Evidence for a Role for ADP-ribosylation Factor 6 in Insulin-stimulated Glucose Transporter-4 (GLUT4) Trafficking in 3T3-L1 Adipocytes*

ADP-ribosylation factors (ARFs) play important roles in both constitutive and regulated membrane trafficking to the plasma membrane in other cells. Here we have examined their role in insulin-stimulated GLUT4 translocation in 3T3-L1 adipocytes. These cells express ARF5 and ARF6. ARF5 was identified in the soluble protein and intracellular membranes; in response to insulin some ARF5 was observed to re-locate to the plasma membrane. In contrast, ARF6 was predominantly localized to the plasma membrane and did not redistribute in response to insulin. We employed myristoylated peptides corresponding to the NH2 termini of ARF5 and ARF6 to investigate the function of these proteins. Myr-ARF6 peptide inhibited insulin-stimulated glucose transport and GLUT4 translocation by ∼50% in permeabilized adipocytes. In contrast, myr-ARF1 and myr-ARF5 peptides were without effect. Myr-ARF5 peptide also inhibited the insulin stimulated increase in cell surface levels of GLUT1 and transferrin receptors. Myr-ARF6 peptide significantly decreased cell surface levels of these proteins in both basal and insulin-stimulated states, but did not inhibit the fold increase in response to insulin. These data suggest an important role for ARF6 in regulating cell surface levels of GLUT4 in adipocytes, and argue for a role for both ARF5 and ARF6 in the regulation of membrane trafficking to the plasma membrane.

Insulin stimulates glucose disposal in peripheral tissues by virtue of the expression of the GLUT4 glucose transporter isoform (1)(2)(3). In the absence of insulin, this transporter is intracellularly sequestered within the elements of the endosomal system, the trans Golgi network and a specialized storage compartment (4 -7). Upon insulin stimulation or muscle contraction, GLUT4 is re-distributed from these intracellular locations to the plasma membrane, resulting in a dramatic increase in the rate of glucose entry into these tissues (4 -7). Several studies have suggested that the insulin-stimulated translocation of GLUT4 to the plasma membrane is mechanistically akin to the fusion of small synaptic vesicles with the neuronal plasma membrane (reviewed in Ref. 2). This has been supported by the identification of a morphologically similar GLUT4 storage compartment within adipocytes, and by the identification of soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs) 1 located in GLUT4 vesicles (the v-SNAREs cellubrevin and vesicle-associated membrane protein 2) (8 -11) which bind in a highly specific manner to t-SNAREs located in the adipocyte plasma membrane (Syntaxin 4 and Syndet) (8,10,(12)(13)(14). Vesicle-associated membrane protein 2 has been shown to be the predominant v-SNARE that targets small synaptic vesicles to the pre-synaptic plasma membrane by interacting with the cognate t-SNAREs, syntaxin1, and synaptosome-associated protein of 25 kDa (SNAP-25) found on the target membrane (15). Recent studies implicating vesicle-associated membrane protein 2 in insulinstimulated GLUT4 translocation further strengthen the mechanistic parallels between regulated exocytosis of small synaptic vesicles and the insulin-stimulated movement of GLUT4-containing vesicles to the adipocyte cell surface (9,16).
ADP-ribosylation factors (ARFs) are a family of GTP-binding proteins (17,18). To date, 6 isoforms have been identified in mouse tissues (19) which fall within three groups, ARFs1, -2 and -3 constitute group I, ARFs4 and -5 group II and ARF6 is the sole member of group III identified to date. ARF proteins have been proposed to play several roles in the control of membrane traffic, including the formation of secretory vesicles at the trans Golgi network, regulating endosome-endosome fusion, and notably in regulating the fusion of secretory vesicles with the plasma membrane in bovine adrenal medulla cells (20 -25). Given the importance of ARF proteins in regulated membrane trafficking, we set out to identify which ARF isoforms were expressed in murine 3T3-L1 adipocytes, how their distribution was modulated by insulin, and whether they played a role in insulin-stimulated GLUT4 translocation.
