Insulin-stimulated GLUT4 Translocation in Adipocytes Is Dependent upon Cortical Actin Remodeling* □ S

Rhodamine-labeled phalloidin staining morphologically differentiated 3T3L1 adipocytes demonstrated that F-actin predominantly exists juxtaposed to and lin-ing the inner face of the plasma membrane (cortical actin) with a smaller amount of stress fiber and/or ruffling actin confined to the cell bottom in contact with the substratum. The extent of cortical actin disruption with various doses of either latrunculin B or Clostridium difficile toxin B (a Rho family small GTP-binding protein toxin) directly correlated with the inhibition of insulin-stimulated glucose uptake and GLUT4 translocation. The dissolution of the cortical actin network had no significant effect on proximal insulin receptor signaling events including insulin receptor autophosphorylation, tyrosine phosphorylation of insulin receptor sub-strate and Cbl, or serine/threonine phosphorylation of Akt. Surprisingly, however, stabilization of F-actin with jasplakinolide also resulted in a dose-dependent inhibition of insulin-stimulated glucose uptake and GLUT4 translocation. In vivo time-lapse confocal fluorescent microscopy of actin-yellow fluorescent protein demonstrated that insulin stimulation initially results in cortical actin remodeling followed by an increase in polymerized actin in the peri-nuclear region. Importantly, the insulin stimulation of cortical actin rearrangements was completely blocked by treatment of the cells with latrunculin B, C. difficile toxin B, and jasplakinolide. g for 20 min to remove insoluble material, and 50 (cid:2) g of the total protein were resolved by SDS-polyacrylamide gel electrophoresis. The gels were then subjected to immunoblotting with either a rabbit polyclonal Akt antibody, a (cid:1) -actin antibody, the monoclonal phospho-Akt (Ser-473) antibody, or the phospho-Akt (Thr-308) antibody and visualized by the SuperSignal Chemiluminescence detec-tion kit (Pierce). For immunoprecipitation, whole cell extracts were incubated for 2 h at 4 °C with 5 (cid:2) g of the monoclonal Cbl antibody. The samples were then precipitated with protein G PLUS-Sepharose (Santa Cruz Biotechnology) and were immunoblotted as described above. GLUT4 protein as described under “Experimental Procedures.” These are representative fields of plasma membrane sheets obtained from three independent experiments.

Insulin stimulates glucose uptake in striated muscle and adipose tissue by inducing the translocation of the insulin-responsive glucose transporter isoform (GLUT4) 1 from intracellular storage site(s) to the plasma membrane (1)(2)(3)(4)(5). In the basal state, GLUT4 cycles slowly between the plasma membrane and one or more intracellular compartments, with the vast majority of the transporter residing within the cell interior (6 -10). Activation of the insulin receptor triggers a large increase in the rate of GLUT4 vesicle exocytosis, with a smaller decrease in the rate of internalization by endocytosis (8, 10 -12). Thus, the overall insulin-dependent shift in the cellular dynamics of GLUT4 vesicle trafficking results in a net increase of GLUT4 protein levels on the cell surface, thereby increasing the rate of glucose uptake.
Over the past several years it has become increasingly apparent that the cell cytoskeleton can have substantial influence over vesicle trafficking events. For example, fast axonal transport of synaptic vesicles to the pre-synaptic membrane requires the microtubule cytoskeleton and motors (13)(14)(15). More recently, several studies (16 -20) have also implicated microtubules in the translocation of GLUT4. The actin cytoskeleton has also been observed to have profound influence over regulated exocytosis; however, the functional role of actin in this process appears to be highly complex. For example, most secretory cells have a dense sheet of F-actin beneath and juxtaposed to the plasma membrane, referred to as cortical actin. Several studies (21)(22)(23)(24)(25) have suggested that this actin functions as a physical barrier to vesicle docking based upon its transient depolymerization during exocytosis and that secretion preferentially occurs at sites where the actin cortex is relatively thin. Furthermore, in some cases disruption of the actin cytoskeleton markedly potentiates agonist-stimulated secretion (26 -29). In contrast, however, in many cell systems depletion of F-actin structures either by sequestering actin monomers or by stimulation of actin severing does not stimulate exocytosis but results in an inhibition of agonist-induced secretion (30 -33).
