Insulin regulates the membrane arrival, fusion, and C-terminal unmasking of glucose transporter-4 via distinct phosphoinositides.

Insulin increases glucose uptake into muscle via glucose transporter-4 (GLUT4) translocation to the cell membrane, but the regulated events in GLUT4 traffic are unknown. Here we focus on the role of class IA phosphatidylinositol (PI) 3-kinase and specific phosphoinositides in the steps of GLUT4 arrival and fusion with the membrane, using L6 muscle cells expressing GLUT4myc. To this end, we detected the availability of the myc epitope at the cell surface or intravesicular spaces and of the cytosol-facing C-terminal epitope, in cells and membrane lawns derived from them. We observed the following: (a) Wortmannin and LY294002 at concentrations that inhibit class IA PI 3-kinase reduced but did not abate the C terminus gain, yet the myc epitope was unavailable for detection unless lawns or cells were permeabilized, suggesting the presence of GLUT4myc in docked, unfused vesicles. Accordingly, GLUT4myc-containing vesicles were detected by immunoelectron microscopy of membranes from cells pretreated with wortmannin and insulin, but not insulin or wortmannin alone. (b) Insulin caused greater immunological availability of the C terminus than myc epitopes, suggesting that C terminus unmasking had occurred. Delivering phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P(3)) to intact cells significantly increased lawn-associated myc signal without C terminus gain. Conversely, phosphatidylinositol 3-phosphate (PI3P) increased the detection of C terminus epitope without any myc gain. We propose that insulin regulates GLUT4 membrane arrival, fusion, and C terminus unmasking, through distinct phosphoinositides. PI(3,4,5)P(3) causes arrival and fusion without unmasking, whereas PI3P causes arrival and unmasking without fusion.

Insulin promotes the uptake of glucose into muscle and fat tissues through a rapid gain in surface-bound glucose trans-porters (1)(2)(3). The muscle-and fat-specific glucose transporter GLUT4 1 cycles continuously between the plasma membrane and intracellular stores, with the steady-state distribution largely favoring the latter. Insulin changes this steady-state resulting in a net gain in surface GLUT4 (4 -6) largely as a result of enhancing the exocytic rate of GLUT4 cycling (7,8). Of significance, insulin resistance and diabetes are accompanied by defective GLUT4 gain at the plasma membrane of muscle and fat cells (9 -11).
It is well established that signaling from class IA phosphatidylinositol (PI) 3-kinase is required for the insulin-dependent net gain in surface GLUT4 (12)(13)(14)(15), but the specific step(s) in GLUT4 cycling that are regulated are not elucidated. We have recently shown that class IA PI 3-kinase is required for the insulin-dependent acceleration of GLUT4 transit through the recycling endosome (16). A second input of class IA PI 3-kinase in muscle cells is the spatial-temporal actin remodeling and its possible contribution to segregating specific signaling molecules (17,18). However, it is not known whether fusion of insulin-sensitive GLUT4 vesicles with the plasma membrane is a regulated step, nor which phosphoinositides participate in this event. Indeed, class II PI 3-kinase C2␣ is also activated by insulin (19). The major product of class IA PI 3-kinase in vivo is PI(3,4,5)P 3 (20,21) and that of class II PI 3-kinase C2␣ (19) is PI3P, whereas both enzymes can lead to the formation of PI 3,4-bisphosphate.
Here we implement the simultaneous detection of exofacially and cytosolically facing epitopes of GLUT4myc to score the arrival and fusion of GLUT4myc-containing vesicles at the plasma membrane of myotubes, and the effect of insulin, PI(3,4,5)P 3 , and PI3P, in these events. An myc epitope in the first exofacial loop of GLUT4myc would face the extracellular milieu in intact cells and the intravesicular space in adhered/ docked vesicles, whereas the C terminus of the transporter would always face the cytosol. Epitopes were immunodetected in membrane lawns and in intact or permeabilized muscle cells. We show that wortmannin (100 nM) and LY294002 (25 M) cause accumulation below the plasma membrane of unfused GLUT4-containing vesicles. These results suggest that class IA PI 3-kinase regulates the membrane fusion step, because class II PI 3-kinase-C2␣ is not affected by these doses of the inhibitors (22). Strikingly, the C terminus of GLUT4myc was more available for antibody recognition after insulin action, suggesting unmasking of this epitope in response to the hormone.
Carrier delivery of PI(3,4,5)P 3 caused significant GLUT4myc arrival and membrane fusion while barely allowing C-terminal detection. Conversely, PI3P caused GLUT4myc arrival and C-terminal exposure but not fusion. We propose that insulin regulates GLUT4 membrane arrival, fusion, and C-terminal unmasking, through distinct phosphoinositides.
Cell Culture, cDNA Transfections, and Phosphoinositide Delivery-L6 rat myoblasts stably expressing GLUT4 with an exofacial myc epitope (L6-GLUT4myc) and L6-wild-type myoblasts were differentiated into myotube monolayers as described (25,26). Where indicated, differentiating L6-wild-type myotubes were transfected with 3 g of GLUT4myc-eGFP cDNA using Lipofectamine 2000 (Invitrogen) as specified by the manufacturer, and analyzed 48 h post-transfection. Distinct phosphatidylinositol phosphates and carrier (neomycin) were prepared and applied to cells essentially as described previously (27). Neomycin was used as carrier to deliver the phosphatidylinositol phosphates. No effect of neomycin alone was observed on either glucose uptake or GLUT4myc surface levels (27). Similarly, we did not observe an effect of neomycin treatment alone on the various GLUT4 traffic events examined in this study.
Immunodetection of GLUT4myc on Plasma Membrane Lawns-Plasma membrane lawns from L6-myotubes were prepared by modification of a protocol developed for 3T3-L1 adipocytes (28,29). Briefly, cells were serum-starved for 4 h then subjected to the diverse pretreatments or stimuli at 37°C for the indicated time periods (the times chosen were those yielding a maximal response to each stimulus). Cells were then placed on ice, washed twice in ice-cold PBS (pH 7.5) and incubated with ice-cold hypotonic swelling buffer (23 mM KCl, 10 mM HEPES, 5 mM MgCl 2 , 1 mM EGTA, pH 7.5) 3 times for 2 min. Cells were then sonicated 3 times for 2 s on ice in sonication buffer (70 mM KCl, 30 mM HEPES, 5 mM MgCl 2 , 3 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 1 M pepstatin A, 1 M leupeptin, and 1 mM dithiothreitol, pH 7.5) using a Hert System sonicator (Misonix Inc., Farmingdale, NY), at 30% output. Lawns were fixed with 4% (v/v) para-formaldehyde (PFA) in PBS for 15 min, and the reaction was quenched with 0.1 M glycine in PBS for 10 min. Intact lawns were used to detect cell surface-and cytosol-exposed epitopes. To detect intra-vesicular myc epitopes, fixed lawns were permeabilized for 15 min at room temperature with 0.1% Triton X-100 in PBS. Where indicated, lawns were incubated with PBS, PBS supplemented with 150 or 500 mM NaCl, or PBS supplemented with 650 mM LiCl but without NaCl for 7 min on ice before or after fixation. Following these steps, lawns were washed quickly three times with PBS and, depending on the primary antibody, blocked with 5% (v/v) goat serum or 3% (w/v) BSA in PBS for 10 min at room temperature. Lawns were labeled for 1 h at room temperature or 7 h at 4°C with primary antibodies (monoclonal or polyclonal anti-myc, 2 g/ml; polyclonal anti-C terminus of GLUT4, 1:200; polyclonal anti-cytosolic loop GLUT4, 4 g/ml; polyclonal anti-exofacial loop GLUT4, 4 g/ml; polyclonal anti-caveolin, 1.5 g/ml) and for 45 min with secondary antibodies (A488-conjugated anti-mouse or anti-rabbit or anti-goat IgGs, 10 g/ml; Cy3-conjugated anti-mouse or anti-rabbit or anti-goat IgGs, 10 g/ml) at room temperature, as indicated. Coverslips were washed twice with PBS and mounted using Dako mounting solution. Lawns from eGFP-GLUT4myc-transfected L6-wild-type cells were treated with or without insulin and labeled with monoclonal anti-myc antibody (2 g/ml) followed by Cy3-conjugated goat anti-mouse IgG (10 g/ml).