Here we show that adipocytes express ARF5 and ARF6, as determined by immunoblotting with ARF-specific antibodies. ARF5 was observed to exhibit modest re-distribution to the plasma membrane in response to insulin. In contrast, ARF6 was predominantly located within plasma membrane subcellular fractions, and its distribution was not altered by insulin treatment. Using myristoylated ARF NH 2 -terminal peptides to inhibit ARF action in permeabilized cells, we show that myristoylated ARF6 peptide partially inhibits insulin-stimulated glucose transport and GLUT4 translocation, whereas ARF1and ARF5-myristoylated peptides were without effect. Insulin stimulates the movement of other proteins to the cell surface, including the transferrin receptor (TfR) and GLUT1 (26,27). Strikingly, we observed marked inhibition of insulin-stimulated TfR and GLUT1 movement to the cell surface in the presence of myristoylated ARF5 peptide, implying an important role for ARF5 in the stimulated delivery of recycling membrane proteins to the cell surface. Myristoylated ARF6 peptide decreased the cell surface abundance of TfR and GLUT1 in both the basal and insulin-stimulated states, but did not inhibit the ability of insulin to increase cell surface levels of these proteins. These data argue for an important role for ARF6 in regulating cell surface levels of GLUT4 in adipocytes, and provide evidence for a role for both ARF5 and ARF6 in the regulation of membrane trafficking to the plasma membrane.

EXPERIMENTAL PROCEDURES
Materials-␣-Toxin was from Calbiochem, United Kingdom, wortmannin from Sigma, UK, and 125 I-transferrin and [ 14 C]sucrose were from NEN Life Science Products Inc. and Amersham International, respectively. All other reagents were as described (16,28).
Cell Culture-3T3-L1 fibroblasts were grown and differentiated into adipocytes exactly as described in Refs. 16 and 28. Cells were used between passages 3 and 10 in all experiments, and between days 8 and 13 after induction of differentiation. Prior to use, cells were incubated in serum-free Dulbecco's modified Eagle's medium for 2 h. Subcellular Fractionation of Adipocytes-Adipocytes were subjected to a differential centrifugation procedure as described previously (28,29). Briefly, cells were scraped and homogenized in ice-cold HES (20 mM HEPES, 1 mM EDTA, 255 mM sucrose, pH 7.4, 5 ml/10-cm plate) containing protease inhibitors (1 g/ml pepstatin A, 0.2 mM diisopropyl fluorophosphate, 20 M L-transepoxysuccinyl-leucylamido-4-guanidiniobutane, and 50 M aprotinin). The homogenate was centrifuged at 19,000 ϫ g for 20 min at 4°C. The pellet from this spin was resuspended in 2 ml HES, layered onto 1 ml of 1.12 M sucrose in HES, and centrifuged at 100,000 ϫ g for 1 h at 4°C in a swing-out rotor. Plasma membranes were collected from the interface by careful aspiration, resuspended in HES, and collected by centrifugation at 41,000 ϫ g for 20 min at 4°C. The supernatant from the 19,000 ϫ g spin was recentrifuged at 41,000 ϫ g to yield a high density microsomal pellet and the supernatant from this spin centrifuged at 180,000 ϫ g for 75 min at 4°C to collect low density microsomes. All fractions were resuspended in equal volumes of HES buffer (cell equivalents), snap frozen in liquid nitrogen, and stored at Ϫ80°C prior to use.
Permeabilization of 3T3-L1 Adipocytes-3T3-L1 adipocytes were washed twice with IC buffer (10 mM NaCl, 20 mM Hepes, 50 mM KCl, 2 mM K 2 HPO 4 , 90 mM potassium glutamate, 1 mM MgCl 2 , 4 mM EGTA, 2 mM CaCl 2 , pH 7.4) at 37°C, then incubated in 500 l of ICR buffer (IC buffer plus 4 mM MgATP, 3 mM sodium pyruvate, 100 g/ml bovine serum albumin, pH 7.4) containing ␣-toxin at 250 hemolytic units/ml for 5 min to permeabilize the plasma membrane. The medium was removed and the cells covered with 500 l of ICR buffer containing peptides, insulin, or vehicle as described in the figure legends. This methodology has been documented in Ref. 30.
