Insulin accelerates inter-endosomal GLUT4 traffic via phosphatidylinositol 3-kinase and protein kinase B.

Insulin enhances plasmalemmal-directed traffic of glucose transporter-4 (GLUT4), but it is unknown whether insulin regulates GLUT4 traffic through endosomal compartments. In L6 myoblasts expressing Myc-tagged GLUT4, insulin markedly stimulated the rate of GLUT4myc recycling. In myoblasts stimulated with insulin to maximize surface GLUT4myc levels, we followed the rates of surface-labeled GLUT4myc endocytosis and chased its intracellular distribution in space and time using confocal immunofluorescence microscopy. Surface-labeled GLUT4myc internalized rapidly (t(12) 3 min), reaching the early endosome by 2 min and the transferrin receptor-rich, perinuclear recycling endosome by 20 min. Upon re-addition of insulin, the t(12) of GLUT4 disappearance from the plasma membrane was unchanged (3 min), but strikingly, GLUT4myc reached the recycling endosome by 10 and left by 20 min. This effect of insulin was blocked by the phosphatidylinositol 3-kinase inhibitor LY294002 or by transiently transfected dominant-negative phosphatidylinositol 3-kinase and protein kinase B mutants. In contrast, insulin did not alter the rate of arrival of rhodamine-labeled transferrin at the recycling endosome. These results reveal a heretofore unknown effect of insulin to accelerate inter-endosomal travel rates of GLUT4 and identify the recycling endosome as an obligatory stage in insulin-dependent GLUT4 recycling.

Transmembrane proteins are constitutively removed from the cell surface and directed to the early endosome. From here, proteins destined for recycling enter the recycling endosome (1). This pathway is thought to work in a constitutive fashion, but little is known about possible regulation of the individual steps, i.e. direction and speed of sorting out of the early endosome, fusion with the recycling endosome, and exit from this compartment.
The mammalian glucose transporter GLUT4 is an integral membrane protein specific to muscle and fat cells that undergoes both constitutive recycling (2) and insulin-regulated exo-cytosis (3). The importance of this phenomenon is highlighted by evidence that GLUT4 externalization is defective in the pathophysiological state of insulin resistance underlying type 2 diabetes (4 -6). Despite extensive work, the intracellular donor compartments remain poorly defined as do the effects of insulin on the route or the velocity of inter-compartmental GLUT4 traffic. This paucity in knowledge is largely because of shortcomings of biochemical approaches to isolate and characterize the diverse intracellular compartments populated by GLUT4 and by the limited intracellular space available for detailed immunolocalization in primary fat and muscle tissue and in cultured adipocytes.
Here we used L6 myoblasts stably expressing GLUT4 bearing an extracellular Myc epitope (GLUT4myc) to examine the time course and route of GLUT4 intracellular transit. The subcellular distribution of GLUT4myc in L6 muscle cells was previously characterized (7)(8)(9). Ninety percent of this protein was sequestered intracellularly (9), more than half of it in compartments that segregate from the constitutively recycling isoform GLUT1 (8). Insulin mobilizes GLUT4myc to attain a new steady state distribution where 30% is exposed at the cell surface (9). This translocation occurred with a t1 ⁄2 of 3.5 min, resembling the rate of GLUT4 exocytosis in adipose cells (2,10). The movement of GLUT4myc to the surface of L6 myoblasts was prevented by expression of dominant negative mutants of phosphatidylinositol (PI) 1 3-kinase and Akt/PKB (11) and was sensitive to agents that prevent actin remodeling (7,12,13). Insulin-dependent translocation of GLUT4myc differed from basal state recycling of GLUT4myc in its sensitivity to tetanus toxin, suggesting that basal and insulin-stimulated GLUT4myc arrive at the plasma membrane on distinct vesicles (14). Like the endogenous GLUT4 of muscle and adipose cells, GLUT4myc also responds to other stimuli such as hyperosmolarity (9) and dinitrophenol (15).
When partially detached from the substratum, L6 myoblasts round up and thus offer the opportunity to discern with clarity the intracellular compartments populated by GLUT4myc using fluorescence microscopy. Here we take advantage of the exofacial exposure of the Myc epitope to follow the journey of GLUT4myc during its recycling back to the cell surface. After labeling with an anti-Myc antibody the transporter summoned to the cell surface in response to insulin, we analyze its temporal and spatial intracellular coordinates during its internalization in the absence and presence of insulin. Strikingly, we observed that insulin expedites GLUT4 transit into and out of the recycling endosome. Dominant-negative mutants of PI 3-kinase or PKB blocked the insulin-dependent acceleration of GLUT4myc traffic. These results provide evidence for a previously unknown action of insulin, i.e. to speed up inter-endosomal GLUT4 traffic. This increase may be the basis for the acceleration of GLUT4 recycling caused by the hormone.