Although it is unclear whether or not GLUT4 transport vesicles have properties more consistent with secretory granules or synaptic vesicles, several studies have suggested a role for F-actin in insulin-stimulated GLUT4 translocation. Treatment with the actin-depolymerizing agent cytochalasin D or the actin monomer-binding Red Sea Sponge toxins latrunculin A or B inhibited insulin-stimulated GLUT4 translocation (19, 34 -37). Insulin has also been observed to induce membrane ruffling (lamellipodia) in a PI 3-kinase-dependent manner (38 -42). Because PI 3-kinase function is also necessary for insulinstimulated GLUT4 translocation, these data suggested that membrane ruffling might play an important regulatory role. * This work was supported in part by Research Grants DK33823, DK59291 and DK25295 from the National Institutes of Health. 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 U.S.C. Section 1734 solely to indicate this fact.
More recently, insulin-stimulated F-actin membrane ruffles in L6 myoblasts were found to accumulate at localized plasma membrane sites and to direct the localization of both the PI 3-kinase and GLUT4 protein (36,43).
Although collectively these data suggest that the actin cytoskeleton plays a role in GLUT4 trafficking events, isolated primary rat adipocytes do not have significant amounts of stress fibers, lamellipodia, or filopodia but instead have a layer of cortical actin similar to that found in secretory cells (37). In this paper, we demonstrate that both disruption and stabilization of adipocyte cortical actin inhibit insulin-stimulated GLUT4 translocation. Furthermore, using time-lapse confocal fluorescent microscopy, we directly demonstrate that insulin induces cortical actin remodeling, and it is the dynamic actin rearrangement process that is necessary for insulin-stimulated GLUT4 translocation.
Cell Culture and 2-Deoxyglucose Transport Assays-3T3L1 pre-adipocytes were cultured in Dulbecco's modified Eagle's medium containing 25 mM glucose, 10% calf serum at 37°C in a 8% CO 2 atmosphere and induced to differentiate into adipocytes as described previously (45). 3T3L1 adipocytes were washed with KRPH buffer (5 mM Na 2 HPO 4 , 20 mM HEPES, pH 7.4, 1 mM MgSO 4 , 1 mM CaCl 2 , 136 mM NaCl, 4.7 mM KCl, and 1% bovine serum albumin). Glucose transport was determined at 4°C by incubation with 50 M 2-deoxyglucose containing 0.5 Ci of 2-[ 3 H]deoxyglucose in the absence or presence of 10 M cytochalasin B. The reaction was stopped after 10 min by washing the cells 3 times with ice-cold phosphate-buffered saline. The cells were then solubilized in 1% Triton X-100 at room temperature for 30 min, and aliquots were subjected to scintillation counting. Protein concentration was determined by the method of Bradford.
Single Cell Microinjection-The microinjection and visualization of single 3T3L1 adipocytes was performed as described previously (46). Briefly, the cells were grown on coverslips, and prior to microinjection, the medium was changed to Lebovitz's L-15 medium containing 0.1% bovine serum albumin. Differentiated 3T3-L1 adipocytes were impaled using Eppendorf model 5171 micromanipulator, and nuclei were injected with 50 g/ml of the YFP-actin cDNA or 200 g/ml of the TC10/ T31N or Cdc42/T17N cDNAs in 100 mM KCl, 5 mM Na 2 PO 4 , pH 7.2, with an Eppendorf model 5246 transinjector. The cells were allowed to recover for 24 h and placed into a perfusion chamber maintained at 37°C and visualized by time-lapse confocal fluorescent microscopy.
Plasma Membrane Sheet Assay-3T3L1 adipocyte plasma membrane sheets were prepared by rinsing the cells in ice-cold phosphate-buffered saline followed by incubation with 0.5 mg/ml poly-D-lysine for 1 min. Cells were then swollen in a hypotonic buffer (23 mM KCl, 10 mM HEPES, 2 mM MgCl 2 , 1 mM EDTA, pH 7.5) by three successive rinses. The swollen cells were sonicated 3 s at power setting 5 with a surface of the cell monolayer in 10 ml of sonication buffer (70 mM KCl, 30 mM HEPES, 6 mM MgCl 2 , 3 mM EGTA, 1 mM dithiothreitol, 0.1 mM phenlymethylsulfonyl fluoride, pH 7.5). The bound plasma membrane sheets were washed twice in sonication buffer, fixed with 2% paraformaldehyde in the sonication buffer, and subjected to confocal fluorescent microscopy.
