The Inactivation of RabGAP Function of AS160 Promotes Lysosomal Degradation of GLUT4 and Causes Postprandial Hyperglycemia and Hyperinsulinemia

Type 2 diabetes mellitus has become prevalent in the last few decades, and a better understanding of its pathogenesis will help to combat this disease. Insulin resistance is the key feature of type 2 diabetes mellitus and causes dysregulation of glucose homeostasis in individuals with this disease (1). One of the functions of insulin is to regulate glucose uptake, the mechanism of which is not completely understood despite intensive studies.

In skeletal muscle and adipose tissues, insulin regulates glucose uptake through controlling translocation of GLUT4 from intracellular GLUT4 storage vesicles (GSVs) onto plasma membrane (2). After the removal of insulin, GLUT4 on the cell surface is internalized back into the cell through endocytosis and sorted into GSVs or recycling endosomes (3). Some of the recycled GLUT4 can also undergo degradation through lysosomes (4). AS160 (Akt substrate of 160 kDa) is a Rab GTPase activating protein (RabGAP), and the Arg⁹⁷³ residue on human AS160 functions as a key residue to maintain its GAP activity (5,6). In 3T3-L1 adipocytes and L6 myocytes, AS160 is required for intracellular retention of GLUT4 under basal conditions (5,7–9). Insulin stimulates the phosphorylation of AS160 via protein kinase B (PKB), and the phosphorylation of AS160 inactivates its GAP activity (10) and promotes GLUT4 translocation onto the cell surface (5,11–13).

It was thought that deletion of AS160 or inhibition of its GAP activity would promote GLUT4 translocation and thereby would improve insulin sensitivity. Surprisingly, deletion of AS160 decreases GLUT4 levels and causes severe insulin resistance in mice (14–17). More importantly, human patients bearing truncation mutations on AS160 (AS160^Arg363Ter or AS160^Arg684Ter, in which arginine is replaced by a stop codon) have postprandial hyperglycemia and hyperinsulinemia (18,19). GLUT4 is also decreased in skeletal muscle of patients harboring the AS160^Arg684Ter mutation (19). Despite the severe insulin resistance, the AS160 knockout (KO) male mice exhibit either normal (14,15,17) or impaired (16) glucose clearance, and have normal plasma insulin levels when intraperitoneally administered with glucose (14,15,17). It is currently unclear whether this difference is due to different administration routes of glucose in human and mice or is inherent between them.

The unexpected decrease of GLUT4 can be due to a tissue-autonomous effect of AS160 deficiency or requires a signal derived from tissues other than adipose and skeletal muscle. Besides the RabGAP domain at the C terminus, AS160 also has two phosphotyrosine binding domains (PTBs) at the N terminus and a calmodulin binding domain in the middle. The PTBs of AS160 can bind phospholipids (20) as well as the insulin-regulated aminopeptidase (IRAP) that associates with GSVs (21). An N-terminal AS160 fragment was detected in transformed lymphocytes from a human patient bearing the AS160^Arg363Ter mutation (18). Overexpression of this fragment inhibits GLUT4 translocation in 3T3-L1 adipocytes, most likely through interference with endogenous AS160 (18). However, it remains unknown which functional domains on AS160 contribute to the decreased GLUT4 expression.

In this study, we used mouse genetic models in combination with isolated tissues and primary cells from those models to address these critical questions.

Recombinant human insulin was bought from Novo Nordisk (Bagsværd, Denmark). MG-132 was from Merck Millipore (Darmstadt, Germany). Microcystin-LR was from Enzo Life Sciences (Farmingdale, NY). 2-Deoxy-d-[1,2-³H(N)]glucose and d-[1-¹⁴C]-mannitol were from PerkinElmer (Waltham, MA). All other chemicals were from Sigma-Aldrich or Sangon Biotech (Shanghai, China). The antibodies are listed in Supplementary Table 1.

