Deletion of Both Rab-GTPase–Activating Proteins TBC1D1 and TBC1D4 in Mice Eliminates Insulin- and AICAR-Stimulated Glucose Transport

In skeletal muscle and adipose cells, insulin stimulation leads to a rapid and reversible redistribution of GLUT4 from intracellular vesicles to the cell surface (1,2). The two related Rab-GTPase–activating proteins TBC1D1 and TBC1D4 (AS160) are phosphorylated in response to insulin, AMPK, and exercise/muscle contraction and have been implicated in important roles in regulating the translocation of GLUT4 (3–9). By positional cloning, we previously identified a naturally occurring loss-of-function mutation in Tbc1d1 as an obesity suppressor in the lean, diabetes-resistant SJL mouse strain (10). In humans, mutations in TBC1D4 (R363×) and TBC1D1 (R125W) have been linked to severe postprandial hyperinsulinemia and obesity, respectively (11–13). TBC1D4 is found mainly in the heart, adipose tissue, and oxidative muscle fibers, whereas TBC1D1 is predominantly expressed in glycolytic skeletal muscle and is nearly absent from fat tissue (10,14,15). Both TBC1D1 and TBC1D4 display a similar domain architecture that includes two N-terminal phosphotyrosine-binding domains and a Rab-GTPase-activating (GAP) domain. The latter is believed to control the activation state of Rab-GTPases by converting them from the active guanosine triphosphate–bound state into the inactive guanosine diphosphate–bound state. Several lines of evidence suggest that the GAP domains in TBC1D1 and TBC1D4 directly regulate the activity of an overlapping set of Rab-GTPases, thereby controlling the subcellular targeting and transport activity of GLUT4 (7). So far, Rab2, Rab8b, Rab10, and Rab14 have been identified as substrates for recombinant GAP domains of TBC1D1 and TBC1D4 in vitro (4,16), but the significance of these findings for GLUT4 translocation in adipose and muscle cells in vivo remains to be further investigated. Moreover, the phosphotyrosine-binding domains may also be involved in TBC1D1/TBC1D4 signaling (17–19). Previous studies of isolated L6 muscle cells and 3L3-L1 adipocytes demonstrated that ablation of either Tbc1d1 or Tbc1d4, or overexpression of mutated proteins, resulted in reduced insulin-stimulated translocation of GLUT4 (20–22). However, knockout mice lacking either Tbc1d1 (15,23) or Tbc1d4 (14,24) displayed only tissue-specific impairments of insulin-stimulated glucose uptake and rather minor alterations in glycemic control, indicating a substantial level of redundant expression and signaling.

The aim of this study was to determine the impact of the combined inactivation of Tbc1d1 and Tbc1d4 on glucose metabolism. Therefore, we characterized energy and substrate metabolism of Tbc1d1/Tbc1d4 double-deficient mice. Our results demonstrate that both TBC1D1 and TBC1D4 operate in concert and play essential roles in insulin-stimulated glucose uptake and substrate preference in skeletal muscle.

Radiochemicals ([9,10(n)-³H]-palmitic acid; 2-[1,2-³H(N)]-deoxy-d-glucose, [1-¹⁴C]-D-mannitol, and [¹⁴C]-D-glucose [U]) were purchased from Hartmann Analytic (Braunschweig, Germany). Human recombinant insulin Actrapid HM Penfill from Novo Nordisk Pharma GmbH (Mainz, Germany) was used throughout the experiments. Collagenase (type I) was from Worthington Biochemical Corp. (Lakewood, NJ). AICAR was purchased from Enzo Life Sciences (Lörrach, Germany). Antibodies against TBC1D1 and GLUT4 were described previously (10). Antibodies against AKT, phospho-AKT (Ser473), AMPKα, phospho-AMPK (Thr172), and phosphoenolpyruvate carboxykinase 2 (PCK2), glycogen synthase, and glycogen synthase kinase 3α were from Cell Signaling Technology (Danvers, MA). Antibodies against PCK1 were from Abcam (Cambridge, UK); antibodies against TBC1D4 were from Millipore (Temecula, CA); and GAPDH antibodies were from Ambion (Austin, TX). Antibodies against insulin-regulated aminopeptidase (IRAP) and GLUT1 were generous gifts from Dr. Susanna Keller (University of Virginia, Charlottesville, VA) and Dr. Annette Schürmann (German Institute of Human Nutrition, Potsdam, Germany), respectively.