Immunoprecipitation and Immunoblotting-Following experimental treatments, the cells were solubilized in 25 mM HEPES, pH 7.4, 1% Nonidet P-40, 100 mM NaCl, 2% glycerol, 5 mM sodium fluoride, 1 mM EDTA, 1 mM sodium vanadate, 1 mM sodium pyrophosphate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 5 g/ml leupeptin, 5 g/ml pepstatin A. After 20 min of rotation at 4°C, the extracts were centrifuged at 13,000 ϫ g for 20 min to remove insoluble material, and 50 g of the total protein were resolved by SDS-polyacrylamide gel electrophoresis. The gels were then subjected to immunoblotting with either a rabbit polyclonal Akt antibody, a ␤-actin antibody, the monoclonal phospho-Akt (Ser-473) antibody, or the phospho-Akt (Thr-308) antibody and visualized by the SuperSignal Chemiluminescence detection kit (Pierce). For immunoprecipitation, whole cell extracts were incubated for 2 h at 4°C with 5 g of the monoclonal Cbl antibody. The samples were then precipitated with protein G PLUS-Sepharose (Santa Cruz Biotechnology) and were immunoblotted as described above.

Adipocyte Differentiation Results in the Conversion of F-actin from Stress Fiber and Lamellipodia to Cortical
Actin Structures-Fully differentiated adipocytes are relatively large round lipid-laden cells that are morphologically quite distinct from their precursor fibroblasts. Because the actin cytoskeleton is dynamically regulated and intimately involved in cell morphology and motility, we examined the organization of F-actin during the differentiation of 3T3L1 fibroblasts to morphologically mature adipocytes (Fig. 1). Prior to differentiation, the precursor adipocytes (fibroblasts) primarily display long organized stress fibers typical of F-actin found in other fibroblast cell types. Even though the fibroblasts are relatively flat cells, stress fibers can be detected throughout the cells but are more pronounced at the cell bottom in contact with the substratum (Fig. 1, panels a and e). However, as the cells differentiate into a large rounded morphology, the long stress fiber F-actin structures are converted into F-actin that lines the inner face of the plasma membrane (Fig. 1, panels b-d). In parallel, after 4 days of adipocyte differentiation the length of F-actin stress fibers visualized at the cell bottom were markedly decreased (Fig. 1, panel f). At longer times, the vast majority of stress fiber F-actin was reduced to small patches of punctate actin (Fig. 1,  panels g and h). The distribution of F-actin in adipocytes is more readily apparent when visualized by stacked z axis confocal images (Fig. 2, panels a-l). These data demonstrate that the major form of F-actin in fully differentiated 3T3L1 adipocytes is cortical actin underlying the plasma membrane.
Previous studies (38 -42, 47-49) have observed that insulin can acutely modulate membrane ruffling and stress fiber Factin in a PI 3-kinase-dependent manner. Therefore, we next examined the relationship between stress fiber, membrane ruffles, cortical actin, and PI 3-kinase activity using the selective PI 3-kinase inhibitor wortmannin (Fig. 3). As expected, insulin stimulation resulted in a decrease in the amount of F-actin stress fibers concomitant to the formation of membrane ruffling in the pre-adipocyte cell population (Fig. 3A, panels a and c). Wortmannin treatment also resulted in decrease in F-actin stress fibers but also completely prevented the insulin stimulation of membrane ruffling (Fig. 3A, panels b and d). Although the relative extent of stress fiber and membrane ruffle types of F-actin was reduced in morphologically differentiated adipocytes, wortmannin had no significant effect on either cortical actin (Fig. 3B, panels a and b) or the punctate actin at the cell bottom ( Fig. 3B, panels e and f). Similar to fibroblast, insulin was able to induced membrane ruffling at the cell bottom in contact with the substratum that was also sensitive to wortmannin (Fig. 3B, panels g and h). Insulin was also able to remodel the cortical actin juxtaposed to the plasma membrane (Fig. 3B, panel c). Although the insulin effect on cortical actin was difficult to detect using rhodamine-labeled phalloidin staining of fixed cells, it was readily apparent when visualized by time-lapse confocal microscopy in living cells (see Fig. 9). In any case, wortmannin pretreatment had no significant effect on the distribution of cortical actin (Fig. 3B, panel d). Together, these data directly demonstrate that in adipocytes the regulation and cellular distribution of cortical F-actin was distinct from F-actin involved in the formation of stress fibers and membrane ruffling.