Deoxyglucose Transport Measurements-␣-Toxin is a 34-kDa protein which inserts into the plasma membrane and oligomerizes to form a 3-nm aqueous pore that allows passage of molecules up to ϳ5 kDa across the cell membrane. We therefore employed this experimental system to determine the role of ARF proteins on [ 3 H]2-deoxy-D-glucose (deoxy-Glc) transport as has been described previously (30). After permeabilization and incubation with peptides and insulin as described in the figure legends, 50 l of radioisotope solution was added to each well of adipocytes such that the final concentration of deoxy-Glc was 50 M and 0.5 Ci/well. Also included in this 50-l aliquot was [ 14 C]sucrose (final concentration 50 M, 0.05 Ci per well) so as to allow estimation of the nonspecific association of sugar with the cells. The transport rates presented have been corrected for this calculation. Uptake was carried out for 3 min, then the cells were rapidly washed three times in ice-cold phosphate-buffered saline and air-dried. Cell associated radioactivity was determined by solubilizing the cells in 1% Triton X-100. Nonspecific association of radioactivity with the cells amounted to less than 20% of the specific uptake under these conditions. Plasma Membrane Lawn Assays for GLUT4 Translocation-After experimental manipulations, coverslips of adipocytes were rapidly washed in ice-cold buffer for the preparation of plasma membrane lawns exactly as described in Ref. 9. After fixation in paraformaldehyde, plasma membrane lawns were incubated with anti-GLUT-4 (1:100 dilution) antibodies for 1 h at room temperature (9). After washing, the coverslips were then incubated with fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG, washed and mounted on glass coverslips. Coversips were viewed using a ϫ 40 objective lens on a Zeiss Axiovert microscope operated in Laser Scanning Confocal mode. Samples were illuminated at 488 nm and the signal at 510 nm collected. Duplicate coverslips were prepared at each experimental condition, and 10 random images of plasma membrane lawns collected from each. These were quantified using MetaMorph (Universal Imaging, CA) software on a DAN PC (Noran Instruments, Surrey, UK). Similar methods were employed to assay GLUT1 levels in plasma membranes.
Cell Surface Transferrin Receptor Binding Assays-Transferrin receptors present at the cell surface were quantified as outlined in Refs. 27 and 31. After experimental manipulations, cells were rapidly chilled by three washes in ice-cold KRP buffer (136 mM NaCl, 4.7 mM KCl, 1.25 mM MgSO 4 , 1.25 mM CaCl 2 , 5 mM NaH 2 PO 4 , pH 7.4) containing 1 mg/ml bovine serum albumin. Thereafter, cells were incubated in the same buffer containing ϳ3 nM 125 I-transferrin for 2 h on ice. After this time, the media was aspirated and the monolayers washed three times with 1 ml of KRP/bovine serum albumin for 1 min. Cells were then solubilized in 1 M NaOH and the radioactivity associated with each well determined by ␥-counting. For each condition, duplicates plates were incubated exactly as above, but in the presence of 1 M transferrin; the radioactivity associated with each well under these conditions was the value of nonspecific binding at each condition, and was found to vary between 5 and 10% of the total counts per well.
Peptides and Antibodies-The peptides employed in this study were prepared either by Thistle Research (Glasgow, UK) or as outlined in Ref. 22. The sequences of the peptides used is shown in Table I. Antibodies to ARF5 were provided by Dr. J. Moss (NHLBI, National Institutes of Health, Bethesda, MD) and Dr. R. A. Kahn (Atlanta, GA) (17,32). Antibodies to ARF6 were provided by Dr. J. Donaldson (NHLBI, National Institutes of Health, Bethesda, MD) (33). Anti-GLUT1 was generously provided by Professor Geoff Holman (University of Bath).

