This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Watson, R. T. Articles by Pessin, J. E. Search for Related Content PubMed PubMed Citation Articles by Watson, R. T. Articles by Pessin, J. E.

Regulated Membrane Trafficking of the Insulin-Responsive Glucose Transporter 4 in Adipocytes

Department of Pharmacological Sciences, State University of New York at Stony Brook, Stony Brook, New York 11794

Correspondence: Address all correspondence and requests for reprints to: Jeffrey E. Pessin, Ph.D., Department of Pharmacological Sciences, State University of New York at Stony Brook, Stony Brook, New York 11794. E-mail: pessin{at}pharm.sunysb.edu

	Abstract

Top Abstract I. Introduction II. The Hexose Transporter... IV. Regulated GLUT4 Exocytosis V. Role of the... VI. The IR: Attenuation... VII. GLUT4 Vesicle Docking... VIII. GLUT4 Endocytosis IX. GLUT4 Activation X. Summary and Future... References

 
Since the discovery of insulin roughly 80 yr ago, much has been learned about how target cells receive, interpret, and respond to this peptide hormone. For example, we now know that insulin activates the tyrosine kinase activity of its cell surface receptor, thereby triggering intracellular signaling cascades that regulate many cellular processes. With respect to glucose homeostasis, these include the function of insulin to suppress hepatic glucose production and to increase glucose uptake in muscle and adipose tissues, the latter resulting from the translocation of the glucose transporter 4 (GLUT4) to the cell surface membrane. Although simple in broad outline, elucidating the molecular intricacies of these receptor-signaling pathways and membrane-trafficking processes continues to challenge the creative ingenuity of scientists, and many questions remain unresolved, or even perhaps unasked. The identification and functional characterization of specific molecules required for both insulin signaling and GLUT4 vesicle trafficking remain key issues in our pursuit of developing specific therapeutic agents to treat and/or prevent this debilitating disease process. To this end, the combined efforts of numerous research groups employing a range of experimental approaches has led to a clearer molecular picture of how insulin regulates the membrane trafficking of GLUT4.

I. Introduction

II. The Hexose Transporter Family

A. Family members, structure, and function

B. GLUT4

III. Intracellular GLUT4 Storage Compartments

A. The specialized GLUT4 storage compartment

B. Insulin-regulated amino peptidase (IRAP)

C. GLUT4 targeting motifs

IV. Regulated GLUT4 Exocytosis

A. The IR: structure and function

B. The IR: knockout mouse models

C. The IR: complex signaling networks

D. The PI3K and GLUT4 translocation

E. Protein kinase B (PKB) and atypical protein kinase C (aPKC)

F. A second insulin-signaling pathway regulating GLUT4 translocation?

V. Role of the Cytoskeleton in Insulin-Stimulated GLUT4 Translocation

A. Actin

B. Microtubules

VI. The IR: Attenuation of Receptor Signaling

VII. GLUT4 Vesicle Docking and Fusion with the Plasma Membrane

A. SNARE proteins

B. SNARE-associated proteins in GLUT4 translocation

C. Exocyst in GLUT4 translocation

D. Rab and ARF proteins in GLUT4 translocation

VIII. GLUT4 Endocytosis

IX. GLUT4 Activation

X. Summary and Future Directions

	I. Introduction

Top Abstract I. Introduction II. The Hexose Transporter... IV. Regulated GLUT4 Exocytosis V. Role of the... VI. The IR: Attenuation... VII. GLUT4 Vesicle Docking... VIII. GLUT4 Endocytosis IX. GLUT4 Activation X. Summary and Future... References

 
GENERATED BY THE carbon fixation or dark reactions of photosynthesis, glucose is extremely common in nature. Indeed, one of the polymeric forms of glucose, cellulose, is likely the most abundant organic molecule in the biosphere. Therefore, it is not surprising that glucose is the primary energy source used in the animal kingdom, and multiple mechanisms have evolved for its metabolism, synthesis, and storage. Because hydrophilic carbohydrate molecules are generally excluded from the hydrophobic core of cellular membranes, numerous transport mechanisms have evolved to move glucose and other sugars across lipid bilayers (1, 2). For example, bacteria employ several symport, antiport, and uniport systems for the uptake of glucose and other carbohydrates. In contrast, mammals use Na⁺-dependent cotransporters and facilitative uniporters. Moreover, many bacteria can survive with glucose as their only carbon source, and several mammalian tissues, notably the brain, are dependent on glucose as their primary energy source. In addition, to accommodate the glucose requirements of the brain and other tissues, finely tuned regulatory systems have evolved in mammals to maintain remarkably stable plasma glucose concentrations despite intermittent food intake. Thus, during fasting periods the liver provides sufficient glucose output to maintain normal circulating glucose levels. In contrast, the increases in plasma glucose levels that occur immediately after a meal are minimized by enhanced glucose uptake in skeletal muscle and adipose tissue. This complex interplay between glucose absorption from the gut, output from the liver, and uptake by peripheral tissues is normally exquisitely balanced such that plasma glucose concentrations are maintained at 4–7 mM in humans. Deviations from this narrow range can have severe consequences. For example, low blood glucose levels can lead to seizures, coma, and death. In contrast, prolonged elevated glucose levels, as occurs in the diabetic state, can result in many complications, including renal failure, blindness, neuropathy, and cardiovascular disease.

There are several types of diabetes, with type 1 or insulin-dependent diabetes mellitus and type 2 or non-insulin-dependent diabetes mellitus accounting for the vast majority of pathological states of dysregulated glucose metabolism. Type 1 diabetes mellitus results from the autoimmune destruction of pancreatic ß-cells of the islets of Langerhans, resulting in absolute insulin deficiency. In contrast, the hallmark of type 2 diabetes mellitus is peripheral insulin resistance, which refers to the inability of insulin to efficiently stimulate glucose uptake into muscle and adipose tissues and to suppress hepatic glucose production. The progression to overt diabetes in these individuals usually results when enhanced insulin secretion from pancreatic ß-cells fails to compensate for peripheral insulin resistance. In addition to environmental risk factors, including diet, age, and exercise habits, type 2 diabetes is a complex polygenic disease. Type 2 diabetes accounts for more than 90% of all cases and is becoming increasing common worldwide, both in developed and developing countries. Basic research into the molecular mechanisms of insulin signaling and glucose uptake thus represents an important research area for the global human community. Since the discovery of insulin in the early 1920s, much has been learned about how this peptide hormone triggers the tyrosine kinase activity of its cell-surface receptor and initiates intracellular signaling cascades. We now know that insulin induces the translocation of the glucose transporter 4 (GLUT4) from intracellular membrane compartments to the plasma membrane, where it catalyzes the uptake of glucose into adipose and muscle cells, the rate-limiting step for glucose metabolism. Research on the mechanisms of insulin-induced GLUT4 translocation has thus progressed toward a convergence of receptor tyrosine kinase-signaling processes and regulated membrane-trafficking events.