Recombinant congenic Tbc1d1-deficient C57BL/6J mice (whole-body D1KO) were described previously (10). Mice with targeted deletion of Tbc1d4 (whole-body D4KO) were obtained from Texas A&M Institute for Genomic Medicine (Houston, TX) and backcrossed to the D1KO strain using a microsatellite marker-assisted (“speed congenics”) selection (25). Heterozygous D1KO/D4KO mice (>97.5% congenic with C57BL/6J) then were intercrossed to generate the four experimental genotypes: wild type (WT), D1KO, D4KO, and Tbc1d1/Tbc1d4 double-deficient D1/4KO. Animals were kept in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals, and all experiments were approved by the Ethics Committee of the State Ministry of Agriculture, Nutrition and Forestry (States of Brandenburg and North Rhine-Westphalia, Germany). Three to six mice per cage (Macrolon type III) were housed at a temperature of 22°C and a 12-h light–dark cycle (lights on at 6 a.m.) with ad libitum access to food and water. After weaning at the age of 19–21 days, animals received a standard chow with 19% (wt/wt) protein (23 cal%), 3.3% fat (8 cal%), and 54.1% carbohydrates (69 cal%) containing 3.06 kcal/g energy (V153 × R/M-H; Ssniff, Soest, Germany).

DNA was isolated from mouse tail tips using the InViSorb Genomic DNA Kit II (Invitek, Berlin, Germany). Mice were genotyped by PCR with three primers for the Tbc1d4 knockout (Fwd: 5′-AGTAGACTCAGAGTGGTCTTGG-3′; Rev-WT: 5′-GTCTTCCGACTCCATATTTGC-3′; Rev-KO: 5′-GCAGCGCATCGCCTTCTATC-3′) and primers for D1KO mice, as described elsewhere (10).

RNA was extracted and cDNA was synthesized as described previously (10). Real-time PCR was performed with a 7500 Fast Real-Time PCR System using TaqMan PCR probes (Applied Biosystems, Foster City, CA) for Tbc1d1, Tbc1d4, and Slc2a4, and data were normalized to Actb according to the ΔCt method (26).

Body weight was measured with an electronic scale (Sartorius, Göttingen, Germany). Body composition was analyzed with a nuclear magnetic resonance spectrometer (Echo MRI, Houston, TX).

Animals were placed in individual cages and respiratory quotient (RQ) was measured by indirect calorimetry after a 24-h adaption phase, as described previously (23). Rates of oxygen consumption (Vo₂) and carbon dioxide production (Vco₂) were monitored for 23 h at 22°C at a flow rate of 30 L/min. Animals had free access to water, and food was removed during the daytime (6 a.m. to 6 p.m.). Whole-body carbohydrate and fat oxidation rates (grams per minute) were calculated using the following equations: carbohydrate oxidation rate = 4.585 × Vco₂ (L/min) − 3.226 × Vo₂ (L/min); fat oxidation rate = 1.695 × Vo₂ (L/min) − 1.701 × Vco₂ (L/min) (27).

For glucose tolerance tests (GTTs), sterile glucose (2 g/kg body weight, 20% solution) was injected intraperitoneally into fasted (16 h) animals. For insulin tolerance tests (ITTs) and AICAR tolerance tests (ATTs), nonfasted mice were injected intraperitoneally either with insulin (1 IU/kg body weight) or AICAR (250 mg/kg body weight), and blood samples were taken from the tail tip at 0, 15, 30, 60 min (intraperitoneal ITT and ATT) and at 120 min (intraperitoneal GTT). Blood glucose was determined with a glucometer (Contour; Bayer, Leverkusen, Germany). Plasma insulin was measured with ELISA (Insulin Mouse Ultrasensitive ELISA; DRG, Marburg, Germany).

Frozen tissue (30 mg; liver and skeletal muscle) was pestled and analyzed using the Triglycerides (TRIGS) GPO-PAP Kit (Randox, Crumlin, UK) according to the manufacturer’s guidelines. Muscle homogenates were tested negatively for adipocyte contamination (Supplementary Fig. 1). Glycogen content was determined using the amyloglucosidase method (28). Tissue homogenates were incubated with 30% potassium hydroxide (wt/vol) at 100°C for 30 min. Subsequently, samples were supplemented with acetic acid and an assay buffer containing sodium acetate and amyloglucosidase and incubated 3 h at 37°C. Glucose content then was measured enzymatically by a glucose oxidase–based colorimetric detection kit (Glucose liquicolor; Human, Taunusstein, Germany) according to the manufacturer’s instructions.