Immunofluorescence of Rounded-up L6-GLUT4myc Myoblasts-L6-GLUT4myc myoblasts pretreated for 20 min with or without 100 nM wortmannin at 37°C were detached from the substratum using Ca 2ϩand Mg 2ϩ -free PBS at 37°C. During re-attachment of cells to glass coverslips (10 min), they were left untreated or treated with insulin and/or wortmannin at 37°C as indicated. Cells were then fixed with PFA and processed for epitope detection in intact cells, or cells were permeabilized with 0.1% Triton X-100 as described previously (16).
Where indicated, L6 myoblast monolayers treated with or without 100 nM insulin for 10 min at 37°C were used to quantify the total cellular signal of immunodetectable myc or C terminus epitopes, after fixation with 4% PFA and permeabilization with 0.1% Triton X-100. Then cells were immunolabeled with monoclonal anti-myc (2 g/ml) or polyclonal anti-C terminus (1:200) antibody followed by A488-conjugated anti-mouse IgG (10 g/ml) or Cy3-conjugated anti-rabbit IgG (10 g/ml), respectively. Quantitative analysis of GLUT4 labeled by antimyc and/or anti-C terminus antibody at the periphery of rounded-up cells was performed by blind counting of at least 300 cells per condition. The number of the positive cells was expressed as a percentage of the total cells.
Confocal Microscopy Image Acquisition, Quantitation, and Statistical Analysis-Images were obtained using a Zeiss Axiovert 100M laser scanning confocal microscopy (Zeiss LSM 510, Carl Zeiss, Thornwood, NY) at room temperature using a ϫ100 objective at the same gain setting unless indicated otherwise. The gain detector was set using cells or plasma membrane lawns labeled with only secondary antibody and images were analyzed using Image software (National Institutes of Health). Lawns were individually outlined, and pixel intensity within each lawn was determined then expressed as intensity per unit area. Background intensity was determined using areas outside of lawns and was subtracted from each reading. At least 50 plasma membrane lawns were examined for each experimental condition. Results are reported as -fold values relative to the indicated control condition in each figure. Statistical analysis was performed using the paired Student's t test or ANOVA with Turkey's multiple comparisons (Graph-Pad Software, Inc., San Diego, CA). Where no error bars are present, the values were too small to show on the columns (except for conditions assigned a value of 1.0).
Detection of GLUT4myc C Terminus by Immunoelectron Microscopy-The method of (30, 31) for immunodetection of GLUT4 on plasma membrane lawns by electron microscopy was used with slight modification. L6 myoblasts differentiated into myotubes on Formvar-coated nickel grids (300 mesh) were serum-deprived, then treated with vehicle (Me 2 SO) or 100 nM wortmannin for 20 min prior to 100 nM insulin (10 min) at 37°C. Lawns were generated as described above from cells grown on coverslips, fixed with 4% PFA and 0.05% glutaraldehyde in PBS for 30 min at room temperature, and blocked with 2% goat serum in PBS. Detection was preceded with C terminus GLUT4 antibody (20 g/ml) in PBS containing 2% goat serum (60 min), followed by 6 nmgold-conjugated anti-rabbit antibody (1:25 dilution, 60 min). Lawns were fixed with 2% glutaraldehyde for 10 min followed by 1% OsO 4 in 0.1 M sodium phosphate buffer (pH 7.4) for 10 min and washed with PO 4 buffer three times for 2 min each followed by washing with dH 2 O for 2 min. Lawns were stained with 2% aqueous uranyl acetate for 10 min prior to dehydration with ethanol. Finally lawns were dried with a critical point dryer CPD 030 (Bal-Tec, Liechtenstein). Transmission electron microscopy examination was done with a Tecnai 20 at room temperature (FEI, Hillsboro, OR).

RESULTS
The experimental premise of this study is that epitopes on different regions of GLUT4myc can be exploited to determine whether GLUT4myc is fully inserted into the plasma membrane, or whether it lies in vesicles adhered/docked to the cytosolic leaflet of the membrane. The myc epitope inserted between transmembrane helices 1 and 2 is expected to face the extracellular milieu when fully fused with the membrane, or the intravesicular space of vesicles docked onto the membrane. Plasma membrane lawns that would retain docked but unfused vesicles would not have the vesicular myc epitope available for antibody decoration unless the vesicles are permeabilized with detergent once the lawns are generated. Conversely, the last 15 amino acids of the C terminus end or the large cytosolic loop between helices 6 and 7 of the transporter would be available for antibody decoration whether GLUT4myc is inserted into the plasma membrane or present on docked vesicles. The availability of the C terminus or cytosolic antigens for antibody recognition on membrane lawns would, however, be affected by endogenous proteins that might bind to these sites. In this case, high salt or other chaotropic agents that might remove the putative bound proteins would increase C terminus (or cytoso-lic loop) antigenicity. Hence, differences between the relative gains in antibody-accessible myc and the C terminus (or cytosolic loop) epitopes in lawns from control and insulin-stimulated cells would suggest differences in the status of GLUT4myc at the plasma membrane.