Expression and Distribution of ARF5 and ARF6 in 3T3-L1
Adipocytes-We have used a panel of ARF-specific antibodies to examine the expression and subcellular distribution of ARFs 5 and 6 in 3T3-L1 adipocytes (Fig. 1). In the basal (non-stimulated) state, ARF5 was predominantly localized to the soluble protein fraction of adipocytes, with some association with intracellular membranes. In response to insulin treatment, ARF5 levels at the plasma membrane were observed to increase with a concomitant decrease chiefly from the soluble protein fraction of the cells. Similar results were obtained using two different anti-ARF5 antibodies. In the same experiments, GLUT4 was observed to translocate to the plasma membrane in response to insulin from intracellular membrane fractions as has been extensively reported (1)(2)(3). In contrast, ARF6 was observed chiefly in the plasma membrane fraction of 3T3-L1 adipocytes (34), and did not exhibit appreciable alteration in subcellular distribution in response to insulin (Fig. 1). The plasma membrane localization of ARF6 is in agreement with studies in a range of cell types (34,35). Although insulin did not appear to modulate the subcellular distribution of ARF6, it is possible that insulin may modify the ARF6-GDP/ARF6-GTP ratio at the plasma membrane, indeed several studies have suggested that unlike other members of the ARF family, ARF6 remains membrane associated even in its GDP-bound state (34,35). Alter-TABLE I Sequence alignment of ARF proteins at the amino terminus Shown is a sequence alignment of the murine isoforms of the ARF family (19). The peptides used in this study are underlined, and in the case of ARF5 and ARF6 were synthesized with or without a myristoyl group at position Gly-2.
natively, insulin may modulate the rate of turnover from membrane to cytosolic states of ARF6 without an apparent alteration in distribution between these fractions. Hence the apparent lack of altered subcellular distribution of ARF6 in response to insulin does not preclude an important role for this protein in insulin action.
Functional Role of ARF Proteins in Insulin-stimulated Glucose Transport and GLUT4 Translocation-Myristoylated peptides corresponding to the amino terminus of ARF proteins have been widely used in many laboratories to probe the function of ARF proteins in intracellular trafficking, including en-doplasmic reticulum to Golgi transport, intra-Golgi transport, and endocytic vesicle fusion (36 -38). Myristoylated peptides corresponding to residues 2 through 13 of the NH 2 terminus of ARF6 have also been shown to inhibit regulated exocytosis in permeabilized chromaffin cells (22), and to inhibit stimulated phospholipase D activity in these cells in response to agents which stimulate secretion (39). In contrast, a corresponding peptide lacking the myristoyl group at Gly-2, or the cognate myristoylated peptide from ARF1 were without effect (22).
We therefore chose to adopt similar methodology to examine the role of ARF5 or ARF6 in insulin-stimulated glucose transport in ␣-toxin-permeabilized 3T3-L1 adipocytes. We synthesized peptides corresponding to residues 2 through 16 of murine ARF5 (with or without a myristoyl group at position Gly-2), and employed myristoylated ARF1 and ARF6 peptides described by one of us previously (22). Permeabilized adipocytes were incubated with the peptides for 10 min, then stimulated with 1 M insulin for a further 20 min. At the end of this time, deoxy-Glc uptake was measured as described under "Experimental Procedures." The results of a typical experiment are presented in Fig. 2A. We consistently observed a diminution in the rate of insulin-stimulated deoxy-Glc uptake in cells incubated with myristoylated ARF6 peptide. In three experiments of this type, the extent of inhibition of insulin-stimulated deoxy-Glc transport was 47 Ϯ 11%. In contrast, neither myristoylated ARF1 nor myristoylated ARF5 peptides inhibited insulinstimulated deoxy-Glc uptake over the same concentration range. In control experiments (not shown) the addition of the peptide to intact 3T3-L1 adipocytes was without effect on either basal or insulin-stimulated deoxy-Glc uptake.
This data argues that ARF6 may play a role in the regulation of plasma membrane GLUT4 levels in response to insulin, as GLUT4 is responsible for the majority of insulin-stimulated glucose uptake in adipocytes. In order to address this directly, we measured insulin-stimulated GLUT4 translocation to the cell surface using the plasma membrane lawn technique. Permeabilized adipocytes were incubated with myristoylated ARF1, ARF5, and ARF6 peptides prior to insulin stimulation and assessment of plasma membrane GLUT4 levels. Data from a typical experiment are presented in Fig. 2B, and the results of three experiments of this type presented in Fig. 2C. We observed marked inhibition of GLUT4 translocation in the presence of 100 M myristoylated ARF6 peptide (42 Ϯ 3%; n ϭ 3). A modest inhibition of GLUT4 translocation was also observed in the presence of 100 M myristoylated ARF5 peptide, but the effect was considerably less than that induced by the ARF6 peptide and did not reach significance in every experiment. 100 M Myristoylated ARF1 peptide was without effect in this assay. A dose-response curve for these peptides on GLUT4 translocation is presented in Fig. 2D. No significant effect on basal (unstimulated) plasma membrane levels of GLUT4 were observed after incubation with the peptides (Fig.  2B), but low signal precludes detailed quantification of the level of GLUT4 at the plasma membrane in the absence of insulin.