Primary adipose cells were isolated by collagenase digestion of epididymal fat pads from 12- to 16-week-old male mice, and basal and insulin-stimulated uptake of [¹⁴C]glucose was performed as described elsewhere (29). Briefly, freshly isolated adipose cells (30) were incubated in Krebs-Ringer bicarbonate HEPES buffer (pH 7.4), 200 nmol/L adenosine containing 5% BSA with and without 120 nmol/L insulin for 30 min before measuring [¹⁴C]glucose uptake. After 30 min, the cells were spun through dinonyl phthalate oil (Merck, Darmstadt, Germany) to remove excess label, and the cell-associated radioactivity was determined by measuring scintillation. The resulting counts were normalized to the lipid weight of the samples (29). Measurements were performed in quadruplicate.

[³H]2-deoxyglucose uptake in intact isolated skeletal muscles was performed as described previously (10). Extensor digitorum longus (EDL) and soleus muscles were removed from anesthetized (500 mg/kg Avertin [2,2,2-tribromoethanol] via intraperitoneal injection) mice. Animals then were killed by heart puncture under anesthesia. Isolated muscles were incubated for 30 min at 30°C in vials containing preoxygenated (95% oxygen/5% carbon dioxide) Krebs-Henseleit buffer (KHB) containing 5 mmol/L HEPES and supplemented with 5 mmol/L glucose and 15 mmol/L mannitol. All incubation steps were conducted under continuous gassing (95% oxygen/5% carbon dioxide) at 30°C and slight agitation. After recovery, muscles were transferred to new vials and incubated for 30 min in KHB/5 mmol/L HEPES/15 mmol/L mannitol/5 mmol/L glucose under basal conditions or in the presence of 120 nmol/L insulin or 2 mmol/L AICAR, throughout the duration of the experiment. Then, muscles were incubated for 10 min in KHB/20 mmol/L mannitol under basal conditions or in the presence of 120 nmol/L insulin or 2 mmol/L AICAR before being transferred to radioactive glucose transport incubation. After 20 min of incubation in the presence of 1 mmol/L [³H]2-deoxyglucose and 19 mmol/L [¹⁴C]mannitol, muscles were immediately frozen in liquid nitrogen and stored at −80°C. Cleared protein lysates were used to determine incorporated radioactivity by counting scintillation. The counts from [¹⁴C]mannitol were used to correct for the extracellular space.

To assess palmitate oxidation, EDL and soleus muscles were incubated in pregassed KHB containing 15 mmol/L mannitol, 5 mmol/L glucose, 3.5% fatty acid–free BSA, [³H]palmitate, and 600 μmol/L unlabeled palmitate with or without 2 mmol/L AICAR at 30°C for 2 h. After absorption of fatty acids into activated charcoal, fatty acid oxidation (FAO) was determined by measuring scintillation of tritiated water.

Affinity photolabeling of GLUTs using the bio-LC-ATB-BGPA compound was performed as described previously (31). Following a 4-h fasting period, mice were anesthetized and EDL and soleus muscles were removed. Muscles were incubated in glass vials in preoxygenated KHB buffer containing 0.1% BSA, 2 mmol/L pyruvate, and 18 mmol/L mannitol for 30 min at 30°C. After recovery, 400 μmol/L bio-LC-ATB-BGPA was added and muscles were incubated with (120 nmol/L) or without insulin at 18°C for 8 min in the dark. After ultraviolet irradiation (6 min, 354 nm, 4°C), muscles were immediately frozen in liquid nitrogen. Recovery of photochemically biotinylated proteins from muscle lysates using streptavidin beads (Pierce, Rockford, IL) was performed as described (31), and total and labeled GLUT4 was determined by Western blot analysis.

Tissues were homogenized (20 mmol/L Tris, 150 mmol/L sodium chloride, 1 mmol/L EGTA, 1 mmol/L EDTA, 1% [v/v] Triton-X-100, 1 mmol/L sodium orthovanadate, 1 mmol/L β-glycerophosphate, 1 mmol/L sodium fluoride, 2.5 mmol/L sodium pyrophosphate tetrabasic, and a proteinase inhibitor and a phosphatase inhibitor cocktail) (Complete and PhosSTOP; Roche, Mannheim, Germany) and centrifuged for 10 min at 16.000 relative centrifugal force at 4°C. Protein content of the supernatant was measured with the BCA Protein Assay Kit (Pierce). Immunoblotting and detection was performed with a ECL Western blot detection analysis system (GE Healthcare, Buckinghamshire, UK), as described previously (23).

