Insulin-responsive Aminopeptidase Trafficking in 3T3-L1 Adipocytes*

The insulin-responsive aminopeptidase (IRAP/VP165/gp160) was identified originally in GLUT4-containing vesicles and shown to translocate in response to insulin, much like the glucose transporter 4 (GLUT4). This study characterizes the trafficking and kinetics of IRAP in exocytosis, endocytosis, and recycling to the membrane in 3T3-L1 adipocytes. After exposure of 3T3-L1 adipocytes to insulin, IRAP translocated to the plasma membrane as assessed by either cell fractionation, surface biotinylation, or the plasma membrane sheet assay. The rate of exocytosis closely paralleled that of GLUT4. In the continuous presence of insulin, IRAP was endocytosed with a half-time of about 3–5 min. IRAP endocytosis is inhibited by cytosol acidification, a property of clathrin-mediated endocytosis, but not by the expression of a constitutively active Akt/PKB. Arrival in an LDM fraction derived via subcellular fractionation exhibited a slower time course than disappearance from the cell surface, suggesting additional endocytic intermediates. As assayed by membrane “sheets,” GLUT4 and IRAP showed similar internalization rates that are wortmannin-insensitive and occur with a half-time of roughly 5 min. IRAP remaining on the cell surface 10 min following insulin removal was both biotin- and avidin-accessible, implying the absence of thin-necked invaginations. Finally, endocytosed IRAP quickly recycled back to the plasma membrane in a wortmannin-sensitive process. These results demonstrate rapid endocytosis and recycling of IRAP in the presence of insulin and trafficking that matches GLUT4 in rate.

The metabolic hormone insulin promotes the disposal of glucose into its peripheral target tissues, adipose and muscle. Insulin-stimulated glucose uptake is mediated primarily by the rapid movement of the fat/muscle-specific glucose transporter (GLUT) 1 4 from a latent intracellular compartment to the cell surface (1). The subcellular trafficking of GLUT4 has been studied using an impermeant photolabel specific for the extra-cellular sugar-binding site of the transporter (2)(3)(4). These studies demonstrated that in the presence of insulin there was constant recycling of GLUT4 between the intracellular compartment and the plasma membrane, although the total amount of GLUT4 in the plasma membrane was elevated and constant. There are data supporting the existence of exocytotic intermediates in which GLUT4 might be present in the membrane but inaccessible to substrate (2). Also, a recent report studying GLUT4 in CHO cells demonstrated endocytotic intermediates similar to those described for a synaptic vesicle protein, synaptophysin (5,6).
Recently, the insulin-responsive aminopeptidase (IRAP) has been cloned, shown to reside in the GLUT4-containing basal compartment, and also translocate to the plasma membrane after insulin stimulation (7,8). The physiological function of IRAP has been hypothesized to be hormonal modulation of circulating vasoactive peptides such as vasopressin (9,10). Indeed, determining the normal function of IRAP might be critical in defining the mechanism of diabetic complications such as hypertension or peripheral vascular disease that might result from impaired translocation of IRAP. Therefore, investigating the normal trafficking and biology of IRAP is critical not only to illuminate insulin-stimulated trafficking but as a first step in eventually defining pathologies in conditions such as diabetes.
The purpose of this study was to describe the trafficking and kinetics of IRAP in the context of GLUT4 in 3T3-L1 adipocytes. Several studies have shown overlapping subcellular distribution of GLUT4 by light microscopy (11,12) and subcellular fraction (13,14). Whereas the steady state location of IRAP is overlapping with GLUT4, its trafficking has not been described in detail. One report suggests that IRAP and GLUT4 traffic differently in response to high glucose or glucosamine (15). Another demonstrates IRAP exclusion from cardiac secretory granules containing atrial natriuretic factor and GLUT4 (11). Finally, a third report studying rat fat demonstrates that IRAP does not internalize in the presence of insulin (16); this is in marked contrast to the constitutive recycling of GLUT4.
In an effort to define the trafficking and kinetics of IRAP movement, we have adapted protocols using biotinylation to develop assays measuring endocytosis and recycling of IRAP back to the cell surface. Additionally, we employed the plasma membrane sheet protocol to compare directly IRAP and GLUT4 levels on the cell surface. By using these techniques we find IRAP is endocytosed rapidly both in the presence and absence of insulin. Moreover, IRAP is endocytosed and exocytosed with rates similar to GLUT4. These techniques have also allowed us to study other details of IRAP as follows: its endocytosis is inhibited by cytosolic acidification, a property of clathrin mediated endocytosis, but not by a constitutively active Akt/PKB. Endocytic intermediates likely occur between exiting the * This work was supported by National Institutes of Health Grants DK39615 (to M. J. B.) and MSTP 5-T32-GM-07170 (to L. A. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  plasma membrane and arrival in the LDM. Finally, we can directly demonstrate endocytosed IRAP rapidly recycles to the plasma membrane in the presence of insulin.