Disruption of Polymerized Actin Inhibits Insulin-stimulated Glucose Uptake and GLUT4 Translocation-Previous studies (19,34,36,37,50) have reported that the actin-depolymerizing agent cytochalasin D and the actin monomer sequestering agents latrunculin A or latrunculin B can inhibit insulin-stimulated GLUT4 translocation in 3T3L1 adipocytes and in primary adipocytes. Consistent with these findings, latrunculin B inhibited insulin-stimulated glucose uptake in a dose-dependent manner (Fig. 4A). As observed previously, rhodamine-labeled phalloidin staining demonstrated that the differentiated adipocytes contain polymerized actin underneath and juxtaposed to the inner face of the plasma membrane (Fig. 4B, panel a). In parallel to the reduction in insulin-stimulated glucose transport activity, latrunculin B treatment resulted in a dosedependent disruption of this cortical actin structure (Fig. 4B, panels b-e). In the absence of latrunculin B only 2% of the cells displayed a discontinuous cortical actin rim, whereas the cortical actin structure was disrupted in 13% of the cells at 1 M, 31% at 5 M, 71% at 20 M, and 91% at 60 M latrunculin B.
It is well established that Rho family members of small GTP-binding proteins play important regulatory roles in the control of actin polymerization (51-54). In addition, we have recently observed that the Rho family member protein TC10 plays a critical function in the insulin-stimulated GLUT4 translocation process (55). We therefore took advantage of C. difficile toxin B, which inactivates Rho family small GTPbinding proteins (56,57). Incubation of adipocytes with toxin B resulted in a slight increase in glucose uptake in the basal state and also resulted in a biphasic response, with a slight enhancement of insulin-stimulated glucose uptake at 10 ng/ml toxin B followed by a dose-dependent inhibition (Fig. 5A). In the basal state, toxin B treatment caused a slight increase in not only glucose uptake but also 125 I-transferrin cell surface binding (data not shown), indicating that toxin B slightly increases general membrane exocytosis and/or inhibits endocytosis. In any case, higher concentrations of toxin B also resulted in a disruption of cortical actin structure (Fig. 5B, panels a-e), suggesting that the Rho family of small GTP-binding proteins primarily maintain the actin cytoskeleton organization in 3T3L1 adipocytes. Similar to the effect of latrunculin B, treatment with 0.01, 0.1, 0.5, and 1.0 g/ml toxin B resulted in 14, 38, 74, and 85% of the cells displaying disrupted cortical actin structures.
To confirm that the inhibition of glucose uptake by the disruption of cortical actin was, in fact, because of a block in GLUT4 translocation, the isolated plasma membrane sheets were examined for the presence of the GLUT4 protein (Fig. 6). As expected, insulin-stimulated a robust translocation of GLUT4 to the plasma membrane as detected by the appearance of a strong GLUT4 immunofluorescent signal (Fig. 6, panels a FIG. 3. Wortmannin inhibits insulin-stimulated membrane ruffling but has no effect on cortical actin in morphologically differentiated adipocytes. A, 3T3L1 fibroblasts were incubated without or with 100 nM of wortmannin for 30 min. The cells were then stimulated without or with 100 nM insulin for 10 min, fixed, labeled with rhodamine-labeled phalloidin, and visualized by confocal fluorescent microscopy. The arrows depict areas of membrane ruffling. B, fully differentiated 3T3L1 adipocytes were incubated without or with 100 nM of wortmannin for 30 min. The cells were then stimulated without or with 100 nM insulin for 10 min, fixed, and labeled with rhodaminelabeled phalloidin. F-actin was then visualized by confocal fluorescent microscopy in sections taken through the middle (panels a-d) and bottom (panels e-h) of the cells. The arrows depict areas of membrane ruffling. These are representative field of cells from three to five independent experiments. and b). However, treatment with either latrunculin B or toxin B markedly inhibited the insulin-stimulated formation of plasma membrane GLUT4 immunofluorescence (Fig. 6, panels c and d).