EXPERIMENTAL PROCEDURES
Materials-The following antibodies were obtained from commercial sources: rabbit (A- 14) and mouse (9E10) antibodies against the c-Myc epitope (Santa Cruz Biotechnology, Santa Cruz, CA), mouse antibodies against human transferrin receptor (TfR) (Zymed Laboratories Inc., San Francisco, CA), fluorescein isothiocyanate, Cy3, and horseradish peroxidase-coupled donkey anti-mouse and anti-rabbit secondary antibodies (Bio/Can Scientific, Mississauga, ON, Canada). 9E10 was also a generous gift of Dr. Mike Moran. Rabbit antibodies against early endosome antigen 1 (EEA1) were a gift of Drs. Heidi McBride and Marino Zerial (EMBL, Heidelberg, Germany). TO-PRO3 and rhodamine-conjugated transferrin were obtained from Molecular Probes (Eugene, OR). DNA encoding the extracellular domain of CD25 and the transmembrane and cytoplasmic domains of furin (16)  Cells and Tissue Culture-A subclone of the L6 rat skeletal muscle cell line stably expressing GLUT4 with an exofacial Myc epitope (L6-GLUT4myc) has been described previously (7-9, 11, 15, 17). For all experimental conditions, cells were serum-deprived for 3 h before treatments. Where indicated, L6-GLUT4myc myoblasts were transiently transfected using the liposome-mediated Effectene reagent (Qiagen, Mississauga, ON, Canada) according to the manufacturer's protocol. Marker DNA (enhanced green fluorescent protein) was transfected in a 1:4 ratio with construct DNA.
Cell Population Assay of GLUT4myc Recycling and Endocytosis-For recycling studies, L6 GLUT4myc myoblasts grown in 24-well tissue culture plates were incubated at 37°C with anti-Myc antibody 9E10 (8 g/ml) in the absence or presence of 100 nM insulin. At the indicated times, cells were washed with ice-cold PBS, fixed with 3% formaldehyde in PBS at 4°C for 20 min, and permeabilized with 0.1% Triton X-100 at room temperature for 30 min. This was followed by incubation in PBS containing 5% goat serum and reaction with horseradish peroxidaseconjugated donkey anti-mouse IgG (1:1,000) at 4°C for 1 h, washing six times with PBS, and incubation with 1 ml/well o-phenylenediamine dihydrochloride (0.4 mg/ml) for 20 -30 min at room temperature. The reaction was stopped by addition of 0.25 ml of 3 N HCl, and the optical absorbance was measured at 492 nm. The total content of GLUT4myc was calculated by permeabilizing the cells using 0.1% Triton X-100 before incubation with 9E10 antibody, as reported earlier (9). The amount of GLUT4myc label at each time point is expressed as the percent of the total GLUT4myc (Fig. 1).
For endocytosis studies, L6-GLUT4myc myoblasts grown in 24-well tissue culture plates were stimulated with 100 nM insulin at 37°C for 30 min. Cells were then rinsed three times with ice-cold PBS and labeled with 9E10 (8 g/ml) for 1 h at 4°C in the absence of insulin. Medium pre-warmed to 37°C was added, and the plates were incubated in the presence or absence of insulin for indicated times. Where indicated, 100 nM wortmannin was added during the 1-h incubation with 9E10 and maintained in 100 nM wortmannin throughout the internalization period. At the indicated times cells were placed on ice, washed once with ice-cold PBS, and fixed briefly with 3% formaldehyde in PBS at 4°C for 10 min. Cells were then incubated with horseradish peroxidase-conjugated donkey anti-mouse IgG (1:1,000) at 4°C for 1 h, followed by washes and reaction with o-phenylenediamine dihydrochloride as described above for GLUT4myc recycling. Cell surface levels of GLUT4myc are expressed as a percentage of the amount of cell surface GLUT4myc on control cells at 0 min of internalization.
Immunofluorescence in Rounded-up Myoblasts-L6-GLUT4myc myoblasts were detached from the substratum using nominally Ca 2ϩand Mg 2ϩ -free PBS. This condition was used instead of trypsin to preserve the GLUT4myc and insulin receptor molecules on the cell surface. Rounded-up (semi-attached) myoblasts offer the opportunity to study topological distribution of intracellular compartments by fluorescence confocal microscopy. Flattened myoblasts (and other cell types) afford considerably less resolution by this approach, especially in dif-ferentiating surface-bound from intracellular epitopes. Total removal of Ca 2ϩ from the extracellular milieu is reported to be inconsequential to GLUT4 translocation (18). After resuspending in HEPES-buffered RPMI, myoblasts were allowed to settle on coverslips for 10 min while treating them with or without 100 nM insulin at 37°C. To determine the overall cellular localization of GLUT4myc in these rounded-up myoblasts, the cells were immediately fixed in ice-cold 4% formaldehyde in PBS containing Ca 2ϩ and Mg 2ϩ for 20 min. The cells were then permeabilized in 0.1% Triton X-100 for 30 min and blocked for 20 min with 5% goat serum. GLUT4 was detected using anti-Myc polyclonal antibody A-14 (1.33 g/ml), and the TfR-containing compartment was detected using monoclonal anti-TfR antibody (1:1,000). These antibodies were added to the cells for 1 h followed by secondary antibodies (1:250) along with the DNA stain TO-PRO3 (2 M). Coverslips were mounted on glass slides and imaged using a Zeiss Axiovert 100M laser scanning confocal microscope 510.
