Rab GAPs AS160 and Tbc1d1 play nonredundant roles in the regulation of glucose and energy homeostasis in mice

Abstract

The related Rab GTPase-activating proteins (Rab GAPs) AS160 and Tbc1d1 regulate the trafficking of the glucose transporter GLUT4 that controls glucose uptake in muscle and fat cells and glucose homeostasis. AS160- and Tbc1d1-deficient mice exhibit different adipocyte- and skeletal muscle-specific defects in glucose uptake, GLUT4 expression and trafficking, and glucose homeostasis. A recent study analyzed male mice with simultaneous deletion of AS160 and Tbc1d1 (AS160^−/−/Tbc1d1^−/− mice). Herein, we describe abnormalities in male and female AS160^−/−/Tbc1d1^−/− mice on another strain background. We confirm the earlier observation that GLUT4 expression and glucose uptake defects of single-knockout mice join in AS160^−/−/Tbc1d1^−/− mice to affect all skeletal muscle and adipose tissues. In large mixed fiber-type skeletal muscles, changes in relative basal GLUT4 plasma membrane association in AS160^−/− and Tbc1d1^−/− mice also combine in AS160^−/−/Tbc1d1^−/− mice. However, we found different glucose uptake abnormalities in isolated skeletal muscles and adipocytes than reported previously, resulting in different interpretations of how AS160 and Tbc1d1 regulate GLUT4 translocation to the cell surface. In support of a larger role for AS160 in glucose homeostasis, in contrast with the previous study, we find similarly impaired glucose and insulin tolerance in AS160^−/−/Tbc1d1^−/− and AS160^−/− mice. However, in vivo glucose uptake abnormalities in AS160^−/−/Tbc1d1^−/− skeletal muscles differ from those observed previously in AS160^−/− mice, indicating additional defects due to Tbc1d1 deletion. Similar to AS160- and Tbc1d1-deficient mice, AS160^−/−/Tbc1d1^−/− mice show sex-specific abnormalities in glucose and energy homeostasis. In conclusion, our study supports nonredundant functions for AS160 and Tbc1d1.

akt substrate of 160 kDa (AS160, also named Tbc1d4) and Tbc1d1 are closely related Rab GTPase-activating proteins (Rab GAPs) (13, 24, 26). They each have two NH₂-terminal PTB domains, a calmodulin-binding domain, and a COOH-terminal Rab GAP domain. Although over their entire length AS160 and Tbc1d1 are only 61% similar (24), their Rab GAP domains are 91% similar and display the same Rab substrate specificity in vitro (21, 24). Rab proteins regulate membrane trafficking (35), and Rab GAPs regulate the activity of their Rab substrates by catalyzing the hydrolysis of Rab-bound GTP to GDP (13). AS160 was discovered originally for its role in the regulation of the trafficking of the glucose transporter GLUT4 (28) before a similar role was also described for Tbc1d1 (24). GLUT4 is the predominant transporter mediating glucose uptake in muscle and fat cells and thus plays a key role in the maintenance of glucose homeostasis (reviewed in Ref. 18). GLUT4 is sequestered in intracellular vesicles under basal conditions and redistributes to the cell surface in response to different stimuli, primarily insulin and exercise. Studies in adipocyte and skeletal muscle cell lines using knockdown and overexpression of wild-type and mutant AS160 and Tbc1d1 established similar roles for the two Rab GAPs in GLUT4 trafficking (reviewed in Ref. 26). Under basal conditions, AS160 and Tbc1d1 are active toward their Rab substrates, thereby keeping the Rabs in their inactive GDP-bound form and sustaining GLUT4 retention. Upon phosphorylation by Akt, AMPK, and related kinases, in response to insulin, exercise, and other stimuli, AS160 and Tbc1d1 Rab GAP activities are inhibited, most likely by binding to the cytosolic protein 14-3-3. This allows the Rabs to become active (GTP-bound) and facilitate GLUT4 vesicle movement to the cell surface and/or docking and fusion of GLUT4 vesicles with the plasma membrane.