ARF5 and ARF6 Regulate Plasma Membrane Transferrin Receptor and GLUT1 Numbers in 3T3-L1 Adipocytes-Insulin treatment of adipocytes results in the movement of other proteins to the plasma membrane, including the TfR, the IGF-II/ cation-independent mannose-6-phosphate receptor (27,40) and GLUT1 (26), albeit to a much lesser extent than is observed for GLUT4 (typically ϳ2-fold compared with 12-15-fold). This is probably due to movement of these proteins from the recycling endosomal system to the plasma membrane. We wished to determine whether the effect of the myristoylated ARF6 peptide to inhibit insulin-stimulated GLUT4 translocation was FIG. 1. Immunological analysis of GLUT4, ARF5, and ARF6 in subcellular membranes of 3T3-L1 adipocytes in response to insulin. 10-cm plates of 3T3-L1 adipocytes were incubated Ϯ 1 M insulin for 15 min then subjected to subcellular fractionation as described under "Experimental Procedures." Samples corresponding to 10% of the total yield of plasma membrane and low density microsomal (LDM) fractions, and 5% of the soluble protein were subjected to SDS-polyacrylamide gel electrophoresis and immunoblot analysis using antibodies against ARF5, ARF6, and GLUT4 as indicated. Data from a representative blot, repeated three times, is shown in Panel A with the approximate position of molecular weight markers indicated at right of the blots. Quantification of three blots of this type is shown in Panel B in which the relative distribution of the indicated protein is expressed as a function of the protein level in basal plasma membranes (note that in Panel A, the amount of soluble protein loaded per lane is 50% of that of the plasma membrane and low density microsomal fractions).

FIG. 2. Myristoylated ARF6 amino-terminal peptide inhibits insulin-stimulated deoxy-Glc transport and GLUT4 translocation in 3T3-L1 adipocytes.
Panel A, 3T3-L1 adipocytes were permeabilized with ␣-toxin as described. After permeabilization, cells incubated with amino-terminal ARF peptides were added (100 M) in ICR buffer for 10 min, followed by insulin addition (1 M) and a further 20-min incubation at 37°C. Deoxy-Glc transport was determined as described under "Experimental Procedures" and the results of a typical experiment of this type is shown. Each point is the mean of three measurements, corrected for nonspecific association of deoxy-Glc with cells.
In three experiments of this type, the myristoylated ARF6 peptide inhibited deoxy-Glc uptake by 47 Ϯ 11%, ARF1 and ARF5 peptides were without effect. * indicates a statistically significant difference from insulin alone, p Ͻ 0.05. Prior incubation with ARF1 or ARF5 peptides did not diminish basal deoxy-Glc uptake; incubation with ARF6 peptide reduced basal deoxy-Glc uptake ϳ15%, but this effect did not reach statistical significance (data not shown). Panel B, 3T3-L1 adipocytes were treated as described for Panel A, then processed for plasma membrane lawn assay for GLUT4 translocation as described under "Experimental Procedures." Shown is data from a typical experiment (Panel B) and the data from three independent experiments of this type, quantified as described under "Experimental Procedures" is shown in Panel C. * indicates significant difference from insulin alone, p Ͻ 0.05. A dose-response curve for the inhibition of insulin-stimulated GLUT4 translocation by myristoylated ARF6 peptide is shown in Panel D. specific for GLUT4, or whether other proteins which traffic between intracellular membranes and the cell surface in an insulin-regulated manner were also effected. We therefore examined the effect of insulin on cell surface levels of TfR and GLUT1 in permeabilized adipocytes incubated with myristoylated ARF peptides (Fig. 3).