Toxin B and Latrunculin B Do Not Affect Proximal Insulin Signaling Events-Previous studies (43, 58, 59) have suggested that insulin receptor downstream signaling effectors (IRS and PI 3-kinase) may interact with the cytoskeleton. Thus, the effect of toxin B and latrunculin B to disrupt cortical actin could potentially result in an inhibition of insulin signaling rather than a direct requirement for actin in the GLUT4 translocation process. To address this issue, we next examined the effect of toxin B and latrunculin B on insulin receptor autophosphorylation and IRS tyrosine phosphorylation (Fig. 7A). Although toxin B treatment resulted in a small decrease in insulinstimulated IRS tyrosine phosphorylation, there was no significant effect on insulin receptor autophosphorylation (Fig. 7A,  lanes 1-4). In addition, latrunculin B had no effect on either insulin receptor autophosphorylation or IRS tyrosine phosphorylation (Fig. 7A, lanes 5 and 6). Because IRS phosphorylation leads to the activation of PI 3-kinase, we also examined the phosphorylation of the downstream serine kinase Akt/PKB that also has been implicated in the regulation of GLUT4 translocation (60 -64). Importantly, neither toxin B nor latrunculin B had any significant effect on insulin-stimulated Akt phosphorylation at the threonine 308 or serine 473 activation sites (Fig. 7B, lanes 1-6).
Recently, we have reported (46, 55) that the insulin activation of the PI 3-kinase pathway is not sufficient to mediate GLUT4 translocation. This pathway appears to function in concert with a second insulin receptor signaling pathway that utilizes the tyrosine phosphorylation of the CAP-Cbl complex. Therefore, we determined the effect of toxin B and latrunculin B on insulin-stimulated Cbl tyrosine phosphorylation (Fig. 7C). Insulin stimulation resulted in the tyrosine phosphorylation of immunoprecipitated Cbl protein (Fig. 7C, lanes 1 and 2). Pretreatment with toxin B had no significant effect on either the basal or insulin-stimulated Cbl tyrosine phosphorylation (Fig.  7C, lanes 3 and 4). Surprisingly, however, disruption of F-actin with latrunculin B resulted in the tyrosine phosphorylation of Cbl in the basal state that was similar to the extent of insulinstimulated tyrosine phosphorylation (Fig. 7C, lanes 5 and 6). This difference was not due to unequal immunoprecipitation of the Cbl protein (Fig. 7C, lanes 1-6). Although the basis for enhanced basal Cbl tyrosine phosphorylation by latrunculin B pretreatment is not known, taken together these data indicate that disruption of cortical actin structure inhibits GLUT4 translocation without preventing proximal insulin receptor signaling events.
Stabilization of Polymerized Actin Also Inhibits Insulin-stimulated Glucose Uptake and GLUT4 Translocation-Because the stimulation of GLUT4 translocation by insulin appears to be dependent upon cortical actin, it is possible that either stable cortical actin structures and/or dynamic cortical actin rearrangements are functionally required. To address this issue, we took advantage of jasplakinolide, a cell-permeable agent that promotes the formation and/or stabilization of actin filaments (65)(66)(67). Similar to latrunculin B and toxin B, jasplakinolide also inhibited insulin-stimulated glucose uptake in a dose-dependent manner (Fig. 8A). Jasplakinolide was also found to prevent insulin-stimulated GLUT4 translocation as assessed by the plasma membrane sheet assay (Fig. 8B, panels a-c). Because jasplakinolide binds to the same site and competes for phalloidin binding to F-actin, we were unable to examine the effect of jasplakinolide on cortical actin structure using rhodamine-labeled phalloidin. However, we were able to examine the effect of jasplakinolide using time-lapse confocal fluorescent microscopy of adipocytes expressing YFP-actin (see Fig. 10). In any case, these data indicate that insulin-induced actin rearrangements but not static actin structures are necessary for GLUT4 translocation.