However, AS160 and Tbc1d1 also have clearly distinctive features. Although each has multiple phosphorylation sites, these are found in different locations and are differentially regulated (6, 34). In addition, AS160 and Tbc1d1 have distinct expression patterns among adipose and muscle tissues. Adipose tissues and heart express high levels of AS160 but little Tbc1d1 (5, 30, 33), whereas different skeletal muscles express variable levels of the two Rab GAPs (14, 17, 30, 33). AS160 is most abundant in soleus and is least expressed in tibialis anterior and extensor digitorum longus (EDL) muscles. In contrast, Tbc1d1 expression is highest in tibialis anterior and EDL and lowest in soleus muscles. In gastrocnemius and quadriceps muscles, AS160 and Tbc1d1 are both relatively well expressed. Among other tissues involved in nutrient and energy homeostasis, brain and pancreas express both Rab GAPs, whereas liver expresses little AS160 or Tbc1d1 (see Refs. 2, 4, 17, 25, and 33 as well as our unpublished data).

The physiological functions of AS160 and Tbc1d1 in glucose and energy homeostasis have been investigated using whole body AS160 knockout (17, 36), Tbc1d1 knockout (4, 9, 14, 30, 32), and AS160 knockin mice (7, 10). These studies, consistent with differential expression of AS160 and Tbc1d1, identified different adipocyte- and skeletal muscle-specific defects in GLUT4 subcellular distribution and expression and glucose uptake and different defects in whole body glucose homeostasis in AS160- and Tbc1d1-mutant mice. No compensatory increases in Tbc1d1 and AS160 expression and phosphorylation were observed in AS160 and Tbc1d1 knockout mice, respectively (3, 4, 9, 17, 30, 32, 36). To investigate possible redundant functions for AS160 and Tbc1d1, we generated and analyzed double-knockout mice for the two Rab GAPs (AS160^−/−/Tbc1d1^−/− mice). While our study was ongoing, another group published the characterization of male AS160^−/−/Tbc1d1^−/− mice (3). We found similar but also different abnormalities between our mice and those in the previously published AS160^−/−/Tbc1d1^−/− study, leading us to different conclusions on the relative roles of AS160 and Tbc1d1. In addition, we also analyzed female AS160^−/−/Tbc1d1^−/− mice in most of the in vivo studies and found clear differences with the male double-knockout mice.

MATERIALS AND METHODS

Generation of AS160^−/−/Tbc1d1^−/− mice.

The generation of mice with single deletions of AS160 (AS160^−/−) and Tbc1d1 (Tbc1d1^−/−) on the C57BL/6N strain background has been described (14, 17). To obtain mice with combined deletions of AS160 and Tbc1d1, AS160^−/− and Tbc1d1^−/− mice were crossed to obtain mice heterozygous for AS160 and Tbc1d1 deletions (AS160^+/−/Tbc1d1^+/−). These heterozygous mice were then bred and offspring genotyped as described before (14, 17). AS160^−/−/Tbc1d1^−/−, AS160^−/−, Tbc1d1^−/−, and wild-type offspring of heterozygous breeders were used for in vivo experiments shown in Figs. 1, 2, A–C, 3, 4, A and B, and 6 (AS160^−/−/Tbc1d1^−/− mice). For metabolic cage studies shown in Fig. 6, AS160^−/− and wild-type mice were offspring of heterozygous AS160^+/− breeder pairs. For all other experiments, offspring of male and female AS160^−/−/Tbc1d1^−/− and male and female wild-type breeders were used. The in vivo phenotype (glucose and insulin tolerance and blood glucose levels after 6 h of fasting) of AS160^−/−/Tbc1d1^−/− offspring of homozygous parents (data not shown) was confirmed to be similar to offspring of heterozygous parents (Figs. 3 and 4). Mice were housed under temperature- (22°C) and humidity-controlled conditions and a constant (14:10-h) light-dark cycle, except for metabolic cage studies (12:12-h light-dark cycle), with free access to water and chow [containing 19.1% (wt/wt) protein (25 cal%), 5.8% fat (17 cal%), and 44.3% carbohydrate (58 cal%), cat. no. 7912 Teklad LM-485; Harlan Laboratories]. In each experiment, mice were matched for age and sex. All animal procedures were approved by the University of Virginia Institutional Animal Care and Use Committee.

Fig. 1.