EXPERIMENTAL PROCEDURES
Materials-Crystalline porcine insulin was a gift of Lilly. Polyclonal antibodies against IRAP cytoplasmic amino terminus were kindly donated by Metabolex, Inc. (Hayward, CA). Polyclonal sheep anti-GLUT4 antibodies were raised against a glutathione S-transferase fusion protein encoding the last 31 amino acids of the GLUT4 carboxyl terminus. Monoclonal mouse anti-IRAP were raised against the 110 amino acids of the IRAP amino terminus. Affinity purified/subtracted rhodamineconjugated donkey anti-sheep and affinity purified/subtracted fluorescein isothiocyanate-conjugated donkey anti-mouse antibodies were purchased from Jackson ImmunoResearch (West Grove, Pa). Bovine serum albumin used in translocation assays was from Calbiochem. Wortmannin was purchased from Sigma and stored as a 10 mM stock in Me 2 SO. 125 I-Protein A was purchased from ICN Radiochemicals (Irvine, CA). A bicinchoninic acid protein assay kit for the determination of protein concentrations was from Pierce as were 2,2Ј-azine-di[3-ethylbenzthiazoline sulfonate], streptavidin-HRP, neutravidin-HRP, sulfo-NHS-LC-LC-biotin, and sulfo-NHS-S-S-biotin. Unless otherwise noted, all other chemicals were from Sigma.
IRAP Biotinylation Assay for Translocation-The presence of IRAP on the cell surface was determined using an IRAP biotinylation assay similar to that described previously (19). 3T3-L1 adipocytes in 6-well dishes were washed twice in PBS, once in Leibovitz-15 media with 0.2% BSA, and left in that media for 2 h at 37°C. Cells were treated with 20 nM insulin for 20 min. All subsequent steps were performed at 4°C. Cells were washed twice in ice-cold KRPH (128 mM NaCl, 4.7 mM KCl, 1.25 mM CaCl 2 , 1.25 mM MgSO 4 , 5 mM NaPO 4 , 20 mM Hepes, pH 7.4) and treated with 1 ml of 0.5 mg/ml sulfo-NHS-LC-LC-biotin in KRPH for 30 min. Each plate was then bathed three times for 10 min each in KRPH containing 20 mM glycine, twice with KRPH, and finally lysed in 1 ml of Solubilization Buffer (1% Triton, 150 mM NaCl, 20 mM Tris-Cl, 5 mM EDTA, 1 mM phenylmethanesulfonyl fluoride, 10 g/ml aprotinin, 10 M leupeptin, 1 M pepstatin A, pH 7.4). The lysate was vortexed briefly, incubated for 15 min, and centrifuged at 23,000 ϫ g for 15 min. After passing the supernatants through a 45-m filter (Millipore), BCA assay was performed to determine protein concentration. 600 g of protein was diluted to 500 l with Solubilization Buffer and reacted with polyclonal anti-IRAP sera overnight, followed by 3-6 h incubation in protein A-Sepharose. These conditions were shown to consistently remove greater than 90% of IRAP from the supernatant. SDS-gels of the boiled and reduced immunoadsorbates were transferred to PVDFϩ membranes, blocked in TBS-T with 6% BSA, treated with 1 g/ml of streptavidin-HRP for 2 h, washed in TBS-T, and developed using ECLϩ (Amersham Pharmacia Biotech) on a STORM 860 Scanner detecting chemifluorescence. The signal intensity of quantitated samples was shown to be within the linear range of detection.