Data are reported as means ± SEMs. Significant differences were determined by one-way or two-way ANOVA (post hoc test, Bonferroni multiple comparison test) or paired two-tailed Student t test, as indicated in the figure legends. P values <0.05 were considered statistically significant.

We crossbred mice deficient in Tbc1d1 (whole-body D1KO) with conventional Tbc1d4 knockout animals (whole-body D4KO) to yield Tbc1d1/Tbc1d4 double-deficient (D1/4KO) mice on a C57BL/6J background. In all tissues analyzed, the abundance of TBC1D4 was not altered in Tbc1d1-deficient mice (Fig. 1A–C and Supplementary Fig. 2). Equally, TBC1D1 protein was not changed in mice deficient in Tbc1d4 (Fig. 1A–C and Supplementary Fig. 2). Male D1KO, D4KO, and D1/4KO mice and WT littermates were raised on a standard diet (8% calories from fat), and body weight and body composition were measured every 3 weeks until week 21. On a standard diet, D1KO, D4KO, and D1/4KO mice showed a lower body weight and a tendency toward reduced fat mass, whereas lean mass was not different compared with WT controls (Fig. 1D–F).

Plasma glucose and insulin concentrations in standard-diet fed male D1KO, D4KO, and D1/4KO mice and WT littermates were determined. D1KO and D4KO mice showed normal plasma glucose and insulin concentrations in both the postprandial and fasted states (Fig. 2A and B). In contrast, postprandial glucose concentrations were decreased in D1/4KO mice (WT vs. D1/4KO: 7.31 ± 0.18 vs. 6.34 ± 0.24 mmol/L; P < 0.01), whereas fasting glucose concentrations were normal. Moreover, fed and fasted plasma triglycerides as well as free fatty acids in plasma were not different between the genotypes (Supplementary Fig. 3). To investigate whole-body glycemic control of the animals, we performed intraperitoneal GTTs, ITTs, and ATTs. D1KO and D4KO mice showed normal blood glucose concentrations, as well as normal plasma insulin concentrations during the intraperitoneal GTT (Fig. 2C and D). In contrast, D1/4KO mice displayed markedly impaired glucose tolerance, whereas insulin concentrations were normal (Fig. 2C and D). Similarly, D1KO and D4KO mice showed normal blood glucose concentrations in response to the intraperitoneal ITT, whereas D1/4KO mice displayed substantially impaired insulin tolerance (Fig. 2E). Last, D4KO mice retained normal blood glucose concentrations in response to intraperitoneal ATT, whereas both D1KO and D1/4KO mice had significantly impaired AICAR tolerance (Fig. 2F).

We investigated whole-body substrate utilization in male D1KO, D4KO and D1/4KO mice and WT littermates fed a standard diet. At the age of 13 weeks, animals were placed in metabolic cages and indirect calorimetry was conducted for 24 h. On a standard diet, D1KO, D4KO, and D1/4KO mice showed a substantial reduction in RQ compared with WT controls during the dark phase (Fig. 3A and B). Whole-body fat oxidation rates were increased substantially, whereas whole-body carbohydrate oxidation rates were significantly reduced only in D1/4KO mice (Fig. 3C and D).

To determine the specific contribution of TBC1D1 and TBC1D4 to glucose uptake in skeletal muscle, EDL and soleus muscles from 16-week-old male mice were isolated and assayed for insulin- and AICAR-stimulated uptake of 2-deoxyglucose, as described in research design and methods. In glycolytic EDL muscle, basal 2-deoxyglucose uptake was not significantly different between the genotypes (Fig. 4A). In contrast, glucose uptake in response to insulin was substantially impaired in EDL muscles from D1KO and D1/4KO mice, whereas glucose uptake in EDL muscle from D4KO mice was normal (Fig. 4A). Similarly, AICAR-stimulated uptake of 2-deoxyglucose was markedly decreased in EDL muscles from D1KO and D1/4KO mice but not in EDL muscles from D4KO mice (Fig. 4B). In oxidative soleus muscle, basal glucose uptake was also not different between the genotypes (Fig. 4C). However, insulin-stimulated glucose uptake was markedly reduced in soleus muscles from D4KO and D1/4KO mice, whereas glucose uptake in the soleus muscles from D1KO mice was normal (Fig. 4C). Again, AICAR-stimulated glucose uptake decreased correspondingly (Fig. 4D).

