|Home | About | Journals | Submit | Contact Us | Français|
The Rab GTPase-activating protein TBC1D4/AS160 regulates GLUT4 trafficking in adipocytes. Nonphosphorylated AS160 binds to GLUT4 vesicles and inhibits GLUT4 translocation, and AS160 phosphorylation overcomes this inhibitory effect. In the present study we detected several new functional features of AS160. The second phosphotyrosine-binding domain in AS160 encodes a phospholipid-binding domain that facilitates plasma membrane (PM) targeting of AS160, and this function is conserved in other related RabGAP/Tre-2/Bub2/Cdc16 (TBC) proteins and an AS160 ortholog in Drosophila. This region also contains a nonoverlapping intracellular GLUT4-containing storage vesicle (GSV) cargo-binding site. The interaction of AS160 with GSVs and not with the PM confers the inhibitory effect of AS160 on insulin-dependent GLUT4 translocation. Constitutive targeting of AS160 to the PM increased the surface GLUT4 levels, and this was attributed to both enhanced AS160 phosphorylation and 14-3-3 binding and inhibition of AS160 GAP activity. We propose a model wherein AS160 acts as a regulatory switch in the docking and/or fusion of GSVs with the PM.
Rab GTPases play a fundamental role in vesicle transport reactions in eukaryotic cells. There are more than 60 Rabs in the human genome, suggesting that these proteins coordinate a complex cascade of events to ensure both fidelity and directionality to intracellular protein sorting (27, 34, 46). Rabs switch between active and inactive forms via GTP loading and GTP hydrolysis, respectively, and these reactions are catalyzed via guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs). Thus, the Rab-regulated step is often considered the major regulatory node in many transport steps. Rabs have been implicated in various stages of vesicle transport, including vesicle budding, vesicle transport along microtubules, and docking of transport vesicles with the target membrane (27, 34, 46). Consistent with a requirement for multiple Rabs in vesicle trafficking pathways, Rab cascades have recently been described in which GDP/GTP loading of sequential Rabs in a pathway is coordinately regulated in a countercurrent manner by cognate GAPs and GEFs (10, 24, 31, 46). This elegant mechanism proposes that for two Rabs that act in series in a pathway the GEF for the downstream Rab binds in an effector-type manner to the upstream Rab, while the GAP for the upstream Rab binds to the GTP-bound form of the downstream Rab. This coupling mechanism provides a potential mechanism for directional transfer in transport pathways. One of the provocative aspects of this model is that it suggests that the regulatory molecules involved play multiple shared roles in the process. For instance, the interaction between the GAP for the upstream Rab with the downstream Rab provides positive information to the pathway, which is somewhat counterintuitive to the negative role that these particular proteins are thought to play.
Certain vesicle transport pathways are subject to exquisite regulation, and so these pathways provide an ideal opportunity to interrogate key regulatory steps. The regulation of glucose transport in muscle and fat cells has become a paradigm for this mode of regulation. In this system, the facilitative glucose transporter GLUT4 is packaged into intracellular storage vesicles (GSVs) that remain disconnected from the plasma membrane and the recycling pathway in the absence of insulin. Insulin causes rapid delivery and fusion of these vesicles with the plasma membrane (PM). The canonical phosphatidylinositide 3-kinase/Akt signaling pathway (44) is a key determinant of this exocytic process.
The discovery of the RabGAP AS160/TBC1D4 as a major insulin-regulated Akt substrate in adipocytes and muscle cells was a major advance in understanding the regulation of GLUT4 translocation to the PM (7, 14, 19, 28, 29, 32, 35, 36, 45). AS160 possesses a RabGAP/Tre-2/Bub2/Cdc16 (TBC) domain at its C terminus flanked by a calmodulin-binding domain and two phosphotyrosine-binding (PTB) domains at the N terminus (14, 32, 45). Akt stimulates AS160 phosphorylation at Thr642, and this triggers 14-3-3 binding (8, 28). Overexpression of an AS160 mutant, in which the phosphorylation sites were mutated to alanine (AS160-4P), inhibited insulin-stimulated GLUT4 translocation (7, 28, 29, 32, 35, 36, 45). Mutating a crucial arginine in the RabGAP domain overcame this inhibitory effect (22, 32), providing a functional link between phosphorylation and AS160 GAP activity (32, 45). Knockdown of AS160 in adipocytes increases basal GLUT4 translocation (2, 7, 19), suggesting that AS160 plays a negative role in insulin-stimulated GLUT4 trafficking. That AS160 interacts with GSVs by binding to vesicle cargo such as the low-density lipoprotein receptor-related protein 1 (LRP1) and insulin-regulated aminopeptidase (IRAP) (12, 19, 26), together with the above findings, gave rise to a model wherein AS160 binds to GSVs in the basal state to maintain its substrate Rab GTPase(s) in a GDP-loaded inactive form, retaining GLUT4 inside the cell. Insulin-dependent phosphorylation of AS160 is thought to inhibit its GAP activity, allowing GTP loading and activation of a Rab that regulates docking and fusion of GSVs with the PM (8, 14, 19, 28, 29, 32, 36).
This model was based upon the canonical view of GAPs, which were thought to principally switch off their substrates by triggering GTP hydrolysis. However, several observations are not entirely consistent with this model: reduction of AS160 in adipocytes using shRNA results in increased GLUT4 at the PM in the basal state but blunted insulin-stimulated translocation of GLUT4 to the PM (2, 7); release of AS160 from GSVs is not necessary for insulin-dependent translocation to the PM, raising the possibility that AS160 remains associated with GSVs until after fusion (17, 35). There is a finite pool of AS160 at the PM that becomes highly phosphorylated with insulin (23), suggesting that AS160 is possibly phosphorylated at the PM. Thus, we hypothesized that, in addition to its negative regulatory role, AS160 might play an additional positive role consistent with the Rab cascade model proposed by Zerial and McBride (46).