Like in skeletal muscle (11,32) insulin usually causes a doubling in surface GLUT4myc in L6-GLUT4myc muscle cells (26), out of a compartment that is largely segregated away from constitutively recycling proteins (26,33). In the present study, we implement the use of membrane lawns to quantify surfaceassociated GLUT4myc. Fig. 1 shows that exposing lawns from wild-type cells to monoclonal anti-myc antibody followed by Alexa 488-conjugated anti-mouse IgG did not produce any fluorescent signal. Under the same settings of confocal fluorescence image acquisition, lawns from L6-GLUT4myc myotubes rendered a substantial signal that was higher in cells pretreated with insulin relative to basal state cells. Omission of the primary antibody eliminated the fluorescent signal (data not shown). Under the conditions of fluorescence acquisition used, the C-terminal epitope of wild-type myotubes was not detectable on membrane lawns. However, as expected, when the detector gain was markedly increased, a fluorescence signal was detected that was higher by 1.88 Ϯ 0.55-fold, n ϭ 5 (p Ͻ 0.05) by t test, in lawns from insulin-pretreated cells than from unstimulated cells assigned a value of 1.00. These results exclude the possibility that endogenous GLUT4 contributes to the gain of C terminus signal observed on lawns of GLUT4myc cells. This behavior is expected given that in L6-GLUT4myc myotubes GLUT4myc is expressed at almost 100-fold higher levels than the endogenous GLUT4 (which is low in these cells compared with muscle or adipose tissue) (34). We have previously determined that GLUT4myc traffic in L6-GLUT4myc cells is not affected by such overexpression (8,26,35), as follows: All of GLUT4myc is available for recycling to the cell surface. Moreover, the t1 ⁄2 of GLUT4myc internalization is 3.5 min (similar to that of GLUT4 or IRAP in adipose cells) and insulin stimulates GLUT4myc recycling by 2-fold (comparable to the 2-fold gain in surface GLUT4 in parental L6 cells and mature skeletal muscle). Immunoprecipitated GLUT4myc-containing compartments via the GLUT4 C terminus (so-called GLUT4 vesicles) exclude GLUT1, and peroxide-dependent ablation of transferrin receptor compartments loaded with horseradish peroxidase-transferrin spare 50% of GLUT4, 2 as in the case of 3T3-L1 adipocytes (36).
Wortmannin and LY294002 Arrest GLUT4myc Vesicle Fusion with the Membrane- Fig. 2A illustrates the gains in myc and C terminus epitopes on membrane lawns derived from basal and insulin-stimulated L6-GLUT4myc myotubes, quantified as described under "Experimental Procedure." By comparison, the immunodetection of caveolin on membrane lawns was not affected by insulin stimulation. The plasma membrane content of caveolin was previously shown not to change in response to insulin (37). The gain in myc epitope caused by insulin pretreatment of cells for 10 min was 1.62 Ϯ 0.05-fold relative to the basal value (p Ͻ 0.005). Wortmannin (100 nM) and LY294002 (25 M) completely inhibit class IA PI 3-kinase activation (12,22). Remarkably, pretreatment of myotubes with 100 nM wortmannin abolished the insulin-dependent gain in myc epitope detection while preserving nearly half of the 2 V. Randhawa and A. Klip, unpublished results.
FIG. 1. Distinction of immunofluorescent signals of endogenous GLUT4 and GLUT4myc C termini through different detector gains. Wild-type L6 myotubes or L6-GLUT4myc myotubes were grown on coverslips were starved of serum 4 h, then treated without or with 100 nM insulin for 10 min. Plasma membrane lawns were prepared and labeled with monoclonal anti-myc or polyclonal anti-C-terminal GLUT4 antibody, followed by incubation with goat anti-mouse A488, goat anti-rabbit Cy3 antibody, respectively, as described under "Experimental Procedures." Immunolabeled lawns were processed and analyzed by confocal fluorescence microscopy. Signals were acquired at low and high detector gain levels. At the detector gain used to acquire the C terminus signal of L6-GLUT4myc myotubes (low gain), there was no detectable signal of the C terminus of wild-type GLUT4 myotubes. The endogenous GLUT4 C terminus of wild-type myotubes was only detectable at the high gain. The results illustrate that the C-terminal signal of the endogenous GLUT4 cannot be detected under the conditions used to acquire the C-terminal signal from the transfected GLUT4myc. Bar, 10 m for 20 min as indicated followed by incubation without (basal) or with 100 nM insulin for 10 min. Plasma membrane lawns were prepared and processed for detection of myc epitope, C-terminal GLUT4, or caveolin by confocal fluorescence microscopy. The fluorescence intensity was quantified from at least 50 plasma membrane lawn pieces for each condition per experiment as described under "Experimental Procedures" and is illustrated as the mean -fold change per lawn over the basal (Ϯ S.E.) obtained from at least five independent experiments. *, p Ͻ 0.005 and #, p Ͻ 0.05 relative to the corresponding basal values (ANOVA).
gain in GLUT4myc C terminus signal (1.42 Ϯ 0.03-fold, p Ͻ 0.005 compared with unstimulated cells equally pretreated with wortmannin). As in the case of 100 nM wortmannin, pretreatment with 25 M LY294002 abolished the insulin-dependent gain in myc epitope detection on membrane lawns, while preserving significant gain in C terminus epitope (1.40 Ϯ 0.03fold, p Ͻ 0.05 compared with unstimulated cells equally pretreated with LY294002) (Fig. 2B).
From the above results we hypothesize that 100 nM wortmannin or 25 M LY294002 prevented the insulin-induced fusion of GLUT4myc-containing vesicles with the plasma membrane while allowing about half of the insulin-induced mobilization of the vesicles to the plasma membrane. Such a situation would render GLUT4myc-containing vesicles docked onto plasma membrane lawns. Triton X-100 is typically used to permeabilize para-formaldehyde-fixed plasma-and endo-membranes by extracting lipids and allowing antibody penetration (38). Accordingly, we examined if Triton X-100 treatment changed the insulin-dependent gain in myc and C terminus epitopes on plasma membrane lawns from L6 myotubes pretreated with 100 nM wortmannin with and without insulin stimulation (Fig. 3). Triton X-100 did not affect the detection of either epitope in plasma membrane lawns from basal or insulin-treated cells. However, Triton X-100 revealed an insulindependent gain in myc epitope in lawns from cells pretreated with wortmannin plus insulin (1.36 Ϯ 0.01-fold above basal cells pretreated with wortmannin, p Ͻ 0.005). By contrast, there was no change in the immunoreactivity of the C terminus upon Triton X-100 treatment of membrane lawns.
Immunoelectron Microscopy Reveals GLUT4 Vesicles on Plasma Membrane Lawns-Collectively, the above results suggest that docked GLUT4myc-containing vesicles are detected in lawns from cells pretreated with wortmannin or LY294002 and stimulated with insulin. To verify this prediction, we explored the presence of docked GLUT4 vesicles on plasma membrane lawns using immunoelectron microscopy. L6 myotubes were pretreated without or with 100 nM wortmannin followed by stimulation without or with insulin. Plasma membrane lawns were generated, and the GLUT4 C terminus was detected by gold-conjugated antibodies and examined by electron microscopy as described in "Experimental Procedures." Fig. 4 shows representative images of lawns derived from cells of each condition. In lawns from unstimulated cells (whether or not pretreated with wortmannin) as well as in insulin-stimulated cells, GLUT4 was largely detected in flat areas of the membrane, in agreement with previous studies using 3T3-L1 adipocytes (29). Strikingly, however, GLUT4-containing vesicles (ϳ100 nm in diameter) were readily visualized in lawns from cells pretreated with wortmannin prior to insulin stimulation, and there was little immunogold-labeled GLUT4 outside the vesicles. Qualitatively similar results were obtained in five separate experiments analyzing at least six fields in each experiment. These results illustrate the preponderance of GLUT4 in flat areas of the lawns from unstimulated cells (without or with wortmannin) or from insulin-stimulated cells, and a converse preponderance of GLUT4 in vesicles in lawns from wortmannin-pretreated, insulin-stimulated cells.