Subcellular Fractionation-Cells were fractionated by standard techniques (2). Briefly, 3T3-L1 adipocytes were resuspended in HES (255 mM sucrose, 20 mM Hepes, pH 7.4, 1 mM EDTA), homogenized by passing through an EMBL Cell Cracker (clearance of 18 m), and subjected to differential centrifugation. The supernatant from the following spins were serially removed and pelleted in a Ti70 rotor as follows: 19,000 ϫ g (20 min), 41,000 ϫ g (20 min), and 180,000 ϫ g (75 min). The first 19,000 ϫ g pellet was resuspended, loaded onto a sucrose cushion (1.12 M sucrose, 20 mM Hepes, pH 7.4, 1 mM EDTA), and isolated from the interface yielding the Plasma Membrane (PM) fraction as the pellet of a 41,000 ϫ g spin (20 min). The last 180,000 ϫ g pellet corresponded to the Low Density Microsome (LDM) fraction. After resuspension of pellets in Solubilization Buffer, protein concentration was determined by bicinchoninic acid protein assay, and 20 g of each fraction was loaded for Western blotting. For immunoprecipitation, 100 g of LDM and 50 g of PM were solubilized in 1% Triton, passed through a 45 m filter, and used for immunoprecipitation of IRAP as detailed above. For experiments measuring total cell IRAP in parallel, aliquots removed after passage through the Cell Cracker but before differential centrifugation were raised to 1% Triton and 100 mM NaCl and subjected to immunoprecipitation of IRAP as detailed above.
Immunofluorescence of Plasma Membrane Sheets-To measure the presence of GLUT4 and IRAP on the cell surface, the plasma membrane sheet assay was used (20,21) with modifications to allow quantitation of triple label. Plasma membrane "sheets" were prepared and processed for indirect immunofluorescence using affinity purified sheep antibodies to the carboxyl-terminal portion of GLUT4 and pooled ascites derived from three separate monoclonal hybridomas raised against the amino terminus of IRAP. Following primary incubation, membranes were incubated with rhodamine-conjugated anti-sheep secondary and fluorescein isothiocyanate-conjugated anti-mouse secondary antisera. To stain for plasma membranes, 0.1 g/l phosphatidylethanolaminebiotin (Molecular Probes) and 0.1 g/l Marina-Blue Neutra-Lite Avidin (Molecular Probes) were included with primary and secondary antibodies, respectively. The images were acquired using a Princeton Instruments cooled CCD camera, and the amounts of GLUT4 and IRAP on the plasma membrane were quantitated by measuring the fluorescence intensity of at least seven fields. Digital image processing was performed as described previously except quantitated in 2 channels of florescence (21,22).
For wortmannin treatment, cells were treated with drug either prior to or after insulin stimulation. For the former, 500 nM wortmannin was added for the final 50 min, and insulin was added for the final 20 min of a 2-h incubation in serum-free media. For the latter, cells were stimulated with 20 nM insulin for 20 min, incubated with 500 nM wortmannin at 4°C for 30 min, and warmed in the presence of 500 nM wortmannin for 10 min.
IRAP Reversible Biotinylation Assay for Endocytosis-IRAP endocytosis was measured by its protection from cleavage by a membraneimpermeable reagent based on published pulse-chase protocols (24,25). Cells were stimulated with insulin, biotinylated, and glycine-treated as for Biotinylation Translocation Assay except for the substitution of sulfo-NHS-S-S-biotin for sulfo-NHS-LC-LC-biotin. After washing twice with KRPH, cells were raised to 37°C by the addition of prewarmed Leibovitz-15 medium containing 0.2% BSA and 100 nM insulin and then floated on a water bath at a setting sufficient to maintain temperature. Following "chase" incubation periods of 0.5-20 min, cells were removed to an ice bath, and washed twice with KRPH at 4°C. All subsequent steps were performed at 4°C. Cells were washed once and incubated for 15-min intervals in Cleavage Buffer (50 mM glutathione, 90 mM NaCl, 1 mM MgCl, 0.1 mM CaCl, 60 mM NaOH, 0.2% BSA, pH 8.6) for a total of 45 min. Subsequent to washing cells in KRPH three times, cells were solubilized and processed as for Biotinylation Translocation Assay using an anti-IRAP immunoprecipitation protocol with the omission of reducing agent in SDS-PAGE sample buffer. Alternatively, cell lysates were incubated with streptavidin-agarose for 1 h and eluted with sample buffer containing reducing agent and developed as for Biotinylation Translocation Assay with the replacement of rabbit anti-IRAP and 125 I-protein A for streptavidin-HRP. Data are expressed either as "Cell Surface IRAP" or "Percent Escaping Cleavage." The former is determined by first subtracting the value of biotinylation after immediate cleavage from all other values, dividing by total uncleaved signal, and subtracting from 1. The latter is determined by first subtracting the value of biotinylation after immediate cleavage from all other values and dividing by maximum protected signal.
Cytosolic Acidification-As described previously (26,27), cells were exposed to a media containing acetic acid in order to acidify cytosol. After biotinylation and glycine treatment cells were washed twice in Leibovitz-15, pH 5.0 (HCl), washed twice in Leibovitz-15, pH 5.0 (acetic acid), and left in a third wash of Leibovitz-15, pH 5.0 (acetic acid), for 30 min at 4°C. Cells were exposed to identical media prewarmed to 37°C, chilled at various times, and cleaved as described under Reversible Biotinylation Assay.