	II. The Hexose Transporter Family

Top Abstract I. Introduction II. The Hexose Transporter... IV. Regulated GLUT4 Exocytosis V. Role of the... VI. The IR: Attenuation... VII. GLUT4 Vesicle Docking... VIII. GLUT4 Endocytosis IX. GLUT4 Activation X. Summary and Future... References

 
A. Family members, structure, and function
In mammals, glucose is cleared from the bloodstream by facilitative transporters (GLUTs), which comprise a family of highly related 12 transmembrane domain-containing proteins (Fig. 1) (3). As facilitative carriers, the GLUT proteins transport glucose down its concentration gradient in an energy-independent manner. Most mammalian cell types are net consumers of glucose and maintain low intracellular glucose concentrations, thus favoring glucose entry. Hepatocytes, however, are net producers of glucose during the periods of reduced food intake (low circulating insulin but high glucagon levels). During fasting periods, hepatic gluconeogenesis and glycogenolysis increase the concentration of intracellular glucose above that in the blood. This results in a net efflux of glucose from the liver and provides the brain and other tissues with a steady supply of glucose despite sporadic food consumption.

View larger version (28K): [in this window] [in a new window]   FIG. 1. Schematic diagram of the predicted membrane topology of GLUT4. Other members of the GLUT family also contain 12-transmembrane domains. The amino- and carboxyl-terminal targeting motifs of GLUT4 are indicated. See text for further details.

GLUT1 is expressed to some degree in most tissues; however, it is highly expressed in endothelial cells lining the blood vessels of the brain and in human erythrocytes (5, 6). Indeed, GLUT1 represents approximately 5% of the total erythrocyte membrane protein content, which facilitated its biochemical purification by Kasahara and Hinkle (7) in 1977. GLUT1 is also expressed in adipose and muscle tissue, which are the insulin-responsive sites for glucose disposal. Here, GLUT1 appears to provide a low, constitutive level of glucose transport required for basal cellular processes, probably in combination with several other GLUT isoforms (5). GLUT2 is a low-affinity [high Michaelis-Menten constant (K_m)] GLUT present in liver, intestine, kidney, and pancreatic ß-cells (8). This transporter functions as part of the glucose sensor system in ß-cells and in the absorption of glucose by intestinal epithelial cells. GLUT3 is expressed primarily in neurons, and, together, GLUT1 and GLUT3 allow glucose to cross the blood-brain barrier and enter neurons (6). GLUT4 is expressed primarily in striated muscle and adipose tissue and, unlike most other GLUT isoforms, is sequestered in specialized intracellular membrane compartments under basal conditions (9). As the major insulin-responsive GLUT isoform, the study of the structure, function, and regulation of GLUT4 has been a major focus of workers in the diabetes field.

B. GLUT4
1. Translocation hypothesis.
In 1980, nearly a decade before the molecular identification and cloning of GLUT4, compelling evidence was presented from two independent groups that strongly supported the hypothesis that insulin causes the redistribution of what was then described as "glucose transport activity" from intracellular membrane compartments to the cell surface (10, 11). By measuring glucose binding and transport activity in membrane fractions isolated from rat adipocytes, both groups found that insulin decreased the number and function of GLUTs in the internal membrane fractions while at the same time causing a concomitant increase in transporter function in the plasma membrane fractions. Although these data were consistent with the translocation of a GLUT to the cell surface, without specific antibodies or a cloned transporter cDNA there was no unequivocal, independent means to verify these results. These early experiments, nevertheless, proved sufficient incentive, and in 1989 five independent groups reported the cloning of the cDNA encoding GLUT4 (12, 13, 14, 15, 16). This enabled the generation of isoform-specific antibodies and confirmed that the glucose transport activity identified in the 1980 experiments corresponded to the translocation of GLUT4 to the cell surface in response to insulin. Indeed, since its initial proposal, more than two decades of accumulated evidence has consistently supported the translocation hypothesis. However, in addition to stimulating translocation, recent work has suggested that insulin may also regulate the intrinsic transport activity of GLUT4 (reviewed in Ref. 17). Nevertheless, the redistribution of GLUT4 to the plasma membrane in response to insulin remains the dominant paradigm and can be readily visualized using a GLUT4-enhanced green fluorescent protein reporter construct transiently expressed in 3T3L1 adipocytes (Fig. 2). The molecular mechanism by which insulin causes this redistribution of GLUT4 remains unknown and is currently under intense investigation. Indeed, this research area holds promise for the identification of therapeutic agents that target specific molecules functioning proximal to the intracellular GLUT4 storage compartment, thereby bypassing the multifarious upstream insulin receptor (IR)-signaling cascades.

View larger version (25K): [in this window] [in a new window]   FIG. 2. Insulin stimulation results in the translocation of GLUT4 from intracellular storage sites to the plasma membrane. Differentiated 3T3L1 adipocytes were transfected with a GLUT4-enhanced green fluorescent protein fusion construct and then incubated in the absence (A) or presence (B) of insulin for 30 min. The cells were fixed and subjected to confocal fluorescent microscopy.

III. Intracellular GLUT4 Storage Compartments
A. The specialized GLUT4 storage compartment
Under basal conditions the majority of GLUT4 is excluded from the plasma membrane of adipose and muscle cells due to efficient endocytosis and intracellular sequestration within insulin-responsive storage compartments. Insulin increases the exocytosis of GLUT4 (9) and slightly inhibits endocytosis (25, 26, 27, 28), resulting in an approximately 10- to 20-fold increase in transporter levels at the plasma membrane. Considerable attention has focused on identifying and characterizing the insulin-regulated GLUT4 storage compartment (29). In addition, several groups (30, 31, 32, 33, 34) have attempted to enhance our understanding of the compartmentalization and regulated trafficking of GLUT4 through comparisons with the insulin-stimulated secretion of other adipocyte proteins. For example, adipocytes secrete several small soluble proteins in response to insulin, including adipsin, leptin, lipoprotein lipase (LPL), and adipocyte complement related protein of 30 kDa (ACRP30; also known as adiponectin or adipoQ), among others. In all cases studied, GLUT4 localized to membrane compartments that were separate from those occupied by these secretory proteins. It therefore appears that adipocytes harbor several intracellular compartments that respond to insulin. Moreover, insulin stimulation results in about a 2-fold increase in the extent of exocytosis of these secreted proteins. In contrast, insulin causes a dramatic 10- to 20-fold increase in GLUT4 at the plasma membrane. Taken together, these data suggest that GLUT4 is targeted to specific membrane compartments that are exceptionally insulin responsive and that are separate from those occupied by secreted peptides.

Although the existence of a specialized insulin-responsive GLUT4 storage compartment is supported by abundant indirect evidence, it has been very difficult to biochemically or structurally characterize this compartment. In part this is because at steady state GLUT4 is distributed, at varying levels, throughout most of the endomembrane system (35). Indeed, after insulin-stimulated translocation, GLUT4 is retrieved from the plasma membrane by endocytosis and routed through a complex trafficking itinerary that appears to include endosome compartments and the trans-Golgi network (TGN), before returning to the insulin-responsive storage compartment. This complex itinerary has made it difficult to reliably distinguish between the insulin-responsive compartment and the general endosome recycling compartments occupied by GLUT4. In addition, there may be multiple interrelated insulin-responsive GLUT4 storage compartments with overlapping properties (36, 37, 38, 39, 40).