Insulin Stimulates Cortical Actin Rearrangements in Intact Cells-Because individual adipocytes are quite distinct morphologically and the cortical actin structures are not as pronounced as stress fibers or membrane ruffles in fibroblasts, it is very difficult to detect dynamic changes in actin organization by comparing different cells following adipocyte fixation. Therefore, to determine whether insulin can induce time-dependent actin rearrangements in vivo, we took advantage of a ␤-actin enhanced yellow fluorescent fusion protein (YFP-actin) to examine the dynamics of actin rearrangements by timelapse confocal microscopy (Fig. 9). It has been reported that ␤-actin-GFP can faithfully reproduce actin polymerization and motility in vitro so long as the GFP-tagged actin was less than 30% of native actin protein (68,69). Therefore, to avoid excess expression of YFP-actin, we microinjected 3T3L1 adipocyte  2,4, and 6) of 100 nM insulin for 5 min. A, whole cell detergent extracts were subjected to SDS-polyacrylamide gel electrophoresis and immunoblotted with phosphotyrosine or actin antibodies. B, the whole cell detergent extracts were immunoblotted with Akt, phosphoserine 3473specific, or phosphothreonine 308-specific Akt antibodies. C, the whole cell detergent extracts were immunoprecipitated with a Cbl antibody and either immunoblotted with the PY20 phosphotyrosine antibody (top) or a Cbl antibody (bottom). These are representative immunoblots independently performed three times. nuclei with a relatively low concentration of YFP-actin cDNA (50 g/ml in the microinjection pipette). The functional properties of the expressed YFP-actin were confirmed under these conditions, as insulin characteristically stimulated membrane ruffling of the YFP-actin in pre-differentiated 3T3L1 fibroblasts (data not shown). In the pseudo color mode scale used to observe changes in fluorescent intensity, expression of YFPactin resulted in substantial background fluorescence (red), characteristic of free monomeric actin, with the more intense fluorescent areas (white) indicative of polymerized actin. In the basal state, there was a low level of polymerized actin present beneath the plasma membrane (cortical actin) with some punctate accumulation in the peri-nuclear region (Fig. 9A, panels a  and b). Insulin stimulation increased cortical actin polymerization and at the 5-min time frame the formation of a large actin protrusion was formed (Fig. 9A, panel c). The increase in cortical actin was subsequently followed by increased actin polymerization in the peri-nuclear region (Fig. 9A, panel d).
The insulin-stimulated dynamic increases in both cortical and peri-nuclear actin can be more readily appreciated by visualization of a time-lapse video (see Supplementary Material) and by quantitation of fluorescent intensity in different regions of the cell marked by the colored boxes (Fig. 9B).
As observed with rhodamine-labeled phalloidin staining, treatment of the YFP-actin expressing adipocytes with latrunculin B or toxin B markedly reduced the amount of polymerized actin (Fig. 10, panels a and e). These treatments also resulted in a complete inhibition of any insulin-stimulated actin rearrangements (Fig. 10, panels b-f and f-h). The complete blockade of actin remodeling is easily visualized in the time-lapse video presented in the Supplementary Material. In contrast, treatment of the adipocytes with jasplakinolide not only stabilized the cortical actin but also induced a marked expansion of polymerized actin concentrated in the peri-nuclear region (Fig.  10, panel i). Despite the large increase in polymerized actin, jasplakinolide-treated cells also had a complete block of insulin-stimulated actin rearrangements (Fig. 10, panels j-l, and Supplementary Material).
The Dominant-interfering TC10/T31N Mutant Disrupts Cortical Actin and Inhibits Insulin-stimulated Actin Rearrangements-Recently, the Rho family member small GTP-binding protein TC10 has been implicated in the regulation of insulinstimulated GLUT4 translocation (55). Because Rho family proteins have been well established to regulate actin structure, we examined the effect of the dominant-interfering TC10/T31N mutant. Following microinjection of the cDNA encoding for a Myc epitope-tagged TC10/T31N, adipocytes expressing this dominant-interfering protein were identified by labeling with a monoclonal Myc epitope-specific antibody (Fig. 11A, panels a  and d). Co-labeling of these cells with rhodamine labeled-phalloidin demonstrated that only the TC10/T31N expressing adipocytes had a marked reduction in cortical actin structures (Fig. 11A, panels b and c). As a control, we also expressed a dominant-interfering mutant of another Rho family member established to regulate actin dynamics in fibroblasts, Cdc42/ T17N (51, 70,71). In contrast to TC10/T31N, expression of Cdc42/T17N had no significant effect on adipocyte cortical actin with only 7% of the cells having any morphological changes in phalloidin labeling (Fig. 11A, panels d-f). This is consistent with the inability of insulin to activate Cdc42 in adipocytes (55).