We present evidence here that AS160 acts as a regulatory switch, playing both an inhibitory and a facilitative role in GLUT4 translocation to the PM. This switch-like mechanism, which is consistent with the Rab cascade model, is shown to be regulated by AS160 phosphorylation. Nonphosphorylated AS160 represses GTP loading of a Rab and GLUT4 translocation, while phosphorylation of AS160 inactivates its GAP activity (22, 32) and allows AS160 to fulfill an active role in facilitating GLUT4 vesicle fusion with the PM. These two separate functions are encoded by the N-terminal PTB domain in AS160, which mediates the interaction between AS160 and GSV cargo proteins, thus encoding the inhibitory function and a separate lipid-binding domain that facilitates interaction of AS160 with the PM.
Polyclonal rabbit antibodies raised against pThr308 Akt, pThr1462 TSC2, pSer21/9 glycogen synthase kinase 3α/β (GSK3α/β), and glutathione S-transferase (GST), monoclonal rabbit antibodies raised against total Akt (11E7) and GAPDH (glyceraldehyde-3-phosphate dehydrogenase; 14C10), and monoclonal mouse antibodies raised against pSer473 Akt (587F11) were purchased from Cell Signaling Technologies (Beverly, MA). Polyclonal rabbit antibody raised against 14-3-3β was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Polyclonal sheep antibodies raised against pThr642 AS160 were obtained from Peter Shepherd (Symansis, Auckland, New Zealand). Rabbit polyclonal antibodies against human AS160 and syntaxin-4 were previously described (40). The FLAG M2 and tubulin antibodies were from Sigma. Horseradish peroxidase (HRP)-conjugated secondary antibodies were from Amersham Biosciences (Buckinghamshire, United Kingdom), and IR dye 700- or 800-conjugated secondary antibodies were from Rockland Immunochemicals (Gilbertsville, PA). Paraformaldehyde was from ProSciTech (Thuringowa, Australia). Dulbecco modified Eagle medium (DMEM) and F-12 medium were from Invitrogen. Fetal calf serum (FCS) was obtained from Trace Scientific (Melbourne, Australia), and the antibiotics were from Invitrogen. Bovine serum albumin (BSA) was from Bovogen (Essendon, Australia). Bicinchoninic acid (BCA) reagent and SuperSignal West Pico chemiluminescent substrate were from Pierce (Rockford, IL). Protease inhibitor mixture tablets were from Roche Applied Science (Indianapolis, IN). The Akt inhibitor, MK-2206, was generously provided by Dario Alessi (University of Dundee, Dundee, United Kingdom). Other materials were obtained from Sigma Chemical Co. (St. Louis, MO).
Protein domains were predicted using SMART (33). The three-dimensional (3D) structure prediction is based on the Dab1 PTB domain in complex with IP3 and the ApoER2 peptide (PDB ID:1NU2). The Phyre2 server was used to identify homology model (16). Subsequent modeling and rendering was done with pyMOL.
3T3-L1 fibroblasts (ATCC, Manassas, VA) were cultured and differentiated to adipocytes as described previously (19). 3T3-L1 fibroblasts were infected with pWZLneo HA-GLUT4 retrovirus only or together with various pBabepuro-AS160 retrovirus. After 24-h recovery period, infected cells were selected with either 2 μg of puromycin/ml or together with 800 μg of Geneticin/ml in DMEM supplemented with 10% FCS for the selection of hemagglutinin (HA)-GLUT4-infected cells or HA-GLUT4/AS160-infected cells, respectively. Surviving 3T3-L1 fibroblasts were then grown to confluence and subsequently differentiated into adipocytes as described above. CHO IR/IRS-1 cells (American Type Culture Collection, Manassas, VA) were cultured in F-12 medium containing 10% fetal calf serum (FCS), 800 μg of G418/ml, 2 mM L-GlutaMAX, 100 U of penicillin/liter, and 100 μg of streptomycin/liter at 37°C in 10% CO2. HEK293E cells (American Type Culture Collection) were cultured in DMEM supplemented with 10% FCS, 2 mM L-GlutaMAX, 100 U of penicillin/liter, and 100 μg/liter streptomycin at 37°C in 10% CO2. CHO IR/IRS-1 cells, and HEK 293E cells were transiently transfected with DNA constructs using Lipofectamine LTX (Invitrogen) or Lipofectamine 2000 (Invitrogen), respectively, according to the manufacturer's instructions.
Cells were washed twice with ice-cold phosphate-buffered saline (PBS) and solubilized in 2% sodium dodecyl sulfate (SDS) in PBS containing phosphatase inhibitors (1 mM sodium pyrophosphate, 2 mM sodium vanadate, 10 mM sodium fluoride) and complete protease inhibitor mixture. Insoluble material was removed by centrifugation at 18,000 × g for 10 min. The protein concentration was measured using the BCA method. Proteins were separated by SDS-PAGE for immunoblot analysis. After the proteins were transferred to polyvinylidene difluoride membranes, the membranes were incubated in blocking buffer containing 5% skim milk in Tris-buffered saline (TBS) and immunoblotted with the relevant antibodies overnight at 4°C in blocking buffer containing 5% BSA–0.1% Tween in TBS. After incubation, the membranes were washed and incubated with HRP-labeled secondary antibodies and then detected by SuperSignal West Pico chemiluminescent substrate. In some cases, IR dye 700- or 800-conjugated secondary antibodies were used and then scanned at the 700- and 800-nm channels using an Odyssey IR imager. Quantification of the protein levels was performed using the Odyssey IR imaging system software or Wright Cell Imaging Facility ImageJ software.
After the indicated treatment, the cells were washed with ice-cold PBS and solubilized in NP-40 buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1% NP-40, 1 mM EDTA, 10% glycerol) containing Complete protease inhibitor mixture and phosphatase inhibitors (2 mM sodium orthovanadate, 1 mM sodium pyrophosphate, 10 mM sodium fluoride). Cell lysates were homogenized 10 times using a 27-gauge needle and centrifuged at 18,000 × g for 20 min at 4°C. One milligram of cell lysate was incubated overnight at 4°C with 2 μg of FLAG antibody. The antibodies were then captured with protein G-Sepharose beads for 2 h at 4°C. Immunoprecipitates were washed three times with ice-cold NP-40 buffer and kept in 2× SDS sample buffer at −20°C.