The AS160 KO-first ES cells (Tbc1d4tm1a(EUCOMM)Hmgu, clone no. HEPD0601_5_F08) were purchased from the European Mouse Mutant Cell Repository (Neuherberg, Germany), and used for blastocyst injection to obtain the AS160^f/f mice (Supplementary Fig. 1A). The AS160^f/f mice were mated with Cre-recombinase–expressing lines to obtain the AS160^E6/7KO and corresponding tissue-specific KO mice. The AS160^R917K knockin mice were generated by the transgenic facility at Nanjing University following the targeting strategy outlined in Supplementary Fig. 4A. The AS160^f/f, AS160^E6/7KO, and AS160^R917K mice were backcrossed to C57BL/6J background for at least five generations before experiments. The AS160^E10KO mice are as previously described (14). The EIIa-Cre (22), Nes-Cre (23), Alb-Cre (24), adipoQ-Cre (25), and myf5-Cre (26) mice are on a C57BL/6J background and were obtained from The Jackson Laboratory.

The Ethics Committee at Nanjing University approved all animal protocols. Unless stated, mice had free access to food and water under a light/dark cycle of 12 h. As for the AS160^E10KO, AS160^E6/7KO, and AS160^R917K mice, het × het mating was used to generate KO or knockin homozygotes and wild-type (WT) littermates. As for tissue-specific KO mice, AS160^f/f × AS160^f/f-Cre mating was used to generate AS160^f/f (control mice) and AS160^f/f-Cre (tissue-specific AS160 KO mice). Genotyping was performed using the primers listed in Supplementary Table 2.

Mouse body composition was measured via DEXA using a Lunar PIXImus-II densitometer (GE Healthcare), as previously described (27).

Mice were fasted for 4 h (prior to insulin tolerance test [ITT]) or overnight (16 h; for glucose tolerance test [GTT]), and basal blood glucose was measured with a Breeze 2 glucometer (Bayer). Afterward, mice were injected with either with glucose (2 mg/g i.p.), for intraperitoneal GTT, or with insulin (0.75 mU/g i.p.) for ITT. As for intragastric GTT, mice were administered glucose (1.5 mg/g) via intragastric gavage. For fasting-refeeding experiments, mice were fasted overnight (16 h) before again being given free access to food. Blood glucose was measured at the indicated time points.

Mouse islets were isolated as previously described (28). Glucose-stimulated insulin secretion assay was performed in islets with comparable sizes for 1 h. Supernatants were collected for insulin measurement using an ELISA kit (EZRMI-13K; EMD Millipore, Billerica, MA), and islets were lysed and used for immunoblotting analysis.

Primary mouse preadipocytes were isolated and cultured as previously described (29). After reaching full confluence, preadipocytes were cultured in induction medium (DMEM supplemented with 850 nmol/L insulin, 0.5 μmol/L dexamethasone, 250 μmol/L isobutylmethylxanthine, and 1 μmol/L rosiglitazone) for 2 days, and further differentiated in DMEM containing 10% FBS and 160 nmol/L insulin for 2 days. After differentiation, adipocytes were cultured in DMEM containing 10% FBS with or without 160 nmol/L insulin for 2 days. For inhibitor treatments, adipocytes after differentiation were cultured in DMEM containing 10% FBS in the presence of bafilomycin-A1 (100 nmol/L), NH₄Cl (12.5 mmol/L), or MG132 (5 μmol/L) for 2 days.

Mouse soleus or extensor digitorum longus (EDL) muscles were isolated and used for glucose uptake, as previously described (11). Briefly, isolated muscles were preincubated for 50 min in Krebs-Ringer bicarbonate buffer (with or without insulin). Afterward, glucose uptake was carried out in Krebs-Ringer bicarbonate buffer (with or without insulin) containing 2-deoxy-d-[1,2-³H(N)]glucose and d-[1-¹⁴C]-mannitol for 10 min. Radioisotopes in muscle lysates were measured using a Tri-Carb 2800TR scintillation counter (PerkinElmer). Cells do not absorb mannitol, whose ¹⁴C-radioactivity was used to calculate extracellular volumes in muscles. Extracellular volumes were used to determine extracellular ³H-radioactivity that was subtracted from total ³H-radioactivity to get intracellular ³H-radioactivity in muscles. Intracellular ³H-radioactivity was normalized with muscle weight and used to calculate the glucose uptake rate.