To determine endocytic transit of cell surface-labeled GLUT4myc, rounded-up myoblasts resuspended in HEPES-buffered RPMI were allowed to settle (semi-attach) on coverslips for 10 min at 37°C in the presence or absence of 100 nM insulin. The coverslips were transferred to 4°C to halt all vesicular traffic and then incubated in the absence of insulin for 1 h at 4°C with Hepes-buffered RPMI medium containing anti-Myc antibodies (1.33 g/ml polyclonal A-14 or 4 g/ml of monoclonal 9E10 as indicated). Coverslips were then washed, maintained at 37°C in the absence or presence of 100 nM insulin for 10 or 20 min, and then immersed in ice-cold 4% formaldehyde in PBS containing Ca 2ϩ and Mg 2ϩ for 20 min followed by quenching with ice-cold 50 mM NH 4 Cl for 10 min. The cells were then permeabilized in 0.1% Triton X-100 for 30 min at room temperature and blocked for 20 min with 5% goat serum. Primary antibodies against TfR (1:1,000) or EEA1 (1:250) were added to the cells for 1 h followed by fluorophore-conjugated secondary antibodies (1:250) along with the DNA stain TO-PRO3 (2 M). Coverslips were mounted on glass slides and imaged using a Zeiss Axiovert 100M laser scanning confocal microscope 510. Secondary antibodies used in this study showed no detectable labeling of L6-GLUT4myc cells when incubated in the absence of primary antibodies. Gain settings for the photomultiplier tube were adjusted for each experiment to bring all images within the linear range of the detector. Images of Cy5 or TO-PRO3 labeling were given a false blue color to aid in visualization. More than 100 cells were observed per condition per experiment, and representative images of each condition are shown. The extent of colocalization between GLUT4myc and EEA1 or TfR signals was quantified using NIH Image (NIH, Bethesda, MD) and Adobe Photoshop (Adobe Systems Inc., San Jose, CA). The yellow signal generated by overlap between the pair of fluors was separated from the remaining image using the Select Color Range option in Adobe Photoshop. For panels i-n in Fig. 8 the green (green fluorescent protein) channel of the RGB Photoshop image was replaced with the blue (TfR) channel for purposes of quantitation. The intensity of the yellow signal selected in this way was then expressed as a percentage of the intensity of the entire GLUT4myc signal.
To determine endocytic transit of rhodamine-conjugated transferrin, rounded-up myoblasts resuspended in HEPES-buffered RPMI were allowed to settle (semi-attach) on coverslips while being loaded for 10 min at 37°C with 100 g/ml of rhodamine-transferrin. Coverslips were transferred to HEPES-buffered RPMI devoid of rhodamine-transferrin and maintained at 37°C in the absence or presence of 100 nM insulin for 10 or 20 min, then placed directly in ice-cold 4% formaldehyde in PBS containing Ca 2ϩ and Mg 2ϩ for 20 min followed by quenching with 50 mM NH 4 Cl for 10 min. The cells were then permeabilized in 0.1% Triton X-100 for 30 min at room temperature and blocked for 20 min with 5% goat serum. Primary antibody against TfR (1:1,000) was added to the cells for 1 h followed by secondary goat anti-mouse alexa 488 antibody (1:500) along with the DNA stain TO-PRO3 (2 M). Coverslips were mounted on glass slides and imaged using a Zeiss Axiovert 100M laser scanning confocal microscope 510.
Immunological Detection of Insulin-dependent Signals-L6-GLUT4myc myoblasts grown to confluency in six-well tissue culture plates were serum-starved for 4 h, then incubated without or with 100 nM insulin at 37°C for 30 min. Some wells were then washed with PBS three times, incubated at 4°C for 1 h in the absence of insulin, washed three times again, and analyzed at this point; the other half was incubated for an additional 10 min at 37°C before analysis. All cells were then rinsed twice with PBS containing 1 mM Na 3 VO 4 and lysed with 300 l/well 2ϫ SDS-polyacrylamide gel electrophoresis sample buffer containing 5 M E-64, 1 M leupeptin, 2.5 M pepstatin, 1 mM Na 3 VO 4 , 0.1 mM phenylmethylsulfonyl fluoride, and 100 nM okadaic acid. After scraping the lysates, they were passed 5 times through a 25-guage syringe and incubated at 65°C for 15 min. Once cooled to 20°C, the lysates were resolved by SDS-polyacrylamide gel electrophoresis, transferred to polyvinylidine difluoride membrane, and immunoblotted with antibodies against phosphotyrosine (PY99, Santa Cruz Biotechnology) or phosphothreonine 308 of PKB (phospho-Akt (Thr-308), New England Biolabs, Mississauga, ON).