Plasma membranes were purified as described previously (3) with some modifications. Briefly, after treatments, the cells were washed twice with ice-cold PBS and twice in ice-cold coating buffer (20 mM morpholineethanesulfonic acid, 150 mM NaCl, 280 mM sorbitol [pH 5.0 to 5.5]). Cationic silica in a final concentration of 1% was added to the cells in coating buffer for 2 min on ice. The cells were then washed with ice-cold coating buffer to remove excess silica. Sodium polyacrylate (1 mg/ml, pH 6 to 6.5) was added to the cells in coating buffer, followed by incubation at 4°C for 2 min. The cells were washed once in ice-cold coating buffer and then washed with modified HES (20 mM HEPES, 250 mM sucrose, 1 mM dithiothreitol [DTT], 1 mM magnesium acetate, 100 mM potassium acetate, 0.5 mM zinc chloride [pH 7.4]) at 4°C and lysed as described above. Histodenz (100%; Sigma) in modified HES buffer was added to the lysate to a final concentration of 50%. The lysate was layered onto 0.5 ml of 70% Hisodenz in modified HES and centrifuged in a swing-out rotor at 25,000 × g for 20 min at 4°C. The supernatant was discarded, and the pellet was resuspended in 0.5 ml of modified HES buffer and centrifuged at 500 × g for 5 min at 4°C. The pellet was resuspended in SDS-PAGE sample buffer and heated to 65°C for 10 min.
3T3-L1 adipocytes stably expressing the indicated constructs were washed with ice-cold PBS and harvested in ice-cold HES buffer (20 mM HEPES [pH 7.4], 1 mM EDTA, 250 mM sucrose) containing Complete protease inhibitor mixture and phosphatase inhibitors (2 mM sodium orthovanadate, 1 mM sodium pyrophosphate, 10 mM sodium fluoride). The cells were lysed with 12 passes through a 22-gauge needle and 6 passes through a 27-gauge needle. The cell lysates were then centrifuged at 500 × g for 10 min at 4°C to remove unbroken cells. The supernatant was centrifuged at 10,080 × g for 20 min at 4°C to yield the following two fractions: the pellet fraction consisting of PM and mitochondria/nuclei and the supernatant fraction consisting of cytosol, low-density microsomes (LDM), and high-density microsomes. The supernatant was then centrifuged at 15,750 × g for 20 min at 4°C to obtain the pellet high-density microsome fraction. The supernatant was again centrifuged at 175,000 × g for 75 min at 4°C to obtain the cytosol fraction (supernatant) and the LDM fraction (pellet). To obtain the PM fraction, the pellet from the first ultracentrifuge spin was resuspended in HES buffer containing phosphatase and protease inhibitors, layered over high-sucrose HES buffer (20 mm HEPES [pH 7.4], 1 mm EDTA, 1.12 m sucrose), and centrifuged at 78,925 × g for 60 min at 4°C. The PM fraction was collected above the sucrose layer, and the pellet was the mitochondrial/nuclear fraction. All of the fractions were resuspended in HES buffer containing phosphatase and protease inhibitors. The protein concentration for each fraction was determined using a BCA assay. Samples were made up in an SDS sample buffer and then kept at −20°C.
3T3-L1 adipocytes were electroporated as described previously (35). Live-cell TIRF microscopy was performed using a Zeiss Axiocam MRm equipped with a heated stage set at 37°C. The cells were randomly selected by bright-field illumination prior to TIRF imaging, and images were analyzed with ImageJ software.
2-Deoxyglucose uptake into 3T3-L1 adipocytes was performed as described previously (43).
HA-GLUT4 translocation to the PM was measured as described previously (9). Briefly, 3T3-L1 adipocytes stably expressing various construct of AS160 and/or HA-GLUT4 in 96-well plates were serum-starved with Krebs-Ringer phosphate buffer (0.6 mM Na2HPO4, 0.4 mM NaH2PO4, 120 mM NaCl, 6 mM KCl, 1 mM CaCl2, 1.2 mM MgSO4, 12.5 mM HEPES [pH 7.4]) supplemented with 0.2% BSA for 2 h. Where applicable, cells were then treated with dimethyl sulfoxide (DMSO) or MK-2206 for 30 min prior to insulin stimulation for 20 min or as indicated. After stimulation, the cells were fixed and immunolabeled with monoclonal anti-HA antibody, followed by Alexa Fluor 488-labeled secondary antibody in the absence or presence of saponin to analyze the amount of HA-GLUT4 at the PM or the total HA-GLUT4 content, respectively.
Escherichia coli BL21(DE3) (Invitrogen) transformed with GST-AS160 fusion proteins or GST-IRAP (residues 1 to 58 of the cytosolic tail) were inoculated in 5 ml of Luria-Bertani (LB) medium and grown overnight at 37°C. Then, 500 ml of LB medium was added to the culture and grown to an optical density at 600 nm (OD600) of 0.6. After stimulation of protein expression by 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 4 h at 30°C, the bacteria were pelleted at 5,000 × g and resuspended at 4°C in 25 ml of PBS (pH 7.4) containing 1 mg of lysozyme/ml, 1% Triton X-100, 50 μg of DNase/ml, 50 μg of RNase/ml, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitors (Roche), ultrasonicated, and clarified by centrifugation at 15,000 × g for 30 min. GSH-Sepharose beads (Amersham Pharmacia) were incubated with GST-containing supernatant and washed extensively with PBS. Beads containing GST-IRAP were used for the IRAP pulldown assay. GST-AS160 fusion proteins were eluted from beads with 20 mM glutathione and buffer exchange with PBS using Amicon centrifuge tubes (10-kDa cutoff). The protein concentration was determined by a BCA assay. For Flag-AS160 protein purification, HEK293E cells were transfected with Flag-AS160 for 16 h. Cell lysate were harvested in NP-40 lysis buffer (50 mM Tris-HCl [pH 7.5], 1% glycerol, 1% NP-40). The lysates were clarified, and immunoprecipitations were performed for 2 h at 4°C. After extensive washes in NP-40 lysis buffer, followed by washes in TBS, Flag-AS160 was eluted with 25 μg of Flag peptide/ml at 4°C for 1 h. PIP Microstrips (Echelon) were processed according to the manufacturer's instructions with some modification. Briefly, the strips were blocked with 5% nonfat milk in TBS–0.1% Tween buffer for 1 h prior to overnight incubation with 1 mg/ml of purified protein in TBS-0.1% Tween containing 2% BSA. After extensive washing, the strips were incubated with anti-mouse GST antibody, washed, and probed with HRP-conjugated anti-mouse antibody. Secondary antibody was detected using enhanced chemiluminescence as described in Western blotting.