Insulin Causes Unmasking of the GLUT4myc C Terminus-Further examination of the results in Fig. 2 showed that the gain in C terminus labeling in lawns from insulin-stimulated cells compared with lawns from basal cells was significantly larger (2.07 Ϯ 0.11-fold, p Ͻ 0.005) than the gain in myc epitope labeling in the same preparations (1.62 Ϯ 0.05-fold, p Ͻ 0.005).
To exclude the possibility that the different gains in myc and C-terminal signals might be due to difference in affinity of each antibody for its epitope or to characteristics of the fluorophores used, various combinations of primary and secondary antibodies were used. Table I shows that the insulin-dependent gain in fluorescence did not depend on the nature of the antibody species or the fluorophore attached to the secondary antibody. In addition, the gain in myc epitope remained 1.62 Ϯ 0.21-fold relative to basal (p Ͻ 0.05) and that in C terminus remained 2.00 Ϯ 0.19-fold relative to basal (p Ͻ 0.005, n ϭ 5) when twice the concentration of primary antibodies was used, suggesting that the differences were not due to lack of epitope saturation.
The antigenic availability of the large cytosolic loop of FIG. 3. Triton X-100 reveals latent myc epitope on plasma membrane lawns from wortmannin-pretreated, insulin-stimulated myotubes. L6-GLUT4myc myotubes were pretreated with vehicle or 100 nM wortmannin (Wm) for 20 min as indicated, followed by incubation without or with insulin (100 nM for 10 min). Plasma membrane lawns were then labeled and processed for detection of myc or Cterminal epitopes of GLUT4myc by confocal fluorescence microscopy. Where indicated, lawns were treated with 0.1% Triton X-100 after fixation (solid bars). The fluorescence intensity per lawn was quantified from at least 50 plasma membrane lawn pieces for each condition per experiment as described under "Experimental Procedures," and is expressed as -fold change per unit area over the basal (mean Ϯ S.E.) obtained from at least five independent experiments. *, p Ͻ 0.005 and #, p Ͻ 0.05 relative to the corresponding basal values (ANOVA).
FIG. 4. Immunoelectron microscopy reveals GLUT4 vesicles on plasma membrane lawns from L6-myotubes treated with wortmannin prior to insulin stimulation. Myotubes are treated with vehicle (Me 2 SO), 100 nM wortmannin for 30 min, 100 nM insulin for 10 min, or 100 nM wortmannin for 20 min prior to 100 nM insulin for 10 min. Plasma membrane lawns were generated, labeled with polyclonal anti-C terminus antibody followed by goat polyclonal anti-rabbit antibody conjugated with 6 nm-gold particles, and processed for electron microscopy as described under "Experimental Procedures." Gold particles in non-vesicular areas are indicated by arrows. Bar, 100 nm.
GLUT4myc was also tested. Insulin caused a gain in immunofluorescent signal of this epitope of 2.01 Ϯ 0.16-fold relative to basal (p Ͻ 0.005), which was comparable to the gain in signal of the C terminus epitope. In contrast, an antibody raised to the first exofacial loop of GLUT4 detected a 1.65 Ϯ 0.15-fold gain in signal on membrane lawns (p Ͻ 0.005), akin to the gain in myc epitope signal. These observations corroborate that the gain in lawn-associated fluorescent signal is larger for cytosolic than for exofacial epitopes of GLUT4.
The gain in antigenicity of the C terminus relative to the myc epitope observed in lawns was also evident in whole cells. L6-GLUT4myc or wild-type myoblasts were permeabilized with Triton X-100 after fixation (see "Experimental Procedures"). The whole cell signal detected by the polyclonal anti-C terminus antibody was elevated by 27% in response to insulin in wild-type cells. Similarly, in L6-GLUT4myc cells, insulin caused a 26% increase in C-terminal signal, without any change in myc signal (Table II).
These results are congruent with the possibility that the cytosolic epitopes on GLUT4myc may be masked by endogenous proteins, thereby hindering their antigenicity in the basal state. Accordingly, we explored whether a chaotropic agent, which might weaken the interaction of the putative interfering protein(s) with GLUT4myc, could increase GLUT4 antigenic recognition by anti-C terminus antibodies in the basal state. Fig. 5 illustrates that treatment of lawns with PBS supplemented with 500 mM NaCl (but not with 150 mM NaCl) for 7 min prior to fixation with PFA and epitope detection, augmented the C terminus signal on lawns from unstimulated cells (1.33 Ϯ 0.04-fold above basal control with PBS, p Ͻ 0.05). As in the case of PBS supplemented with 500 mM NaCl, washing the lawns with PBS supplemented with 650 mM LiCl (without NaCl) also increased the C terminus antigenicity in the basal state (1.37 Ϯ 0.03-fold compared with unwashed lawns, p Ͻ 0.005, results not illustrated). Notably, the high salt treatment did not affect the magnitude of the C terminus signal in lawns from insulin-stimulated cells, suggesting that the interfering agent had only been present in lawns from basal (unstimulated) cells. Furthermore, neither the myc nor caveolin signals were affected by treatment of the lawns with the high salt treatment. These results are congruent with the interpretation that a putative masking agent (protein) restricting C terminus recognition by its cognate antibody is removed upon insulin action in intact cells or upon high salt washing of unfixed lawns.
Consistent with this view, there was no change in C terminus reactivity in membrane lawns fixed with PFA prior to the high salt wash (results not shown), suggesting that fixation prevented the removal of the putative masking protein.
As with GLUT4myc myotubes, the C-terminal antigenicity of

TABLE II Insulin increases GLUT4 C-terminal antigenicity in L6 cells
Myoblast monolayers of parental L6-cells or -GLUT4myc cells were serum-starved for 4 h, then stimulated with 100 nM insulin for 10 min.