For glucose uptake in in vitro differentiated primary adipocytes, cells were cultured in medium free of serum for 3 h and stimulated with or without 100 nmol/L insulin for 30 min. Glucose uptake was performed in HEPES-buffered saline buffer containing 2-deoxy-d-[1,2-³H(N)]glucose for 10 min. After lysis, radioisotopes in cell lysates were determined and normalized with cellular protein contents.

Subcellular fractionation was performed in mature primary adipocytes that were isolated from epididymal fat pads, as previously described (11).

Mouse tissues and primary adipocytes were lysed as previously described (14). Western blot analysis was performed as previously described (27).

GTPase activity of recombinant GST-Rab8a was measured in the presence or absence of immunoprecipitated AS160 WT/mutant proteins as previously described (27).

Total RNA isolated from tissues or primary adipocytes was reverse transcribed into cDNAs using a PrimeScript RT Reagent kit (Takara). mRNA levels of target genes were determined via quantitative PCR using an Applied Biosystems StepOnePlus Real-Time PCR system (Life Technologies) and the primers listed in Supplementary Table 3.

Hematoxylin-eosin and immunofluorescence staining were performed on pancreatic sections as previously described (30). Photographs were taken using an Olympus optical microscope (hematoxylin-eosin staining) or a Leica confocal microscope (immunofluorescence staining).

After differentiation, primary mouse adipocytes were infected with lentivirus-expressing HA-GLUT4-GFP. After infection, adipocytes were cultured in DMEM containing 160 nmol/L insulin for 2 days, and then in DMEM without insulin for 6 h. After fixation and permeabilization, cells were stained with LAMP1 antibody and Cy3-conjugated secondary antibody. Images were taken using a Δ Vision OMX system (GE Healthcare).

Data were analyzed using GraphPad Prism software (GraphPad Software, San Diego, CA). Two-group comparisons were performed via t test, and multiple-group comparisons were performed with two-way ANOVA. Differences were considered statistically significant at P < 0.05.

We first sought to find out whether different administration routes might cause different responses in GTT assays in our previously reported AS160^E10KO mice (in which the 10th exon of AS160 is deleted) (14). In agreement with previous reports (14,15,17), the AS160^E10KO mice displayed normal glucose tolerance when glucose was intraperitoneally administered (Fig. 1A). Similarly, they only exhibited moderate glucose intolerance and had normal plasma insulin levels when glucose was administered via intragastric gavage (Fig. 1B and C). Previous reports (14,15,17) show that blood glucose levels are normal in the AS160 KO mice after an overnight fast (16 h), but are significantly lower after partial fasting (4–6 h). In agreement with these reports, the basal blood glucose level after an overnight fast was normal in the AS160^E10KO mice in this study (Fig. 1A and B). We then used a fasting-refeeding protocol to investigate changes in postprandial blood glucose and insulin levels, in which mice were refed with normal chow diet after an overnight fast (16 h). This protocol is in a more physiological setting than GTT assays in mice. In agreement with their insulin resistance phenotype (14), the AS160^E10KO mice had a nearly twofold increase in their postprandial blood glucose levels at 60 min after refeeding, despite normal food intake compared with the WT littermates (Fig. 1D, E, and G). Blood glucose remained higher at 120 min after refeeding in these mice (Fig. 1D). Interestingly, refeeding caused an over eightfold increase in plasma insulin levels in the AS160^E10KO mice, in contrast with only a twofold increase in the WT littermates (Fig. 1F).

Free fatty acid, triglycerides, cholesterol, and amino acids in the plasma did not show an increase after refeeding in the AS160^E10KO mice (Supplementary Tables 4 and 5). The islets of the AS160^E10KO mice were normal, containing unaltered insulin contents (Fig. 1H–K). Moreover, glucose-stimulated PKB phosphorylation and insulin secretion in isolated islets were comparable between the two genotypes (Fig. 1L and M).