Data are expressed as means ± the standard deviations (SD) or ± the standard errors (SE), and P values were calculated by using a two-tailed Student t test and Microsoft Excel unless stated otherwise in the figure legends.
Most of the known functions of AS160 map to the GAP domain or the phosphorylation sites. Little is known about the function of the N-terminal PTB domains. We used the SMART database to detect and align the two AS160 PTB domains with the hidden Markov model (HMM) for PTB domains and the Phyre2 server for secondary structure prediction and identification of tertiary structural homologs (Fig. 1A). Although both domains displayed the classical PTB secondary structure aligning well with the SMART HMM (E-values < 10−30), the second domain was somewhat atypical. The closest tertiary structural homolog identified by the Phyre2 server was the Dab1 PTB domain. Interestingly, residues 190 to 239 at the N terminus of the second PTB domain in AS160 were not recognized by Phyre2 but were included in the SMART alignment, which also contains sequences corresponding to the PTB domain of Dab1, as well as both PTB domains of the Drosophila AS160 ortholog Pollux. Closer inspection of the SMART HMM alignment in the context of the Dab1 tertiary structure revealed an insertion of 107 amino acids in the α2-β2 loop, indicative of a “split” architecture with residues 190 to 239 corresponding to the α1 helix, β1 strand, and α2 helix and residues 347 to 450 corresponding to the remainder of the Dab1 PTB domain (Fig. 1A). Although the Dab1 PTB domain structure contains a PtdIns(4,5)P2 head group located at a site proximal to the α2-β2 loop (Fig. 1A), the basic residues contacting the bound head group are not conserved in AS160.
In view of our previous observation that a finite pool of AS160 is located at the PM in adipocytes (23), we postulated that this might be mediated via an interaction between the AS160 N terminus and phospholipids. To test this, we examined the binding of the AS160 N terminus to phospholipids in vitro. GST-AS160190-365, a region that consists of the α1-β1-α2 segment, as well as the insert and β2 strand (Fig. 1A and Table 1). This region interacts with most phosphorylated forms of phosphatidylinositol (PI) but not with PI itself (Fig. 1B), which is not unexpected since PTB domains bind to various phospholipids and can localize to various subcellular compartments (18). To determine whether the predicted insertion in the second PTB domain could mediate this interaction, we examined the lipid binding of GST-AS160239-365. AS160239-365 did not bind any phospholipids (data not shown), suggesting that the predicted insertion alone is incapable of binding to phospholipids. Positively charged lysines and arginines play a crucial role in the interaction of PTB domains with lipid (18, 20). The putative lipid-binding domain in AS160 contains several lysines and arginines (Fig. 1C) that likely form an essential part of this domain based on structure prediction (Fig. 1A). K209, K215, K216, K228, and R236 within the lipid binding module of AS160 are highly conserved across species and are also found in the AS160 homologue TBC1D1 and the Drosophila ortholog Pollux (Fig. 1B). Mutation of two lysines at 215 and 216 to alanine abrogated lipid binding (Fig. 1C). The phospholipid binding site in the N terminus of the second PTB domain is not located on the same surface as in Dab1 or Shc PTB domains and does not correspond to the canonical site in the structurally related Pleckstrin homology (PH) domain (data not shown). Furthermore, the conserved arginine and lysines in AS160 orthologs (Fig. 1C) are not conserved in PTB domain paralogs (data not shown). The lipid-binding domain of AS160 appears to map to residues 190 to 365 since full-length AS160 displayed similar lipid binding profile to the 190-365 fragment (Fig. 1B). Interestingly, full-length AS160 displayed reduced binding to PI(3,4,5)3P (Fig. 1B) compared to AS160190-365. This suggests that the conformation of the N-terminal lipid-binding domain may be modified in the full-length protein, possibly due to homodimerization (see below) or posttranslational modifications that may occur in the case of the full-length protein made in HEK cells but not for the truncation mutants produced in bacteria. Regardless, we do not see any evidence for insulin-dependent recruitment of AS160 to the membrane (data not shown), which would be indicative of PI(3,4,5)P3 binding. These data indicate that this novel phospholipid-binding domain in AS160 involves at least two conserved lysines (KK215/216) in the predicted β1-α2 loop of the second PTB domain. We provide a hypothetical model of this domain in AS160 where Ins(1,4,5)P3 is modeled in an arbitrary orientation consistent with the prediction that K215 and K216 likely interact with the 4-and/or 5-phosphates (Fig. 1A), though presumably in a nonstereoselective manner, as suggested by the lipid-binding data (Fig. 1B).
Phospholipids are found in multiple cellular locations (18), and so we next wanted to determine whether the lipid-binding domain in the N terminus of AS160 was synonymous with its targeting to the PM. A cationic silica method was used to isolate PM from HEK cells transfected with AS160 constructs. This method of PM isolation was validated by blotting for a cytosolic protein (GAPDH), which showed a very low PM/total cell lysate (TCL) ratio, whereas a bona fide PM protein (syntaxin-4) displayed a much higher ratio consistent with enrichment in the PM fraction. In agreement with our previous data (23), we detected a pool of AS160 at the PM in HEK293 cells, and this interaction mapped to the AS160 N terminus (residues 1 to 924) (Fig. 2B). Further truncation analysis revealed that PM binding was encoded within the N-terminal half of the second PTB domain (Fig. 2C). Consistent with the lipid-binding data (Fig. 1B), residues 190 to 239 of AS160 associated with the PM, whereas residues 239 to 365 did not (Fig. 2C and Table 1). Moreover, an AS160 mutant lacking residues 190 to 239 (AS160Δα1β1α2) displayed reduced PM association compared to full-length AS160 (Fig. 2C). Further evidence that PM binding was encoded via the lipid-binding domain was provided by the observation that mutation of Lys215 and 216 within AS160190-365 caused a significant reduction in PM binding (Table 1), whereas mutation of other conserved lysine and arginine residues had no effect (Table 1). These observations were confirmed using a separate subcellular fractionation approach in 3T3-L1 adipocytes (data not shown).
