Characterization of the insulin-regulated endocytic recycling mechanism in 3T3-L1 adipocytes using a novel reporter molecule.

The endocytic trafficking of the GLUT4 glucose transporter and the insulin-regulated aminopeptidase (IRAP) are regulated by insulin. We have used a chimera between the intracellular domain of IRAP and the extracellular and transmembrane domains of the transferrin receptor (vpTR) to characterize IRAP-like trafficking in 3T3-L1 adipocytes. Our data demonstrate that the cytoplasmic domain of IRAP is sufficient to target vpTR to the insulin-regulated, slow recycling pathway in adipocytes and that the dynamic retention of vpTR is dependent on a di-leucine motif. Our kinetic analysis demonstrates that vpTR recycles as a single kinetic pool and that vpTR is very efficiently sorted from endosomes to the insulin-regulated recycling pathway. An implication of these findings is that the key step in the dynamic retention of vpTR occurs within the early endosomal system. We have previously shown that vpTR is trafficked by an insulin-regulated pathway in Chinese hamster ovary cells (Johnson, A. O., Subtil, A., Petrush, R., Kobylarz, K., Keller, S., and Mc Graw, T. E. (1998) J. Biol. Chem. 273, 17968-17977). The behavior of vpTR in Chinese hamster ovary cells is similar to its behavior in 3T3-L1 adipocytes. The main difference is that insulin has a larger effect on the trafficking of vpTR in the adipocytes. We concluded that the insulin-regulated slow recycling endocytic mechanism is expressed in many different cell types and therefore is not a unique characteristic of cells that express GLUT4.

Insulin stimulation of glucose uptake by fat and muscle is an important mechanism for regulating whole body glucose homeostasis (reviewed in Ref. 1). Insulin regulates glucose uptake by inducing the translocation of GLUT4, a glucose transporter isoform expressed in fat and muscle, from intracellular compartments to the plasma membrane, shifting the distribution of GLUT4 from less than 10% on the surface to greater than 50% on the surface (2,3). Considerable efforts have focused on understanding the molecular mechanism that regulates GLUT4 distribution (e.g. Refs. 4 -9). The basal state intracellular sequestration of GLUT4 and redistribution in the pres-ence of insulin are dynamic processes determined by the rate of internalization and the rate of return to the plasma membrane (reviewed in Refs. 10 and 11).
Many membrane proteins continually cycle between the cell surface and endosomal compartments (for review see Ref. 12). The transferrin receptor (TR) 1 is the model receptor of choice for studies of constitutive endocytic internalization and recycling. An advantage of using the TR is that the complete endocytic cycle can be monitored using the receptor's native ligand, transferrin (Tf). Although in no single study has the trafficking of GLUT4 and TR been directly compared, two characteristics are believed to distinguish GLUT4 trafficking from the trafficking of the TR (and more aptly the general endocytic trafficking) (1,(13)(14)(15). One is the more pronounced intracellular accumulation of GLUT4 in the basal state as follows: ϳ90 versus ϳ70% for GLUT4 and TR, respectively. The other is the large effect that insulin has on the surface expression of GLUT4, which increases 5-10-fold, whereas surface expression of the TR increases by less than 2-fold. The greater intracellular accumulation of GLUT4 relative to TR is believed to reflect slower recycling of internal GLUT4 to the cell surface. The major effect of insulin on the distribution of GLUT4 is believed to be an increase in the recycling rate.
The molecular mechanisms underlying the dynamic intracellular retention of GLUT4 and the effects of insulin on recycling are not known. Most data support a mechanism in which GLUT4 is targeted to an insulin-regulated, retention compartment distinct from the general endosomal recycling system, although this compartment has not been well characterized (16 -19). There is evidence that both amino and carboxyl cytoplasmic domains of GLUT4 are required for insulin-regulated GLUT4 trafficking (7)(8)(9)20). Those data indicate that these domains contain the information (i.e. trafficking motif) that targets GLUT4 to the specialized compartment, although the specific amino acids that comprise this motif(s) have not been identified.
A type II membrane protein with aminopeptidase activity, IRAP, was identified as a major component of immunopurified GLUT4-containing vesicles (21)(22)(23). In subsequent studies it was found that IRAP co-localizes with GLUT4 in fat and muscle cells and that insulin stimulates a large increase in the surface expression of IRAP (24 -28). IRAP is the only other protein identified that has trafficking characteristics like GLUT4. The physiologic role of IRAP is not known.
To develop a reporter molecule for studies of insulin-regulated trafficking, we have created a chimera, referred to as vpTR, containing the cytoplasmic domain of IRAP and the transmembrane and extracellular domains of the human TR (29). The advantage of using vpTR as a reporter molecule is that the methods developed for studying TR trafficking can be used. We have previously shown that vpTR expressed in CHO cells is more slowly recycled than the TR and that insulin specifically induces an ϳ2-fold increase in the recycling rate of the chimera (29). Those studies demonstrate that the cytoplasmic domain of IRAP contains information that can direct trafficking when transferred to the TR. Finding that CHO cells have a specialized, insulin-regulated trafficking mechanism was unexpected because insulin-regulated trafficking was considered a characteristic of fat and muscle cells.