The myc or C-terminal epitopes were detected by immunofluorescence after cell permeabilization with 0.1 % Triton X-100, and quantification was performed as described under "Experimental Procedures." Results from five independent experiments are expressed as the mean Ϯ S.E. relative to the corresponding unstimulated cells (assigned a value of 1). -, no signal.  I Insulin induces gain in GLUT4 epitopes on the plasma membrane lawns derived from L6-GLUT4myc myotubes by labeling with various antibodies L6-GLUT4myc myotubes were incubated without (basal) or with insulin (100 nM, 10 min). Plasma membrane lawns were prepared, fixed, and incubated with different combinations of primary and fluorophore-conjugated secondary antibodies precisely as stated. The fluorescence intensity per unit area of at least 50 lawn pieces was quantified for each condition per experiment. The results are expressed as the mean fold change per unit area over the basal (Ϯ S.E.) obtained from at least five independent experiments. endogenous GLUT4 in wild-type, unstimulated L6 myotubes was also elevated upon treatment of the corresponding lawns with PBS supplemented with 500 mM NaCl (to a value of 1.26 Ϯ 0.04-fold above the untreated control, p Ͻ 0.005, n ϭ 3, t test). Insulin caused a gain in endogenous C terminus of 1.81 Ϯ 0.05-fold above basal control (p Ͻ 0.005), and this gain was not affected by high salt treatment of the lawns (remaining at 1.88 Ϯ 0.04-fold of above basal control; p Ͻ 0.005, n ϭ 3). The above results support the possibility that the larger gain in lawn-associated C terminus than myc signals observed in response to insulin arises from unmasking of the C terminus, which is hindered from antibody recognition in the basal state. However, it is also possible that the myc epitope becomes antigenically compromised in the insulin-stimulated state. The following experiment with a GLUT4myc-eGFP chimera addresses this possibility. Presumably, the putative C terminus unmasking of GLUT4 would not affect the signal of eGFP linked to the GLUT4myc C terminus. The insulin-induced gains in myc and eGFP signals associated with membrane lawns would be expected to be equivalent unless the myc epitope were antigenically compromised. Accordingly, parental L6 myoblasts were transiently transfected with GLUT4myc-eGFP and stimulated with insulin (100 nM for 10 min). Thereafter, membrane lawns were generated and reacted with monoclonal anti-myc antibody and Cy3-conjugated anti-mouse IgG. Despite the variable expression levels of GLUT4myc-eGFP among cells, by analyzing multiple fields and quantifying 80 membrane lawns for each condition we estimated the gains in myc and eGFP signals after insulin treatment to be 1.41 Ϯ 0.25-and 1.43 Ϯ 0.17-fold above basal values, respectively (p Ͻ 0.05, results not shown). The ratio of immunodetected myc to eGFP signal on plasma membrane lawns was then assigned a value of 1.00 in unstimulated cells. This ratio was not affected by insulin stimulation (1.03 Ϯ 0.03). Hence, GLUT4myc-eGFP responds to insulin, and the eGFP fluorescence per molecule is not altered by insulin treatment, in contrast to the heightened immunoreactivity of the C terminus epitope described above. Collectively, the experiments described in this section suggest the possibility that insulin exerts an unmasking effect on the C terminus of GLUT4 by removing a bound protein.

PI(3,4,5)P 3 and PI3P Increase GLUT4 Presence by the Membrane but Only PI(3,4,5)P 3 Causes GLUT4 Fusion-
The major product of class IA PI 3-kinase is PI(3,4,5)P 3 , and this lipid is elevated in response to insulin in L6 myotubes (39). We have recently shown that exogenous administration of PI(3,4,5)P 3 (delivered via coupling with cationic carrier) suffices to cause significant GLUT4myc arrival at and fusion with the plasma membrane in intact L6 myoblasts, without increasing glucose uptake (27). It was therefore important to assess whether GLUT4myc membrane insertion in response to PI(3,4,5)P 3 is detectable in the lawns, and subsequently to determine whether this phosphoinositide can also cause unmasking of the GLUT4 C terminus. In addition to PI(3,4,5)P 3 , PI3P is produced by muscle and fat cells in response to insulin (40). Therefore, we compared the effect of the two phosphoinositides on GLUT4 arrival at the membrane, and on the availability of the myc epitope for antigenic recognition. L6 myotubes were incubated with PI(3,4,5)P 3 or PI3P and carrier (10 M each) for 20 min, and membrane lawns were generated and labeled with monoclonal anti-myc and secondary antibody. Consistent with our published results in intact myoblasts (27), carrier-mediated PI(3,4,5)P 3 delivery caused significant gain in myc epitope on myotube lawns (1.58 Ϯ 0.04-fold relative to untreated cells, p Ͻ 0.005). These results were not changed when PI(3,4,5)P 3 and carrier were tested at a concentration of 20 M each (1.55 Ϯ 0.03-fold above untreated cells, p Ͻ 0.005). In contrast, carrier-mediated deliveries of PI3P, of PI(4,5)P 2 or of carrier alone, failed to change the lawn levels of myc epitope. The results using 10 M concentrations of each agent were as follows: 1.04 Ϯ 0.23-, 0.99 Ϯ 0.09-, and 0.99 Ϯ 0.20-fold above untreated controls, respectively. No effects were seen either using 20 M concentrations of the carrier and PI3P or PI(4,5)P 2 , as follows: 1.10 Ϯ 0.09 and 0.99 Ϯ 0.11, respectively. These results indicate that input from PI(3,4,5)P 3 could insert GLUT4 into the plasma membrane, whereas PI3P, PI(4,5)P 2 and carrier alone could not (Fig. 6A).
The above results suggest that PI3P may have not mobilized GLUT4myc to the membrane at all, or that GLUT4myc-containing vesicles may be docked but unfused, keeping the myc epitope unavailable for antigenic recognition. We therefore treated the plasma membrane lawns with 0.1% Triton X-100 after fixation, prior to myc immunodetection (Fig. 6A). This treatment revealed a PI3P-dependent gain of previously latent myc epitopes (1.40 Ϯ 0.02-fold relative to untreated controls, p Ͻ 0.05), presumably present in docked, unfused vesicles. In contrast, permeabilizing the lawns with Triton X-100 barely affected the gain in myc signal observed in intact lawns from FIG. 6. PI(3,4,5)P 3 causes a gain in myc epitope on membrane lawns; the gain caused by PI3P is revealed upon treatment with Triton X-100. L6-GLUT4myc myotubes were incubated without or with insulin (100 nM for 10 min), carrier alone (10 M and 20 min), or carrier plus PI(3,4,5)P 3 or PI3P or PI(4,5)P 2 (10 M, 20 min), as indicated. Plasma membrane lawns were generated and fixed, labeled with monoclonal anti-myc antibody followed by goat anti-mouse A488 antibody (A) or polyclonal anti-C terminus of GLUT4 antibody followed by goat anti-rabbit Cy3 (B) and processed for confocal fluorescence microscopy. Where indicated, lawns were treated with 0.1% Triton X-100 after fixation (solid bars). The fluorescence intensity per lawn was quantified from at least 50 plasma membrane lawn pieces for each condition per experiment as described under "Experimental Procedures." Results are the mean -fold change per unit area over the basal (Ϯ S.E.) obtained from at least five independent experiments. *, p Ͻ 0.005; #, p Ͻ 0.05 relative to the corresponding basal values (ANOVA). cells pretreated with PI(3,4,5)P 3 (1.62 Ϯ 0.04-fold relative to untreated controls, p Ͻ 0.005). The myc signal on lawns from cells pretreated with PI(4,5)P 2 plus carrier or carrier alone was not affected by Triton X-100 treatment (0.93 Ϯ 0.12-and 0.99 Ϯ 0.11-fold relative to untreated controls). The above results suggest that both PI(3,4,5)P 3 and PI3P increase the presence of GLUT4myc at the cell periphery, but whereas PI(3,4,5)P 3 leads to GLUT4 vesicle fusion with the plasma membrane, PI3P does not.