Intriguingly, the C terminus of AS160, while displaying reduced PM binding compared to full-length AS160, retained a PM/TCL ratio higher than the cytosolic protein GAPDH. This suggests that the C terminus may also possess PM binding activity independently of the lipid-binding domain. AS160 has been shown to homodimerize (5, 17), and we wondered whether the homodimerization domain could comprise the AS160 C terminus, in which case PM binding of this mutant might be due to homodimerization with endogenous AS160. To explore this, we coexpressed various Flag-AS160 truncation mutants with Myc-tagged full-length AS160. The N-terminal fragments (i.e., the 1-190, 1-440, and 1-865 fragments) interact poorly with myc-tagged full-length AS160 (Fig. 3), but fragments from this region retained their PM interaction (Fig. 2C). This supports the conclusion that PM targeting of the N terminus of AS160 is not mediated via homodimerization. The AS160 C terminus containing the GAP domain, on the other hand, displayed a strong interaction with full-length AS160 (Fig. 3). Hence, we surmise that the PM binding observed with the AS160 C terminus is likely due to homodimerization with endogenous AS160 in HEK cells.
To confirm that AS160 binds to the PM in adipocytes, AS160 mutants were tagged with enhanced green fluorescent protein (EGFP), expressed in 3T3-L1 adipocytes, and analyzed using live-cell total internal reflection fluorescence microscopy (TIRFM) imaging. This resolves fluorescently tagged proteins at or just beneath the PM. The fluorescence intensity in the TIRF zone was normalized to the total cellular fluorescence, measured via epifluorescence, to correct for differences in total expression. Consistent with the biochemical data and the lipid-binding data (Fig. 1B and and2C;2C; Table 1), EGFP-AS160190-365 was localized to the TIRF zone to a greater extent than EGFP-AS160239-365 or the EGFP controls (Fig. 4). Although EGFP-AS160190-239 was also enriched in the TIRF zone (Fig. 4), this was less than was observed for EGFP-AS160190-365, indicating that the predicted insertion (i.e., 239-365) contributes to the AS160 PM binding domain (Table 1). Mutation of KK215/216 to alanine (EGFP-AS160190-365KK215/216AA) led to a reduction in the fluorescence detected in the TIRF zone (Fig. 4 and Table 1), a finding consistent with the lipid-binding data (Fig. 1B) and the PM isolation approach (Table 1). The observation that full-length AS160 KK215/216AA displayed reduced localization to the PM provided further confirmation that lipid binding is a major mechanism for the interaction of AS160 with the PM (Table 2).
The putative PM targeting signal in AS160190-239 is highly conserved in TBC1D1 and Pollux (Fig. 1C). We reasoned that if these TBC proteins behaved in a similar manner to AS160, this would further support the conclusion that a lipid-binding domain in these proteins encodes PM targeting. Hence, we next expressed TBC1D1 and Pollux and two unrelated TBC proteins, RabGAP1 and TBC1D13, in HEK293 cells and determined their localization. TBC1D1 and Pollux displayed a similar enrichment at the PM to AS160, whereas this was not the case for RabGAP1, which also contains a PTB domain, or TBC1D13 (data not shown). These data indicate that the second PTB domain of AS160 regulates PM targeting and that this function is conserved in TBC1D1 and Pollux.
Since the GSV cargo protein IRAP also binds to the second PTB domain in AS160 (26, 28), we next sought to determine whether there is an overlap between the IRAP and PM binding domains. GST-IRAP interacted with AS1601-365 but not with AS160365-1299 (Fig. 5A) and only weakly to AS160Δα1β1α2 (Fig. 5B), suggesting that, analogous to the PM binding domain, the AS160 α1β1α2 domain is sufficient to mediate IRAP binding. Interestingly, AS160KK215/215AA, which bound weakly to the PM (Table 2), retained a significant interaction with IRAP (Fig. 5B and Table 2). This suggests that the critical lysine residues (KK215/216) required for PM binding are dispensable for the IRAP interaction. Intriguingly, the region within the second PTB domain that contains the lysines was highly conserved in Pollux, consistent with its strong interaction with the PM, whereas the remainder of the N terminus was less conserved (data not shown). Hence, we postulated that the IRAP binding domain might not be conserved in the Drosophila paralog. Consistent with this, Pollux displayed considerably less IRAP binding than AS160 (Fig. 5B). Moreover, a Pollux-AS160 chimera comprised of Pollux1-334 and AS160365-1299 retained PM but not IRAP binding (Fig. 5B and andCC and Table 2). These data indicate that distinct structural determinants within the second PTB domain define phospholipid/PM and GSV cargo interactions.
It was shown that AS160 binds to another GSV cargo protein LRP1 and the cytosolic tail of LRP1 shares sequence similarity with IRAP (12). Consistent with a shared mode of binding, we observed that LRP1 coimmunoprecipitated with full-length AS160 but not with AS160 lacking the α1β1α2 domain (Fig. 5D). These data indicate that AS160 targeting to GSVs is likely mediated via its interaction with multiple proteins in GSVs.
The inhibitory function of nonphosphorylated AS160 is thought to involve its interaction with GSVs via binding to cargo proteins such as IRAP and LRP1 (12, 26, 28). To confirm that the in vitro binding data recapitulate AS160 localization to GSVs, we examined AS160 binding to GSVs isolated by density sedimentation. Consistent with the IRAP pulldown data, a significant pool of wild-type AS160 and AS160KK215/215AA cofractionated with GLUT4 vesicles, whereas this was not the case for AS160365-1299, Pollux-AS160, and AS160-Δα1β1α2 (Fig. 5E). These data indicate that the second PTB domain in AS160 interacts with the PM and with GSVs via discrete binding modes.
