amount of recombinant glucose transporter present on the cell surface increased 2.5-fold over 2 h of cell culture (Fig. 5A). Similarly, a 3.5-fold increase in basal HA-GLUT4 levels was observed in dynamin-K44A-transfected cells (Fig. 5B). In insulin-stimulated cells expressing HA-GLUT4 only, cell surface glucose transporters doubled within 2 h of incubation (Fig. 5C). The addition of wortmannin to the latter cells leads to a decrease in cell surface HA-GLUT4 as described before (Fig. 3). In contrast, the addition of wortmannin to basal cells did not affect the protein synthesis-dependent increase in cell surface HA-GLUT4 either in the absence or presence of dynamin-K44A expression (Fig. 5, A and B, respectively). Thus, whereas wortmannin blocked GLUT4 translocation from the intracellular compartment to the plasma membrane, it had no effect on the observed protein synthesis-associated increase in cell surface glucose transporters. DISCUSSION To study the subcellular trafficking of GLUT4 in an insulin target cell, we have transfected rat adipose cells with a recom-binant glucose transporter containing an HA epitope tag in the first exofacial loop (30). The HA-GLUT4 was detected on the cell surface of transfected cells by the binding of an antibody against the HA epitope. The observed insulin response of HA- GLUT4 translocation to the plasma membrane is markedly reduced when the amount of expression plasmid is increased and/or the time period of protein expression is extended (the latter reflecting the experimental conditions as described in the original protocol; cf. Ref. 30). Thus, the magnitude of the insulin response is a function of the total amount of GLUT4 present in the cells, suggesting a saturation of the GLUT4 sorting and trafficking system. Likewise, an increase in basal cell surface GLUT4 levels in muscle and adipose cells was also observed in transgenic mice overexpressing GLUT4 (41). However, it remains unclear which component(s) of the GLUT4 trafficking system is (are) saturated by an excess of cellular GLUT4. It could be speculated that at least two trafficking steps might be affected by an excess of recombinant glucose transporters: (i) a hypothetical retention mechanism, responsible for the relatively low cell surface levels of GLUT4 in the basal state, and/or (ii) a sorting mechanism, responsible for the segregation of GLUT4 in the insulin-sensitive compartment. In the first scenario, a disproportionally greater amount of GLUT4 would be directed toward the plasma membrane in the basal state, as the expression of GLUT4 increased. The latter scenario predicts a disproportionally smaller amount of GLUT4 on the cell surface in the insulin-stimulated state with increasing GLUT4 expression. However, the molecular mechanisms of GLUT4 sorting and trafficking are as yet too poorly understood to distinguish between these two possibilities.
A possible involvement of the GTPase dynamin in GLUT4 trafficking has been studied by co-transfecting rat adipose cells with HA-GLUT4 and either wild-type dynaminϪ1 or a GTP binding-deficient dynaminϪ1 mutant (K44A). A plasmid ratio (HA-GLUT4:dynamin) of 1:3 to 1:9 was used to ensure an efficient coupling of the co-transfected plasmids. Expression of both low amounts (3.5 h of protein expression) and high amounts (24 h of protein expression) of recombinant wild-type dynamin decreases cell surface HA-GLUT4 in both the basal and insulin-stimulated states (Fig. 2). This suggests that the endogenous dynamin activity is rate-limiting for endocytosis in rat adipose cells. It could be argued that the endogenous dynamin activity in transfected cells is saturated by an excess of recombinant GLUT4, leading to an increased HA-GLUT4 internalization in wild-type dynamin-transfected cells. Although we cannot exclude this possibility, it appears rather unlikely since 3.5 h of protein expression using 0.5 g of plasmid/cuvette led to only a Ͻ2-fold increase in total cellular GLUT4 level/ transfected cell as judged by Western blots using an anti-GLUT4 antibody and correcting for transfection efficiency (data not shown).
In contrast, high levels of expression of mutant dynamin led to a dramatic increase in basal cell surface glucose transporters (Fig. 2B), whereas intermediate levels of expression led to partial effects ( Fig. 2A). Moreover, the absolute amount of cell surface HA-GLUT4 was increased by about 30 -40% compared with the insulin-stimulated control. This latter increase is in accord with a previous observation that insulin stimulation leads to the net steady state translocation of only Ϸ50% that of the total cellular GLUT4 to the plasma membrane, whereas Ϸ50% of the glucose transporters remains inside the cell, even in the continuous presence of insulin (1,3). Thus, high expression levels of the mutant dynamin lead to an effective accumulation of glucose transporters on the plasma membrane where most or all of the cellular HA-GLUT4 is present on the cell surface even in the basal state. Measurements of the kinetics of HA-GLUT4 endocytosis were not sufficiently sensitive to demonstrate an increased endocytosis with overexpression of wildtype dynaminϪ1. However, expression of dynamin-K44A even at intermediate levels clearly shows that this dynamin mutant effectively blocks internalization of the glucose transporters. These results provide strong evidence that functional dynamin is required for the internalization of the GLUT4.