Nevertheless, several approaches including subcellular fractionation, vesicle immunoabsorption, membrane compartment ablation, and immunoelectron microscopy have begun to distinguish between the general versus specialized GLUT4 recycling compartments. For example, a horseradish peroxidase-transferrin receptor conjugate has been used in compartment ablation studies to show that the insulin-responsive pool of GLUT4 protein partitions into vesicular compartments that are largely separate from those occupied by cellubrevin and the transferrin receptor (41). Moreover, vesicle immunoabsorption experiments have demonstrated that although some vesicle populations contain both GLUT4 and general endosomal markers, other vesicles are enriched for GLUT4 and appear to exclude endosomal proteins (41, 42, 43). Similarly, studies using quantitative immunoelectron microscopy have demonstrated that the insulin-responsive pool of GLUT4 apparently consists of a unique population of vesicles that is largely devoid of constitutively recycling proteins such as the cation-dependent mannose-6-phosphate receptor (CD-MPR) (44). Moreover, cell fractionation studies employing iodixanol (Optiprep) gradients showed that, although GLUT4 was sorted into multiple intracellular compartments, those responsive to insulin were distinct from endosomes or the TGN (45).

B. Insulin-regulated amino peptidase (IRAP)
IRAP was identified as a major protein that colocalizes with GLUT4 in insulin-responsive storage vesicles (Refs. 46, 47, 48, 49 and reviewed in Ref. 50). IRAP is a type II integral membrane protein with an amino-terminal cytoplasmic tail of 109 amino acids, a transmembrane domain, and an extracellular domain of 894 amino acids. Interestingly, the cytoplasmic domain of IRAP harbors dileucine and acidic cluster motifs that resemble those in the carboxyl terminus of GLUT4 (51). Membrane aminopeptidases are known to process regulatory peptides, and IRAP has been shown to cleave vasopressin, oxytocin, lys-bradykinin, met-enkaphalin, dynorphin, and angiotensins III and IV (50). In addition, IRAP has recently been identified as the angiotensin IV receptor (52). A possible connection between IRAP and GLUT4 was recently uncovered through the targeted ablation of IRAP (53). Interestingly, the IRAP^–/– mice showed decreased levels of GLUT4 protein in skeletal muscle, heart, and adipose tissues, despite being normal for glucose homeostasis (53). Although the mechanism for the observed decrease in GLUT4 protein levels remains to be elucidated, these results nevertheless hint at a potential role for IRAP in regulating GLUT4 expression.

C. GLUT4 targeting motifs
The data described above support a model in which GLUT4 is partitioned into specialized insulin-responsive storage compartments that represent the primary site of insulin action. In addition, this model predicts that GLUT4 contains intrinsic targeting domains that direct its localization to the insulin-responsive compartment. Consistent with this prediction, the membrane compartment distribution of GLUT4 is dramatically different from the very similar GLUT1 transporter (65% amino acid identity), which localizes predominantly to the plasma membrane, even under basal conditions (54). This difference in intracellular localization suggests that the two transporters undergo differential targeting/sorting processes. In addition, when GLUT4 is heterologously expressed in fibroblasts, it is efficiently retained in intracellular compartments. This suggests that GLUT4 may harbor intrinsic targeting signals for efficient sequestration within the cell. Understanding the mechanism for targeting and retaining GLUT4 within intracellular compartments may provide important insights regarding the molecular mechanism by which insulin mobilizes GLUT4 vesicles. Several groups have therefore examined potential targeting motifs within the GLUT4 protein, both by mutating candidate motifs and by transplanting them onto heterologous reporter constructs (55). Attention has largely focused on the N and C termini of GLUT4, which contain sequence elements that are absent from the GLUT1 protein. However, results from several laboratories have been controversial and at times conflicting. For example, an early study found that neither the N nor the C termini contributed to the overall trafficking dynamics of GLUT4 in 3T3L1 adipocytes (56). Instead, two domains, one between residues 80 and 101 and the other between residues 316 and 343, were found to contribute to the intracellular sequestration of GLUT4. In contrast, a significant body of research has generally supported a role for either the N or the C termini in the intracellular trafficking of GLUT4 (Fig. 1). In general, the interpretation of these data is complicated by the fact that various researchers have used different cell lines and different heterologous GLUT4 chimeric constructs, making direct comparisons among investigators challenging.

1. The amino-terminal targeting motif.
Studies carried out in the early 1990s suggested that a cytosolic N-terminal FQQI motif was important for the intracellular sequestration of GLUT4. Mutations in this sequence, particularly the aromatic phenylalanine residue, resulted in the accumulation of GLUT4 at the plasma membrane (57). Furthermore, when the FQQI motif was transplanted onto GLUT1 or the H1 subunit of the asialoglycoprotein receptor, the resulting heterologous reporter constructs were excluded from the plasma membrane and were instead retained in intracellular compartments (57). Initially, these results were interpreted as evidence that the FQQI motif played a necessary and sufficient role in targeting GLUT4 to intracellular compartments. However, by closely examining the endocytotic behavior of GLUT4-chimeric constructs, it was subsequently discovered that the FQQI motif functions during the endocytosis of GLUT4 from the plasma membrane, rather that in its intracellular retention per se (58, 59). Because wild-type GLUT4 undergoes efficient endocytosis, in the absence of insulin the transporter is primarily localized to intracellular compartments. Because mutations in the FQQI motif inhibited internalization, over time GLUT4 accumulated at the plasma membrane under basal conditions. Conversely, transplanting the FQQI motif led to the efficient internalization of heterologous reporter constructs from the plasma membrane, resulting in an intracellular localization under steady-state conditions.

Although the FQQI motif is required for efficient endocytosis, some mutations in this motif do not completely block the uptake of reporter constructs. Thus, it has been possible to investigate whether the FQQI motif plays additional roles in intracellular GLUT4 vesicle trafficking. For example, it was recently shown that the FQQI motif functions in the sorting of internalized cell surface GLUT4 protein into specific intracellular compartments (60). Thus, mutation of FQQI to SQQI caused the mislocalization of the mutant transporters to late endosome and lysosome compartments. Moreover, it was independently shown that mutation of the FQQI to AQQI also caused GLUT4 to localize to endosomes (in this case recycling endosomes) (61). These results suggest that, in addition to playing an important role in endocytosis, the FQQI motif may also function during the intracellular trafficking of GLUT4 from endosome compartments into the insulin-responsive perinuclear storage compartment. However, these results must be interpreted with caution because the mutant transporters may be incorrectly folded and thus targeted for degradation in lysosomes. In addition, these data are based on the steady-state distribution of reporter constructs and may therefore reflect net changes in the trafficking rate constants between membrane compartments of the normal trafficking pathway, rather than mislocalization to compartments not normally occupied by the transporter.

2. Carboxyl-terminal targeting motifs.
In addition to the N-terminal FQQI motif, the cytoplasmic C-terminal 30-amino-acid region of GLUT4 was also initially found to be essential for the intracellular localization of the transporter (61, 62, 63). However, similar to the FQQI sequence described above, subsequent work demonstrated that a dileucine (LL) motif located within this region functioned during the endocytosis of GLUT4 from the plasma membrane (64, 65). Thus, mutating the LL motif to AA significantly inhibited internalization, resulting in the cell-surface accumulation of the mutant transporter. Moreover, studies employing chimeric proteins consisting of the extracellular domain of the transferrin receptor fused to the intracellular dileucine motifs of either GLUT4 or IRAP also support a role for this sequence during endocytosis (66). In addition to its role in endocytosis at the plasma membrane, the LL motif has also been reported to function during exit of GLUT4 from the TGN (60, 67). However, a recent study using primary adipocytes isolated from rat epididymal fat pads found that mutation of the LL motif resulted in no detectable alteration in the trafficking of a GLUT4 reporter construct (58). The apparent conflicting results regarding the LL motif may reflect important differences in the trafficking properties of GLUT4 reporter constructs expressed in different cell types.