To further study the relationship between AS160 deficiency and the development of postprandial hyperglycemia and hyperinsulinemia, we generated an AS160^f/f mouse line by flanking the 6th and 7th exons of AS160 with loxP sites (Supplementary Fig. 1A). We mated this AS160^f/f mouse with an EIIa-Cre line (22) to generate another whole-body AS160 KO mouse. Neither full-length AS160 nor the putative N-terminal fragment could be detected in this KO line (Fig. 2A and Supplementary Fig. 1B and C). We named this line as the AS160^E6/7KO mice. Expression of TBC1D1, a RabGAP highly related to AS160 (31), was normal in the AS160^E6/7KO mice (Fig. 2A). The AS160^E6/7KO mice displayed glucose intolerance, and had higher postprandial blood glucose levels when refed with chow diet after an overnight fast (Fig. 2B and C). Furthermore, these mice exhibited postprandial hyperglycemia and hyperinsulinemia when refed with pure glucose after an overnight fast (Fig. 2D and E).

Together, these data demonstrate that AS160 deficiency also causes postprandial hyperglycemia in mice, which further leads to postprandial hyperinsulinemia.

We next deleted AS160 in tissues that are important for whole-body glucose homeostasis to identify the key tissues causing postprandial hyperglycemia. To this end, we mated the AS160^f/f mice with Nes-Cre (23), Alb-Cre (24), adipoQ-Cre (25), and myf5-Cre (26) lines to obtain neuron-specific, liver-specific, adipose-specific, and skeletal muscle–specific AS160 KO mice, respectively (Supplementary Fig. 2A). Tissue-specific AS160 deletion was confirmed in these lines (Supplementary Fig. 2B–D and Fig. 3E). The neuron-specific and liver-specific AS160 KOs had similar postprandial blood glucose levels as the AS160^f/f controls (Supplementary Fig. 2E). The adipose-specific AS160 KO mice (AS160-aKO) exhibited only a mild increase in postprandial blood glucose levels (Fig. 3A and B). Interestingly, the skeletal muscle–specific AS160 KO mice (AS160-smKO) were glucose intolerant, and displayed postprandial hyperglycemia and hyperinsulinemia (Fig. 3A–D). Together, these data show that the deletion of AS160 in skeletal muscle makes a significant contribution to the development of postprandial hyperglycemia and hyperinsulinemia.

We next determined GLUT4 protein levels in the AS160-aKO and AS160-smKO mice. GLUT4 protein levels were significantly decreased in a muscle type–dependent manner but remained normal in adipose tissues in the AS160-smKO mice (Fig. 3E and Supplementary Fig. 3A). In contrast, GLUT4 proteins were markedly diminished in adipose tissues but unaltered in skeletal muscle in the AS160-aKO mice (Fig. 3E and Supplementary Fig. 3A). In contrast to GLUT4, levels of three other membrane proteins, namely GLUT1, IRAP, and transferrin receptor (TfR) proteins, were normal in the AS160-aKO and AS160-smKO mice (Fig. 3F and G). Diminished GLUT4 protein levels were not due to changes in mRNA levels, which were normal in both AS160-aKO and AS160-smKO mice (Fig. 3H–J). Furthermore, glucose uptake rates were significantly lower in isolated soleus muscle or adipocytes where GLUT4 protein levels were decreased (Fig. 3K and L). Insulin-stimulated PKB phosphorylation remained normal in the AS160-aKO and AS160-smKO mice (Supplementary Fig. 3B and C). Together, these data suggest that AS160 deficiency lowers GLUT4 levels in a tissue-autonomous manner.