IRAP Biotinylation Assay Avidin Accessibility-The accessibility of biotin to avidin was performed as described (28) with minor modifications. All steps were performed at 4°C. Following biotinylation and glycine treatment, cells were treated with avidin buffer (KRPH, 0.2% BSA, 250 g/ml avidin) for 1 h, washed twice in KRPH, and incubated for 30 min in biocytin buffer (KRPH, 0.2% BSA, 250 g/ml biocytin) to quench remaining available avidin sites. Cells were rinsed three times in KRPH and processed as in Biotinylation Assay for Translocation with the omission of boiling step before loading samples for SDS-PAGE to preclude disassembly of the avidin-biotin complexes.
Reappearance Assay-The recycling of IRAP back to the cell surface was monitored based on a published protocol examining recycling of antigen in B cells (29). Cells were stimulated with insulin, biotinylated, glycine-treated, and incubated at 37°C for 5 min as described for the IRAP Reversible Biotinylation Protocol with the substitution of LC-LC biotin for cleavable S-S biotin. Cells were then avidin-treated and biocytin-quenched as described above. Cells were warmed again to 37°C for indicated times and returned to 4°C by the addition of cooled KRPH. All subsequent steps were performed at 4°C. Next, cells were treated with Neutravidin Buffer (KRPH, 0.2% BSA, 25 g/ml neutravidin-HRP) for 1 h and quenched with biocytin a second time. Lysates were prepared, and IRAP was immunoprecipitated from 1 mg of protein as described above. 2,2Ј-Azine-di[3-ethylbenzthiazoline sulfonate] (ABTS) was added to immunoprecipitates, incubated for 10 min at 37°C, pelleted, and the reaction product measured at A 405 nm .

RESULTS
IRAP Translocates like GLUT4 from the LDM to the PM-IRAP recently has been identified (8), cloned (7), and its subcellular distribution shown to be overlapping with GLUT4 (11)(12)(13)(14). In an attempt to characterize IRAP trafficking in 3T3-L1 adipocytes, we measured insulin-stimulated translocation to the cell surface by three complimentary methods. The first was a quantitative biotinylation protocol adapted from published studies (8,14). Insulin treatment induced a large increase in IRAP available at the cell surface for biotinylation (Fig. 1A). We next applied techniques that are useful for measuring GLUT4 translocation as well as that of IRAP. After stimulating cells with insulin or maintaining them in a basal state, we homogenized cells and fractionated them into Plasma Membrane (PM), and Low Density Microsome (LDM) fractions ( Fig.  1B) (30). Insulin elicited a dramatic redistribution of IRAP from the LDM fraction to the PM fraction, similar to GLUT4 (31). Finally, to compare directly trafficking of GLUT4 and IRAP in the same cells, we labeled membrane sheets for the presence of IRAP and GLUT4 (Fig. 1C). Quantitation of images revealed a similar degree of translocation between the two proteins. These results demonstrating insulin-stimulated translocation of IRAP are consistent with previous reports (12,14,32) and serve to emphasize the parallel between IRAP and GLUT4 trafficking.
In the Presence of Insulin IRAP Is Rapidly Endocytosed and Continues to Recycle to the Plasma Membrane-A recent report demonstrated that IRAP does not endocytose in the presence of insulin in rat fat (16). To measure internalization in our cells, we adapted a reversible biotinylation technique to pulse-chase labeled IRAP into cells and determine the amount remaining on the surface by its susceptibility to a membrane-impermeable cleavage reagent. Cells were stimulated with insulin, biotiny- C, cells were stimulated as in A, and the extent of GLUT4 or IRAP translocation was determined by the plasma membrane sheet assay. Images were captured with a cooled CCD camera; a composite photomicrograph of a typical plasma membrane sheet experiment depicting translocation is shown. Images such as these were quantitated to generate the shown graph. Values in the presence of insulin were set to 100%, and the other values were normalized accordingly. The translocation graph represents the mean Ϯ S.E. of three experiments. PE-biotin, phosphatidylethanolamine-biotin. lated at 4°C to label IRAP, and rewarmed in the presence of insulin for lengthening time. At the indicated time points, cells were chilled again and treated with an impermeant reducing agent, which cleaves the disulfide bond joining the biotin moiety with the protein-binding NHS ester. Thus, at time 0 most of the total label should be removed, and if internalization occurs, at later time points the biotin label should be protected. By using this method we indeed saw rapid and extensive internalization of IRAP (Fig. 2, A and B). In Fig. 2A, lysates were precipitated with avidin-agarose and submitted to Western blot analysis with antibodies against IRAP. In Fig. 2B lysates were immunoprecipitated with anti-IRAP antibodies, and the biotinylated IRAP was detected with streptavidin-HRP. Results were quantitated and expressed as cell surface IRAP. By using either method of detection, IRAP was found to be rapidly internalized in the presence of insulin with a t1 ⁄2 of approximately 3 min.