Here we report our findings on the analysis of the endocytic trafficking of vpTR in 3T3-L1 adipocytes. We find the vpTR is trafficked by a specialized insulin-regulated mechanism in these cells. For the first time we have directly measured the internalization and recycling rate constants of a protein trafficked by the insulin-regulated pathway in adipocytes. The vpTR chimera is recycled at approximately one-fifth the rate of the TR in the basal state, and insulin stimulates recycling of the chimera by ϳ4-fold, whereas insulin increases recycling of the TR by only ϳ1.3-fold. These data demonstrate that the cytoplasmic domain of IRAP is sufficient for targeting a heterologous protein to the insulin-regulated pathway in adipocyte cells. Site-directed mutagenesis demonstrates that the dileucine sequence at position 76,77 of the cytoplasmic domain of IRAP is necessary for the slow recycling of the chimera. Kinetic analysis demonstrates that vpTR is efficiently sorted from early endosomes to the insulin-regulated recycling pathway. In the basal state vpTR is localized to a peri-centriolar compartment that only partially overlaps with the TR-containing pericentriolar recycling compartment, indicating that the insulinregulated recycling endosomes are distinct from general recycling endosomes.

MATERIALS AND METHODS
Ligands and Chemicals-Human Tf was obtained from Sigma and further purified by Sephacryl S-300 gel filtration. Diferric Tf and 125 I-Tf were prepared as described previously (30). 125 I and 55 Fe 3ϩ were purchased from NEN Life Science Products. Tf was labeled with the fluorescent dye, Cy3 (Biological Detection Systems, Pittsburgh), according to the manufacturer's instructions. All chemicals were from Sigma unless otherwise specified.
Plasmids and Transfection-DNA coding for the human TR, vpTR, and the LL53AA and LL76AA vpTR chimera mutants were subcloned from a pUC8 vector (29) into pMexNeo vector (7) using BamHI restriction sites. 3T3-L1 fibroblasts were transfected using LipofectAMINE (Life Technologies, Inc.). Two days after transfection 1 mg/ml G418 was added to the culture medium. Individual clones were isolated using cloning rings, grown in 24-well clusters and tested for expression of the transfected genes as discussed below.
Down-regulation of Endogenous Mouse TR-On the day before the experiment, a rat monoclonal antibody (antibody 220, ATCC) that specifically binds the extracellular domain of the mouse TR was added to the culture medium. This antibody does not recognize the human TR and consequently does not bind the vpTR. The amount of antibody required to reduce Tf binding by non-transfected 3T3-L1 adipocytes to less than 10% of control level was determined for each batch of antibody. For most experiments a dilution of hybridoma culture medium was used as a source of 220 antibody. Identical results were observed when the antibody was purified from the conditioned medium. The antibody was also present in the medium during the experiments to ensure that the background from the endogenous mouse TR was less than 10%. To identify clonal lines, cells were plated, differentiated, the endogenous TR down-regulated, and cells incubated with iodinated Tf.
Glucose Uptake-Glucose uptake was measured on differentiated cells grown in 24-well plates using 2-deoxy-D- [2,  Microscopy-Cells, grown on coverslip bottom dishes, were incubated in med1 with 220 antibody and 3 g/ml Cy3-Tf for 4 h at 37°C to achieve steady-state occupancy of the receptors. For experiments with insulin, 700 nM insulin was added for the last 15 min of the incubation. The cells were washed twice with med 2, fixed for 30 min in 3.7% formaldehyde in med 2, and labeled with a polyclonal antibody specific for the C-terminal domain of GLUT4 (a kind gift of Giulia Baldini, Columbia University). The secondary antibody was an Alexa488-labeled anti-rabbit antibody (Molecular Probes, Eugene, OR). Images were collected with an Axiovert 100M inverted microscope equipped with an LSM 510 laser scanning unit and a 63 ϫ 1.4 NA plan Apochromat objective (Carl Zeiss, Inc.). 543 and 488 nm light were used used to stimulate Cy3 and Alexa 488 fluorescence, respectively. Emissions were selected with a 560-nm long pass filter for Cy3 and a 505-530-nm band pass filter for Alexa 488 and collected sequentially to prevent crossover.
Steady-state Distribution of vpTR and TR-For each assay, cells were grown in two 6-well clusters. Cells were incubated in med 1 with 220 antibody and 3 g/ml iodinated Tf for 4 h at 37°C to achieve steady-state occupancy of the receptors. 1 M insulin was added in 1 plate in the last 30 min of incubation. Unless otherwise noted, all experiments were performed using 1 M insulin. Identical results were obtained using 100 nM insulin. The plates were then placed on ice and washed 4 times with 150 mM NaCl, 20 mM HEPES, 1 mM CaCl 2 , 5 mM KCl, 1 mM MgCl 2 , pH 7.4 (med 2), at 4°C. The cells were incubated for 5 min at 4°C in 1 ml/well 0.15 M NaCl, 0.02 M sodium citrate, pH 5.0, followed by 5 min at 4°C in med 2 supplemented with 3 g/ml unlabeled Tf and 100 M of the iron chelator deferroxamine. This acid/ neutral wash was repeated twice. These washes remove greater than 90% of the Tf bound to the surface. The washes were pooled, and the radioactivity was measured. This value represents the Tf bound to the cell surface. The cells were solubilized, and the radioactivity was measured. This value represents the Tf in intracellular compartments. Two wells per plate were treated as the others except that a 200-fold excess of unlabeled Tf was added with the iodinated Tf. The radioactivity in the acid/neutral washes and associated with the cells represents nonspecific binding and was subtracted from the acid/neutral washes and from the cell-associated radioactivity, respectively.