We next sought to find out whether the loss of the GAP function of AS160 caused insulin resistance and postprandial hyperglycemia and hyperinsulinemia. To this end, we generated a knockin mouse model in which Arg⁹¹⁷ (equivalent to Arg⁹⁷³ on human AS160) on AS160 was mutated to a lysine residue to inactivate the GAP function of AS160 (Supplementary Fig. 4A and B). The knockin mutation neither affected the expression of AS160^R917K and TBC1D1 nor altered the phosphorylation of PKB and GSK3 in the AS160^R917K mice (Fig. 4A and B). Furthermore, immunoprecipitated WT AS160 exhibited significant GAP activity toward Rab8a, a known substrate for AS160 (6), and increased the GTPase activity of Rab8a in an in vitro assay (Fig. 4C). Immunoprecipitated AS160^R917K protein lost ∼70% of its GAP activity toward Rab8a compared with WT protein (Fig. 4C). Previous reports (6) show that a recombinant human AS160^R973K GAP domain exhibits a complete loss of its GAP activity. AS160 can interact with TBC1D1 when they are coexpressed in cells (31). We suspect that the incomplete loss of GAP activity for immunoprecipitated AS160^R917K mutant protein might be due to other RabGAP proteins, such as TBC1D1 in the immunoprecipitates. Collectively, these data validate the suitability of the AS160^R917K knockin mice and their derived cells/tissues for studying the specific in vivo and in vitro functions of the GAP of AS160.

The AS160^R917K knockin mice grew normally with no apparent change in their body weight and composition (Fig. 4D–F). The mRNA levels of glut4 (also known as Slc2a4) were normal in adipose and skeletal muscle of the knockin mice (Fig. 5A). However, GLUT4 proteins were diminished in adipose tissues of the knockin mice and also decreased in their skeletal muscle again in a muscle type–specific manner (Fig. 5B and Supplementary Fig. 4C). Insulin-stimulated glucose uptake was strongly inhibited in the AS160^R917K soleus muscle, but not in the EDL muscle (Fig. 5C and D), and GLUT4 translocation was also impaired in the knockin adipocytes (Fig. 5E–G). Normal GLUT4 expression and glucose uptake in EDL muscle of the AS160^R917K knockin or AS160 KO mice (14–17) are most likely a result of the presence of TBC1D1, which regulates GLUT4 expression in white skeletal muscle (32–34). The AS160^R917K knockin mice displayed significant insulin resistance and exhibited intolerance to glucose administration (Fig. 5H and I). Importantly, these AS160^R917K knockin mice had postprandial hyperglycemia and hyperinsulinemia when refed with chow diet after an overnight fast despite normal food intake (Fig. 5J–L). Glucose-stimulated insulin secretion was normal in isolated knockin islets (Fig. 5M). Together, these data show that the AS160^R917K knockin mice phenocopy the AS160 KO mice, suggesting that the loss of GAP activity of AS160 underlies the unexpected phenotype of AS160 KO mice.

We next used in vitro differentiated primary adipocytes to study how AS160 deficiency lowers GLUT4 levels (Fig. 6A). GLUT4 protein levels were normal in AS160 KO or AS160^R917K knockin adipocytes after being differentiated in the presence of insulin (Fig. 6B and C). Interestingly, GLUT4 protein levels became significantly lower in AS160 KO or AS160^R917K knockin adipocytes than those in WT cells 2 days after the removal of insulin (Fig. 6B and C) despite comparable glut4/Slc2a4 mRNA levels (Fig. 6D). A time-course study showed that GLUT4 protein was significantly decreased by ∼30% in AS160 KO adipocytes 8 h after insulin deprivation, whereas it remained unchanged in WT cells within the same period (Fig. 6E and F). In contrast, glut4/Slc2a4 mRNA dropped by ∼50% in both WT and AS160 KO adipocytes 6 h after insulin deprivation (Fig. 6G). IRAP remained unaltered in both genotypes of cells within 12 h after the removal of insulin (Fig. 6E). The withdrawing of insulin lengthens the half-life of GLUT4 (35), which may help to maintain GLUT4 protein levels in WT adipocytes when glut4/Slc2a4 mRNA levels were dropped within 12 h after the removal of insulin. However, GLUT4 protein levels could not be maintained in AS160 KO adipocytes within the same period when glut4/Slc2a4 mRNA dropped to levels similar to those in WT cells, suggesting that GLUT4 degradation might be accelerated in AS160 KO adipocytes after insulin deprivation. Treatment with lysosome inhibitors bafilomycin-A1 or NH₄Cl, or with a proteasome inhibitor, MG132, increased GLUT4 protein levels in WT cells (Fig. 7A). Interestingly, treatment with bafilomycin-A1 or NH₄Cl, but not MG132, prevented GLUT4 degradation in AS160 KO adipocytes after the removal of insulin (Fig. 7B and C). The glut4/Slc2a4 mRNA levels were significantly decreased but still remained comparable between the two genotypes of cells upon bafilomycin-A1 treatment (Fig. 7D). The differentiation status of primary adipocytes was comparable between the different genotypes, as evidenced by adipogenic markers peroxisome proliferator–activated receptor-γ and perilipin (Figs. 6B and C and 7B). Moreover, the colocalization of GLUT4 with the lysosome became more evident in AS160 KO adipocytes than that in WT cells after insulin withdrawal (Fig. 7E). Together, these data show that AS160 deficiency accelerates lysosome-dependent GLUT4 degradation and that the inhibition of lysosome function can restore GLUT4 protein levels.