We determined the impact of Tbc1d1/Tbc1d4 deficiency on glucose uptake in isolated white adipose cells. Adipose cells from 16-week-old male mice were isolated by collagenase digestion and assayed for insulin-stimulated uptake of [¹⁴C]-d-glucose, as described in research design and methods. In basal adipose cells, glucose uptake was not different between the genotypes (Fig. 4E). However, insulin-stimulated glucose uptake was markedly reduced in adipose cells from D4KO and D1/4KO mice, whereas glucose uptake in cells from D1KO mice was normal (Fig. 4E).

We next investigated whether insulin or AMPK signaling were impaired in skeletal muscle from D1KO, D4KO, and D1/4KO mice. Basal and insulin-stimulated (120 nmol/L; 60 min) EDL muscle was homogenized, and expression of AKT and phosphorylation of AKT^Ser473 was analyzed by Western blotting. In addition, basal and AICAR-stimulated (2 mmol/L; 60 min) soleus muscle was homogenized, and both expression of AMPK and phosphorylation of AMPK^Thr172 were analyzed by Western blotting. Quantitative assessment of replicate skeletal muscle samples revealed that neither expression nor phosphorylation of AKT or AMPK differed between genotypes (Fig. 4F). Moreover, we found no evidence of compensatory changes in the phosphorylation of TBC1D1 and TBC1D4 (TBC1D1^Ser237, TBC1D4^Thr588, phospho(Ser/Thr)-AKT substrate (PAS) phosphorylation) in insulin-stimulated skeletal muscle from knockout mice (Supplementary Fig. 4).

To determine the role of TBC1D1 and TBC1D4 in lipid utilization in skeletal muscle, EDL and soleus muscles from 16-week-old male mice were isolated and assayed for basal and AICAR-stimulated oxidation of palmitate, as described in research design and methods. In glycolytic EDL muscle, basal palmitate oxidation was increased substantially (∼twofold) in mice deficient in either Tbc1d1, Tbc1d4, or both (Fig. 5A). In intact isolated EDL muscle from control animals, AICAR stimulation led to a more than twofold increase in FAO (Fig. 5A). In contrast, in EDL muscle from D1KO, D4KO, and D1/4KO mice, the rates of FAO did not increase further in response to AICAR (Fig. 5A). In oxidative soleus muscle, basal palmitate oxidation was markedly increased (∼1.4-fold) in mice deficient in Tbc1d1, whereas no changes in FAO were observed in mice lacking either Tbc1d4 or in double-deficient mice (Fig. 5B). For AICAR-stimulated FAO in soleus muscle, no significant differences between the genotypes were observed (Fig. 5B). The copy numbers of mitochondrial DNA in gastrocnemius muscle were not different between genotypes (Supplementary Fig. 5A). However, citric acid synthase activity was moderately elevated in gastrocnemius muscle from D1KO mice (Supplementary Fig. 5B).

In skeletal muscle, the mRNA levels for GLUT4 were not different between the genotypes (Supplementary Fig. 6). To further examine the mechanism of the reduced insulin- and AICAR-stimulated glucose uptake in skeletal muscle and adipose cells, tissues from 16-week-old male mice were homogenized, and the abundance of GLUT4 protein was determined by Western blot analysis. D1KO mice displayed the strongest reduction (∼40–50%) in GLUT4 content in EDL and tibialis anterior muscle, whereas the abundance of GLUT4 was reduced by only ∼30% in quadriceps muscle (Fig. 6A, B, and D). In D4KO mice, GLUT4 was reduced to ∼75% in soleus muscle and to ∼60% in white adipose tissue (WAT) (Fig. 6C and F). Correspondingly, D1/4KO mice showed reduced GLUT4 content in all tissues analyzed. The abundance of the GLUT4-associated aminopeptidase IRAP (32) also was significantly reduced in EDL and soleus muscle from D1/4KO mice (Supplementary Fig. 7).