Since AS160 possesses distinct binding sites for GSV cargo and phospholipids, we surmised that this likely demarcates separate functions. The best-described function of AS160 is its inhibitory role in GLUT4 translocation. This was established by overexpressing an AS160 mutant (AS160 4P) in adipocytes that was unable to be phosphorylated and so presumably locked into a GAP active state (32). We reasoned that if this negative role is mediated by binding to GSVs via the N-terminal cargo binding domain in AS160, then abolition of the cargo binding domain but not the PM binding domain in AS160 should overcome its inhibitory effect on GLUT4 trafficking. We therefore generated 4P mutants (S318A, S588A, T642A, and S751A) of a series of AS160 constructs and 3P mutants of those lacking Ser318. Consistent with this hypothesis, neither Pollux-AS160-3P nor AS160 365-1299-3P, both of which fail to bind to IRAP, exerted any inhibitory effect on insulin-stimulated GLUT4 translocation (Fig. 6A and Table 2). Strikingly, both AS160-Δα1β1α2-4P and AS160 KK215/216AA-4P, which retain IRAP but not PM binding, inhibited insulin-stimulated GLUT4 translocation (Fig. 6A and Table 2). Furthermore, there was a strong correlation (R2 = 0.87) between the degree of cargo binding and the inhibition of insulin-stimulated GLUT4 translocation across these mutants (Fig. 6B). These data indicate that the inhibitory function of AS160 on GLUT4 trafficking is encoded via its interaction with GSVs and not with the PM.
Since GSV binding seems to mediate the negative regulatory effect of AS160 on GLUT4 translocation, we next wondered whether the interaction of AS160 with the PM could facilitate a downstream step in the movement of GLUT4 vesicles to the PM. Consistent with a positive role for AS160 in this process, some studies have shown that knockdown of AS160 in adipocytes inhibits insulin-dependent GLUT4 translocation (2, 7). Hence, in addition to its inhibitory role under basal conditions, AS160 may have a positive role on insulin-regulated GLUT4 trafficking that may be attributed to its interaction with the PM. To explore this, we targeted AS160 to the PM by tagging it with the myristoylation/palmitoylation signal from Lyn, which is referred to as Lyn-AS160 from this point on. Lyn-AS160 was enriched at the PM in adipocytes (Fig. 7A). PM levels of HA-GLUT4 were increased by ~2-fold in nonstimulated adipocytes expressing Lyn-AS160, and insulin-dependent GLUT4 translocation was potentiated at submaximal insulin levels (Fig. 7B). Moreover, GLUT4 translocation to the PM was significantly higher over the whole time course in cells expressing Lyn-AS160 (Fig. 7C). Glucose transport in adipocytes expressing Lyn-AS160 was also increased (data not shown). This was not due to increased GLUT4 expression (data not shown) or to reduced GLUT4 endocytosis, since the latter would have led to a potentiation of cell surface GLUT4 levels at maximal insulin stimulation (Fig. 7B). One possibility is that Lyn-AS160 dimerizes with endogenous AS160, sequestering it at the PM away from GSVs and thus overriding its inhibitory effect. However, we did not observe a significant change in the subcellular distribution of endogenous AS160 in cells overexpressing Lyn-AS160 (Fig. 7A). This supports a positive regulatory role for AS160 at the PM in regulating GLUT4 translocation.
To confirm that targeting of AS160 to the PM alone was sufficient to mediate a physiologically relevant increase in glucose transport, we constructed an assay that coupled cellular glucose transport to sensitivity in response to cellular stress. The uptake of the nonmetabolizable glucose analogue 2-deoxyglucose into cells inhibits glycolysis, reduces ATP levels, activates the stress kinase AMP-activated kinase (AMPK), and ultimately triggers apoptosis (42). Wild-type cells or cells overexpressing Lyn-AS160 were incubated with 2-deoxyglucose, and AMPK phosphorylation was measured as a marker of cell stress. As shown (Fig. 7D), Lyn-AS160-overexpressing cells exhibited a significant increase in their stress response, as indicated by elevated levels of phosphorylated AMPK and the AMPK substrate ACC. This clearly indicates that targeting AS160 to the PM causes a physiological increase in cellular glucose uptake.
The increase in GLUT4 at the PM in Lyn-AS160 overexpressing cells may be attributed to enhanced insulin signaling. However, we did not observe any significant change in the phosphorylation of Akt, GSK3, or TSC2 (Fig. 8A). Intriguingly, however, the phosphorylation of Lyn-AS160, as well as endogenous AS160, at Thr642 was increased under basal conditions, whereas this was not observed in cells expressing Flag-AS160 (Fig. 8A). The Akt inhibitor MK-2206 (38) partially blocked phosphorylation of Lyn-AS160 and totally inhibited phosphorylation of endogenous AS160 (Fig. 8A). The partial inhibition of Lyn-AS160 phosphorylation under basal conditions may denote slow turnover of phospho-AS160 at the PM or the presence of a small pool of active Akt at the PM that is refractory to MK-2206. Alternatively, another kinase at the PM may phosphorylate AS160. These data indicate that constitutive targeting of AS160 to the PM is sufficient to facilitate its phosphorylation (Fig. 8A) and enhance GLUT4 translocation in the absence of insulin (Fig. 7B). In addition, Lyn-AS160 may interact with endogenous AS160 at the PM (5, 17) to facilitate phosphorylation of both proteins. Consistent with this model, endogenous AS160 was coimmunoprecipitated with Lyn-AS160 and with wild-type Flag-AS160 in 3T3-L1 adipocytes (Fig. 8B). This likely represents a low proportion of endogenous AS160 that was recruited to the PM (Fig. 7A and and8B8B).