Rates of Recycling of GLUT4myc in the Absence and Presence
of Insulin-L6-GLUT4myc myoblasts were incubated in the absence or presence of insulin, and any GLUT4myc molecule arriving at the cell surface was labeled with anti-Myc antibody 9E10 during the incubation. In the absence of insulin, halfmaximal recycling occurred after 2 h, whereas insulin sharply increased the rate of GLUT4myc recycling, reducing the halftime to about 40 min (Fig. 1). This result confirms that GLUT4myc behaves as reported for the endogenous GLUT4 in adipose cells (2,19). Moreover, the rate of recycling of GLUT4myc is similar to that determined for a chimeric protein encoding the exofacial portion of the TfR and the cytosolic tail of the insulin-regulated aminopeptidase vpTR (or IRAP) (20,21). Fig. 1 also shows that virtually all the cellular GLUT4myc molecules were available for surface labeling within 6 h, establishing that the antibody-labeled transporter is able to recycle without undergoing degradation. Insulin accelerated this recycling so that all GLUT4myc molecules were labeled within 3 h.
Rate of Disappearance of GLUT4myc from the Cell Surface-To measure the rate of removal of GLUT4myc from the cell surface, L6-GLUT4myc myoblasts were stimulated with insulin for 30 min to expose GLUT4myc at the cell surface. This treatment consistently increases cell-surface GLUT4myc by 2-3-fold in these cells (7,8,11). Such an increase is consistent with the change observed in isolated skeletal muscles (22,23). The exofacially exposed Myc epitope was then labeled at 4°C with anti-Myc antibody for 1 h and then allowed to internalize upon warming to 37°C. Insulin and/or wortmannin were added during the internalization period to examine their effects on disappearance of GLUT4myc from the cell surface. Fig. 2 shows that labeled GLUT4myc disappeared from the cell surface with a t1 ⁄2 of 3 min for all treatments, fitting a one-phase exponential decay curve (r 2 values: control, 0.970; insulin, 0.928; wortmannin, 0.999; insulin plus wortmannin, 0.999). Tukey's pairwise analysis of variance of all values at each time point indicated that only the 30-min insulin re-addition and 30-min insulin re-addition plus wortmannin points were significantly different from one another (p Ͻ 0.05). These results suggest that there is no evident effect of insulin on the initial rate of removal of GLUT4myc from the cell surface. The separation of the insulintreated curve from the curves in other conditions at longer time points is likely the result of the enhanced recycling of labeled transporters back to the cell surface.
Insulin Signaling Is Significantly Reduced after Insulin Removal and Incubation for 1 h at 4°C-The lack of effect of insulin on GLUT4myc removal from the cell surface raised the question of whether insulin signals are maintained during incubation of myoblasts for 1 h at 4°C so that re-addition of insulin to the cells after this time period would not have any further input. To address this possibility, myoblast monolayers were incubated as described in Fig. 1 but without the addition of primary antibody during the 1-h incubation at 4°C. The cells were then lysed in SDS sample buffer, resolved by SDS-polyacrylamide gel electrophoresis, and immunoblotted to probe for phosphotyrosine-containing proteins and for phosphothreonine 308 of PKB. The results shown in Fig. 3 indicate that insulindependent increases in a 175-kDa phosphotyrosine protein (likely IRS-1) and phosphorylated (on threonine 308) PKB (the site phosphorylated by phosphoinositide-dependent protein kinase-1) had returned to near basal levels after 1 h of incubating cells in the absence of insulin at 4°C. These results indicate that during the internalization period, any effects of prior insulin addition are not preserved, and therefore subsequent addition of insulin represents re-stimulation of the cells with the hormone.
Insulin Causes Relocation of GLUT4myc in Rounded-up Myoblasts-The increase in GLUT4myc recycling without reduction in GLUT4myc removal from the cell surface suggests that insulin stimulation increases the rate of return of GLUT4myc to the cell surface. It is therefore important to examine the mechanism leading to increased recycling in space and time. For this purpose, we were required to maximize our ability to discern the location of GLUT4myc vis à vis known organellar markers. When spread out on substratum, the cytosol of myoblasts (as well as fibroblasts) is distributed very thinly over a large area, making it difficult to distinguish juxtamembrane localization of immunofluorescent signals from more internal signals. To overcome this problem we lifted L6-GLUT4myc myoblasts off tissue culture flasks by incubating them in the absence of Ca 2ϩ and Mg 2ϩ for ϳ20 min. Once detached, myoblasts were allowed to settle onto coverslips for 10 min in the absence and presence of insulin, fixed, and immunocytochemically labeled for GLUT4myc and TfR as well as stained for DNA with TO-PRO3 (Fig. 4). In the absence of insulin, a large fraction of the GLUT4myc label was intracellular, with a loose perinuclear pattern. At this steady state distribution, a significant portion of intracellular GLUT4myc escaped colocalization with the TfR-containing compartment. Insulin addition caused translocation of a fraction of the GLUT4myc signal to the cell periphery. On the other hand, virtually all of the TfR was detected intracellularly in a compact region about the nucleus in the absence and presence of insulin, with no discernible changes caused by the presence of the hormone. DNA labeling was included to add perspective and emphasize the perinuclear localization of the TfR-containing compartment. Fig. 4 illustrates the ability to determine intracellular and surface GLUT4myc in rounded-up myoblasts, and that GLUT4myc responds to insulin in rounded-up L6 myoblasts by translocating to the cell surface.