We investigated alterations in the subcellular distribution of GLUT4 using bio-LC-ATB-BGPA, a cell surface impermeant bis-glucose photolabeling reagent (31). Intact isolated EDL and soleus muscles from D1KO mice were incubated with bio-LC-ATB-BGPA in the absence or presence of insulin, as described in 2Research Design and Methods. After ultraviolet irradiation to cross-link the photoactivated diazirine group of bio-LC-ATB-BGPA to cell surface–localized GLUTs, photolabeled GLUT4 was recovered from cell membranes using streptavidin–agarose and quantified by immunoblotting with GLUT4 antibodies. Compared with the WT controls, the amount of photolabeled cell surface GLUT4 was significantly reduced in insulin-stimulated EDL muscles (Fig. 7A). In contrast, basal levels of photolabeled cell surface GLUT4 from Tbc1d1-deficient EDL muscles were not different from controls. In soleus muscle from D1KO mice, however, the levels of photolabeled GLUT4 were similar in the basal and insulin-stimulated states (Fig. 7B). When the amount of total GLUT4 is taken into account, the proportion of GLUT4 at the cell surface is increased in EDL but not in soleus muscle from D1KO mice (Fig. 7C and D). Similar changes occurred in D4KO mice (24).

We investigated the impact of Tbc1d1/Tbcd4 deficiency on the amount of triglycerides and glycogen in skeletal muscle and liver from mice fasted for 4 h. Hepatic glycogen concentrations were substantially decreased in D1KO, D4KO, and D1/4KO mice compared with WT controls (Fig. 8A). In contrast, fasting glycogen was increased in gastrocnemius (Fig. 8B) and quadriceps muscle (data not shown) of D1/4KO mice compared with WT controls. Abundance of glycogen synthase and phosphorylation (Ser⁶⁴¹) in liver and muscle were not significantly different between the genotypes (Supplementary Fig. 8). Also, no differences in the abundance of glycogen synthase kinase 3α and phosphorylation (Ser²¹) were observed in liver and muscle (Supplementary Fig. 8). Hepatic triglyceride concentrations were increased in D1KO, D4KO, and D1/4KO mice (Fig. 8C), whereas triglycerides in muscle were not changed (Fig. 8D). Expressions of both forms of the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (Pck1 and Pck2) were increased in liver from D1KO, D4KO, and D1/4KO mice (Fig. 8E and F).

In this study we investigated the impact of the combined inactivation of Tbc1d1 (D1KO) and Tbc1d4 (D4KO) on glucose and lipid metabolism in Tbc1d1/Tbc1d4 double-deficient (D1/4KO) mice. Our results demonstrate that both RabGAPs play critical roles in GLUT4 trafficking and glucose and lipid metabolism in muscle and adipose tissue.

On a standard diet, D1/4KO mice showed moderately reduced body weight and a trend toward reduced fat mass. Moreover, D1/4KO mice showed a reduced RQ, which resulted from a substantial increase in whole-body FAO and decreased use of carbohydrates as fuel. Both D1KO and D4KO mice displayed a similar reduction in body weight and RQ, suggesting that both Tbc1d1 and Tbc1d4 may contribute to energy substrate preference in the animals. In a previous study, animals expressing a signaling-deficient Tbc1d4-(T642A) mutant (33) also displayed reduced body weight, whereas Tbc1d4 knockout mice generated by Nestin-Cre/Lox-mediated deletion of the gene showed no significant change in body weight (14). In contrast, Tbc1d1 deficiency substantially reduced body weight and RQ in obese mice fed a high-fat diet, indicating a possible interaction of the gene with dietary fat and/or obesity (10).

Similar to the single knockout mice, D1/4KO mice had normal fasting glucose concentrations. In contrast to D1KO and D4KO animals, however, D1/4KO mice displayed markedly impaired tolerance to glucose, insulin, and AICAR, indicating an important role of both RabGAPs in glucose disposal in vivo. Thus, the impact of a single RabGAP knockout on glycemic control (i.e., D1KO and D4KO) seems to be largely compensated for by the residual RabGAP activity of the functioning gene, whereas inactivation of both RabGAPs leads to a pronounced impairment of glucose disposal. Interestingly, D1KO mice had reduced AICAR tolerance compared with D4KO animals, indicating an increased relevance of TBC1D1 in the AICAR response. This might reflect tissue-specific differences in the abundance of RabGAP and/or sensitivity of the RabGAPs to AICAR. The reduced postprandial glucose concentrations in D1/4KO mice could be related to a proposed function of the RabGAPs in insulin secretion (34,35). However, we did not observe genotype-specific differences in circulating insulin concentrations.