One interpretation of these data is that AS160 is phosphorylated principally at the PM and not on GSVs (19, 32). To test this, we determined the phosphorylation status of the mutant AS160 365-1299 that does not localize to the PM (Fig. 8C). Since AS160 is phosphorylated rapidly (39), we stimulated cells with insulin (1 nM) at 28°C. The mutant was phosphorylated at a slower rate compared to endogenous AS160 (Fig. 8C). To determine whether these effects were specific to AS160 at the PM rather than some other membrane, we next targeted AS160 to GLUT4 vesicles by fusing it to GLUT4 (35). This construct exhibited very low levels of Thr642 phosphorylation in the absence of insulin compared to Lyn-AS160 (data not shown). These data suggest that AS160 is primarily phosphorylated at the PM.
To determine whether the positive effect of Lyn-AS160 could be due to increased phosphorylation of endogenous AS160, we made use of the knowledge that phosphorylation of endogenous AS160 is inhibited by the Akt inhibitor, whereas PM-targeted AS160 is less affected under both basal and insulin-stimulated conditions (Fig. 8A). In cells expressing Flag-AS160, both endogenous AS160 and Flag-AS160 were highly phosphorylated with insulin, and both were robustly inhibited by MK-2206 (Fig. 9A), as was insulin-stimulated GLUT4 translocation (Fig. 9B). In Lyn-AS160 expressing cells, MK-2206 inhibited phosphorylation of endogenous AS160 in the basal state with minimal effect on Lyn-AS160 phosphorylation (Fig. 9A). Inhibition of endogenous AS160 phosphorylation did not affect the stimulatory effect of Lyn-AS160 on GLUT4 translocation (Fig. 9B). Consistent with this, the stimulatory effect of Lyn-AS160 on GLUT4 translocation was retained in insulin-stimulated cells treated with MK-2206 (Fig. 9B). A similar trend was observed by measuring glucose uptake in wild-type 3T3-L1 adipocytes expressing Lyn-AS160, indicating that this effect was independent of HA-GLUT4 expression (data not shown). This indicates that the positive effect of Lyn-AS160 on GLUT4 translocation is not due to increased phosphorylation of endogenous AS160 and is likely associated with Lyn-AS160 phosphorylation at the PM.
We next sought to determine whether increased AS160 phosphorylation at the PM was required for its facilitative effect on GLUT4 translocation. To test this, GLUT4 trafficking was examined in cells expressing the AS160 mutants AS160-4P or Lyn-AS160-4P. Although neither of these mutants are phosphorylated in response to insulin, the latter mutant is enriched at the PM. AS160-4P inhibited GLUT4 translocation under basal and insulin-stimulated conditions (Fig. 10A), an observation consistent with previous findings (32). Whereas Lyn-AS160 increased basal HA-GLUT4 levels (Fig. 10A), this was not observed in cells expressing Lyn-AS160-4P. Lyn-AS160-4P decreased HA-GLUT4 translocation in insulin-stimulated cells, albeit to a lesser extent than AS160-4P (Fig. 10A). This suggests that phosphorylation of AS160 is required for the positive regulatory role of AS160 at the PM. To determine whether this phosphorylation-dependent positive effect is due to changes in AS160 GAP activity, similar studies were performed using AS160 mutants in which an arginine residue known to be crucial for GAP activity was mutated (22, 25) (Lyn-4P-R/A). Intriguingly, cells expressing Lyn-4P-R/A retained the stimulatory effect on HA-GLUT4 translocation under basal conditions and, if anything, this mutation potentiated the stimulatory effect on GLUT4 translocation observed in cells expressing Lyn-AS160 (Fig. 10A and Table 3). This suggests that inactivation of the GAP, which is normally considered to be encoded by AS160 phosphorylation, facilitates the positive function of AS160 at the PM. One interpretation of these data is that targeting of an inactive GAP per se is sufficient to achieve this positive effect. We tested this by targeting the GAP domain of AS160 (Lyn-GAP) or its inactive counterpart (Lyn-GAP R/A) to the PM. Neither of these mutants had a significant effect on GLUT4 translocation (data not shown). These data indicate that in addition to requiring an inactive GAP, the positive effect of AS160 requires other domains in AS160.
We have previously shown that insulin-dependent AS160 phosphorylation triggers binding of 14-3-3 (19, 28). We next wanted to test whether the positive effects of AS160 at the PM are due to phosphorylation per se or to the concomitant binding of 14-3-3 proteins to AS160. To test this, we introduced a constitutive 14-3-3 binding site into the AS160 4P mutant (4P-R18). We have previously shown that this mutation increases 14-3-3 binding in AS160 and that this mutation overcomes the inhibitory effects of the AS160-4P mutant on insulin-stimulated GLUT4 translocation (28). Strikingly, cells expressing Lyn AS160-4P-R18 displayed higher levels of cell surface GLUT4 in the absence of insulin than was observed in cells expressing Lyn-AS160 (Fig. 10B). This effect was reversed by mutating two lysines in the R18 14-3-3 binding site that disrupt 14-3-3 binding (Lyn-AS160 4P-R18KK) (Fig. 10B and Table 3). These data indicate that there is a functional link between AS160 phosphorylation, 14-3-3 binding, and inhibition of AS160 GAP activity at the PM that facilitates GLUT4 translocation.
We provide evidence that the RabGAP AS160 plays both negative and positive regulatory roles in vesicle transport. This supports the existence of Rab regulatory networks whereby individual components play active roles both in promoting and repressing flux through the pathway. This is consistent with the notion of Rab cascades where the function of multiple Rabs that act in series in a pathway can be coupled by a countercurrent mechanism encoded both by GEFs and GAPs (10, 24, 31, 46). The current studies extend this model by showing that phosphorylation and/or 14-3-3 binding switches AS160 from a negative to a positive regulator of vesicle fusion. In view of the vast number of different RabGAPs found in the human genome, these studies have broad implications for the role of this family of proteins in eukaryotic vesicle transport.
In addition to the TBC domain, RabGAPs possess a range of modular domains, the function of which in many cases has not been ascertained. A key observation in the present study was the identification of a lipid-binding domain encoded within the second PTB domain in the N terminus of AS160 that conferred its localization to the PM. This was a striking observation because previous studies had suggested that AS160 acted principally as a negative regulator of GLUT4 trafficking by binding to intracellular GLUT4 vesicles and inhibiting GTP loading of a Rab that was required to facilitate docking of the vesicles at the PM. The identification of a PM binding domain was also intriguing in light of previous observations identifying a pool of highly phosphorylated AS160 at the PM in adipocytes (23).