To determine the kinetics of the entire population of IRAP, cells were biotinylated at 37°C in the presence of insulin for increasing times, and afterward biotinylated material was precipitated from cell lysates with avidin-agarose as in Fig. 2A. Precipitated material was then assayed for IRAP by Western blot analysis and compared with total cellular IRAP. Fig. 2C demonstrates a time-dependent increase in IRAP biotinylation such that virtually all IRAP in the cell has translocated to the cell surface at least once by 1 h. Furthermore, since cell surface levels of IRAP are constant after 10 min of insulin stimulation (Fig. 5A), these data indicate that intracellular stores of IRAP are constantly driven to the cell surface to replace the internalized IRAP demonstrated in Fig. 2, A and B. IRAP Endocytosis Is Inhibited by Cytosol Acidification but Not by myr-Akt-Cytosolic acidification has been shown to inhibit potently clathrin-mediated endocytosis by disruption of the clathrin lattice (33). Insulin-treated, biotinylated cells were warmed for 2 or 5 min in regular or cytosolic acidification media and then exposed to cleavage buffer (Fig. 3A). For the final washout lane, cells were warmed for 5 min in acidification media, chilled, washed, and warmed for the final 15 min in pH 7.4 media. Cytosolic acidification prevented any sequestration from cleavage buffer, but its removal allowed biotinylated IRAP to endocytose and escape cleavage. These data support a role for clathrin in the endocytosis of IRAP.
We next tested the mechanism of Akt-induced increases of plasma membrane IRAP. A constitutively active myristoylated Akt results in a large increase in IRAP and GLUT4 present at the plasma membrane (Fig. 3B, inset) (21). To test whether Akt might be inhibiting endocytosis, cells expressing a constitutively active myr-Akt construct were compared with empty vector controls in a reversible biotinylation endocytosis assay as in Fig. 2B. Both cells demonstrated similar rates indicating that inhibition of endocytosis does not account for the ability of myr-Akt to stimulate translocation of IRAP to the cell surface.
IRAP Arrival in the LDM Is Slower Than Escape from Cleavage-Because of the likelihood that development of biotinylated IRAP's resistance to cleavage is a very early event in its endocytosis, the rate of escape from cleavage was compared with the rate of arrival into LDM. Sensitivity to cleavage by impermeant reducing agent was determined from experiments performed as in Fig. 2B, with the data expressed as percent escaping cleavage. Arrival into the LDM was determined by fractionating cells treated as in Fig. 2, A and B, immunoprecipitating IRAP from the LDM pellet, and assaying for biotinylated IRAP with streptavidin-HRP. At 5 min after warming there was a smaller quantity of IRAP present in LDM compared with that which is inaccessible to extracellular cleavage (Fig. 4A). To verify this delay, the escape from cleavage was determined in the same cells used to measure arrival in LDM. As shown in Fig. 4B, after 1 min of internalization, there was a statistically significant difference in the percent escaping cleavage (40%) versus that arriving in LDM (15%). This discrepancy does not simply reflect contamination of PM with LDM as indicated by its disappearance at later time points. Thus, after a 40-min chase, 60% more of the internalized IRAP is pelleted in the LDM fraction compared with the recovery of internalized IRAP in the LDM after a 1-min chase.
IRAP Exocytic and Endocytic Rates Are Similar to GLUT4 in the Presence and Absence of Insulin, Respectively-To compare IRAP and GLUT4 trafficking directly, plasma membrane sheets were used to assay the quantity of these proteins on the same cell surfaces at different time points after the addition or removal of insulin. The data represented in Fig. 5A demonstrate cells treated with 20 nM insulin at time 0 and warmed for lengthening periods before surface levels of IRAP and GLUT4 were measured. IRAP and GLUT4 exocytotic rates are very similar, with a half-time of approximately 5 min. To measure endocytotic rate in the absence of insulin, a similar analysis was performed as shown in Fig. 5B. Insulin was withdrawn, and surface levels of IRAP and GLUT4 were measured after lengthening times. Again, rates of endocytosis correlate closely, with a half-time of approximately 5 min. Finally, we compared the effect of wortmannin on IRAP and GLUT4 trafficking. Wortmannin potently inhibited translocation of the two proteins if added before insulin treatment (Fig. 5C), similar to previous reports (12). When wortmannin was added after insulin stimulation in parallel to insulin withdrawal, the surface levels of IRAP and GLUT4 decreased identically. Moreover, the rates of internalization were identical to those observed when insulin was removed in the absence of wortmannin (Fig. 5C).