In mice, Tbc1d1 is predominantly expressed in glycolytic muscle fibers, whereas Tbc1d4 is expressed more strongly in oxidative muscle and in adipose cells (14,23). Consistent with the tissue-specific expression profile of both RabGAPs, insulin-stimulated glucose uptake in D1KO mice was substantially reduced in intact isolated EDL muscle but was normal in soleus muscle and isolated adipose cells. Conversely, D4KO mice showed severely impaired insulin-stimulated glucose uptake in soleus muscle and adipose cells, whereas glucose uptake was normal in EDL muscle. The combined ablation of both RabGAPs in D1/4KO mice essentially resulted in the elimination of insulin- and AICAR-stimulated glucose uptake in EDL and soleus muscle and in a substantial reduction of insulin-stimulated glucose uptake in adipose cells. However, in the double knockout mice, deletion of one of the RabGAPs (e.g., Tbc1d1 in EDL, Tbc1d4 in soleus and WAT) is sufficient to abolish insulin- and AICAR-stimulated glucose uptake without any further reduction in glucose transport in these tissues. In D1KO mice, GLUT4-associated glucose transport is clearly deficient in glycolytic muscle, yet GTT and ITT responses are normal. This suggests that glycolytic muscle types alone are not essential for maintaining whole animal glucose homoeostasis, possibly because of background maintenance of TBC1D4 activity and normal glucose transport activity in nonglycolytic muscle. Consistent with these findings, previous studies of GLUT4-deficient mice also demonstrated substantial compensation for reduced glucose transport in muscle and adipose tissue by altering hepatic fuel metabolism and increasing utilization of fatty acids (36). Thus, taken together, Tbc1d1 and Tbc1d4 play an essential role in insulin- and AICAR-stimulated glucose uptake in both skeletal muscle and adipose cells.

The reduced glucose uptake in response to insulin and AICAR was paralleled by a considerable reduction in GLUT4 abundance in muscle and adipose cells. Because the mRNA levels of GLUT4 were unaltered in the knockout mice, the decrease in GLUTs is likely the result of posttranslational events and may reflect missorting of the protein, presumably caused by the altered function of one or more Rabs downstream of TBC1D1/TBC1D4 (14,23,24,37). Consistent with a defect in protein sorting, the abundance of the insulin-regulated aminopeptidase IRAP, a resident protein of GLUT4 vesicles (38,39), binding partner of TBC1D4 (40), and potential regulator of GLUT4 sorting (41), also was significantly reduced in both EDL and soleus muscle from D1/4KO mice.

Previous studies showed that basal glucose transport is strongly correlated with the abundance of GLUT4 protein in adipose cells and skeletal muscle. Deletion of GLUT4 in these tissues led to a profound reduction in basal glucose transport, whereas overexpression of GLUT4 resulted in a substantial increase in basal uptake of glucose into the cells (42–45). In contrast, in skeletal muscle and adipose cells from D1KO, D4DO, and D1/4KO mice, basal glucose transport was normal despite substantially reduced abundance of cellular GLUT4 protein. Thus the dissociation of basal glucose transport and GLUT4 abundance in mice suggests that a lack of RabGAPs leads to a redistribution of GLUT4 from intracellular compartments to the cell surface in the basal state. We therefore investigated the subcellular distribution of GLUT4 by cell surface photolabeling of the protein in EDL and soleus muscle from D1KO mice and compared the surface concentration with the total GLUT4 concentration. Consistent with the glucose transport data, the amount of basal cell surface GLUT4 in the EDL muscles of D1KO mice was comparable with that of WT mice despite a lower abundance of total GLUT4. Thus, in unstimulated D1KO cells, a higher proportion of GLUT4 was present on the cell surface. In the insulin-stimulated state, both glucose uptake and cell surface GLUT4 were equally reduced compared with concentrations in WT mice, indicating that the fraction of GLUT4 present on the plasma membrane in insulin-stimulated cells is relatively normal. Thus, while it is not known at what stage in GLUT4 trafficking TBC1D1 and TBC1D4 participate at the molecular level, our findings are consistent with the RabGAPs being involved in intracellular retention of GLUT4 vesicles in the basal state (22,46).