Insulin stimulation of permeabilized adipocytes results in a ϳ2-fold increase in plasma membrane TfR levels, in agreement with published studies (27,40). Prior incubation with myristoylated ARF1 peptide did not reduce the magnitude of this response (Fig. 3A). In contrast, prior incubation of 3T3-L1 adipocytes with myristoylated ARF5 peptide significantly in-hibited the ability of insulin to stimulate TfR levels at the cell surface with no effect on the basal (unstimulated) TfR levels. In contrast, myristoylated ARF6 peptide reduced cell surface TfR levels in both basal and insulin-stimulated cells significantly compared with control cells, without significantly reducing the fold increase in cell surface TfR levels observed in response to insulin (Fig. 3A). Similar data were also observed for GLUT1 (Fig. 3B), with the exception that the reduction in plasma membrane GLUT1 levels in the presence of the ARF6 peptide were not as extensive as those observed for TfR (Fig. 3B). Consistent with this, we observed a diminution of basal (unstimulated) deoxy-Glc uptake in cells incubated with the ARF6 peptide (ϳ15% inhibition), but this effect did not reach statistical significance, presumably because the rate of basal transport is low (data not shown). Nevertheless, these data argue that both ARF5 and ARF6 are intimately involved in the trafficking of membrane proteins between the plasma membrane and the recycling endosomal system in this cell type.

DISCUSSION
Here we have shown that insulin stimulation of 3T3-L1 adipocytes results in the redistribution of ARF5 from the soluble protein (cytosolic) fraction to the plasma membrane. We hypothesize that ARF5 recycles between the plasma membrane and intracellular (cytosolic) fractions presumably as a consequence of GDP/GTP exchange, and that insulin stimulates GTP loading of this protein. Despite this insulin-stimulated translocation of ARF5, our data is not consistent with a role for this protein in insulin-stimulated GLUT4 translocation. Rather, we suggest that ARF5 plays a role in the insulin-stimulated trafficking of membrane proteins between the recycling endosomal system and the plasma membrane, as evidenced by the ability of myristoylated ARF5 peptides to inhibit insulin-stimulated TfR and GLUT1 movement to the cell surface (Fig. 3). To our knowledge, this is the first demonstration of a functional role for ARF5 in membrane trafficking.
In contrast to the data relating to ARF5, we show that a myristoylated peptide corresponding to the amino terminus of ARF6 partially inhibits insulin-stimulated GLUT4 translocation and deoxy-Glc transport in ␣-toxin permeabilized 3T3-L1 adipocytes (Fig. 2), implicating ARF6 as a key component of this response. Furthermore, we show that incubation of 3T3-L1 adipocytes with myristoylated ARF6 peptide reduces TfR number at the cell surface both in the basal state and after insulin stimulation, without decreasing the fold increase in cell surface levels in response to insulin (Fig. 3A); similar results were observed for GLUT1, except the reduction in plasma membrane GLUT1 levels were not as great (Fig. 3B). Collectively, these data argue that ARF6 plays a fundamental role in trafficking between intracellular membranes and the cell surface, indeed ARF6 has been proposed to mediate the targeting of recycling vesicles to the plasma membrane, either from a perinuclear compartment (CHO cells (21)) or from a unique tubular-vesicular compartment (HeLa cells (33)). With this in mind, several models may be proposed to explain the inhibitory effect of myristoylated ARF6 peptides on insulin-stimulated GLUT4 translocation.
GLUT4 has been proposed to populate at least two distinct intracellular compartments, one of which corresponds to the recycling endosomal system, the other a specialized intracellular storage compartment referred to as GLUT4 storage vesicles (reviewed in Refs. 1, 2, and 41). Hence, the partial inhibition of insulin-stimulated GLUT4 translocation by myristoylated ARF6 peptides may be explained by the selective inhibition of translocation of one of the proposed multiple intracellular GLUT4 pools. This is consistent with previous studies which FIG. 3. Effects of ARF peptides on insulin-stimulated cell surface transferrin receptor levels. 3T3-L1 adipocytes were permeabilized, incubated with myristoylated ARF peptides, and incubated Ϯ 1 M insulin as described in the legend to Fig. 2. After this, cell surface transferrin receptor levels were determined by radioactive Tf binding (Panel A) or cell surface GLUT1 levels measured using plasma membrane lawns (Panel B). Shown are the results of a typical experiment using myristoylated ARF1, ARF5, and ARF6 peptides. In Panel A, * indicates a statistically significant difference from insulin alone (p ϳ 0.01); ** indicates statistically significant difference from basal cells (p Ͻ 0.05); *** indicates a statistically significant difference from basal ϩ myr-ARF6 peptide-loaded cells (p ϭ 0.01). In Panel B, * indicates a statistically significant difference from insulin alone (p ϭ 0.010); ** indicates statistically significant difference from basal cells (p ϳ 0.05); *** indicates a statistically significant difference from basal ϩ myr-ARF6 peptide-loaded cells (p ϭ 0.05).
implicate ARF6 in the exocytosis of secretory vesicles in other cell types (22).