Other studies utilizing GLUT1/GLUT4 chimeras stably expressed in 3T3L1 adipocytes found that the carboxyl-terminal 30 amino acids of GLUT4 were sufficient to correctly target chimeras to the insulin-responsive compartment (68). However, although the LL motif within this 30-amino-acid region was important for endocytosis from the plasma membrane, this motif was not critical for targeting chimeric constructs to the insulin-responsive compartment. These data are consistent with recent results implicating an acidic cluster motif (TELEYLGP) in GLUT4 trafficking. The TELEYLGP sequence is located downstream of the LL motif and appears to function in the targeting of GLUT4 from endosomes to a subdomain of the TGN enriched in the t-SNARE proteins syntaxins 6 and 16 (51, 70). Once recycled from the cell surface back to the TGN, GLUT4 may undergo additional sorting processes, perhaps involving syntaxins 6 and 16, before becoming insulin responsive (70). However, further work is needed to clarify the functional roles of syntaxin 6 and 16 in GLUT4 sorting processes.

	IV. Regulated GLUT4 Exocytosis

Top Abstract I. Introduction II. The Hexose Transporter... IV. Regulated GLUT4 Exocytosis V. Role of the... VI. The IR: Attenuation... VII. GLUT4 Vesicle Docking... VIII. GLUT4 Endocytosis IX. GLUT4 Activation X. Summary and Future... References

 
A. The IR: structure and function
The IR is a member of the family of transmembrane receptors with intrinsic tyrosine kinase activity (71). Encompassing more than 120 kb, the IR gene harbors 22 exons, 11 of which encode the -subunit, with the remaining 11 exons encoding the ß-subunit. Synthesized as a single polypeptide, the proreceptor is cleaved in the TGN by the serine protease furin. The mature cell surface IR is composed of two extracellular - and two transmembrane ß-subunits disulfide linked into an ₂ß₂-heterotetrameric structure (Fig. 3) (72, 73). Moreover, owing to alternative splicing of exon 11 of the IR transcript, the human IR exists in two isoforms, IR-A and IR-B, that differ by the inclusion (IR-B) or exclusion (IR-A) of 12 amino acids at the carboxyl terminus of the -subunit (74). IR-A is expressed predominantly in the developing fetus as well as a few adult tissues, including pancreatic ß-cells (Refs. 75 and 76 ; reviewed in Refs. 77 and 78). In contrast, IR-B is expressed predominantly in adult tissues and mediates the metabolic effects of insulin in muscle, adipose, and liver. During development, IR-A is controlled mainly by IGF-II. In adult tissues, insulin is the major ligand for IR-B; however, in ß-cells insulin appears to activate both IR-A and IR-B and may regulate different cellular processes through these receptor isoforms. Regardless of the receptor isoform, insulin binds to the extracellular -subunits and generates a conformational change that allosterically regulates the intracellular ß-subunit tyrosine kinase domain. Subsequently, the ß-subunits undergo a series of intermolecular trans-autophosphorylation reactions that generate multiple phosphotyrosine sites, some of which serve distinct functional roles (73). For example, tyrosine phosphorylation at the juxtamembrane Y960 residue is necessary for appropriate substrate recognition (e.g., insulin receptor substrates, IRS1–4), whereas tyrosine phosphorylation at residues Y1146, Y1150, and Y1151 in the kinase activation domain relieves pseudosubstrate inhibition, thereby further enhancing the receptor’s tyrosine kinase activity (73).

View larger version (28K): [in this window] [in a new window]   FIG. 3. Structure of the IR. The left side depicts IR-B, which includes the 12-amino acid alternatively spliced exon 11 (Ex11) at the carboxyl terminus of the -subunit. The right side depicts IR-A. The extracellular - and intracellular ß-subunits are indicated. The horizontal black bars represent disulfide linkages. CR, Cysteine-rich domain; JM, juxtamembrane domain; KD, kinase domain; CT, carboxyl-terminal domain. The positions of the tyrosine autophosphorylation sites are indicated. [Adapted from L. Jongsoon and P. F. Pilch: Am J Physiol 266:C319–C334, 1994 (361 ) and P. De Meyts and J. Whittaker: Nat Rev Drug Discov 1:769–783, 2002 (362 ).]

C. The IR: complex signaling networks
Many receptor tyrosine kinases directly recruit downstream effector molecules to their phosphorylated cytoplasmic domains. In contrast, the insulin receptor phosphorylates several scaffolding proteins that in turn recruit various downstream effector proteins (91). A number of scaffolding proteins downstream of the activated IR have been identified, including the four members of the insulin receptor substrate family (IRS1, -2, -3, and -4), Gab1, Shc, SIRPS, Cbl, and adaptor protein containing PH and SH2 domains (APS). The best studied of these scaffolding proteins is the IRS family (92), and the genetic ablation of IRS2 resulted in a diabetic phenotype (93). This resulted from a combination of peripheral insulin resistance and impaired ß-cell function, similar to what occurs in typical type 2 diabetes. IRS1 knockouts showed mild insulin intolerance (94, 95), whereas ablation of IRS3 resulted in no apparent abnormalities (96). However, IRS3 does not appear to be functional in humans (97). IRS4 knockout mice showed mild growth defects and were slightly glucose intolerant (98). The insulin-dependent tyrosine phosphorylation of IRS proteins generates docking sites for several downstream effectors, including the p85-regulatory subunit of the type 1A phosphatidylinositol 3-kinase (PI3K), the SH2-containing protein tyrosine phos-phatase (SHP2), the Src family member kinase Fyn, and the small adapter proteins Grb2 and Nck (91). Although the precise signaling function of these various adapters/effectors remains poorly understood, they appear to serve distinct yet overlapping biological roles in insulin signaling. For example, Grb2 primarily functions in Ras activation by insulin through engagement of the Ras guanylnucleotide exchange factor Son-of-Sevenless (SOS). This pathway leads to ERK activation (via Raf and MAPK kinase) and is an important cascade regulating several transcription events and eventually mitogenesis. In addition, Ras can also regulate the activity of the PI3K through interactions with its p110 catalytic subunit (99). The PI3K functions in many cellular responses, including the regulation of transcription, mitogenesis, antiapoptosis, protein synthesis, glycogen synthesis, and glucose transport (100, 101, 102). Moreover, PI3K functions in the regulation of the actin cytoskeleton, which is under the control of Rac and Rho family members (103). Defining the interrelationship between all these possible signaling events and the in vivo specificity required for a defined biological response represents a challenging problem for future investigators. Here our discussion will be limited to two independent signaling pathways that function downstream of the IR to regulate GLUT4 translocation (Fig. 4). A key player of the first, and more established, pathway is PI3K. The second pathway is characterized by the APS/CAP (Cbl-associated protein)/Cbl complex, which is compartmentalized at specialized microdomains at the plasma membrane.

View larger version (29K): [in this window] [in a new window]   FIG. 4. Schematic model for the in-terrelationship between the two IR-mediated signaling pathways. Insulin stimulation results in the tyrosine phos-phorylation of IRS proteins, which induces the association, activation, and/or targeting of the PI3K. Production of PI(3 4 5 )P3 results in the recruitment and activation of PDK1, which phosphorylates and activates both aPKCs and PKB. In parallel, insulin appears to phosphorylate Cbl through the adaptor protein APS. CAP also associates with Cbl and targets it to lipid raft microdomains through its interaction with the caveolar protein flotillin. Cbl then recruits the CrkII-C3G complex, resulting in the regulation of the lipid raft-associated Rho GTP-binding protein, TC10.