Plasma Membrane IRAP Is Biotin-and Avidin-accessible in the Presence and after the Withdrawal of Insulin-Recent reports (6,34) have indicated the presence of synaptic vesicle proteins in compartments continuous with the plasma membrane but restricted by thin-necked invaginations limiting accessibility to the large (69 kDa) probe avidin. A recent report (5) has extended these findings to CHO cells exogenously expressing GLUT4 showing that GLUT4, but not transferrin receptor, was selectively sequestered in a compartment near the plasma membrane as determined by immunofluorescence. We therefore sought evidence for the presence of IRAP in such a compartment in 3T3-L1 adipocytes. To test IRAP accessibility to biotin, cells were exposed to insulin and then withdrawn from the hormone for increasing times before being chilled and biotinylated to determine the amount of accessible IRAP. Shown in Fig. 6A is a comparison of the amount of cell surface IRAP as determined by biotin accessibility and plasma membrane sheets. Given the close correlation, it is likely that the immu-nostained IRAP on sheets is all accessible to biotin. To determine avidin accessibility, cells either in the presence of insulin or 10 min after removal were biotinylated as above, maintained at 4°C, and then exposed to avidin to complex all available biotin sites. The remaining avidin was quenched with biocytin, and IRAP was immunoprecipitated and Western-blotted for streptavidin-HRP to detect free biotin on IRAP. When 3T3-L1 adipocytes were exposed to avidin immediately after biotinylation, greater than 90% of the biotin sites on the protein were exofacially disposed and available for binding to avidin. Similar data were obtained after allowing the cells to internalize biotinylated IRAP for 10 min. Thus, IRAP appears accessible to both biotin and avidin at all time points tested.
Rapid Wortmannin-sensitive Recycling of IRAP Back to the Cell Surface-To ascertain whether there is continuous recycling of IRAP in the presence of insulin, IRAP was biotinylated, allowed to internalize for 5 min, chilled, and all remaining biotin sites on the cell surface were quenched with avidin. Cells were then rewarmed for the indicated times to allow exocytosis of internalized label. Finally, cells were chilled and exposed to neutravidin-HRP to bind newly exposed biotin residues. IRAP was immunoprecipitated and assayed for HRP activity. Fig. 7A demonstrates a time-dependent increase in HRP activity associated with IRAP. After 3 min of warm-up, 60% of maximum measured levels were achieved. In Fig. 7B, cells were exposed to 500 nM wortmannin during the final warm-up, demonstrating wortmannin sensitivity of the reappearance process.

DISCUSSION
Originally identified as a constituent of GLUT4-containing vesicles, the physiological role and trafficking pathways of IRAP still remain undefined. The current study represents an attempt to define the trafficking and kinetics of IRAP. We demonstrate four novel findings. First, IRAP trafficking resembles closely that of GLUT4. IRAP displays similar exocytosis and internalization rates, also undergoing constant recycling between the plasma membrane and the LDM fraction in the presence of insulin. Moreover, IRAP is likely endocytosed via clathrin-coated pits, the means proposed for GLUT4 endocytosis (35)(36)(37)(38). Second, the candidate insulin effector Akt does not inhibit IRAP endocytosis, indicating a truly insulinomimetic activity of the kinase on exocytosis. Third, endocytic intermediates likely occur between exit from the cell surface and arrival in LDM fraction. Fourth, IRAP is unlikely to be endocytosed via plasma membrane continuous compartments invoked for synaptic vesicle trafficking. These results compliment previous reports demonstrating the steady-state colocalization of IRAP with GLUT4 and lend support to the idea of a family of insulin-responsive proteins that traffic similarly (11)(12)(13)(14).