PI3P, but Not PI (3,4,5)P 3 , Promotes GLUT4 C Terminus Unmasking-In stark contrast to the gain in myc epitope, no significant gain in C terminus immunoreactivity was elicited by carrier-delivered PI(3,4,5)P 3 (1.15 Ϯ 0.08-and 1.17 Ϯ 0.03-fold with 10 M and 20 M concentrations, respectively) (Fig. 6B). The gain in C terminus immunoreactive epitope on membrane lawns evoked by carrier-dependent delivery of 10 or 20 M PI3P was 1.4 Ϯ 0.07-and 1.4 Ϯ 0.11-fold above untreated controls, respectively (p Ͻ 0.005). Conversely, and as expected, there was no change in C terminus epitope in lawns from myotubes pretreated with 10 M or 20 M PI(4,5)P 2 (1.01 Ϯ 0.09-and 1.00 Ϯ 0.1-fold relative to untreated controls) or with carrier alone (1.02 Ϯ 0.09 and 1.09 Ϯ 0.03) (Fig. 6B). Interestingly, and in contrast to the myc epitope, the gain in C-terminal epitope was unaffected by lawn permeabilization with Triton X-100 under any condition examined. Instead, high salt treatment of lawns from cells stimulated with PI(3,4,5)P 3 elevated C-terminal antigenicity by 1.80 Ϯ 0.03-fold relative to PI(3,4,5)P 3 stimulation without additional NaCl treatment (p Ͻ 0.005) (Fig. 7). Together, these results suggest that PI(3,4,5)P 3 causes arrival and fusion of GLUT4 with the membrane but is insufficient to cause C terminus unmasking. Conversely, PI3P causes arrival of GLUT4 and C terminus unmasking but does not allow fusion of GLUT4 with the plasma membrane.
Following the results described above, it was of interest to examine the combined effect of PI(3,4,5)P 3 and PI3P on GLUT4myc (10 M each). This treatment produced a gain in myc epitope on membrane lawns of 1.65 Ϯ 0.08-fold above basal, p Ͻ 0.005, and a gain in C terminus of 1.80 Ϯ 0.08, p Ͻ 0.005. Hence, although the myc epitope did not increase significantly under these conditions compared with the effect of PI(3,4,5)P 3 alone, the gain in C terminus signal was augmented by the presence of PI3P. These results are consistent with the possibility that PI3P causes unmasking of the C terminus of GLUT4myc molecules recruited by PI(3,4,5)P 3 . Moreover, PI(3,4,5)P 3 added along with insulin did not affect the insulin-dependent gain in C terminus (results not shown), suggesting that PI(3,4,5)P 3 itself does not mask this epitope.
Finally, as presented in Table III, in cells with class IA PI 3-kinase inhibited by 100 nM wortmannin, addition of exogenous PI(3,4,5)P 3 was still able to cause GLUT4 translocation to the membrane (middle column). In cells pretreated with 100 nM wortmannin and stimulated with insulin, PI(3,4,5)P 3 caused GLUT4 recruitment and this GLUT4 underwent C-terminal unmasking (third column), presumably through PI3P produced in response to insulin.
GLUT4myc Fusion and Unmasking in Rounded-up L6 Myoblasts-The proposed three stages in GLUT4 translocation were also analyzed in rounded-up L6 myoblasts. This preparation enables one to label the surface-exposed myc epitope in intact cells, or the intravesicular myc epitope and cytosolicfacing C terminus in permeabilized cells. Hence it offers the possibility to confirm the observations made in lawns without resorting to sonication. Fig. 8A illustrates that insulin caused a gain in myc epitope at the surface of intact cells, which was fully prevented by pretreatment with 100 nM wortmannin. Cellular permeabilization with Triton X-100 reveals the intracellular depots of GLUT4myc, whether perinuclear or at the cell periphery. Such permeabilization enabled detection of myc epitope along the periphery in wortmannin-treated, insulinstimulated cells (Fig. 8B), despite the lack of gain in myc signal in intact cells. Quantitative analysis was performed as described under "Experimental Procedures," and the results are illustrated in Fig. 8C. These results corroborate the observations made in membrane lawns suggesting that unfused vesicles are arrested at the membrane under these conditions. PI(3,4,5)P 3 also provoked a significant gain in surface myc epitope in intact myoblasts, suggesting that PI(3,4,5)P 3 alone can mobilize GLUT4 toward the membrane. The peripheral FIG. 7. High salt treatment of plasma membrane lawns increases the C terminus antigenicity of GLUT4 mobilized by PI(3,4,5)P 3 . L6-GLUT4myc myotubes were incubated without (basal) or with insulin (100 nM for 10 min), carrier plus PI(3,4,5)P 3 or PI(4,5)P 2 (10 M each, for 20 min), as indicated. Plasma membrane lawns were generated and incubated with PBS without or with additional 500 mM NaCl for 7 min (solid bars) on ice prior to fixation, labeling with monoclonal anti-myc antibody followed by goat anti-mouse A488 antibody or polyclonal anti-C terminus of GLUT4 antibody followed by goat anti-rabbit Cy3 and processing for confocal fluorescence microscopy. The fluorescence intensity per lawn was quantified from at least 50 plasma membrane lawn pieces for each condition per experiment as described under "Experimental Procedures." Results are the mean -fold change per unit area over the basal (Ϯ S.E.) obtained from at least five independent experiments. *, p Ͻ 0.005; #, p Ͻ 0.05 (ANOVA).  (3,4,5)P 3 increases GLUT4 C terminus gain on the plasma membrane lawns from the cells treated with 100 nM wortmannin prior to insulin stimulation L6-GLUT4myc-myotubes were serum-starved for 4 h, followed incubation with 100 nM wortmannin (Wm) for 20 min. Cells were incubated with additional 100 nM insulin (I) and/or PI(3,4,5)P 3 plus carrier (10 M each) for 20 min, and plasma membrane lawns were generated and immunolabeled with the indicated antibody, and the immunofluorescent signal was quantified as described under "Experimental Procedures." Results are expressed as the mean Ϯ S.E. from the four independent experiments. myc signal detected in PI(3,4,5)P 3 -treated, then permeabilized myoblasts must arise from GLUT4myc fused with the membrane. In contrast, PI3P did not cause significant gain in surface myc signal unless the cells were permeabilized with Triton X-100 treatment. Again, this observation suggests that PI3P might have mobilized GLUT4myc vesicles to the cell surface that however fail to fuse with the plasma membrane. Moreover, these results suggest that the effect of PI3P is similar in myoblasts (used in the rounded-up cell assay) and myotubes (used for the lawn assay). Consistent with these results, the C terminus epitope of GLUT4myc was readily detectable at the inner periphery of permeabilized, rounded-up, wortmannintreated, insulin-stimulated myoblasts (Fig. 8B, bottom row, and quantified in Fig. 8C), presumably arising from docked, unfused GLUT4 vesicles.