Insulin stimulation results in the tyrosine phosphorylation of the IRS proteins generating docking sites for the SH2 domains of the p85-regulatory subunit (78). This interaction apparently provides a dual function by both activating the PI3K and targeting it to the plasma membrane localized substrate PI(4, 5)P2. In turn, the local increase in PI(3, 4, 5)P3 in the membrane provides a lipid-based platform for the recruitment and anchorage of downstream signaling molecules that contain pleckstrin homology (PH) domains (100). PH domains are 100- to 120-amino acid modules with a characteristic ß-sandwich fold that binds, with varying degrees of specificity and affinity, to membrane phosphoinositides (104). Although most PH domains studied to date preferentially bind various phosphoinositides, the PH domain of IRS1 has been found to mediate interaction with a recently identified protein termed PHIP for PH domain-interacting protein (105). Interestingly, this interaction appears to be required for several insulin-dependent processes, including GLUT4 translocation.

In any case, it has been well established that the insulin-stimulated production of PI(3, 4, 5)P3 is essential for GLUT4 translocation. For example, inhibition of PI3K activity with relatively selective pharmacological inhibitors (wortmannin and LY29004) prevents insulin-dependent GLUT4 translocation (106, 107, 108). Similarly, introduction of PI3K-blocking antibodies or expression of dominant-interfering PI3K mutants also inhibits insulin-stimulated GLUT4 translocation (109, 110). On the other hand, expression of a constitutively active p110 subunit or addition of a PI(3, 4, 5)P3 analog to cells induces GLUT4 translocation, albeit not to the same extent as insulin (111, 112). Surprisingly however, p85-regulatory subunit knockout mice paradoxically display enhanced insulin sensitivity and increased glucose transport in adipose and skeletal muscle, leading to a hypoglycemic phenotype (113, 114). This appears to result, at least in part, from the preferential use of the p50-regulatory subunit splice variant in mice lacking full-length p85. It has also been suggested that the p85 subunit may function as a competitive inhibitor of PI3K signaling in the monomeric state (115). Reduced levels of p85 may thus improve the p85/p110/IRS stoichiometry and enhance insulin signaling through this pathway.

Additional evidence favoring PI3K function in insulin action comes from investigations on the mechanisms for inactivating the PI(3, 4, 5)P3 second messenger. Attenuation of PI3K signaling occurs through the activity of type-II SH2-domain-containing inositol 5-phosphatase (SHIP2), a phosphatase that removes the 5'-phosphate from PI(3, 4, 5)P3, generating phosphatidylinositol-3,4-bisphosphate [PI(3, 4)P2] (116). SHIP2 negatively regulates insulin signaling, and its genetic ablation resulted in enhanced insulin sensitivity (117). Although SHIP2-null animals died shortly after birth and were severely hypoglycemic, heterozygotes showed improved glucose tolerance and enhanced insulin sensitivity. This appeared to be due, at least in part, to enhanced insulin-dependent recruitment of GLUT4 to the plasma membrane (117). In contrast to SHIP, phosphatase and tensin homolog (PTEN) is a 3'-specific PI(3, 4, 5)P3 phosphatase that generates PI(4, 5)P2, and overexpression of PTEN prevents the accumulation of PI(3, 4, 5)P3 and also inhibited insulin-stimulated GLUT4 translocation in 3T3L1 adipocytes (118). Together, these data provide compelling evidence that the PI3K lipid product, PI(3, 4, 5)P3 is an essential intermediate in mediating insulin-stimulated GLUT4 translocation.

E. Protein kinase B (PKB) and atypical protein kinase C (aPKC)
Insulin rapidly activates the serine/threonine kinase PKB (also called Akt). This occurs through PI3K activity and the generation of PI(3, 4, 5)P3, which recruits PKB to the plasma membrane through specific interactions with its amino-terminal PH domain (119). At the plasma membrane, PKB is activated when two key residues are phosphorylated, threonine 308 (T308) and serine 473 (S473) (119, 120). T308 lies within the kinase activation loop and is phosphorylated by another PH domain-containing kinase, the 3'-phosphoinositide-dependent kinase 1 (PDK1) (121). The carboxyl-terminal PH domain of PDK1 binds PI(3, 4, 5)P3 with high affinity, which also recruits PDK1 to the plasma membrane (121). Although PDK1 appears to be maintained in an active, phosphorylated state even under basal conditions, binding of PI(3, 4, 5)P3 may also potentiate the catalytic activity of PDK1 (122, 123). S473 is located near the carboxyl terminus of PKB and may undergo autophosphorylation (124). Alternatively, there has been an extensive search for a putative PDK2, and several kinases including the integrin-linked kinase, MAPK-activated protein kinase 2, protein kinase C-related kinase 2, and others have been implicated in the phosphorylation of S473 (reviewed in Ref. 119).

Although the precise activation mechanism of PKB remains unclear, results from cell culture systems have generally supported a role for PKB function in insulin-stimulated GLUT4 translocation (125, 126, 127, 128, 129, 130), although others have reported evidence against such a role (131, 132). Nevertheless, stable expression of a constitutively active membrane-bound form of PKB in 3T3L1 adipocytes resulted in increased glucose transport and persistent localization of GLUT4 to the plasma membrane (72, 128, 133). In addition, coexpression of an epitope-tagged GLUT4 with dominant-interfering PKB mutants was also reported to inhibit insulin-stimulated GLUT4 translocation (134). Moreover, there are three PKB isoforms, , ß, and , and knockout of PKBß resulted in mild glucose intolerance, impaired glucose uptake in skeletal muscle, and the failure of insulin to suppress hepatic glucose production (135, 136, 137). In contrast, the ablation of PKB resulted in no observable defects in glucose homeostasis (138). These results are consistent with an important role for PKBß in insulin signaling and glucose uptake. However, although numerous substrates are known for PKB (119, 120, 139), how PKB activity influences GLUT4 translocation remains undetermined (139).

In addition to PKB, the aPKC isoforms, and /, have also been implicated in insulin-dependent GLUT4 translocation (see Ref. 140 for discussion of isoforms). Unlike the conventional and novel PKCs, aPKCs are not activated by diacylglycerol. However, although aPKCs lack PH domains, PI(3, 4, 5)P3 and other acidic phospholipids bind to the regulatory domains of these kinases and apparently induce conformational changes that lead to aPKC activation (141, 142). Activation of aPKC activity has also been reported to occur through autophosphorylation, phosphorylation by PDK1, and relief from pseudosubstrate inhibition (143, 144). As with PKB, results from cell culture experiments also support a role for aPKC- and -/ in insulin-regulated GLUT4 translocation (140). These include inhibition of aPKC activity with a pseudosubstrate peptide (145) and with general PKC inhibitors (145, 146). In addition, expression of kinase-dead or activation-resistant aPKCs inhibited GLUT4 translocation and glucose transport, whereas expression of constitutively active aPKCs promoted glucose transport (131, 147). However, knockout of the aPKC isoform resulted in mice without gross abnormalities but with some impairment of immune system function (148).