Growth and body composition. A: body weights (BW). BW were determined at indicated ages for male (top) and female (bottom) wild-type (WT) (open bars; n = 7–8), Tbc1d1^−/− (light gray bars; n = 4–6), Akt substrate of 160 kDa (AS160)^−/− (dark gray bars; n = 5–6), and AS160^−/−/Tbc1d1^−/− mice (black bars; n = 8). Means ± SE are shown. Similar results were obtained with a second group of male mice at 9 and 12 wk of age. The other time points were not evaluated in this second group of mice. B: body composition. Body composition was evaluated by dual-energy X-ray absorptiometry (DEXA) in 18 wk-old male (top) and female (bottom) WT (open bars; n = 8), Tbc1d1^−/− (light gray bars; n = 6), AS160^−/− (dark gray bars; n = 5–6), and AS160^−/−/Tbc1d1^−/− mice (black bars; n = 8). Fat mass in %total body mass (means ± SE) is shown. *P < 0.05, unless indicated otherwise, comparing respective data with unpaired 2-tailed t-tests.

Fig. 2.

Glucose uptake (GU) ex vivo and glucose transporter GLUT4 expression in adipocytes and skeletal muscles and GLUT4 subcellular distribution in skeletal muscles. A: GU in isolated adipocytes. Epididymal adipocytes were isolated from 14- to 15-wk-old male WT (open bars) and AS160^−/−/Tbc1d1^−/− mice (black bars) and GU in amol glucose·cell⁻¹·min⁻¹ determined in the absence (Bas) and presence of 10 nM insulin (Ins). Data (means ± SE) were plotted as %WT basal (%+/+ Bas) (n = 4 assays). *P < 0.05 for comparisons of data with WT Bas (1-sample t-tests with WT Bas assigned a value of 100); ^P < 0.05 for comparison of AS160^−/−/Tbc1d1^−/− Bas and AS160^−/−/Tbc1d1^−/− Ins with WT Ins; #P < 0.05 for comparison between AS160^−/−/Tbc1d1^−/− Bas and AS160^−/−/Tbc1d1^−/− Ins using unpaired 2-tailed t-tests. AS160^−/−/Tbc1d1^−/− and WT mice used for the GU assays had similar BW (AS160^−/−/Tbc1d1^−/− 25.1 ± 0.55 g vs. WT 25.7 ± 0.69 g, P = 0.49, n = 9), fat pad weights (AS160^−/−/Tbc1d1^−/− 0.377 ± 0.0395 g vs. WT 0.430 ± 0.0554 g, P = 0.44, n = 9), and fat cell sizes (AS160^−/−/Tbc1d1^−/− 3,459 ± 295 vs. 3,757 ± 381 arbitrary units, P = 0.55, n = 5, with 100 cell circumferences sized for each genotype and experiment). B and C: GU in isolated soleus (Sol) and extensor digitorum longus (EDL) muscles. Sol (B) and EDL (C) muscles were isolated from the same 14- to 15-wk-old male WT (open bars) and AS160^−/−/Tbc1d1^−/− mice (black bars) used in A, and 2-deoxyglucose uptake was determined under Bas conditions and in the presence of 13 nM Ins. GU data in μmol·g⁻¹·20 min⁻¹ represent means ± SE (n = 8 individual muscles). *P < 0.05, comparing data with WT Bas; ^P < 0.05, comparing data for WT Ins with AS160^−/−/Tbc1d1^−/− Bas and AS160^−/−/Tbc1d1^−/− Ins; #P < 0.05, comparing data for AS160^−/−/Tbc1d1^−/− Bas with AS160^−/−/Tbc1d1^−/− Ins using unpaired 2-tailed t-tests. D: GLUT4 expression in adipose tissues and skeletal muscles. GLUT4 immunoblots were performed with epididymal white adipose tissue (WAT), interscapular brown adipose tissue (BAT), Sol, EDL, gastrocnemius (gastro), and tibialis anterior (TA) muscle homogenates (40 μg total protein/lane) obtained from 9-wk-old male WT (+/+) and AS160^−/−/Tbc1d1^−/− (−/−) mice. Left: representative results for 2 samples for each genotype. Right: quantitations of GLUT4 signals on immunoblots (means ± SE; n = 3–4) in %WT (%+/+, with one of the WT samples for each tissue assigned a value of 100%). *P < 0.05, comparing data for AS160^−/−/Tbc1d1^−/− and WT tissues with unpaired 2-tailed t-tests. E–G: GLUT4 plasma membrane localization. Male WT (+/+, open bars) and AS160^−/−/Tbc1d1^−/− mice (−/−, black bars) (16–18 wk old) were fasted for 6 h (8 AM to 2 PM), anesthetized, and treated with no insulin (Bas) or Ins (21 mU/g body wt) for 30 min. Gastro and quadriceps (quad) muscles were isolated, pooled (gastro/quad), and subjected to subcellular fractionation. Immunoblots for GLUT4 were performed on plasma membranes (PM GLUT4; 5 μg protein/lane) and total membranes (Total GLUT4; 10 μg protein/lane), and representative immunoblots for 1 subcellular fractionation are shown in E and F, top. Signals on immunoblots were quantified and expressed as relative levels of WT Bas (assigned a value of 1). Resulting data (means ± SE, n = 4–5) are shown in bar graphs below the respective immunoblots in E and F. Ratios of normalized PM GLUT4 to total GLUT4 were calculated (PM GLUT4/total GLUT4), and data (means ± SE; n = 4–5) are shown in G. Ratio of PM GLUT4 to total GLUT4 in gastro/quad. *P < 0.05 for comparisons of data with WT Bas (1 sample t-tests with WT Bas assigned a value of 1); ^P < 0.05 for comparisons of AS160^−/−/Tbc1d1^−/− Bas and AS160^−/−/Tbc1d1^−/− Ins with WT Ins using unpaired 2-tailed t-tests. No statistically significant differences were observed between AS160^−/−/Tbc1d1^−/− Bas and AS160^−/−/Tbc1d1^−/− Ins for any of the data. Plasma membrane protein yields from total membranes were similar between WT and AS160^−/−/Tbc1d1^−/− muscles for all conditions.