Binding of 125 I-Tf to the Cell Surface-For each assay, cells were grown in two 6-well clusters. Cells were incubated in med 1 with 220 antibody for 4 h at 37°C, and insulin was added to 1 plate during the last 30 min of incubation. The plates were placed on ice and incubated for 2 h at 4°C with 3 g/ml 125 I-Tf, washed 4 times with med 2 (4°C), solubilized, and the radioactivity measured. Two wells per plate were treated as the others except that a 200-fold excess of unlabeled Tf was added with the iodinated Tf. The radioactivity represents nonspecific binding and was subtracted from the cell-associated radioactivity.
Internalization Assay-For each assay cells were grown in four 6-well clusters. Cells were incubated for 4 h in med 1 with 220 antibody at 37°C, and insulin was added in 3 wells per plate at the last 30 min of incubation. 3 g/ml 125 I-Tf was then added in every well, and after the indicated times of incubation at 37°C the plates were placed on ice and washed 4 times with med 2 at 4°C. The cells were incubated for 5 min at 4°C in 1 ml/well of 0.15 M NaCl, 0.02 M sodium citrate, pH 5.0, followed by 5 min at 4°C in med 2 supplemented with 3 g/ml unlabeled Tf and 100 M of the iron chelator deferroxamine. This acid/ neutral wash was repeated twice. These washes remove greater than 95% of the Tf bound to the surface. The washes were pooled, and the radioactivity was measured. This value represents the Tf bound to the cell surface for each time point. The cells were solubilized, and the radioactivity was measured. This value represents the Tf internalized during the incubation at 37°C. One additional plate was treated as the others except that a 200-fold excess of unlabeled Tf was added with the iodinated Tf before incubating the plate for 5 min at 37°C. The radioactivity in the acid/neutral washes and associated with the cells represent nonspecific binding and was subtracted from the acid/neutral washes and from the cell-associated radioactivity, respectively, for ev-ery time point. The slope of a plot of the ratio of internal Tf to surface Tf versus time is the internalization rate constant.
Recycling Assay-For each assay cells were grown in two 24-well clusters. Cells were incubated in med 1 with 220 antibody and 3 g/ml iodinated Tf for 4 h at 37°C to achieve steady-state occupancy of the receptors. The cells were washed with med 2, incubated for 2 min at 37°C in 0.2 M NaCl, 50 mM MES, pH 5.0, followed by 2 washes with med 2. Incubation in the acidic medium releases Tf bound to TR on the cell surface. The cells were incubated at 37°C in med 1 with 3 g/ml unlabeled Tf, 100 M of the iron chelator deferroxamine (efflux medium). Insulin was added to the efflux medium in one plate. At the times indicated the medium was collected, and the cells were solubilized. The radioactivity in the medium is the Tf released from the cells during the incubation, and the cell-associated radioactivity is the Tf remaining inside cells. All data have been corrected for nonspecific binding, determined using a 100-fold excess of unlabeled Tf, which was typically less than 10% of the total. The recycling rate constant is the slope of the natural logarithm of the percent radioactivity remaining cell-associated versus time. 55 Fe Uptake- 55 Fe-Chloride was neutralized by adding 0.05 ml of 1 N NaOH to the radioactive vial (1 mCi in 0.1 ml 0.5 M Hcl), and 0.5 ml of 100 mM disodium nitrilotriacetate and 0.5 ml of 9.5 mg/ml apotransferrin in 10 mM NaHCO 3 , 0.25 M Tris-Cl, pH 8.0, were added. The solution was incubated for 1 h at room temperature before being passed over a PD-10 column equilibrated in phosphate-buffered saline. The A 465 nm /A 280 nm of the Tf fractions was 0.0486, which corresponds to 98% Fe saturation. To measure 55 Fe uptake, cells were incubated for 2 h in med 1 before adding 5 g/ml 55 Fe-Tf. After incubation for the indicated times the cells were washed 3 times in med 2 and solubilized in 0.1% SDS, and cell-associated radioactivity was determined. Data were corrected for nonspecific binding that was measured by adding a 200-fold excess of nonlabeled Tf.
Cell Permeabilization and Tf Efflux Assay-Adipocytes were permeabilized using a modification of a previously described procedure (32). Differentiated cells in 24-well clusters were incubated in med 1 with 3 g/ml 125 I-Tf for 4 h at 37°C. The plates were put on ice, washed twice in cold transport buffer (140 mM potassium glutamate, 5 mM NaCl, 5 mM MgCl 2 , 5 mM EGTA, 20 mM HEPES, pH 7.2) before adding 0.2 ml/well of 0.4 units/ml streptolysin O (Murex Diagnostics, Atlanta, GA) in cold transport buffer. The plates were transferred to 37°C for 2 min, placed on ice, and washed twice with cold transport buffer before adding 0.2 ml/well of transport buffer with 3 g/ml unlabeled Tf, 2 mM ATP, 10 mM creatine phosphate, 50 units/ml creatine phosphokinase, and 5 mg/ml bovine serum albumin. The cells were incubated at 37°C, and at the indicated times the medium was collected, and the cells were solubilized. The radioactivity in the medium is the Tf released from the cells during the incubation, and the cell-associated radioactivity is the Tf remaining inside cells. Data were analyzed as described (33).