GLUT4myc Travels through the Early Endosome to the Recycling Endosome-Morphologically, different endosomes are defined by the presence of marker proteins. EEA1 is a putative tethering protein that helps to bring vesicles in close proximity with the early endosome and is found solely on early endosomes (24). Numerous proteins undergo constitutive recycling between the plasma membrane and the recycling endosome. The TfR is responsible for iron entry into the cell via binding to transferrin and is constitutively recycled back to the cell surface (25). The relative abundance of TfR on the plasma membrane and in the recycling endosome varies among cell types. Greater than 90% of the TfR resides in the recycling endosome in L6 myoblasts, 2 making it a suitable marker for this compartment.
To follow the transit of GLUT4myc through the early endosome and recycling endosome, surface GLUT4myc labeled with anti-Myc antibodies in intact myoblasts was allowed to internalize at 37°C for different times up to 20 min. At each time point studied, the localization of internalized GLUT4myc was compared with the steady state distribution of endosomal markers. Fig. 5a shows the localization of labeled GLUT4myc and EEA1 at the onset of rewarming. Two min after initiation of internalization, some GLUT4myc could be detected in a compartment positive for EEA1 staining (Fig. 5b) and remained in this compartment for at least 5 min (Fig. 5c). At 10 min, there was no detectable labeled GLUT4myc in the perinuclear, TfR-positive compartment (Fig. 5d), but by 15 min (Fig.  5e) GLUT4myc began to collect there, and by 20 min (Fig. 5f) a large portion of the labeled GLUT4myc overlapped with TfR.
This distribution remained for up to 30 min (longer times could not be analyzed because of flattening of the myoblasts).
Insulin Re-addition Accelerates the Transit of GLUT4 through the Endosomal System-GLUT4myc was allowed to internalize and chased in the absence or presence of re-added insulin. The hormone did not appear to affect the rate of appearance of surface-labeled GLUT4myc in the EEA1-positive compartment (Fig. 6, a-d). Approximately 10% of the GLUT4myc was found in the EEA1-positive compartment (Fig.  6, a-d) for up to 8 min (results not shown). Insulin did not affect the staining patterns of EEA1 or TfR (steady state distribution). In unstimulated cells after a 10-min chase, 10.2% of surface-labeled GLUT4myc was detected in the recycling endosome (Fig. 6e), and by 20 min this colocalization increased to 30.1% (Fig. 6f). In contrast, in the continued presence of insulin, 36.9% of labeled GLUT4myc was already present in the TfR-positive compartment at 10 min after initiation of internalization (Fig. 6g). Moreover, by 20 min, only 9.9% of the 2 L. Foster and A. Klip, unpublished observation. labeled GLUT4myc internalized in the continued presence of insulin was detectable in the TfR-containing endosomes (Fig. 6h).
To begin to explore the transit of other recycling proteins, we investigated the routing of internalized rhodamine-transferrin (Fig. 7). For these experiments, rhodamine-transferrin was provided to the cells during the 10 min of attachment to the coverslip at 37°C to allow enough label to enter the cells. After this 10-min pre-loading period, external rhodamine-transferrin was removed, and the localization of the intracellular ligand was chased in space and time. At this point, most of the internalized rhodamine-transferrin was observed in a layer below the plasma membrane, presumably in early endosomes (Fig.  7a). After a subsequent 10 min of internalization (for a total of 20 min from the beginning of transferrin loading), significant rhodamine-transferrin was detected in the perinuclear compartment characterized by the bulk of the TfR, i.e. the recycling endosome (Fig. 7b). Importantly, the presence of insulin during the internalization of rhodamine-transferrin did not appear to alter the arrival of this cargo molecule at the recycling endosome (Fig. 7c). Upon a further 10 min of internalization (for a total of 30 min), the internalized rhodamine-transferrin was absent from this compartment either in the presence or absence of insulin (not shown). The label could be found in the cytosol of unstimulated cells but had largely exited all insulinstimulated cells. These results are consistent with an increase in transferrin recycling caused by the hormone, leading to rhodamine-transferrin re-externalization. An in-depth study of the coincident transit of GLUT4myc and rhodamine-transferrin is called for through future research. At present, the absolute times of transit of each molecule are not directly comparable because of the need to load with rhodamine-transferrin for 10 min at 37°C to obtain sufficient label. Nonethe-less, the results of Figs. 6 and 7 suggest that insulin differentially affects the inter-endosomal transit of these two molecules. Insulin accelerates the arrival of surface-labeled GLUT4myc at the recycling endosome but not that of rhodamine-transferrin.