F. A second insulin-signaling pathway regulating GLUT4 translocation?
Although PI3K appears to be necessary for insulin-stimulated GLUT4 translocation, substantial evidence indicates that the generation of PI(3, 4, 5)P3 is not sufficient to recruit GLUT4 to the plasma membrane (Refs. 112, 149, 150, 151, 152 ; reviewed in Ref. 153). Furthermore, expression of an active PDK1 mutant was found to phosphorylate and activate both PKB and aPKC/ without any significant effect on glucose uptake (154). Together, these data imply that additional signaling pathways may function in parallel with the PI3K pathway. Recently, a candidate second signaling cascade that is functionally organized within caveolin-enriched lipid raft microdomains has been partially characterized (Fig. 4) (155, 156, 157, 158, 159). Caveolae are 50- to 100-nm invaginations of the plasma membrane and represent a subset of lipid raft domains (160). In addition to the caveolin and flotillin structural proteins, these liquid-ordered domains are enriched in palmitoylated signaling proteins, glycosylphosphatidylinositol (GPI)-anchored proteins, glycolipids, sphingolipids, and cholesterol (161). Lipid rafts are thought to spatially organize signaling cascades, and the tyrosine kinase activity of the IR is reportedly enhanced by virtue of its association with caveolae (162). In addition, it has recently been shown that the tyrosine phosphorylation of another IRS, the protooncogene c-Cbl, resulted in the insulin-dependent recruitment of Cbl into caveolin-enriched lipid raft compartments (155). This appears to involve two adapter proteins, APS and CAP (163, 164). CAP contains three SH3 domains that stably interact with proline-rich regions of Cbl. The CAP/Cbl heterodimer is recruited to activated IRs through interactions with APS (164). APS is also an IRS, and tyrosine-phosphorylated APS facilitates the phosphorylation of Cbl by the IR (164, 165, 166). Once Cbl is phosphorylated, the CAP/Cbl heterodimer appears to dissociate from the IR and accumulate in lipid raft domains through interactions between the Sorbin homology (SoHo) domain of CAP and the caveolae-resident protein flotillin. Dominant interfering mutants of CAP that prevent the localization of Cbl to lipid rafts specifically block insulin-stimulated GLUT4 translocation and glucose uptake, without affecting the PI3K signaling pathway (155).

Recent work has shown that Cbl recruitment to several activated tyrosine kinase receptors leads to receptor ubiquitination and down-regulation (167, 168, 169). In contrast, in the insulin signaling pathway, the phosphorylated and raft-associated Cbl recruits the SH2-containing adaptor protein CrkII to lipid microdomains. CrkII exists in a heterodimeric complex with C3G, a guanylnucleotide exchange factor (GEF) capable of activating the Rho-family GTPase TC10 (156). Recent results from cell culture experiments have implicated TC10 in both insulin-dependent and osmotic shock-induced GLUT4 translocation and glucose uptake (156, 157, 170, 171). Most Rho family members undergo geranylgeranylation and interact with GDP dissociation inhibitors. However, TC10 has a CAAX targeting domain very similar to H-Ras and is likely to be lipid modified with farnesyl and palmitoyl moieties. These posttranslational modifications target TC10 specifically to lipid raft domains (157, 170). Thus, the insulin-dependent recruitment of Cbl to lipid rafts results in the stepwise assemblage of several adapter proteins that function to target C3G to lipid microdomains. Because TC10 is a resident component of lipid rafts, this results in the insulin-dependent activation of TC10 by C3G. Indeed, the experimental mistargeting of TC10 into nonlipid-raft regions of the plasma membrane prevents its activation by insulin and abrogates the ability of TC10 to modulate insulin-stimulated GLUT4 translocation (157). Furthermore, disruption of lipid rafts with cholesterol-extracting drugs or by expression of inhibitory forms of caveolin effectively blocks insulin-dependent TC10 activation and GLUT4 translocation (157). Together, these results suggest that TC10 may be a key molecule in a signaling cascade that functions independently of the PI3K pathway. Nevertheless, a key question remains: How is insulin-dependent TC10 activation linked to GLUT4 translocation? Rho-family proteins are well known for their roles in controlling actin dynamics, and TC10 directly interacts with several effector molecules that regulate actin cytoskeletal function. The potential for TC10 to regulate the actin cytoskeleton during insulin-dependent GLUT4 translocation is discussed below in Section VI.

	V. Role of the Cytoskeleton in Insulin-Stimulated GLUT4 Translocation

Top Abstract I. Introduction II. The Hexose Transporter... IV. Regulated GLUT4 Exocytosis V. Role of the... VI. The IR: Attenuation... VII. GLUT4 Vesicle Docking... VIII. GLUT4 Endocytosis IX. GLUT4 Activation X. Summary and Future... References

 
It is generally recognized that the cell cytoskeleton plays important roles in various membrane trafficking events and in the retention of organelles at specific locations within cells (172, 173, 174, 175). In the case of GLUT4, the integrity of the GLUT4 storage compartment is disrupted by cytoskeleton-perturbing reagents such as nocodazole (for microtubules) and latrunculin (for actin filaments) (176, 177), and subsequently insulin-induced glucose uptake and GLUT4 translocation are severely impaired by these treatments (158, 177, 178, 179, 180, 181, 182, 183, 184). Nevertheless, recent studies have revealed the importance of the actin cytoskeleton on the acute metabolic actions of insulin (185, 186). Furthermore, the PI3K and TC10 insulin signaling pathways have profound influence on the actin cytoskeletal organization in various cell types (111, 187). Although a substantial amount of evidence strongly supports a critical role of the cytoskeleton in insulin-induced GLUT4 translocation, the precise functional roles and cytoskeleton-regulatory mechanisms remain enigmatic.

A. Actin
Monomeric globular actin (G actin) is highly conserved and is the most abundant cytoskeletal protein in mammalian cells. G actin polymerizes in a head to tail fashion to form filamentous actin (F actin) (188). Currently, more that 150 actin-binding proteins have been identified, which not only modulate the behavior of actin but also directly impinge upon the function of intracellular signaling cascades (189, 190). Stimuli such as attachment to the extracellular matrix and growth factors including insulin integrate these various actin-binding proteins to regulate the actin-based cytoskeletal system. Spatial and temporal regulation of dynamic actin rearrangements play important roles in a wide array of cell functions such as motility, chemotaxis, cytokinesis, phagocytosis, and intracellular vesicle trafficking, as well as for cell polarity and differentiation processes (191, 192, 193, 194).

In fibroblasts, F actin is well known to form stress fibers, lamellipodia, and filipodia. However, the actin cytoskeleton is dramatically changed during the differentiation of pre-adipocytes (fibroblast-like) into adipocytes (Fig. 5). Although preadipocytes contain well-defined stress fibers, after differentiation into adipocytes this F actin converts to a cortical actin lining the inner face of the cell surface membrane (158, 195). Concurrent with the changes in F actin during this differentiation process, the levels of caveolin mRNA and protein expression increase 20-fold, and the number of caveolae increases 10-fold (196, 197). This marked induction of caveolin, and hence caveolae, results in the clustering of the individual caveolae (50–80 nm) into large ring-like arrays (caveolae rosettes) that can be visualized by fluorescent microscopy (157, 195, 197, 198). A recent report demonstrates that these caveolae-rosette structures are a complex of both caveolae and other membrane components that form large caves (199). Intriguingly, fully differentiated 3T3L1 adipocytes display patches of punctate F actin that emanate from the organized caveolae-rosette/cave structure composing the relatively thick cortical actin (Fig. 5) (158, 195). In addition, the F actin localized along the inner circumference of the caveolae-rosettes/caves disappears after the disruption of caveolae by depleting cholesterol from the cells. This unique F actin structure has therefore been designated as caveolin-associated F actin (Cav-actin) (195).