3T3-L1 Cells Expressing the Human TR or vpTR-
The advantage of using vpTR as a reporter molecule is that insulinregulated trafficking can be studied with assays previously developed to study the TR (e.g. Ref. 29). To use the chimera for studies in 3T3-L1 adipocytes the endogenous mouse TR, which binds human Tf with high affinity (15), must be "down-regulated." Expression of the mouse TR can be down-regulated in fibroblasts by incubating cells with a rat monoclonal antibody that specifically binds the mouse TR extracellular domain (antibody 220) (34). Incubation of 3T3-L1 adipocytes with this antibody reduces the amount of Tf uptake by ϳ10-fold (Fig.  1A). In all experiments, unless otherwise noted, the cells were incubated overnight with 220 antibody to down-regulate endogenous mouse TR.
3T3-L1 cells were transfected with vpTR or the human TR. Clonal lines expressing the transgenes were identified by screening for 125 I-Tf (iodinated-Tf) uptake following down-regulation of the mouse TR (Fig. 1A). Treatment of cells with the antibody does not affect expression of vpTR because the chimera contains the human TR ectodomain, and the 220 antibody does not bind these sequences. Clones expressing the constructs to levels similar to the endogenous mouse TR were chosen for further studies. The vpTR55 cell line, based on anti-IRAP Western blot analysis of total cell extracts, expresses the chimera at a level similar to the endogenous IRAP, and the highest expressing line examined (vpTR85) expresses the chimera at ϳ2-fold the level of endogenous IRAP (not shown). We found no differences between the trafficking of the endogenous mouse TR (analyzed in non-transfected, non-down-regulated 3T3-L1 cells) and the human TR expressed in 3T3-L1 adipocytes (not shown), indicating that the down-regulation protocol does not affect general endocytic trafficking.
The effect of insulin on glucose uptake was examined in clonal lines differentiated into adipocytes (Fig. 1B). Insulin induced a large increase in glucose uptake, a response similar to that in non-transfected 3T3-L1 adipocytes. Although there is some clone-to-clone variation in response, this variation does not correlate with the level of expression of the chimera, and most likely reflects clonal variation unrelated to expression of vpTR. These data indicate that the insulin recruitment of GLUT4 to the plasma membrane is not affected by the expression of vpTR. Importantly, incubation with the 220 antibody does not affect insulin-stimulated glucose uptake (Fig. 1B), documenting that the down-regulation protocol does not affect insulin-regulated trafficking.
Insulin Stimulates the Translocation of vpTR to the Cell Surface-To determine whether vpTR expression on the surface was regulated by insulin, cells were mock-treated or incubated with insulin for 30 min at 37°C, and the binding of iodinated Tf to cells at 4°C was determined. Insulin induced a 2-4-fold increase in surface expression of vpTR and only a 1.3-fold increase in surface expression of the TR ( Fig. 2A). Since TR is a marker for general endocytic recycling, these data demonstrate that the surface expression of vpTR is specifically regulated by insulin. Although the extent of translocation of vpTR varied among clones, in each case the effect of insulin on the surface expression of vpTR was significantly greater than the effect of insulin on the surface expressing the TR.
The increased amount of vpTR on the surface could be due to a redistribution of a cycling pool of vpTR or to the recruitment of a storage pool (not in equilibrium with the cell surface in the basal state). To distinguish between these possibilities the total cycling pool of vpTR was measured in the presence and absence of insulin. If insulin recruits a storage pool of vpTR to the surface, then the total amount of vpTR accessible to binding Tf in the medium will be larger than in the basal conditions. If insulin stimulates a redistribution of a single vpTR cycling pool, then insulin will have no effect on the total cycling pool. Cells were incubated with iodinated Tf at 37°C in the absence or presence of insulin, and the total cell-associated radioactivity as a function of time was determined (Fig. 2B). Tf rapidly binds to vpTR on the cell surface, and the cell-associated iodinated Tf increases over time until all the cycling vpTRs are occupied. The time-dependent increase in cell-associated Tf reflects the recycling of intracellular vpTRs, which are unoccupied with iodinated Tf, to the surface (35). The observation that the total amount of Tf cell-associated at steady state (i.e. plateau level) is the same for both the insulin and basal conditions demonstrates that in the basal state vpTR continually cycles between the surface and intracellular compartments, and therefore the increase in vpTR on the cell surface is a redistribution of the chimera. In the basal state, total cell-associated Tf plateaus at ϳ3 h, when steady-state occupancy of vpTR was reached. In the presence of insulin, Tf accumulated faster, reaching the plateau level by ϳ90 min. These data indicate that one effect of insulin is to increase the recycling of vpTR back to the plasma membrane, since the rate of approach to steady state is determined by the rate of appearance of unoccupied vpTR at the surface (see below).
To measure the steady-state distribution of vpTR in the basal and insulin-stimulated conditions, cells were incubated with iodinated Tf at 37°C for 4 h to achieve steady-state occupancy of the chimera with Tf. Insulin was included during the last 30 min of the 4-h incubation with iodinated Tf to measure the distribution in the presence of insulin. The distribution of the chimera was determined by measuring the amount of Tf on the surface and inside cells. In the basal state less than 20% of the chimera is on the cell surface, whereas ϳ40% of the TR is on the cell surface (Fig. 2C). Following a 15-min incubation with insulin, there is a large shift in the distribution of the vpTR to the surface and only a minor shift in the distribution of the TR to the cell surface.
Studies with overexpression of GLUT4 have raised concerns about the potential saturation of the intracellular retention mechanism (7). This is not a concern with the clones used in this study, since clones expressing the chimera to the lowest and highest level (vpTR6 and vpTR85, respectively) showed the same distribution of the chimera in the basal state, indicating that the constructs were expressed at levels below a potential saturation level.