Finally, as observed in membrane lawns, peripheral GLUT4myc that is fully fused with the plasma membrane in response to PI(3,4,5)P 3 was inaccessible to detection by anti-Cterminal antibodies in permeabilized cells, cementing the notion that this phosphoinositide suffices to cause arrival and fusion but not unmasking of GLUT4 at the cell surface. The gain in C terminus epitope at the periphery of PI3P-treated myoblasts is further consistent with the results using plasma membrane lawns suggesting that, like insulin, PI3P causes GLUT4 C terminus unmasking. DISCUSSION A Model for GLUT4 Arrival, Fusion, and C Terminus Unmasking-The results presented here support a model whereby insulin causes arrival, membrane fusion, and the removal of a putative masking protein from GLUT4. Although it is well documented that class IA PI 3-kinase is activated by insulin and that dominant-negative mutants of the enzyme prevent the gain in surface GLUT4, the precise step(s) in GLUT4 cycling regulated by the enzyme remained to be defined. Here we establish a strategy to determine the degree of incorporation of GLUT4 with the plasma membrane, by following simultaneously the changes in antigenic availability of exofacially and endofacially facing epitopes on the transporter. The myc, C-terminal, and middle loop epitopes of GLUT4myc should change in parallel unless there are constraints to the recognition of either one. Under the fluorescence acquisition conditions used the contribution of endogenous GLUT4 is negligible, and therefore only GLUT4myc is detected by the antibodies used. Membrane lawns were used as a platform containing membrane-inserted and membrane-docked vesicles that is amenable to quantitative fluorescence assessment, and rounded-up myoblasts, intact or permeabilized, are used to visualize inserted from docked GLUT4, respectively.
GLUT4 Fusion with the Membrane-The first proposition of this study is that class IA PI 3-kinase input regulates the fusion of insulin-sensitive, GLUT4myc-containing vesicles with the plasma membrane. This possibility is raised by three complementary observations as follows: (a) In intact, rounded-up myoblasts, 100 nM wortmannin and 25 M LY294002 (result not shown) prevent the insulin-induced gain in myc epitope availability from the extracellular milieu, yet, upon permeabilization, the gain in GLUT4 C terminus is readily visualized in the periphery of the cell. (b) In plasma membrane lawns derived from wortmannin-or LY294002-pretreated cells, there was no insulin-dependent gain in antigenically available myc epitope, although, as above, a substantial gain in C terminus was detected. (c) Lawn permeabilization with Triton X-100, expected to permeabilize any vesicular bodies present on the lawns, exposed significant myc epitope to match the insulindependent gain in C terminus. Consistent with this observation, vesicles containing GLUT4 were detected by immunoelectron microscopy on lawns from wortmannin-pretreated, insulin-stimulated cells.
We interpret these results to indicate that, at the concentrations used, wortmannin (100 nM) or LY294002 (25 M) allow significant arrival of GLUT4 at the vicinity of the plasma membrane but prevent the fusion of GLUT4 vesicles. With one exception (41), all previous studies in the literature examining the role of PI 3-kinase in GLUT4 translocation used only high concentrations of wortmannin (300 to 1000 nM) followed by insulin stimulation and immunofluorescent detection of membrane GLUT4 via its C terminus (42)(43)(44) and failed to detect any gain in membrane GLUT4 under these conditions. Indeed, we observed that the insulin-dependent gain in C terminus availability declined gradually in lawns from cells pretreated with higher concentrations of wortmannin (down to 1.2 Ϯ 0.3fold gain in the presence of 1 M wortmannin). High concen-FIG. 8. GLUT4myc arrival, fusion, and unmasking in rounded-up L6-myoblasts. L6-GLUT4myc myoblasts were pretreated with wortmannin (Wm) where indicated, then detached from the substratum as described under "Experimental Procedures." During re-attachment, cells were incubated without (basal) or with insulin (100 nM for 10 min), wortmannin (100 nM), wortmannin (100 nM) plus insulin (100 nM, 10 min; 100 nM Wm plus insulin); or 10 M PI(3,4,5)P 3 or PI3P, or PI(4,5)P 2 , each with carrier. myc-epitope detection with monoclonal anti-myc antibody in intact cells was performed immediately following fixation, without cell permeabilization (A). Cells were permeabilized with 0.1% Triton X-100 treatment (15 min) following fixation but prior to labeling with monoclonal anti-myc or polyclonal anti-GLUT4 C terminus as described under "Experimental Procedures" (B). Shown are representative images obtained by confocal fluorescence microscopy from five independent experiments. Bar, 10 m. A, at least 300 rounded-up cells for each condition as indicated were counted and scored blindly for positive or negative peripheral signal of myc or C terminus epitopes, as described under "Experimental Procedures" (C). The percentage of cells positive for myc or C terminus peripheral signal was calculated from three independent experiments (mean Ϯ S.E.). trations of LY294002 also virtually eliminated the gain in C terminus epitope on the lawns from insulin-stimulated cells (1.14 Ϯ 0.09-fold gain in the presence of 75 M LY294002). Collectively, these observations suggest that migration of GLUT4 toward the plasma membrane (the actual translocation) is less sensitive to these inhibitors than the insertion of GLUT4 into the membrane. When this study was being completed, Bose et al. (45) reported that the fluorescence of GLUT4myc-GFP transfected into 3T3-L1 adipocytes increased in cells pretreated with LY294002 and insulin, yet the myc epitope was unavailable from the cell exterior. Similarly, van Dam et al. (46) reported that Akt activation correlates with GLUT4 insertion rather than arrival at the membrane. Our study expands these observations by demonstrating that the peripheral myc epitope is trapped in Triton X-100-sensitive structures, likely docked but unfused GLUT4 vesicles. Indeed, using immunoelectron microscopy we directly visualize GLUT4-containing vesicles on the lawns of wortmannin-pretreated, insulin-stimulated cells. To our knowledge, this is the first visualization of GLUT4 vesicles adhered to the plasma membrane. The fact that GLUT4 is found in membrane-adhered vesicles in the presence of wortmannin suggests that fusion is a regulated step. Moreover, because no membrane-adhered vesicles were observed in the absence of wortmannin, GLUT4 arrival rather than fusion appears to be the rate-limiting step in basal and insulinstimulated cells.
Unmasking of the GLUT4 C Terminus-The second major observation of this study is that the insulin-dependent gain in myc epitope detected on membrane lawns is routinely and significantly lower than the gain in C terminus detection. The difference holds irrespective of the nature of the antibodies used to detect each epitope (monoclonal or polyclonal) or of the type of fluorophore associated with the secondary antibody. The gain in myc signal was, however, comparable to that of an exofacial epitope between transmembrane helices 1 and 2 prior to the myc epitope insert in GLUT4myc, whereas the gain in C terminus was comparable to that of the middle cytosolic loop of the transporter. We interpret these results to indicate that either the exofacial epitopes are less available for antigenic recognition in the insulin-stimulated state than in the basal state relative to the cytosolic ones or that the cytosolic epitopes are less available for recognition in the basal state. Several observations support the latter possibility, as follows: (a) Treatment of lawns with Triton X-100, shown above to permeabilize docked vesicles, failed to increase the gain in myc signal, suggesting that the lesser myc signal is not due to unavailability due to an abundance of docked, unfused vesicles in the insulinstimulated state. (b) Treatment of lawns with high salt solutions increased the C-terminal antigenicity in the basal, but not in the insulin stimulated state, suggesting that a C terminus obstructing protein might have been removed from the lawns of basal-state cells. This change was precluded if the lawns were fixed with PFA prior to exposure to high salt. (c) The insulin-dependent gain in myc epitope on lawns was similar to that in eGFP signal, analyzed in cells transiently transfected with GLUT4myc-eGFP. (d) Conversely, the C terminus signal was of similar magnitude in GLUT4mycoverexpressing myotubes as in lawns from L6-wild-type myotubes, and high salt treatment of lawns also increased the basal-state signal of endogenous GLUT4. (e) As in membrane lawns, the C-terminal signal from the entire cell was elevated by insulin treatment. In contrast, the myc epitope signal of the entire cell remained unaffected by the treatment with the hormone.