An important finding of this study is the continuous endocytosis of IRAP in the presence of insulin. By using a pulse-chase reversible biotinylation technique, we demonstrate the recy-cling of IRAP in the presence of insulin is rapid and robust, such that within 5 min of insulin-stimulated translocation, 80% of surface-labeled IRAP is intracellular. During this period when steady-state levels are unchanged, intracellular IRAP is mobilized to the cell surface to replace the endocytosed population. These data correlate well with the reported constant recycling of GLUT4 in the presence of insulin (2,4), but the data disagree with a recent study in rat fat cells in which no cell surface-labeled IRAP appeared in the LDM over the course of 30 min of insulin stimulation (16). To reduce the likelihood that technical differences might account for this discrepancy, we have used two methods to detect endocytosed IRAP. First, we isolated biotinylated IRAP by avidin-agarose beads and stained blots with anti-IRAP antibodies comparing total IRAP protein to the amount escaping cleavage reagent. Second, we immunoprecipitated IRAP and stained blots with avidin-HRP to compare total biotin signal in IRAP to the biotin signal in IRAP FIG. 4. Comparison of IRAP appearance in the LDM versus escape from cleavage. A, 3T3-L1 adipocytes were stimulated with insulin, biotinylated, quenched, and warmed for increasing times and exposed to biotin cleavage buffer as in Fig. 2B. Cells were then homogenized in HES, and subcellular fractions were prepared. IRAP was immunoprecipitated from resuspended LDM pellets, and the immunoprecipitate was blotted for biotin-IRAP with streptavidin-HRP (solid line). Results are presented as the mean Ϯ S.D. of three experiments. Shown also are experiments performed as in Fig. 2B with IRAP immunoprecipitated from whole cells (shaded line). Results are presented as the mean Ϯ S.D. of three experiments. Data are expressed as percent escaping cleavage with maximum values set to 100%. B, cells were treated exactly as above, warmed for 1 min, and then homogenized. Before fractionation, portions of sample were removed and raised to 1% Triton to measure total cellular-protected IRAP. Remaining homogenates were fractionated to yield LDM pellets for immunoprecipitation to determine protected IRAP in LDM. The asterisk denotes the statistical significant (p Ͻ 0.05) difference between whole cell biotin-IRAP and LDM biotin-IRAP. Results are presented as the mean Ϯ S.E. of three experiments. Data are expressed as above.

FIG. 5. Comparison of IRAP and GLUT4 rates of exocytosis and internalization.
A, 3T3-L1 adipocytes were incubated for 2 h in serum-free media and stimulated with insulin (20 nM) for increasing lengths of time. IRAP and GLUT4 translocation were determined by plasma membrane sheet assay. The graph represents the mean Ϯ S.E. of three experiments. B, 3T3-L1 adipocytes were stimulated with insulin (20 nM) for 20 min as in Fig. 1C. Cells were briefly chilled and stripped of insulin with an acid wash. After warming for increasing times, GLUT4 and IRAP translocation was measured by plasma membrane sheet assay. The graph represents the mean Ϯ S.E. of three experiments. C, 3T3-L1 adipocytes were treated as in B except exposed to 500 nM wortmannin either for 30 min prior to insulin treatment (ϩ Wortmannin) or during 10 min of warm-up (10 min Wortmannin). The graph represents the mean Ϯ S.E. of three experiments. escaping cleavage. Both techniques demonstrated endocytosis in the presence of insulin. Perhaps differences in 3T3-L1 adipocytes versus rat fat might account for the discrepancy. In agreement with the above report, we do also detect endocytosis of transferrin receptor in the presence of insulin. 2 Although no previous reports have quantified IRAP internalization in adipocytes, GLUT4 internalization rates have been measured in the presence of insulin (4). The rate of internalization in 3T3-L1 adipocytes was ascertained using a labeled substrate analog covalently bound to the GLUT4 receptor by photoactivation. By following this kinetic pool in the presence of insulin, the authors determined a half-time for endocytosis of 4.2 min for GLUT4 exit from the plasma membrane fraction. This is very similar to the 3-min half-time measured for IRAP using techniques described above. Although the differences are probably within experimental variation, a functional measurement of escape from a membrane cleavage reagent could be different from movement between homogenized cell fractions, as used for GLUT4. We indeed detected a delay when comparing rates of escape from cleavage reagent to rates of arrival in the LDM. Within the same samples, there was more than a 2-fold difference in the amount of IRAP internalized into the whole cell lysate versus that recovered in the LDM after a 1-min chase. This is reflected in a significantly lower recovery of endocytosed signal in the LDM at early time points compared with ones at equilibrium; 60% more of the endocytosed signal was recovered in LDM after a 40-min chase compared with a 1-min chase. The absence of endocytosed signal from the LDM at early time points might indicate that the protected IRAP is either in a closed but pre-fission state or perhaps in an immature vesicle of sedimentation properties different than those of LDM.