A second model, however, should also be considered which may also explain the experimental data presented here. Previous studies of ARF6 function in other cell types have shown that mutants of ARF6 disrupt receptor-mediated endocytosis (42). Overexpression of ARF6 redistributed TfR to the cell surface, while a dominant negative mutant of ARF6 was shown to redistribute TfR to intracellular membranes and inhibit TfR recycling to the cell surface, suggesting that ARF6 is an integral component of the endocytic apparatus and that its GTP cycle/nucleotide status regulate progression through the endocytic pathway (42). In agreement with these studies, we have shown that myristoylated ARF6 peptide decreases the cell surface levels of TfR both in the presence and absence of insulin without affecting the magnitude of the insulin-dependent increase in TfR at the cell surface (Fig. 3A), consistent with ARF6 regulating progression of TfRs through the endocytic pathway. Hence, an alternative explanation of the data presented here is that the myristoylated ARF6 peptide functions in a fashion similar to a dominant negative ARF6, resulting in a change in the steady-state distribution of TfR such that the intracellular/ plasma membrane ratio is increased. It is possible that after insulin stimulation when cell surface GLUT4 levels are increased, ARF6 function regulates the internalization/recycling of GLUT4 in a similar manner to that of the TfR. When ARF6 function is disrupted by the myristoylated peptide, intracellular levels of GLUT4 and TfR are increased, resulting in decreased plasma membrane levels of both of these proteins. This effect is only manifest for GLUT4 in the insulin-stimulated state as plasma membrane levels of GLUT4 in the absence of insulin are already very low. Hence, in this model, the role of ARF6 is not specific for any particular GLUT4 compartment, but rather is manifest at the level of GLUT4 recycling between the plasma membrane and intracellular compartments, which is known to occur even in the presence of insulin (1,41,43). Distinguishing between these (or other) models for ARF6 action represents an important goal.
These data also offer further insight into the ability of insulin to regulate membrane trafficking in adipocytes. We have shown that myristoylated ARF5 peptide inhibits insulin-stimulated GLUT1 and TfR translocation to the plasma membrane without significantly inhibiting GLUT4 translocation. This result implies that the pathways by which these two proteins reach the plasma membrane after insulin stimulation are distinct. Several studies have identified a GLUT4 compartment which is relatively devoid of TfR in both adipocytes and muscle (4, 44 -46). This (GLUT4 storage vesicles) compartment has been suggested to be a GLUT4 storage compartment which is rapidly mobilized in response to insulin (2). Although there is ample data implicating some overlap between GLUT4 and the TfR in endosomes, it is possible that this overlap is mainly a consequence of these proteins sharing the same components of the endocytic arm of the recycling pathway. Indeed studies using mutant dynamin molecules has suggested that the slowly recycling (GLUT1/TfR positive) GLUT4 compartment contributes minimally to insulin-stimulated GLUT4 translocation (47). In response to insulin when traffic through the recycling endosomal system is clearly increased, some GLUT4 will move to the cell surface from this location. We would suggest that this represents a modest proportion of the total insulin-stimulated GLUT4 translocation, as evidenced here by the lack of significant inhibition of GLUT4 translocation under conditions when GLUT1 and TfR translocation are significantly compromised (i.e. in the presence of ARF5 peptide). Hence we suggest that the main effect of insulin is to recruit GLUT4 from the post-endosomal storage compartment (GLUT4 storage vesicless), and that unlike insulin-stimulated movement through the recycling endosomes, this is independent of ARF5 function.
In conclusion, we suggest that both ARF5 and ARF6 regulate trafficking of membrane proteins to and from the cell surface, and show that ARF6 plays an important regulatory role in the steady-state levels of GLUT4 at the adipocyte cell surface after insulin stimulation. We speculate that the actions of ARF5 and ARF6 are chiefly mediated by effects on the recycling endosomal system, at least in this insulin-responsive cell type.