The difference in the basal state distributions of vpTR and the TR and the large difference in the effect of insulin on their distributions demonstrate that vpTR is specifically and dynamically retained within cells in the basal state and redistributed to the surface by insulin, two defining characteristics of GLUT4 trafficking.
Insulin Specifically Increases the Rate of vpTR Recycling-Insulin could increase the amount of vpTR chimera on the cell surface by reducing the internalization rate, by increasing the recycling rate, or by changes in both rates. The effect of insulin on the internalization of vpTR was measured using the In/Sur method (36). The chimera is rapidly internalized, with a rate constant (ϳ0.2 min Ϫ1 ) similar to the internalization constants of other membrane proteins (not shown). Insulin did not affect the internalization rate of vpTR. Thus, the insulin-induced redistribution of vpTR is not due to a decrease in the internalization rate of the chimera. We next measured the rate of release of iodinated Tf from adipocytes expressing vpTR or the human TR. Cells were incubated with iodinated Tf at 37°C for 4 h to achieve steadystate occupancy of the chimera with Tf, washed free of surfacebound Tf, re-incubated at 37°C in medium containing unlabeled Tf (efflux medium), and the radioactivity released into the medium and the remaining cell-associated radioactivity were determined. Since Tf remains associated with the receptor (or chimera) until it is returned to the plasma membrane, the release of Tf from cells reflects the rate of return of the receptor (37). Tf was released more slowly from adipocytes expressing the vpTR than from cells expressing the human TR (Fig. 3A). The released Tf could be precipitated at 4°C with 15% trichloroacetic acid, indicating the Tf is released intact from both cell types (not shown). In both cases, the loss of cell-associated Tf fits a monoexponential decay with ϳ20% of the Tf retained in cells after 4 h. These data demonstrate that the recycling of vpTR is slower than the TR and that the same fraction of internalized Tf is ultimately recycled from both cell types. A similar size (ϳ20%) non-released pool of Tf was observed previously in studies of the endogenous TR in 3T3-L1 adipocytes (38). Since the size of this non-recycled pool of internalized Tf does not differ between vpTR, TR,and mouse TR, it is not a confounding variable in our studies.
To determine the effect of insulin on the recycling of vpTR, adipocytes were treated as in Fig. 3A, except that insulin was added to the efflux medium. Insulin stimulated a large increase in the rate of release of Tf from vpTR-expressing cells (Fig. 3B). These data demonstrate that the recycling of vpTR is regulated by insulin and are consistent with the finding that the increase in vpTR surface expression is due to a redistribution of a slowly cycling pool. A summary of the recycling rate constants measured for untransfected and different clonal adipocyte lines expressing either the vpTR or the human TR are presented in Fig. 3C. In the basal state the chimera is recycled ϳ4-fold more slowly than the TR. The effect of insulin on recycling of vpTR is large (4 -5-fold), whereas the effect on the TR recycling rate constant is modest (1.3-fold). The response to insulin is very rapid since the same recycling rates were measured when the insulin was added to cells during the final 30 min of incubation with iodinated Tf (not shown). The recycling rate of vpTR in the presence of insulin is not significantly different from the recycling rate of the TR in the presence of insulin. The human TR recycles identically to the endogenous mouse TR, confirming that the down-regulation procedure is not affecting general endosomal trafficking.
The above experiments indicate that the intracellular distribution of vpTR in the basal state is due to a slow recycling rate, which is accelerated by insulin. However, the possibility remained that the slow Tf efflux rate was due to an inefficient release of iron from Tf internalized by vpTR. In that case, the Tf would remain bound to its receptor as it returns to the cell surface, and the rate of Tf release would be slower than the actual recycling rate of vpTR (e.g. Ref. 35). To exclude this possibility we measured the accumulation of 55 Fe in control and insulin-treated cells (not shown). Total cellular 55 Fe uptake was linear over more than 100 min of incubation and was increased by insulin, reflecting an increase in vpTR on the cell surface. Therefore, the accumulation of 55 Fe per Tf cycle is the same in control and insulin-treated cells, demonstrating that the iron release efficiency is identical in both conditions. This experiment shows that the differences in recycling measured cannot be attributed to differences in 55 Fe release efficiency.
vpTR Recycling Is Inhibited by Wortmannin and Stimulated by GTP␥S-Insulin stimulation of GLUT4 translocation to the surface requires phosphatidylinositol 3Ј-kinase activity (25, 39 -42). The insulin-induced increase in vpTR recycling was completely inhibited by wortmannin, indicating that vpTR translocation, like that of GLUT4 and IRAP, requires phosphatidylinositol 3Ј-kinase activity (not shown).
GTP␥S stimulates GLUT4 translocation to the cell surface (32,43,44). To test the effect of GTP␥S on the vpTR recycling rate, cells were incubated with iodinated Tf, permeabilized using streptolysin O toxin, and the recycling of Tf from the permeabilized cells was monitored (e.g. Refs. 32 and 45). GTP␥S increased Tf recycling in vpTR-expressing cells but not in TR-expressing cells, demonstrating that GTP␥S specifically stimulates vpTR recycling (not shown).
The results of the above studies indicated that the trafficking characteristics of vpTR are similar to those of GLUT4 and IRAP and distinct from those of the TR. Based on this analysis we conclude that GLUT4, IRAP, and vpTR are trafficked by a common insulin-regulated mechanism.