To determine whether the expressed TC10/T31N protein also prevented insulin-stimulated cortical actin rearrangements, we co-microinjected adipocytes with TC10/T31N and YFP-actin (Fig. 11B). In the absence of TC10/T31N, expression of YFPactin displayed the typical polymerization pattern being distribution between cortical and peri-nuclear regions. As observed previously (Fig. 9), insulin stimulation increased the amount of both polymerized cortical and peri-nuclear actin. In contrast, expression of TC10/T31N completely disrupted the cortical actin localization of YFP-actin consistent with the rhodaminelabeled phalloidin staining of fixed cells (Fig. 11B, panel a). Furthermore, in the presence of TC10/T31N insulin was unable to induce the appearance of or change in cortical actin structure (Fig. 11B, panels b-d). The dynamics of YFP-actin in adipocytes expressing TC10/T31N can be more readily observed in the time-lapse imaging presented in the Supplementary Material. In any case, the disruption of cortical actin remodeling by expression of TC10/T31N is consistent with the inhibition of insulin-stimulated GLUT4 translocation (55). DISCUSSION One of the most intensively studied actions of insulin is its ability to enhance glucose uptake through the translocation of the insulin-responsive GLUT4 glucose transporter from intracellular storage sites to the plasma membrane. This is a highly complex and dynamic process that appears to require at least two independent but cooperative signal transduction pathways (46,55). The GLUT4 protein continuously cycles through various endomembrane compartments in both the basal and insulin states (1-3, 5). Although it is well established that insulin increases the rate of exocytosis, whether this occurs from a direct trafficking of a specialized GLUT4 compartment, the release of an intracellular sequestered vesicle population or the entrance of the GLUT4 protein into the constitutively recycling vesicle population has not yet been established. Furthermore, the specific sites of action and/or the molecular targets of the insulin signaling cascades leading to any of these potential events have remained highly controversial and incomplete.
Recently, several studies have suggested that both the microtubule and actin-based cytoskeleton networks may play important roles by providing an organized network for the movement or trafficking of the insulin-stimulated vesicle compartments to the cell surface (16 -20, 34 -37, 43, 58, 59). These findings are primarily based on the observation that disruption of F-actin and/or microtubules inhibits GLUT4 translocation and that a portion of the GLUT4 compartments appears to co-localize with these cytoskeleton structures. These data are consistent with the cytoskeleton providing a molecular scaffold allowing for the organized trafficking of insulin-stimulated GLUT4-containing compartments to the plasma membrane.
However, the actin cytoskeleton is quite complex and different cell types display multiple distinct organization patterns of F-actin with each undergoing different modes of regulation and extents of polymerization/depolymerization. This problem has been further complicated by studies examining the role of Factin in different secretory and membrane transport processes. In many cases, a negative or barrier function for F-actin in membrane transport has been proposed as actin-depolymerizing agents can markedly potentiate agonist-stimulated secretory events (21)(22)(23)(24)(25). In other contexts, F-actin appears to provide a critical scaffolding function, and in some cases the rapid polymerization/depolymerization on membrane compartments can act as molecular motors (72)(73)(74). In the case of GLUT4 translocation, this problem is further exacerbated by the observation that different model cell systems appear to display different cellular organizations of F-actin. Furthermore, insulin has been observed to markedly induce membrane ruffling in a PI 3-kinase-and Rac1-dependent manner (38 -42, 75-77). However, membrane ruffling is a characteristic of cells undergoing active motility, whereas insulin-responsive adipocytes and skeletal muscle are typically non-motile cells.