View larger version (60K): [in this window] [in a new window]   FIG. 5. Adipocyte-differentiation is accompanied by reorganization of F actin. 3T3L1 Fibroblasts express relatively low levels of caveolin 1, which is dispersed along the plasma membrane (panel A). Phalloidin labeling of F actin demonstrates that these cells display typical stress fibers (panel B) that do not colocalize with caveolin 1 (panel C). In contrast, after adipocyte differentiation, caveolin 1 expression is induced and is concentrated in large ring-like rosette structures at the plasma membrane (panel D). F actin is also remodeled and concentrates at the sites of the caveolin 1 rosettes (panels E and F). [Adapted with permission from M. Kanzaki and J. E. Pessin: J. Biol. Chem.277:25867–25869, 2002 (195 ).]

The role of F actin in relation to insulin action has been examined using actin depolymerizing agents such as cytochalasin D, latrunculin A and B, and all these compounds inhibited insulin-induced glucose uptake and GLUT4 translocation (178, 179, 180, 181, 182). More recently, using enhanced yellow fluorescent protein-tagged G actin and real-time imaging techniques, insulin was shown to induce dynamic actin rearrangements at both the cortical and perinuclear regions in differentiated 3T3L1 adipocytes (158). Consistent with these findings, latrunculin, which directly binds to the monomeric G actin to prevent its polymerization (200), has a more potent inhibitory effect on GLUT4 translocation than cytochalasin D, which inhibits further polymerization of the existing F actin by capping the elongating ends (181). These results suggest that de novo actin polymerization may play an important role in insulin-dependent GLUT4 translocation. Furthermore, jasplakinolide, a cell-permeable F actin stabilizer, results in massive actin polymerization at both cortical and the perinuclear regions and inhibits insulin-induced dynamic actin rearrangements and GLUT4 translocation (158). The necessity of actin dynamics has been further supported by the recent observation that myosin 1C (Myo1c) function is required for insulin-stimulated GLUT4 translocation (201). In a series of experiments, it was shown that expression of wild-type Myo1c enhanced insulin-dependent GLUT4 translocation, whereas a dominant-interfering mutant form of Myo1c inhibited GLUT4 translocation. Moreover, small interfering RNAs were used to reduce the expression of endogenous Myo1c, and this maneuver inhibited insulin-stimulated glucose uptake.

In addition to the importance of nonconventional myosin families, multiple mechanisms have been proposed for the function of actin in the control of vesicle-trafficking events (174). For example, several studies have shown that a proper actin regulation is required for the TGN exit (202, 203), spatial targeting of secretory proteins, and the maintenance of the Golgi complex structure (204, 205, 206) including the GLUT4 storage compartment (177). In different cell systems, regulated secretion events apparently require distinct properties of actin with some fusion events promoted by F actin stabilization (207) and others by F actin disassembly (208) or dynamic actin rearrangements such as depolymerization/repolymerization (209, 210). Recently, one exciting mechanism has emerged from the study of the N-WASP isoform of the Wiskott-Aldrich syndrome protein (WASP) (211). N-WASP, a ubiquitously expressed member of the WASP family, contains several multifunctional domains including the WASP homology/Ena vasodilator-stimulated phosphoprotein homology domain, IQ motif, basic region, Cdc42/Rac interactive binding/GTPase binding domain, proline-rich region, and VCA (verpolin, cofilin-like, and acidic) region (212, 213). In concert with PI(4, 5)P2, N-WASP is regulated by direct binding of the Rho family small GTP binding proteins such as Cdc42 (211, 212, 214), TC10 (187, 215), and recently identified RhoT (216). The interaction of activated Rho proteins with N-WASP induces a conformational change in N-WASP that exposes its VCA region. In turn, this activates the Arp2/3 complex, resulting in a burst of de novo actin polymerization in response to extracellular stimuli (188, 213). This N-WASP-dependent actin polymerization, referred to as actin-comet tails, has been shown to provide a driving motor force to propel vesicles (159, 217, 218, 219), viruses, and infectious bacteria (220, 221, 222).

In vitro reconstitution with endogenous vesicles and exogenous artificial vesicles containing PI(4, 5)P2 or PI(3, 4, 5)P3 resulted in the nucleation of actin comet-tails and actin-based motility in a manner dependent upon Cdc42, N-WASP, Arp2/3, and tyrosine phosphorylation (214, 217, 219). Similarly, TC10 appeared to be capable of inducing the nucleation of actin-comet tails on the vesicles in an N-WASP-dependent manner (187). Intriguingly, using the same in vitro reconstituted system, GLUT4-containing vesicles were also found to induce actin-comet tails and actin-based motility stimulated by insulin stimulation (159). Furthermore, GTPS plus sodium vanadate treatment is known to induce insulin-like actions including glucose uptake and GLUT4 translocation in adipocytes and muscle cells (223, 224, 225). These agonists also induced actin-comet tails on the GLUT4 vesicle and actin-based motility in the in vitro reconstituted system. The GLUT4 vesicle mobility was inhibited by latrunculin B, toxin B, and a dominant-interfering N-WASP/VCA-lacking VCA region that is necessary for Arp2/3-dependent actin polymerization (159). Furthermore, expression of N-WASP/VCA in 3T3L1 adipocytes partially, but significantly, inhibited insulin-stimulated GLUT4 translocation (159, 226). Taken together, these data suggest that the N-WASP-dependent actin polymerization on the GLUT4-containing compartments somehow plays a role in the process of the translocation of GLUT4 protein from the intracellular storage compartments to the plasma membrane.

Although these findings support a required role for actin function in the regulation of insulin-stimulated GLUT4 vesicle trafficking, they also raise several questions, especially with regard to the insulin-signaling pathways involved. Actin is an exceedingly complex molecule in terms of its dynamics and interactions with other proteins; however, two key regulators are well established: as mentioned above, one is phosphoinositides and another is the Rho family small GTP-binding proteins (191, 227). The insulin-induced dynamic actin rearrangements necessary for GLUT4 translocation apparently result from a combination of the PI3K signal and spatially compartmentalized TC10 signal that both are activated by insulin stimulation.

It is well established that phosphoinositides play critical roles in modulating actin cytoskeleton and membrane trafficking (227, 228). The concentration and composition of cellular phosphoinositides are tightly regulated by both inositol lipid kinases including the type IA p85/P110 PI3K and phosphatases such as PTEN and SHIPs. In particular, the PI(4, 5)P2 and PI(3, 4, 5)P3 appeared to be important regulators of the actin cytoskeleton in response to multiple extracellular stimuli (229). In many cell types, activation of PI3K and subsequent production of PI(3, 4, 5)P3 from PI(4, 5)P2 stimulated by growth factors/hormones including insulin induce dynamic actin rearrangements (111, 230). The most recent link of PI3K to the actin cytoskeleton is a role for these enzymes in the regulation of chemotaxis, the migration of cells along a chemical gradient (231, 232, 233). There are several potential pathways for these lipid products to regulate the actin cytoskeleton. For example, Rac-GEFs such as Vav-1 and Tiam-1 interact with PI(3, 4, 5)P3 through their PH domains and are thereby activated (234, 235). Consistent with these results, treatment of cells with PI3K inhibitors blocks Rac activation by Vav-1 and Tiam-1 and PI3K-dependent actin rearrangements in fibroblasts (235, 236).