The Di-leucines at Position 76,77 of IRAP Are Required for Slow Recycling-One class of endocytic trafficking signals is based on di-leucine sequences (46). The IRAP cytoplasmic domain contains 2 di-leucine sequences, one at position [53][54] and the other at position 76 -77. We have previously shown that the di-leucine motif at position 76 -77 is necessary for the slow recycling of the chimera in CHO cells (29). To investigate the role of these motifs in the trafficking of the chimera in adipocytes, mutants of the chimera, in which alanines were substituted for the leucines at positions 53-54 or at positions 76 -77, were transfected in 3T3-L1 cells. Both mutants were efficiently internalized, consistent with our previous results in CHO cells that neither of these di-leucine motifs are necessary for internalization (not shown). The LL 53,54 AA chimera was recycled at a slow rate near that of the vpTR, and insulin stimulated a large increase in the recycling rate (Fig. 4), indicating that this di-leucine is not important in regulating the trafficking of the chimera. The vpTRLL 76,77 AA was rapidly recycled, and insulin had a small effect on the recycling rate constant. These results demonstrate that the LL 76,77 motif is required for the dynamic retention of vpTR.
vpTR Traffics as a Single Kinetic Pool-It has been proposed that at steady state ϳ40% of GLUT4 is in TR-containing endosomal compartments, and the rest is in a specialized compartment, presumed to be the insulin-responsive retention compartment (IRC) (13,14,17). It is not known if these two pools of GLUT4 behave as a single kinetic pool or two distinct kinetic pools. For example, the GLUT4 in the Tf-containing endosomes may only be returned to the surface by first trafficking to the IRC, in which case there would be a single, slowly recycling kinetic pool of GLUT4, even though nearly half of the GLUT4 is in a compartment from which Tf rapidly recycles. Alternatively, there could be two kinetic pools, with GLUT4 in the Tf-containing endosomes recycling rapidly back to the surface along with other endosomal proteins and GLUT4 in the IRC recycling slowly. We can distinguish between these two possibilities by measuring the recycling rate of vpTR following a 30-min chase after the 4-h incubation with iodinated Tf (Fig.  5A). If a significant fraction of vpTR is recycled from the fast endosomal pool, then the rate measured after the 30-min chase would be slower than the rate measured without the chase. The measured rate of recycling would decrease because a greater percentage of the Tf remaining cell-associated would be in the IRC, since Tf in endosomes would largely have been recycled during the chase.
The rate of Tf release from cells expressing vpTR following a 30-min chase was not significantly different from the rate measured at steady state, indicating that vpTR traffics as a single, slowly recycling pool (Fig. 5B). The effect of insulin on the recycling was the same following the 30-min chase as it was without the chase, consistent with vpTR trafficking back to the cell surface only after having trafficked to the IRC.
The vpTR Intracellular Retention Mechanism Is Very Efficient-Sorting of membrane proteins to a specific intracellular compartments can occur by an iterative process in which several cycles of internalization and recycling back to the plasma membrane are necessary for a protein to reach its final destination, in which case, iteration of an inefficient sorting step results in an overall efficient sorting (47)(48)(49). Alternatively, intracellular targeting may be efficient, with the majority of the protein being properly targeted with each passage through endosomes (e.g. Ref. 50). To determine the mechanism for dynamic retention of vpTR, we measured the recycling rate of vpTR after a 15-min pulse with iodinated Tf (Fig. 6). With this short pulse of iodinated Tf a greater fraction of cell-associated radioactivity will be in early endosomes as compared with the steady-state labeling conditions. Consequently, if the vpTR must cycle between endosomes and the plasma membrane a number of times before it reaches the IRC, then the recycling rate measured after this short pulse would be faster than the rate measured at steady state. This is not the case. We found that there was a minor, although statistically significant, increase in the recycling rate constant when cells had been loaded for 15 min, from 0.007 min Ϫ1 Ϯ 0.001 (ϮS.E., n ϭ 4) to 0.011 min Ϫ1 Ϯ 0.002 (ϮS.E., n ϭ 4; p ϭ 0.02; paired Student's t test). By fitting the 15-min pulse efflux data to a double exponential with the recycling rate of the TR for one rate and the recycling rate constant of vpTR for the other, we find that  Fig. 3, are shown. The recycling rate constants of vpTR and TR are shown for comparison.

FIG. 5. Efflux of Tf after 30 min chase in vpTR-expressing adipocytes.
A, a cartoon of the endosomal system in fat cells. TR recycles back to the surface by the rapid pathway (from endosomes). The vpTR chimera could recycle from by the fast endosomal and/or slow IRC pathways. B, schematic of experiment protocol. For analysis of recycling following a 30-min chase the cells were incubated for 4 h to achieve steady-state occupancy of the chimera with iodinated Tf, and the cells were washed and incubated for 30 min in efflux medium without insulin. At the end of this incubation (chase), the medium was discarded and replaced with fresh efflux medium, and the recycling time course was begun. C, the release of iodinated Tf from vpTR85 adipocytes was measured either as described in Fig. 3 (no chase) or after a 30-min chase as noted above. Data are from a representative experiment. a maximum of ϳ20% of vpTR internalized during the 15-min pulse is rapidly recycled. These data indicate that sorting of vpTR from endosomes to the slow recycling pathway is very efficient. Consistent with the efficient sorting of the chimera to the insulin-regulated pathway is the observation that insulin has a large stimulatory effect on the recycling of Tf in cells loaded for 15 min (Fig. 6).