Consistent with the reduction in RQ, D1KO, D4KO, and D1/4KO mice exhibited elevated basal FAO in EDL muscle. In the oxidative soleus muscle, FAO was elevated only in D1KO mice and was not increased further by AICAR stimulation. Unexpectedly, soleus muscle from D1/4KO mice did not show increased FAO despite the lack of TBC1D1, indicating that additional factors that modulate lipid oxidation in this type of muscle might be involved. In contrast, D1KO, D4KO, and D1/4KO mice had equally increased levels of FAO in glycolytic EDL muscle, which also was not further increased by AICAR. Thus, in glycolytic muscle, inactivation of each RabGAP equally contributes to the elevated FAO. Moreover, elevation of FAO is not directly related to impaired glucose uptake because Tbc1d4 knockouts have normal glucose uptake but increased FAO in EDL muscle. The mechanism responsible for the elevated FAO in glycolytic skeletal muscle of RabGAP-deficient mice remains to be clarified. We previously showed that knockdown Tbc1d1 in cultured skeletal muscle cells also increases fatty acid uptake and FAO, whereas overexpression of the gene has the opposite effect (10). However, overexpression of a RabGAP-deficient mutant, Tbc1d1-R941K, had no effect on fatty acid uptake and FAO, indicating the involvement of Rab-GTPases in this process (10). Interestingly, knockout of both Tbc1d1 and Tbc1d4 did not exert additive effects on FAO in EDL muscle, which suggests that both RabGAPs may regulate the same downstream target.

RabGAP-deficient D1/4KO mice phenocopy adipose- and muscle-specific GLUT4 knockout mice; these animals show substantially reduced glucose uptake in response to insulin and increased use of lipids as an energy source (36). In fact, the reduced abundance of GLUT4 explains, at least in part, the decreased glucose uptake in response to a stimulus. The inability to use glucose in muscle and adipocytes is associated with elevated hepatic triglycerides and reduced hepatic glycogen content, whereas fasting glycogen concentrations in skeletal muscle tend to be increased. Although hepatic glycogen concentrations in adipose- and muscle-specific GLUT4 knockout mice were not reported, conversion of glucose to fatty acids in the liver was substantially increased in these animals (36). Consistent with our data, muscle-specific GLUT4 knockout mice also displayed increased fasting glycogen content in skeletal muscle despite substantially reduced glucose uptake (47). A complex counterregulation of increased glycogen synthesis and decreased glycogen breakdown was proposed to explain this effect (47). We could not confirm a substantial increase in glycogen synthesis in skeletal muscle from fasted D1/4KO mice; however, the effect size might be larger in muscle completely lacking GLUT4. Thus, the mechanistic basis for the underlying tissue crosstalk remains to be further explored.

In summary, our results indicate that both RabGAPs play important roles in GLUT4 trafficking and glucose and lipid metabolism in muscle and adipose cells. So far, of the many Rab-GTPases previously associated with GLUT4-containing compartments (48,49), only Rab8, Rab10, and Rab14 were identified as substrates for both TBC1D1 and TBC1D4 in vitro; furthermore, they have been implicated in roles in trafficking of GLUT4 in vivo (50–53). Future studies investigating the specific contribution of the Rabs to the regulation of metabolic flexibility are required.

Acknowledgments. The authors thank Carolin Borchert, Diana Schulze, Ines Grüner, Angelika Horrighs, Anette Kurowski, Ilka Römer, Annette Schober, Antonia Osmers, and Jennifer Schwettmann for expert technical assistance, as well as Susanna Keller and Annette Schürmann for generously providing antibodies.

Funding. This work was supported by the Ministry of Innovation, Science and Research of the State of North Rhine-Westphalia (MIWF NRW) and the German Federal Ministry of Health (BMG) and was funded in part by grants from the Deutsche Forschungsgemeinschaft (GRK1208, AL452/4-1), the German Academic Exchange Service (DAAD) (to G.D.H.), the EFSD/Lilly European Diabetes Research Program, and the Swiss National Science Foundation (310000-122243/1, 310030-143929/1).

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

Author Contributions. A.C. and H.A.-H. conceived the experiments, analyzed data, and wrote the manuscript. A.C., A.I., C.d.W., C.S., Z.Z., T.S., and D.L.-C. performed the experiments and analyzed data. A.C., G.D.H., J.L., H.-G.J., and H.A.-H. contributed to the discussion and reviewed and edited the manuscript. H.A.-H. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.