When measured in parallel, GLUT4 and IRAP showed almost identical exocytic and endocytic rates, prompting the question of whether IRAP is endocytosed by a similar mechanism as GLUT4. GLUT4 has been shown by electron microscopy (39) and cytosolic acidification (40) to bud from clathrincoated pits. IRAP possesses similar intracellular dileucine motifs to GLUT4, but these were not necessary for endocytosis of IRAP in chimerae with transferrin receptor expressed in CHO cells (41). Therefore, we assayed IRAP endocytosis under conditions inhibiting clathrin lattice formation, and we found that cytosolic acidification completely and reversibly ablated internalization, supporting a role for clathrin-mediated endocytosis of IRAP. Considering these similarities in mechanism and rate, it seems likely IRAP and GLUT4 are endocytosed by similar means.
A recent report concluded that GLUT4 endocytosis in CHO cells (5) exhibits characteristics similar to those proposed for synaptic vesicles (6,34). These latter studies suggest that synaptic vesicle biogenesis might occur from compartments continuous with the plasma membrane but with thin-necked constrictions such that whereas biotin may access this 2 L. A. Garza and M. J. Birnbaum, unpublished observations. FIG. 6. Biotin and avidin accessibility to IRAP in the presence and after the withdrawal of insulin. A, 3T3-L1 adipocytes were stimulated with insulin, then withdrawn for increasing amounts of time as in Fig. 5B, and chilled. Cells were biotinylated, quenched, and lysed for IRAP immunoprecipitation and Western blotting with streptavidin-HRP (dashed line). The graph represents the mean Ϯ S.E. of three experiments. The IRAP data from Fig. 5B are replotted (shaded line). B, 3T3-L1 adipocytes were treated as in A except, following biotinylation at 0 or 10 min, cells were exposed to excess avidin at 4°C and then quenched in excess biocytin before homogenization, immunoprecipitation of IRAP, and Western blotting with streptavidin-HRP. The graph represents the mean Ϯ S.E. of three experiments.

FIG. 7. Recycling of endocytosed IRAP back to the cell surface.
A, 3T3-L1 adipocytes were stimulated with insulin, biotinylated, and quenched as in Fig. 1A and then warmed for 5 min at 37°C before being chilled and treated with avidin as in Fig. 6B to quench all available surface sites. Cells were then rewarmed for increasing times before being chilled again and incubated with neutravidin-HRP to bind any newly available sites. Finally, cells were lysed, and immunoprecipitated IRAP was incubated with ABTS to measure HRP activity. Colorimetric changes were measured at 405 nm. B, cells were treated as above except incubated in the presence or absence of 500 nM wortmannin during final warm-up. compartment, avidin may not (6). Since IRAP and GLUT4 show similar means of endocytosis, it seemed possible that the insulin-responsive vesicle might also bud directly from the plasma membrane, particularly in light of the recent demonstration of the presence of GLUT4 in peripheral structures in CHO cells (5). However, we found that IRAP was both biotin-and avidinaccessible in the presence of insulin and, after its withdrawal, inconsistent with the hypothesis of a plasma membrane-continuous donor compartment as formulated in PC12 cells (6). A second prediction of the work in synaptic vesicles and also reported for GLUT4 in CHO cells is that cooling of cells to 15-18°C would selectively inhibit endocytosis of IRAP but not transferrin receptor. Attempts to demonstrate this phenomenon for IRAP in 3T3-L1 cells have also been unsuccessful. 2 Developing an assay for IRAP endocytosis has allowed us to ask whether the putative downstream signaling molecule from the insulin receptor Akt is stimulating exocytosis or inhibiting endocytosis. A constitutively active construct of the Akt kinase did not inhibit endocytosis of IRAP indicating that Akt increases cell surface levels of IRAP by a truly insulinomimetic capacity of stimulating exocytosis.
In conclusion, this study provides evidence for similar if not identical trafficking in rate and path between IRAP and GLUT4. The endocytosis of IRAP appears be clathrin-mediated and not analogous to models proposed for synaptic vesicle recycling. We also provide novel evidence for endocytic intermediates and the role of Akt in stimulating exocytosis. The correlation in trafficking between GLUT4 and IRAP imply that investigations of the mechanisms of GLUT4 retention and recycling are equally pertinent to IRAP. Indeed, given the capacity of microinjected IRAP amino-terminal fragments to stimulate GLUT4 translocation (42), the molecular machinery responsible for sorting these molecules will likely be identical. Future studies might further delineate these broadly outlined steps of endocytosis and recycling of IRAP into more detailed movements between specific compartments.