Acceleration of GLUT4myc Traffic because of Insulin Is Dependent on PI 3-Kinase and PKB-PI 3-kinase activity is re- FIG. 7. Effect of insulin on inter-endosomal transit of transferrin. L6-GLUT4myc myoblasts were rounded-up and incubated for 10 min with rhodamine-transferrin at 37°C during the re-attachment period. Internalization was allowed for 10 additional minutes (see "Experimental Procedures"). The cells were then fixed, permeabilized, and incubated with antibody to the TfR, followed by a fluorescein isothiocyanate-conjugated secondary antibody. Shown are confocal fluorescence microscopy images of rounded-up myoblasts pre-loaded with rhodamine-transferrin, analyzed immediately upon termination of the 10 min of loading (a) or a subsequent 10 min internalization (for a total of 20 min) (b and c) in the absence (b) or presence (c) of 100 nM insulin (con and ins, respectively). Red staining is rhodamine-transferrin, blue staining is DNA, green staining is TfR. The scale bar represents 10 m. Shown are representative cells from one of four independent experiments. The degrees of overlap (mean Ϯ S.E.) calculated for any given condition, using NIH Image and Adobe Photoshop, are expressed as the percentage of colocalization out of the total rhodamine-transferrin (red) signal and are indicated below each panel (see "Experimental Procedures"). quired for insulin-dependent translocation of GLUT4 to the plasma membrane (26 -30), but the exact steps regulated by the enzyme are not known. To explore the role of PI 3-kinase in GLUT4 endocytic traffic, surface-labeled GLUT4myc was chased during its internalization for 10 ( Fig. 8, a-d) or 20 (Fig.  8, e-h) min in the absence (Fig. 8, a, b, e, and f) or presence (Fig.  8, c, d, g, and h) of insulin, without (Fig. 8, a, c, e, and g) or with (Fig. 8, b, d, f, and h) cellular pretreatment with the PI 3-kinase inhibitor LY294002. In the absence of insulin, LY294002 slightly slowed the time course of movement of labeled GLUT4myc through the recycling endosome (compare Fig. 8, a  with b, e with f). In contrast, in insulin-stimulated cells LY294002 prevented the accelerated movement of GLUT4myc through this compartment (compare Fig. 8, c with d, g with h).
In fact, the time course of colocalization of labeled GLUT4myc with TfR was similar for control unstimulated cells as for LY294002-pretreated, insulin-stimulated cells.
To further demonstrate that PI 3-kinase is involved in insulindependent acceleration of inter-endosomal traffic and to test the role of PKB/Akt in this process, dominant-negative DNA constructs of the two enzymes were transiently transfected into L6-GLUT4myc cells. A construct coding for the p85 subunit of PI 3-kinase lacking the intervening SH2 domain that binds the catalytic p110 subunit has been shown to override the insulinstimulated activation of endogenous PI 3-kinase (31) and to inhibit insulin-induced arrival of GLUT4myc to the cell surface (11,32). We have shown that a construct encoding PKB with three point mutations (K179A, T308A, and S473A) named FIG. 8. Acceleration of inter-endosomal traffic by insulin re-addition is dependent on PI 3-kinase and PKB. L6-GLUT4myc myoblasts transfected to express dominant negative p85 or AAA-PKB were rounded-up and incubated for 10 min in the presence of 100 nM insulin at 37°C during the re-attachment period. The cells were then chilled, rinsed, and incubated for 1 h with anti-Myc antibody in the absence of insulin as described under "Experimental Procedures." After the wash steps, internalization was allowed by switching the cells to 37°C for the indicated time periods in the absence or presence or re-added insulin. At each time point, the cells were fixed and stained with secondary antibody along with antibodies to the indicated markers. Shown are confocal fluorescence images of surface-labeled GLUT4myc (red) in relation to whole-cell TfR (green in a-h, blue in i-n) after a 10 (a-d)-or 20 (e-n)-min endocytosis in the absence (a, b, e, f, i, k, and m) or presence (c, d, g, h, j, l, and n) of 100 nM re-added insulin, as indicated (con and ins, respectively). Untransfected cells were treated with Me 2 SO (a, c, e, and g) or 25 M LY294002 (LY, b,  d, f, and h)  AAA-PKB overrides the insulin-stimulated activation of cotransfected PKB␣ (11) and PKB␤ 3 and also inhibits insulin-dependent exocytosis of GLUT4myc in L6 myoblasts (11). Transfection of empty pcDNA3 vector had no effect on either the unstimulated arrival of labeled GLUT4myc to the recycling endosome marked by the TfR (Fig. 8i) or the insulin-stimulated movement of labeled GLUT4myc through this compartment (Fig. 8j) (images were taken 20 min after initiation of internalization). Although neither ⌬p85␣ nor AAA-PKB affected the arrival of GLUT4myc to the TfR-containing recycling endosome in unstimulated cells (Fig. 8, k and m, respectively), expression of either mutant prevented the accelerated transit of GLUT4myc through this compartment in response to insulin (Fig. 8, l and n). These results suggest that both enzymes participate in the insulin-dependent regulation of this aspect of GLUT4 traffic.