Distribution of vpTR and of the TR in 3T3-L1 Adipocytes-We next used confocal microscopy to examine the subcellular localization of vpTR. Cells expressing vpTR or the human TR were loaded to steady state with fluorescent Tf, and the distributions of these proteins were compared with the distribution of endogenous GLUT4 (Fig. 7, A-C). In the basal state the majority of GLUT4 is localized to a peri-centriolar compartment which in many cells rings the nucleus. In addition, there are numerous GLUT4-containing vesicles dispersed throughout the cytoplasm. The intracellular compartments containing vpTR, which were labeled by internalization of Cy3-Tf from the medium, co-localize with the GLUT4 in the peri-centriolar region as well as in the peripheral punctate structures. The distributions of GLUT4 and vpTR are not dramatically altered by insulin, except that in many cells an increase in plasma membrane fluorescence is apparent (Fig. 7B, arrows). GLUT4 and vpTR remain co-localized in the presence of insulin. These microscopy data are consistent with vpTR and GLUT4 trafficking through the same compartments in 3T3-L1 adipocytes.
The human TR expressed in 3T3-L1 adipocytes is also localized in the peri-centriolar region (Fig. 7C). There is some overlap between GLUT4 and the TR, as expected from previous studies (16,17,38,(51)(52)(53)(54), but it is also apparent that much of the TR does not co-localize with GLUT4. These data are consistent with previous studies of the degree of co-localization of GLUT4 and the TR. Importantly, expression of vpTR does not alter the distribution of GLUT4, since the GLUT4 pattern in cells expressing vpTR were the same as in cells expressing the TR.

DISCUSSION
In this report we contrast the endocytic behaviors of vpTR and the human TR expressed in 3T3-L1 adipocytes. This is the first time that identical experimental methods have been used to compare directly the kinetics of insulin-regulated trafficking to general endosomal trafficking in adipocytes. In the absence of insulin vpTR is dynamically retained within cells. The greater intracellular concentration of vpTR, relative to the TR, is due to slow recycling back to the cell surface. Insulin induces a translocation of vpTR to the plasma membrane by increasing the recycling rate constant by ϳ4-fold. The effect of insulin on the chimera is not a result of comprehensive changes in general endosomal trafficking because insulin has only a minor effect on the recycling of the TR, increasing the rate constant by 1.3-fold. The effects of insulin on chimera trafficking are limited to recycling because insulin does not have a significant effect on the internalization rate of the chimera. These results demonstrate that vpTR is trafficked by a specialized, insulinregulated mechanism in 3T3-L1 adipocytes.
Our results indicate that vpTR is trafficked by the same insulin-regulated pathway as IRAP and GLUT4. One, the kinetics of trafficking of vpTR in the basal and insulin-stimulated states are similar to those reported for GLUT4, and the greater intracellular concentrations of vpTR and GLUT4, relative to the TR, reflect a difference in recycling rates (5,6). Two, the effects of wortmannin and GTP␥S on vpTR trafficking are the same as their effects on GLUT4 trafficking. These findings support the use of vpTR as a reporter molecule for insulinregulated membrane trafficking.
Our data demonstrate that the 109-amino acid cytoplasmic domain of IRAP is sufficient for targeting IRAP to the insulinregulated recycling pathway. A di-leucine sequence at position 76,77 is part of the motif that regulates the slow recycling of the chimera, since mutation of these leucines to alanines releases FIG. 6. Tf efflux after 15 min or 4-h loading in vpTR85 adipocytes. Cells were either incubated for 4 h in serum-free medium containing iodinated Tf or for 4 h in serum-free medium with iodinated Tf added for the last 15 min. The release of iodinated Tf was measured as described in Fig. 3. The data are plotted as the natural log of the fraction Tf cell-associated. The recycling rate constant is the slope of the line. Data are from a representative experiment. retention and blunts insulin responsiveness. These results suggest that the region containing the di-leucine 76,77 interacts with the machinery responsible for the insulin-regulated intracellular retention. A number of proteins that are sorted from the endosomal pathway to other membrane compartments contain di-leucine sorting motifs (46). Additional mutagenesis studies are required to delineate further the "retention motif" and to identify the cytoplasmic sequences that determine rapid internalization.
In agreement with our data on the requirement of the dileucines at position 76,77 are the previous results that a peptide encompassing residues 55-82 of the IRAP cytoplasmic domain shifts GLUT4 to the surface of fat cells by saturating the retention mechanism (55). These competition data also indicate that the same machinery is involved in the retention of both IRAP and GLUT4, thereby providing further support for the use of vpTR as a model for studies of insulin-regulated trafficking. Previous studies have shown that a di-leucine motif in the carboxyl cytoplasmic domain of GLUT4 plays a role in its trafficking, although the data suggest that in GLUT4 this motif is involved in internalization and therefore may be playing a different role in GLUT4 than in vpTR and IRAP (7,8,57,58). One interpretation of these data is that different motifs target IRAP and GLUT4 to the insulin-regulated pathway. There is precedent for different motifs targeting proteins to the same intracellular destination. For example, different motifs are used to target furin and TGN38 from the plasma membrane to the trans-Golgi network (59 -61). However, it is likely that residues other than the di-leucine are part of the motif which regulates recycling, and until those sequences are identified it is not possible to state rigorously that different motifs determine the trafficking of IRAP and GLUT4.