Thus, to examine further the relationship between actin structure, insulin action, and GLUT4 translocation, we initially compared the F-actin structures during adipocyte differentiation. As expected, the pre-adipocytes have a fibroblast morphology and displayed typical actin stress fibers and membrane ruffles. Insulin stimulation in these cells resulted in increased membrane ruffling that was inhibited by the selective PI 3-kinase inhibitor wortmannin. In contrast, morphologically differentiated adipocytes primarily express F-actin around the cell cortex (cortical actin) that lines the inner face of the plasma membrane. This cortical actin did not display insulin-stimulated membrane ruffling but appeared to undergo dynamic remodeling (polymerization/depolymerization). These results are in excellent agreement with the analysis of F-actin in primary rat adipocytes, which were also found to predominantly display a cortical actin network (37). In addition, the morphologically fully differentiated adipocytes also contained a smaller amount of punctate F-actin localized to the cell bottom in contact with the substratum. Similar to fibroblasts, the adipocyte cell bottom also underwent an insulin-stimulated membrane ruffling in a wortmanninsensitive manner. However, this actin does not appear to be responsible for insulin-stimulated GLUT4 translocation, as this is only a minor fraction of the adipocyte F-actin localized to a small portion of the cell surface membrane. In addition, expression of constitutively active and dominant-interfering Rac1 mutants markedly modulates actin at the cell bottom but does not have any significant effect on insulin-stimulated GLUT4 translocation. 2 These findings are also consistent with previous reports (78 -80) indicating that Rac-mediated changes in the actin cytoskeleton do not affect insulin-stimulated glucose uptake or GLUT4 translocation in adipocytes. Nevertheless, other members of the Rho family of small GTP-binding proteins have also been implicated in various aspects of F-actin regulation, and we have recently (55) reported a role for TC10 in the insulin regulation of GLUT4 translocation. In this regard, Rho family-specific toxin (C. difficile toxin B) was also a potent disrupter of cortical actin structures and prevented insulin-stimulated actin rearrangements. Importantly, the degree of cortical actin derangement by both latrunculin B and toxin B directly correlated with the extent of glucose transport inhibition. Although the interpretation of data obtained with various pharmacological agents must be made with caution, we have used two highly specific modulators of actin polymerization as well as specific Rho family toxin. Furthermore, timelapse actin fluorescent microscopy clearly revealed an insulinstimulated rearrangement of cortical actin in vivo that was completely blocked following expression of a dominant-interfering Rho family member GTP-binding protein mutant. This model of cortical actin remodeling is also consistent with a recent report (81) indicating direct function of the actin cortex in the transport of PC-12 cell secretory granules.
Previously, we have also demonstrated that the insulin regulation of GLUT4 translocation requires at least two signaling transduction events, one dependent on activation of PI 3-kinase signaling and the other on the activation of TC10 (46,55). The activation of these two pathways are distinct in that wortmannin blocks PI 3-kinase activation but not TC10, whereas expression of a dominant-interfering CAP mutant inhibits TC10 activation but not PI 3-kinase signaling. The independence between the TC10 and PI 3-kinase pathways is also consistent with our observation that disruption of cortical actin with either latrunculin B or toxin B does not impair the insulin stimulation of Akt/PKB activation. However, these findings are in contrast to fibroblasts in which disruption of F-actin with cytochalasin D was found to inhibit the insulin stimulation of Akt/PKB (82). In any case, our data are consistent with the insulin regulation of cortical actin structures in fully differentiated adipocytes occurring through a TC10-dependent pathway. In addition, the specificity for TC10 function was demonstrated by the observation that the closely related Rho family member Cdc42 does not play a significant role in the regulation of actin dynamics in adipocytes, consistent with the inability of insulin to activate Cdc42 or to regulate GLUT4 translocation (55). Thus together, these data specifically implicate adipocyte cortical actin remodeling through TC10 function as an essential process in GLUT4 translocation.
In summary, fully differentiated adipocytes primarily express cortical actin that must undergo active insulin-stimu-lated remodeling to allow for insulin-stimulated GLUT4 translocation. This cortical actin network appears to be regulated by TC10, a member of the Rho family of small GTP-binding protein. Although the insulin stimulation of membrane ruffling in fibroblasts and at the cell bottom of adipocytes is clearly PI 3-kinase-dependent, insulin regulation of cortical actin rearrangements appears to be a necessary event in GLUT4 translocation. More importantly, our data are inconsistent with a model of cortical actin functioning as a passive molecular network allowing for the trafficking of the GLUT4-containing compartments to the plasma membrane. Instead, these data demonstrate that cortical actin remodeling plays an integral positive role in the trafficking of GLUT4 vesicles. Further studies will be necessary to determine whether cortical actin rearrangements in vivo reflect actin-based motility of GLUT4 vesicles or some other dynamic change in actin-based trafficking function.