Internalized GLUT4myc Does Not Travel to the Trans-Golgi Network (TGN)-Because GLUT4myc travels through the recycling endosome and retrograde transport of some proteins takes place from the recycling endosome to the TGN (33), it was conceivable that internalized, surface-labeled GLUT4myc might cycle through the TGN. Indeed, some models of GLUT4 segregation at steady state have suggested the presence of some GLUT4 in the TGN. To address whether during its endocytosis GLUT4 reaches the TGN, L6-GLUT4myc cells were transiently transfected with a construct encoding the transmembrane, and cytosolic domains of the endopeptidase furin, a TGN-resident protein, fused to the extracellular domain of the high affinity interleukin-2 receptor (CD25) as an epitope (16). After allowing surface-labeled GLUT4myc in transfected myoblasts to internalize, the cells were labeled with anti-CD25 antibody to reveal the TGN and TO-PRO3 to reveal the nucleus. Internalized, surface-labeled GLUT4myc did not colocalize appreciably with furin-CD25 staining at either 0 (Fig. 9a), 10 ( Fig. 9b), or 20 (Fig. 9c) min of endocytosis. Instead, within the limits of our detection system, surface-labeled GLUT4myc appeared in a furin-negative perinuclear compartment after internalization. This site is likely the TfR-positive compartment seen in Figs. 4 -6, which can be distinctively resolved from the furin-containing compartment (Fig. 9d). DISCUSSION Biochemical and morphological approaches have been used to date in an effort to characterize the endomembranes populated by GLUT4. Gradient centrifugation of adipose cells has revealed at least two if not three distinct intracellular compartments containing GLUT4 (34,35) as have mathematical models built from measurements of steady state distributions of GLUT4 (36,37). Moreover, the soluble N-ethylmaleimide-sensitive factor attachment protein receptor proteins (SNAREs) vesicle-associated membrane protein-2, syntaxin 4, and synaptosome-associated protein of 23 kDa are required for about 50% of the insulin-induced GLUT4 translocation in muscle and fat cells (see Ref. 38), suggesting that insulin draws GLUT4 from two pools distinguishable by their complement of fusogens. Consistent with this scenario, oxidative ablation of transferrin-containing compartments by transferrin-coupled horseradish peroxidase obliterates 30 -50% of the intracellular GLUT4 of 3T3-L1 adipocytes (39,40), indicating that the recycling endosome is one of the compartments populated by GLUT4. Until now there had been no information on whether insulin regulates the traffic through intracellular compartments in either muscle or fat cells. Answering this question is key to our understanding of GLUT4 availability for translocation.
For this study we selected L6 myoblasts stably expressing GLUT4 tagged at its first exofacial loop with an Myc epitope. We have previously shown that GLUT4myc is effectively sequestered intracellularly in L6 myoblasts and myotubes such that at steady state only 10% of the transporter is found at the cell surface (9). In its intracellular location, GLUT4myc largely segregates away from GLUT1 or GLUT3 but co-segregates with IRAP, based on immunoadsorption of membranes with antibodies to each of these antigens (8). Moreover, GLUT4myc responds to insulin by mobilizing to the plasma membrane of L6 myotubes and in this respect mimics quantitatively the movement of endogenous GLUT4 (7). Additional studies confirm that GLUT4myc is a reliable marker of GLUT4 traffic and, in response to insulin, its mobilization to the cell surface requires the SNARE VAMP2 and input from PKB/Akt (11,14). In the present study we further characterize the recycling rate of this molecule in L6 myoblasts and confirm that insulin increases its recycling rate to the cell surface (Fig. 1). Moreover, we show by confocal microscopy that GLUT4myc is concentrated in a perinuclear compartment and that it markedly increases at the cell surface upon insulin stimulation in rounded-up myoblasts (Fig. 4). This behavior is highly reminiscent of the sequestration and movement of GLUT4 in rat and mouse adipose cells and suggests that rounded-up L6 myoblasts expressing GLUT4myc constitute a useful and reliable cell system to study spatial/temporal coordinates of GLUT4 traffic in response to insulin and other stimuli.
In the present study we measured the rates of GLUT4myc recycling and disappearance of GLUT4myc from the cell surface and followed the localization of the transporter once internalized in L6 myoblasts. Insulin markedly increased the rate of GLUT4myc recycling (Fig. 1) without altering the rate of GLUT4myc removal from the cell surface within 10 min of initiation of endocytosis (Fig. 2). Given that insulin markedly increases GLUT4myc recycling back to the cell surface, this observation suggests that the primary effect of insulin in muscle cells is to promote the externalization of GLUT4myc. Studies with rat primary adipocytes have yielded conflicting results, some failing and others detecting a small effect of insulin on the rate of removal of GLUT4 from the cell surface (2,19). There is evidence for a small reduction in the rate of GLUT4 internalization caused by insulin, detected by either photolabeling of surface 3 R. Somwar and A. Klip, unpublished results. transporters in 3T3-L1 adipocytes followed by immunoprecipitation (41) or by proteolytic cleavage of an exofacially exposed site in isolated rat adipocytes followed by subcellular fractionation (42). In all cases, however, by far the major effect of insulin was a stimulation of the exocytic arm of its recycling (2,19,41,42). In the present study GLUT4myc internalization was measured by a simpler procedure not requiring either immunoprecipitation or subcellular fractionation, techniques that have complications of incomplete recoveries. It is not clear whether the use of different techniques or a cell-type specificity are responsible for the lack of evidence for an insulin-dependent reduction in GLUT4myc removal from the cell surface in L6 myoblasts. Consistent with the present results, insulin did not alter the rate of removal from the cell surface of IRAP (21) or of a chimera expressing the cytosolic IRAP region and the extracellular TfR region (20) in 3T3-L1 adipocytes.