Following insulin stimulation the number of GLUT4 molecules on the surface of 3T3-L1 adipocytes increases 3-20-fold (see e.g. Refs. 32 and 62-64). We have found a discordance between the effects of insulin on vpTR trafficking and glucose uptake. Glucose uptake was stimulated ϳ10-fold by insulin, whereas vpTR on the cell surface was increased 2-4-fold. There are a number of possibilities for these differences. First, there is evidence that GLUT4 transport activity is regulated by insulin (64 -69), in which case the increase in glucose uptake will reflect this increase in addition to translocation of GLUT4 to the plasma membrane. Unlike glucose uptake, analysis of vpTR directly measures translocation. Second, 3T3-L1 adipocytes express GLUT1 transporters (62), and although insulin induces only a small translocation of GLUT1, the increased GLUT1 on the surface will contribute to the overall stimulation of glucose uptake. Third, GLUT4 retention in intracellular compartments in the basal state may be more efficient than vpTR retention, which could result in a larger insulin effect. In this regard, it is known that IRAP and GLUT4 are not identically distributed among intracellular compartments in all cell types (24). Regardless, our data clearly demonstrate that vpTR is trafficked by a specialized insulin-responsive mechanism in 3T3-L1 adipocytes.
Biochemical studies indicate that at steady-state GLUT4 is found in compartments that contain Tf as well as compartments that do not (16,17,38,51,53). The GLUT4-containing compartments that lack Tf are believed to be the insulin-responsive retention compartment (IRC). A current model for the trafficking of GLUT4 is depicted in Fig. 8. Proteins trafficked by the insulin-regulated mechanism (i.e. GLUT4, IRAP, and vpTR) are internalized into endosomes (k INT ), from which they are specifically sorted to the IRC (k SORT ). In the basal state these proteins are slowly trafficked from the IRC back to the plasma membrane (k IRC ). Insulin increases the rate of traffick-ing from the IRC to the plasma membrane, thereby increasing the concentration of these proteins on the cell surface. TR, which is not efficiently sorted to the IRC, is recycled back to the cell surface at a rapid rate from endosomes (k REC ). We can further refine this model using the kinetic data derived from studies of the vpTR. Our data indicate that the k SORT step is very efficient, with very little of the chimera trafficking back to the cell surface by the fast endosomal recycling pathway. Therefore, we propose that the ϳ40% of GLUT4 previously detected in Tf-containing endosomes (16,17) does not traffic directly back to the cell surface via the general endosomal system but is efficiently retrieved from endosomes and targeted to the IRC. Since sorting from endosomes to the IRC is very efficient, it is surprising that such a large fraction of GLUT4 is found in Tf-containing endosomes (70). One interpretation of these findings is that although the efficiency of targeting from endosomes to the IRC is high, the rate of this reaction is not. Therefore, at steady state a significant fraction of vpTR would be in endosomes; this fraction, however, traffics to the IRC and not back to the cell surface. Inherent in this model is the ability of endosomes to sequester and transiently retain vpTR (GLUT4 and IRAP) before they are trafficked to the IRC. An alternative explanation consistent with the data is that in the basal state vpTR continually traffics between endosomes and the IRC. In either case, our data demonstrate the importance of the endosomal sorting step in maintaining the basal state retention of proteins that are trafficked by the insulin-regulated pathway, and therefore suggest that understanding this mechanism is as important as understanding the characteristics of the IRC.
Our results question whether insulin-regulated trafficking is a specific characteristic of fat and muscle. We have previously characterized the behavior of vpTR in CHO cells (71). The vpTR chimera behaves similarly in both CHO cells and 3T3-L1 adipocytes as follows: 1) the chimera is slowly recycled with respect to the TR; 2) insulin stimulates a large increase in the recycling rate of the chimera and has a small effect of the recycling of the TR; 3) GTP␥S increases recycling of the chimera but not the TR; 4) wortmannin blocks the effect of insulin on the recycling of the chimera; and 5) mutation of the dileucines at positions 76,77 releases retention of the chimera (29). The major difference in the behavior of the chimera in CHO and 3T3-L1 adipocytes is that insulin increases recycling of vpTR in fat cells by ϳ4-fold, whereas insulin has a smaller effect on the chimera in CHO cells (ϳ2-fold increase). Although the trafficking of vpTR in fat cells is more responsive to insulin, it is clear that CHO cells have a bona fide insulin-regulated recycling mechanism, since in both cell types the trafficking of vpTR is considerably more responsive to insulin than is the FIG. 8. A model for the insulin-regulated membrane recycling pathway in 3T3-L1 adipocytes. IRC is the insulin-regulated retention compartment. k int is the internalization pathway through clathrincoated pits; k rec is the fast endocytic recycling pathway utilized by TR; k sort is the putative sorting step from endosomes to the IRC; k IRC is the pathway from the IRC back to the surface; k IRC-endo is the putative pathway from the IRC to endosomes. See text for details.
trafficking of the TR. At present we do not know whether this is a characteristic of all cells or peculiar to CHO cells. In this regard, it has recently been shown that the insulin-responsive vesicular compartment exists in 3T3-L1 cells early in differentiation (72). Early in differentiation, preceding the expression of GLUT4, the distributions of GLUT1 and TR are shifted toward the interior of cells (i.e. increased sequestration), and in these conditions insulin stimulates a significant redistribution of these proteins to the cell surface. These data establish a precedent for the existence of insulin-responsive trafficking compartments in cells that do not express GLUT4 and are therefore in agreement with our conclusion that the insulinregulated, slow recycling endocytic mechanism is not a unique characteristic of cells that express GLUT4.