Another important molecular link between PI3K and actin includes members of the ARF (ADP ribosylation factor) family of small GTP-binding proteins. ARFs are well established as important regulators of intracellular membrane trafficking through controlling the assembly and disassembly of vesicle coat proteins (237, 238). Recent evidence has emerged that ARF1 and ARF6 are involved in reorganizing the actin cytoskeleton (239, 240, 241, 242). ARFs are regulated by ARF-GEFs and ARF-GAPs (GTPase-activating proteins) (237, 238). Two classes of ARF-GEFs can be distinguished on the basis of their sensitivity to brefeldin A. Although members of the brefeldin-sensitive type do not contain a PH domain and control Golgi apparatus traffic and integrity, the cytohesin family of ARF-GEFs that is insensitive to brefeldin contains a PH domain, a coiled-coil domain, and a Sec7 catalytic exchange factor domain (243). The PH domains of ARF nucleotide-binding site opener, cytohesin-1, and general receptor for phosphoinositides (also known as cytohesin-3) preferentially bind PI(3, 4, 5)P3 over PI(3, 4)P2 and other phosphoinositides (244, 245). On the other hand, the PH domain-containing ARF-GAPs have recently been grouped in the centaurin family (246). These proteins display a PI3K-dependent recruitment to the plasma membrane through their PH domains upon activation of PI3K (247). Although the precise roles played by ARF1 and ARF6 during actin reorganization remain unclear, some evidence indicates that ARF1 may potentiate Rho-stimulated stress fiber formation (242). In contrast, ARF6 may function in Rac-stimulated membrane ruffling (248).

In addition to these possible roles for PI(3, 4, 5)P3 function, PI(4, 5)P2 also plays an important role in regulating actin organization through modulating the activities of multiple actin-binding proteins. For example, PI(4, 5)P2 inhibits the activities of actin-severing proteins (e.g., ADF/cofilin), capping proteins (e.g., gelsolin), and profilin. In contrast, PI(4, 5)P2 activates actin cross-linking proteins (e.g., -actinin) and links actin to the plasma membrane through vinculin, talin, and ezrin/radixin/moesin proteins (228). Recent studies have revealed that WASP family proteins are activated by PI(4, 5)P2 in conjunction with Cdc42 and TC10 leading to Arp2/3 complex-induced actin nucleation (187, 249, 250). As mentioned above, phosphatidylinositol 4-phosphate 5-kinase expression induces actin-comet tails that are predominately nucleated from sphingolipid-cholesterol-rich rafts (217). Moreover, local increases in PI(4, 5)P2 production might facilitate vesicle budding by enhancing membrane curvature (251).

As indicated above, phosphoinositide-based actin polymerization appears to function in cooperation with small GTP-binding proteins of the Rho family, particularly, RhoA, Rac, and Cdc42. As previously discussed, TC10 is an unusual member of this family and has been implicated in the insulin regulation of GLUT4 translocation (156, 157). In vitro binding assays have indicated that active GTP-bound TC10 can bind several potential effectors that were originally identified as binding partners for Cdc42 and/or Rac. These include mixed-lineage kinase 2 (MLK2), myotonic dystrophy-related Cdc42 kinase (MRCK), p21-activated protein kinases (PAK), the Borg family of interacting proteins, the mammalian partition-defective homolog Par6, and N-WASP (215, 252, 253, 254, 255). To date, more than 20 Cdc42 target proteins have been identified in mammalian cells, and most of these are also potentially able to interact with TC10. However, despite their high degree of sequence similarity, TC10 and Cdc42 may play distinct functional roles in adipocytes and other cell types (156, 157, 187, 256).

For example, overexpression of Cdc42 wild type (Cdc42/WT) in fibroblasts caused a decrease in actin stress fibers along with the formation of plasma membrane microspikes, whereas overexpression of wild-type TC10 (TC10/WT) was without effect (255). However, expression of constitutively active TC10 (TC10/Q75L) is fully capable of inducing plasma membrane microspikes in fibroblasts. However, expression of constitutively active TC10 (TC10/Q75L) is fully capable of inducing plasma membrane microspikes. These data suggest that the signaling pathways required for TC10 activation are not present and/or functional in fibroblasts. This is in marked contrast to adipocytes where expression of Cdc42 has no significant effect on the actin cytoskeleton or insulin-induced GLUT4 translocation (156, 187). In contrast, a more recent report has indicated that expression of a constitutively active mutant version of Cdc42 (V12) can induce GLUT4 translocation (257). In any case, overexpression of TC10/WT, TC10/Q75L, or even the GDP-bound inactive mutant (TC10T31N) results in a total disruption of cortical and Cav-actin, leading to a marked inhibition of insulin-induced GLUT4 translocation. One interpretation of these data is that the ability of TC10 to regulate actin function is dependent upon specific cell types that express unique subsets of actin regulators. In addition, the existence and amount of lipid raft/caveolae are also crucial for actin regulation by TC10, because targeting to these specialized microdomains through C-terminal lipid modification is necessary for TC10 activation and its subsequent biological effects (157). Consistent with a role of TC10 in the control of Cav-actin, the Rho family-specific inhibitor, Clostridium difficile toxin B, blocks TC10 activation, disrupts Cav-actin, and prevents insulin-stimulated GLUT4 translocation without affecting PI3K signaling (158).

B. Microtubules
Microtubules are composed of repeating subunits of and ß-tubulin polymerized into long, hollow cylinders. Typically, microtubules have one end attached to a single microtubule-organizing center (MTOC) called a centrosome. Microtubules play critical roles in the long-range transport of intracellular organelles and vesicles in neurons (258). The balance of molecular motor activity directed toward the plus ends (cell periphery) vs. the minus ends (perinuclear region) of microtubules is an important mechanism implicated in the regulation of membrane trafficking (172).

Several studies have implicated the microtubule cytoskeletal network as important structural and regulatory elements in insulin-induced GLUT4 translocation (176, 177, 183, 184, 259). For example, -tubulin and the intermediate filament protein vimentin have been copurified with intracellular GLUT4-containing vesicles (176), and expression of a dominant-interfering vimentin peptide dispersed the perinuclear localized GLUT4 protein. Furthermore, inhibition of the microtubule motor proteins dynein and kinesin reduced insulin-stimulated GLUT4 translocation (176, 184). Most recent work has demonstrated that conventional kinesin KIF5B is highly expressed in adipocytes, and dominant negative mutants of conventional kinesin light chain blocked GLUT4 translocation in response to insulin (260). In addition, microtubule-depolymerizing agents (nocodazole, colchicine, and vinblastine) dispersed the perinuclear localized GLUT4 protein and partially inhibited insulin-stimulated glucose uptake and GLUT4 translocation (177, 183, 184, 259). Furthermore, the microtubule minus end motor dynein has been found to function with Rab5 in regulating the endocytosis of GLUT4 (28). These data support a model wherein the microtubule cytoskeleton participates in insulin-induced GLUT4 translocation.

In contrast, other studies have demonstrated that disruption or stabilization of microtubule structure