The stimulation of the externalization of GLUT4myc could conceivably involve events of intracellular sorting/transit as well as events of mobilization of vesicles directly to the plasma membrane. To begin to explore these possible phenomena, we resorted to a spatial and temporal analysis of the localization of surface-labeled GLUT4myc. For these studies, insulin was given to cells for 10 min to summon GLUT4myc to the cell surface, where it was labeled with antibody at 4°C, and then its intracellular transit was examined in space and time under conditions where insulin signals had waned or with re-added insulin. Once antibody-labeled GLUT4myc was internalized in the absence of insulin, the transporter was chased and found to be present in a compartment marked by the early endosomal marker EEA1. The presence of insulin did not appear to alter the time course of appearance of labeled-GLUT4myc in this organelle.
Within 10 min of initiation of its internalization, some GLUT4myc reached the TfR-containing compartment, but a more significant accumulation was noted at 20 min. Strikingly, insulin cut by half the time required for labeled-GLUT4myc to accumulate in the TfR-containing compartment. Moreover, by 20 min, GLUT4myc internalized in the presence of insulin had exited from the recycling endosome. We interpret these results to reveal a novel, insulin-induced acceleration of GLUT4myc through the endosomal system. In contrast, internalized rhodamine-transferrin was detected in this compartment 20 min after initiation of loading with this ligand, whether in the absence or presence of insulin. Importantly, the PI 3-kinase inhibitor LY294002 and a dominant negative form of the p85␣ subunit of PI 3-kinase prevented the early arrival and depar-ture of GLUT4myc to and from the recycling endosome stimulated by insulin. Quantitation revealed that treatment of cells with LY294002 alone had a slightly retarding effect on the movement of GLUT4myc through the endosomal system. This is consistent with previous reports that wortmannin slows inter-endosomal movement of TfR (43,44).
Although finding surface-labeled GLUT4myc in the recycling endosome was not surprising, the observed effect of insulin on the inter-endosomal traffic of GLUT4 was unexpected. At steady state, intracellular GLUT4myc was found both in the recycling endosomes and in punctate structures within the cytosol (Fig. 4). Immunoadsorption of GLUT4-containing bodies results in segregation of over half of GLUT4myc away from GLUT1 (8), and arrival of basal and insulin-stimulated GLUT4myc vesicles are differentiated by their sensitivity to tetanus toxin (9,14). These results suggest that, as in rodent adipocytes, a portion of GLUT4 segregates from the recycling compartment. However, the site of such segregation and the origin of the specialized segregated compartment were unknown. The findings reported here lead us to propose a revised model of intracellular GLUT4 traffic and its regulation by insulin as presented in Fig. 10. In this model, internalized GLUT4 travels through the early endosome defined by the presence of EEA1 and progresses to the recycling endosome defined by TfR. Insulin accelerates GLUT4 arrival at the recycling endosome. Our results also suggest that insulin accelerates the exit of GLUT4 from this compartment since the residence time of labeled GLUT4myc in the TfR-positive endosome was Ͼ10 min in the absence of insulin but Ͻ10 min in the presence of the hormone. The above results suggest that insulin input is required for at least two distinct functions, movement of GLUT4 into the recycling endosome (Fig. 10) and budding out of the recycling endosome. We propose that, from the recycling endosome GLUT4myc would travel to the cell surface, possibly via generation of specialized exocytic vesicles. Our model also proposes that sorting of GLUT4 occurs in the recycling endosome but does not rule out that a portion of the exocytic vesicle pool may form directly from the early endosome, since there was always a fraction of the internalized GLUT4myc that did not colocalize with TfR at either 10 or 20 min after internalization. That the TGN does not appear to be involved in insulin-dependent GLUT4 traffic is borne out by previous studies showing that brefeldin A does not prevent insulin-dependent GLUT4 exocytosis (45-49) with one exception (50). Yet GTP␥S caused GLUT4 budding out of a TGNenriched preparation in vitro (51). Our study suggests that the FIG. 10. Model of GLUT4 traffic and sites of insulin input. The results of this study suggest the following model. After removal from the plasma membrane, GLUT4 enters the early endosome (EE) characterized by EEA1. From the early endosome GLUT4 can travel to the juxtanuclear, recycling endosome (RE) marked by TfR or to specialized vesicles. Transit to the recycling endosome is regulated by a PI 3-kinaseand PKB-dependent signal from insulin. Future studies should investigate the localization of GLUT4 after its exit from the recycling endosome in an attempt to define the genesis of the specialized exocytic vesicles.