These results are consistent with the possibility that a so far undefined protein that partially obstructs cytosolic epitopes from recognition by antibodies might be released in response to insulin. Indeed, two separate studies have reported insulin-dependent increased reactivity of the C terminus of GLUT4 in adipocytes (47) and skeletal muscle (48), presumably as a result of removal of an interfering protein or a conformational change of the C terminus. It remains possible that the increase in C terminus immunoreactivity may be elicited by a conformational change in GLUT4. Such conformational change would have to be induced by insulin stimulation and reproduced by the high salt treatment of lawns. Several proteins can bind to the C terminus of GLUT4 (49 -51), but so far none of these was shown to be sensitive to insulin stimulation. On the other hand, it was recently reported that the protein TUG binds to GLUT4 (52), and this binding is diminished in response to the hormone (52). Future work should investigate the nature of the protein purported to determine antigenic availability of cytosolic epitopes on GLUT4. Differential Regulation of GLUT4 Fusion and Unmasking by Distinct Phosphoinositides-The third major observation of this study is that PI(3,4,5)P 3 and PI3P delivery each promotes significant arrival of GLUT4myc at the cell periphery. These results, along with the relative insensitivity to 100 nM wortmannin of GLUT4myc arrival at the cell periphery, raise the possibility that the class IA PI 3-kinase product PI(3,4,5)P 3 and the class II PI 3-kinase C2␣ product PI3P might be jointly responsible for the insulin-dependent arrival of GLUT4myc at the vicinity of the plasma membrane. However, whereas PI(3,4,5)P 3 causes GLUT4myc externalization, presumably via vesicle fusion with the membrane, the GLUT4myc molecules summoned to the periphery by addition of PI3P do not result in productive externalization. We surmise that only PI(3,4,5)P 3 but not PI3P can cause GLUT4myc-vesicle fusion with the membrane. Recently, Maffucci et al. (40) also reported that carrier delivery of PI3P promotes GLUT4 arrival at periphery of the cells but interpreted this result to be equivalent to GLUT4 insertion. Our approach comparing detection of the C terminus and the myc epitope on GLUT4myc affords the distinction between these two steps and reveals that each phosphoinositide has distinct actions on GLUT4 externalization. Similarly, Chaussade et al. (53) recently reported that overexpression of myotubularin, a PI3P phosphatase, impaired insulin-induced GLUT4 translocation to the plasma membrane presumably because of reduced PI3P production by insulin in 3T3-L1 adipocytes. As in the study of Maffucci et al., GLUT4-GFP was used, precluding differentiation between adhered and fused GLUT4-vesicles at the plasma membrane. In those studies, as in the present case, the effect of phosphatidylinositol phosphates was determined on the steady-state levels of surface GLUT4 and not on the individual rates of exocytosis or endocytosis. It is unlikely that PI(3,4,5)P 3 increased surface GLUT4 levels by reducing the internalization of preexisting GLUT4 molecules. This is because preventing GLUT4myc internalization should have elevated the myc and C-terminal signals by the same proportion, yet PI(3,4,5)P 3 only increased the myc signal. However, it is theoretically possible that PI3P reduces the progression of membraneattached coated vesicles containing GLUT4 toward the early endosome. These interesting possibilities must be addressed in the future.
Importantly, while PI(3,4,5)P 3 suffices to promote fusion of the GLUT4-containing vesicles arriving at the plasma membrane, it does not emulate the C-terminal unmasking provoked by PI3P and insulin in these muscle cells. In contrast, in lawns of 3T3-L1 adipocytes treated with PI(3,4,5)P 3 we observed a gain in GLUT4 C terminus availability (27), but it was not possible to compare such a gain with that of alternative epitopes. Hence, the question of unmasking of the C terminus in 3T3-L1 adipocytes requires further investigation, as this matter has been controversial in other studies as well (47,54).
In the muscle cells, unmasking of the C terminus of GLUT4 was caused by both insulin and PI3P, but not by PI(3,4,5)P 3 alone. The unmasking of the C terminus was readily detected in whole cells (see Table II) as well as on membrane lawns. Based on the limited sensitivity to low concentrations of wortmannin or LY294002 sensitivity, class II PI 3-kinase C2␣ is a likely candidate to effect the insulin-dependent C-terminal unmasking. Indeed, as presented in Table III, in cells with class IA PI 3-kinase inhibited by 100 nM wortmannin and stimulated by insulin, addition of exogenous PI(3,4,5)P 3 was still able to cause GLUT4 translocation to the membrane, and this GLUT4 underwent C-terminal unmasking. This result further supports the concept that a 100 nM wortmannininsensitive input (presumably class II PI 3-kinase C2␣) can unmask the GLUT4 C terminus. However, class IA PI 3-kinase may also have some input into unmasking, because the gain in C terminus labeling was not higher than that in myc epitope in lawns from cells treated with insulin and 100 nM wortmannin (Fig. 3). In this case, it is plausible that unmasking is signaled preferentially through PI(3,4)-bisphosphate rather than PI(3,4,5)P 3 .
A more accurate discernment of the participation of each class of PI 3-kinase will require selective gene silencing of each isoform. As well, the molecular mechanism of C-terminal unmasking, its functional implications, and the identity of the putative masking protein are the subject of future investigations.
Taken together, our results reveal that insulin regulates GLUT4 arrival at the plasma membrane, its fusion with the plasma membrane, and unmasking of its C terminus. Distinct phosphoinositides are shown to affect these steps differentially, so that PI(3,4,5)P 3 causes GLUT4 arrival and fusion but not unmasking; conversely, PI3P causes GLUT4 arrival and unmasking but not fusion (Fig. 9). We hypothe-size that some combination of these two phosphoinositides contributes to the insulin-dependent regulation of all three events: arrival, fusion, and unmasking. FIG. 9. Proposed model of GLUT4 arrival, fusion, with the plasma membrane (PM) and C terminus unmasking regulated by insulin or distinct phosphoinositides. Insulin-promoted GLUT4 arrival at the PM is partly inhibited by 100 nM wortmannin (Wm). This arrival is also promoted by carrier delivery of the PI(3,4,5)P 3 or PI3P. Insulin-induced GLUT4 fusion with the PM is prevented by 100 nM Wm, and carrier delivery of the PI(3,4,5)P 3 but not PI3P causes GLUT4 fusion. Unmasking of the C terminus of GLUT4 (through a putative GLUT4-binding protein) is effected by insulin or carrier delivery of PI3P but not by PI(3,4,5)P 3 . The insulin-dependent C-terminal unmasking is partly inhibited by 100 nM Wm. The specific location at which unmasking occurs remains unknown.