During preparation of this manuscript, Omata et al. (42) reported the effects of overexpression of dynaminϪ1 wild type and a dynaminϪ1-K44E mutant in Chinese hamster ovary cells co-transfected with insulin receptors and GLUT4. The subcellular distribution of GLUT4 was analyzed nonquantitatively by immunofluorescence microscopy. Consistent with our findings, expression of dynaminϪ1-K44E leads to an increase in cell surface GLUT4 compared with dynaminϪ1 wild type in cells not exposed to insulin. Likewise, wortmannin fails to decrease cell surface GLUT4 in insulin-stimulated K44E-transfected cells. However, in contrast to the present study, dynamin wild type increases the cell surface levels of GLUT4 in insulinstimulated cells. Despite this apparently inconsistent result, the authors did not investigate this effect further nor did they offer any explanation. Hence, it is difficult to interpret these data based on the information provided. So far, no comparable effect of dynamin on the cell surface levels of any membrane protein has been reported. Thus, we conclude that this observed phenomenon appears more likely to represent an artifact originating from this particular expression system than the physiological role of dynamin in the cell. After all, it has yet to be established that the minimal insulin effect on glucose transporter subcellular trafficking in heterologous expression systems such as Chinese hamster ovary cells is in any way related to the dramatic effects of insulin on GLUT4 in insulin target cells.
Early morphological studies in brown adipose tissue have revealed the presence of GLUT4 in clathrin-coated vesicles as well as in noncoated vesicles (3,4). A study in our laboratory does not show an extensive co-localization of GLUT4 with clathrin (5), but due to the dynamics of GLUT4 trafficking, a large fraction of GLUT4 would not be expected to be associated with clathrin-coated vesicles. On the other hand, recent reports present evidence for a clathrin-dependent step in the endocytosis of GLUT4 on the basis of biochemical methods. Potassium (K ϩ ) depletion, known to inhibit the formation of clathrincoated vesicles, inhibits GLUT4 endocytosis in rat adipose cells (13). However, inhibition of endocytosis of the insulin receptor, thought to be internalized by a clathrin-mediated process (43)(44)(45) is not observed with K ϩ depletion, despite an apparently enhanced insulin sensitivity of GLUT4 translocation to the plasma membrane (13). These data suggest that the effects of K ϩ depletion cannot be attributed only to an inhibition of clathrin-coated pit formation.
GLUT4 glucose transporters are reported to co-purify with clathrin from brefeldin A-treated 3T3-L1 adipocytes (14) using differential centrifugation of a Triton X-100-insoluble cell fraction. Nevertheless, VIP21-caveolin (46), the main constituent of caveolae (review in Ref. 47), is also isolated by its detergent insolubility (48). Caveolae are highly abundant in rat adipose cells and 3T3-L1 adipocytes (49,50), and an involvement of caveolae in GLUT4 trafficking is still debated (50 -52). In addition, the study by Corvera and co-workers (14) reveals a 40 -80% decrease in glucose transport activity in both the basal and insulin-stimulated states after brefeldin A treatment, but others (53,54) report no effects at all. Several recent studies demonstrate that expression of dominant-negative dynaminϪ1 inhibits the internalization of transferrin receptors, epidermal growth factor receptors, and ␤2-adrenergic receptors in a variety of cultured mammalian cells (17,18,22,25). Since these receptors are known to be internalized by a clathrin-mediated mechanism, it is possible that GLUT4 are also internalized by a clathrin-dependent pathway. However, it has also been reported that the dynaminϪ1 homologue shibire may participate in clathrin-independent endocytosis in Drosophila melanogaster (55,56). The precise roles of the dynamins in coated pit function and endocytosis via nonclathrin-coated vesicles are unknown. Thus, the sensitivity of GLUT4 endocytosis to a dominant-negative mutant dynaminϪ1 does not explicitly favor either of the two pathways.
In view of the well established effect of wortmannin to block the insulin-stimulated exocytosis of GLUT4 (Figs. 3B and 5C), we were surprised to observe that addition of wortmannin to insulin-stimulated, dynamin-K44A-expressing adipose cells synthesizing HA-GLUT4 did not change the amount of cell surface glucose transporters (Fig. 3C). Likewise, wortmannin did not inhibit the increase over time in cell surface HA-GLUT4 in basal cells synthesizing only tagged glucose transporters (Fig. 5A). Apparently, the pathway involved in targeting the newly synthesized recombinant glucose transporters to the plasma membrane is insensitive to wortmannin and thus different from the pathway involved in the insulin-induced translocation of GLUT4 to the cell surface. Two distinct post-Golgi biosynthetic pathways have been proposed for the delivery of newly synthesized membrane proteins to their endosomal/lysosomal compartments: from the TGN directly and indirectly by way of the plasma membrane. Major late endosomal and lysosomal membrane proteins (Lgps/Lamps) are transported from the TGN directly to endosomes and lysosomes (57). Asialoglycoprotein H1 and transferrin receptors traverse endosomes on their way from the TGN to the cell surface (58,59), whereas the major histocompatibility complex class II molecules enter endosomes via the cell surface in HeLa cells (60). In analogy to the latter pathway, our findings suggest a biosynthetic route of GLUT4 where the glucose transporters are first directed from the TGN to the plasma membrane and then internalized by a dynamin-dependent pathway. Subsequently, GLUT4 are sorted into their intracellular compartment and thereby enter the insulin-sensitive GLUT4 recycling system. Newly synthesized GLUT4 leaving the TGN bypass the insulin-sensitive compartment and are targeted to the cell surface by means of a phosphatidylinositol 3-kinase-independent pathway, making the appearance of newly synthesized GLUT4 on the cell surface insensitive to wortmannin. However, once GLUT4 enters the insulin-sensitive compartment, their targeting to the plasma membrane becomes sensitive to wortmannin.
Taken together, the data show that the GTPase dynamin plays an important functional role in the endocytic portion of the GLUT4 trafficking pathway. Further work is needed to identify the specific components allowing dynamin to interact with the machinery required for GLUT4 internalization.