PTB domains bind to phosphopeptides and phospholipids, although the locations of the phospholipid binding sites differ drastically between PTB domains (6). For example, the PI(4,5)P2 binding site in Dab1 is not likely to be a canonical binding site for phospholipids, as indicated by the lack of conservation of the relevant basic residues in paralogs. Indeed, the phospholipid binding site in the Shc PTB domain is located on a completely different surface compared to Dab1 (30). Using the Dab1 PTB domain as a homology model, it is predicted that the phospholipid binding site in the second PTB domain of AS160 is located on a surface proximal to the peptide binding groove and not where the PI(4,5)P2 site is located in the Dab1 (Fig. 1A) or Shc PTB domains. Intriguingly, the lysine and arginine residues conserved in AS160 orthologs are not broadly conserved in other PTB domain paralogs. Further, the second PTB domain of AS160 carries an insertion of 107 amino acids not found in other PTB domains. Therefore, it is likely that the second PTB domain of AS160 is unique. Also noteworthy is that the second PTB domain in AS160 regulates binding to GSV cargo proteins, although this function is distinct from phospholipid binding (Fig. 6). Nevertheless, the fact that both of these functions are encoded within the same domain suggests that the inhibitory and facilitative roles that are encoded by these distinct interactions coevolved. By examining a series of AS160 mutants with various degrees of IRAP binding, we mapped the negative regulatory function of AS160 to the IRAP binding domain (Fig. 5 and and6).6). However, given that this domain also interacts with other proteins in GSVs, such as LRP1 (Fig. 5), this suggests that the interaction with GSVs is mediated via interactions with multiple cargo components. This is consistent with the observation that the intracellular distribution of AS160 is unaltered in adipocytes from IRAP−/− mice and 3T3-L1 adipocytes that have reduced IRAP expression (13, 15).
Thus, the interaction of AS160 with GSVs likely confers an inhibitory effect on GLUT4 translocation by inhibiting a Rab associated with GSVs. Evidence indicates that the inhibition of AS160 RabGAP activity is mediated by AS160 phosphorylation and 14-3-3 binding (28, 32), although this has not been formally proven. The present study extends this model. We propose that AS160 associated with GSVs might become phosphorylated at the PM when it encounters active Akt at this location. Consistent with this, it has been shown that Akt functions principally at the PM (1) and that insulin stimulates GSV trafficking to the PM in an Akt-independent manner (43). Moreover, AS160 is highly phosphorylated at the PM and not at other locations (23). Notably, constitutive targeting of AS160 to the PM enhanced its phosphorylation in the absence of insulin, and this was accompanied by increased GLUT4 translocation. This surprising result suggests that under basal conditions there must be a small amount of active Akt at the PM that under normal circumstances is insufficient to phosphorylate endogenous AS160. By targeting AS160 to the PM, we have likely shifted the equilibrium in favor of AS160 phosphorylation. These findings indicate that phosphorylation of AS160 at the PM positively regulates GLUT4 trafficking. AS160 phosphorylation at the Thr642 site encodes 14-3-3 binding (28), and here we show that constitutive binding of 14-3-3 to AS160 was sufficient to replicate its facilitative role in GLUT4 trafficking. Collectively, these findings suggest that phosphorylation and 14-3-3 binding not only suppress the GAP activity of AS160 but that this also confers an additional facilitative regulatory function. We have yet to resolve the nature of this role, but we speculate that phosphorylated/14-3-3-bound AS160 at the PM plays an active role in the docking of GSVs at the PM. It will be intriguing to determine whether other RabGAPs display a similar dual role. Hence, we conclude that AS160 possesses two separate and mutually exclusive functions that can be interchangeably regulated by Akt-dependent phosphorylation and 14-3-3 binding. An elegant feature of the use of its PTB domain for targeting AS160 to the PM is that by strategically localizing a very small pool of AS160 to the precise site of vesicle docking at the PM, this makes it relatively easy for Akt to disarm the inhibitory function of AS160 only at this location to promote the facilitative role of AS160 on the docking and fusion reaction (Fig. 11).
The fact that we detected a stimulatory effect of PM targeted AS160 on GLUT4 trafficking in the absence of any other perturbation was noteworthy. First, insulin-dependent GLUT4 translocation requires many steps other than phosphorylation of AS160, including trafficking of GLUT4 vesicles to the adipocyte cortex, actin rearrangement (21), recruitment of the exocyst complex to the PM (11), and posttranslational modification of SNARE proteins (4) or their regulators such as Munc18c (37, 40, 41). Second, in addition to disarming the GAP, one imagines that in the absence of an active effect on the Rab GEF this might have a relatively minor effect. Hence, the impact of constitutive targeting of AS160 to the PM, in the absence of any of these other changes, on GLUT4 trafficking probably points to the highly significant dual role of AS160 in this process.
In summary, we have mapped out the critical residues for the association of AS160 with phospholipids at the PM and unravel an additional role for AS160 in GLUT4 translocation. These data implicate AS160 as a major fork in the pathway that determines the probability that GSVs will either fuse with the PM or recycle back to the cell interior. This is a highly efficient way of coupling the activity of the Rab on the vesicle to the correct location in the cell, to which the vesicle is destined to fuse, and the nutrient status of the cell, which is encoded by the activity state of Akt also found at the same location.
We thank Morris Birnbaum (University of Pennsylvania) for providing the Pollux cDNA. The Akt inhibitor, MK-2206, was generously provided by Dario Alessi (University of Dundee). The LRP1 plasmid and antibody were kindly provided by Joachim Herz (University of Texas, Southwestern Medical Center). We thank Roger Daly and Antony Cooper (Garvan Institute of Medical Research) for providing invaluable feedback on the manuscript.
This study was supported by grants from the NHMRC of Australia and Diabetes Australia Research Trust (to D.E.J.) and the National Institutes of Health (DK060564 to D.G.L.). D.E.J. is an NHMRC Senior Principal Research Fellow.
Published ahead of print 8 October 2012