Our findings shed light on how AS160 regulates GLUT4 and glucose homeostasis, and are consistent with a model in which GAP deficiency of AS160 promotes lysosome-dependent GLUT4 degradation, and the loss of GAP activity of AS160, particularly in skeletal muscle, causes insulin resistance and postprandial hyperglycemia and hyperinsulinemia.

The fasting-refeeding protocol is more effective in revealing postprandial hyperglycemia and hyperinsulinemia in the AS160 mutant mice than intraperitoneal or intragastric GTT. Unlike the refeeding protocol, both GTTs in mice cause substantial stress and are different from oral GTTs in humans (36). Stress responses during GTT might suppress the hyperglycemic and hyperinsulinemic phenotype of the AS160 mutant mice. Although AS160 has been implicated in regulating insulin secretion in cultured Min6B1 cells and dispersed primary mouse β-cells (37), we did not observe the apparent changes of insulin secretion in islets from the AS160 KO or AS160^R917K knockin mice. Different experimental systems and/or conditions in the two studies might cause this discrepancy, although the exact reasons are unclear. Moreover, the AS160-smKO mice also displayed postprandial hyperglycemia and hyperinsulinemia. Therefore, postprandial hyperinsulinemia in these AS160 mutant mice may be secondary to postprandial hyperglycemia.

AS160 has two splicing isoforms, and the short isoform lacks exon 11 and 12, which are present in the long isoform. It is worthy of note that the human AS160^Arg684Ter mutation only affects the long isoform that is dominant in human skeletal muscle (19), whereas the AS160^E6/7KO, AS160^E10KO, and AS160^R917K knockin mutations affect both isoforms in mice. Nevertheless, GLUT4 levels were decreased in all of the AS160 mutant mice (14–17) and human AS160^Arg684Ter patients (19). One possibility for this unexpected decrease of GLUT4 in mutant mice and human patients is that the putative N-terminal fragments of AS160 might be the cause for this phenomenon. However, we could not detect the putative N-terminal fragments in our AS160^E10KO (14) and AS160^E6/7KO mice. Alternatively, the loss of PTBs or calmodulin binding domain on AS160 in mutant mice and human patients might decrease GLUT4 protein levels. However, our data show that the loss of GAP activity of AS160 most likely accounts for the unexpected decrease of GLUT4 through promoting its lysosomal degradation. The GAP-inactive AS160^R917K mice phenocopied the AS160 KO mice (this study and [14–17]), showing that the primary function of AS160 in vivo is a RabGAP and that the other domains of AS160 may be aid in the regulation of its GAP function.

A number of studies have shown that AS160 is required for intracellular retention of GLUT4 in cultured adipocytes (5,7,9) and myocytes (8). In AS160^R917K knockin adipocytes, surface GLUT4 remained normal despite the reduction of total GLUT4 in the basal state, which also supports this retention role of AS160. Therefore, it is possible that the lack of GAP activity of AS160 renders cells unable to retain GLUT4 within intracellular compartments, which may promote GLUT4 to undergo constant translocation and endocytosis. After endocytosis, GLUT4 proteins are further sorted, and some of them are degraded via the lysosome. Consequently, this may lead to the accelerated lysosomal degradation of GLUT4. In line with this possibility, there is such a precedent in which the TUG (Tether, containing a UBX domain, for GLUT4) protein retains GLUT4 within unstimulated adipocytes (38), and its depletion increases GLUT4 translocation and accelerates lysosomal degradation of GLUT4 (4). Alternatively, AS160 may additionally regulate endocytosis or lysosomal degradation of GLUT4 besides its role for intracellular retention of GLUT4. In line with this possibility, it has been proposed that AS160 may regulate both GLUT4 trafficking into and release from GSVs (39). We are currently unable to distinguish these two possibilities, and further experimentation is required to address this question in the future. It is well known that the other protein residing in GSVs, IRAP, traffics together with GLUT4 in response to insulin (40), and its expression is decreased in various GLUT4-null cells (41,42). Interestingly, in contrast to GLUT4, IRAP levels were normal in tissues or primary adipocytes from the AS160 KO mice, suggesting that the degradation of these two proteins might be differentially regulated.

Our finding on the lysosome-dependent GLUT4 degradation by the inactivation of RabGAP function of AS160 suggests that certain Rabs may be involved in this process. Four Rabs, Rab2, Rab8a, Rab10, and Rab14, have been shown as substrates of AS160 in an in vitro assay (6), among which Rab10 and Rab8a regulate GLUT4 exocytosis in 3T3-L1 adipocytes (43) and L6 myocytes (44), respectively. Interestingly, a recent report (45) shows that Rab14 may regulate the endocytic trafficking of GLUT4 in 3T3-L1 adipocytes. Furthermore, Rab14 regulates autophagosome-lysosome fusion through interacting with the PX domain-containing kinesin Klp98A in Drosophila (46). Therefore, these potential substrates of AS160 may have distinct roles in regulating GLUT4 exocytosis and endocytosis, which also raises the possibility that AS160 may regulate both exocytosis and lysosomal degradation of GLUT4 in vivo. Besides these four Rabs, AS160 may regulate other Rabs that might also be involved in GLUT4 exocytosis and/or endocytosis. Although detailed mechanisms for the lysosome-dependent GLUT4 degradation by the inactivation of RabGAP function of AS160 remain to be elucidated, our findings show that the inhibition of lysosomal degradation of GLUT4 can effectively restore GLUT4 expression in primary AS160 KO adipocytes. These findings have translational implications in that the inhibition of lysosomal degradation of GLUT4 might have therapeutic benefits for human patients bearing the AS160^Arg684Ter mutation, although it remains to be determined whether their glut4/Slc2a4 mRNA levels are normal.

Acknowledgments. The authors thank Yanqiu Ji and members of the resource unit at Nanjing University for technical assistance. The authors also thank Professor Geoffrey Holman (University of Bath, Bath, U.K.) for the GLUT1 and GLUT4 antibodies, and Professor Xiaowei Chen (Peking University, Beijing, China) for the lentivirus carrying HA-GLUT4-GFP.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Funding. This research was supported by the Ministry of Science and Technology of China (grants 2014CB964704 and 2014BAI02B01 to the National Key Scientific Research Program of China and grant 2014AA021104 to the National High Technology Research and Development Program of China), the National Natural Science Foundation of China (grants 31571211 and 31271498), and the Ministry of Education of China (grants 20120091120048 and NCET-13-0270).

Author Contributions. B.X., Q.C., L.C., and Y.S. performed the experiments, analyzed the data, and reviewed the manuscript. H.Y.W. and S.C. designed the experiments, analyzed the data, and wrote the manuscript. H.Y.W. and S.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.