Fig. 3.

Blood glucose (BG) and insulin levels under random-fed and fasting conditions. BG and plasma Ins were measured in random-fed (at 9 wk of age; A and B) and 6-h-fasted (at 11 wk of age; C and D) male (top) and female (bottom) WT (open bars; n = 8), Tbc1d1^−/− (light gray bars; n = 6), AS160^−/− (dark gray bars; n = 5–6), and AS160^−/−/Tbc1d1^−/− mice (black bars; n = 8). Data represent means ± SE. *P < 0.05, comparing respective data with unpaired 2-tailed t-tests. P values for close to statistically significant differences are as indicated. Note that for female fasting plasma Ins, several of the values were below the range of the standard curve (<0.188 ng/ml), and the data shown represent means ± SE only for values within the range of the standard curve for WT (n = 5), Tbc1d1^−/− (n = 3), AS160^−/− (n = 2), and AS160^−/−/Tbc1d1^−/− mice (n = 7).

Fig. 4.

Glucose tolerance, insulin sensitivity, and insulin secretion. A: glucose tolerance tests were performed after overnight fasting with 15-wk-old male (left) and female (right) WT (■ and solid black line; n = 8), Tbc1d1^−/− (inverted gray triangles and dotted gray line; n = 6), AS160^−/− (gray triangles and solid gray line; n = 5–6), and AS160^−/−/Tbc1d1^−/− mice (⧫ and dotted black line; n = 8). BG levels were measured before (time 0) and after intraperitoneal injection of glucose (1 mg/g body wt) at 10, 20, 30, 60, 90, and 120 min. BG (means ± SE) is graphed. Data are compared with 2-way ANOVA. For male mice, statistically significant differences were found between WT and AS160^−/−/Tbc1d1^−/− (P < 0.0001) and AS160^−/− mice (P = 0.0007) but not between WT and Tbc1d1^−/− mice (P = 0.53). Data for AS160^−/−/Tbc1d1^−/− and AS160^−/− mice were not different from each other (P = 0.28) but were significantly different from data for Tbc1d1^−/− mice (P < 0.0001 and 0.0002, respectively). For females, results for AS160^−/− mice were significantly different from WT (P = 0.03), Tbc1d1^−/− (P = 0.03), and AS160^−/−/Tbc1d1^−/− mice (P = 0.02). Bar graphs above curves show areas under the curve (AUC). Glucose tolerance tests were also performed with 2 mg glucose/g body wt, as in Ref. 3. Similar abnormalities were observed, as shown for male and female mice, but several of the BG levels were above the reading range of the glucometer (>600 mg/dl). B: insulin tolerance tests were performed at 2 PM on 10-wk-old random-fed male (left) and female (right) WT (n = 6–8), AS160^−/− (n = 5), Tbc1d1^−/− (n = 5–6), and AS160^−/−/Tbc1d1^−/− (n = 6–7) mice. For the description of symbols and lines for different genotypes, see A. BG levels were measured before (time 0) and after an intraperitoneal injection of insulin (0.75 U/kg body wt) at 15, 30, 45, and 60 min. BG in %BG at time 0 [BG (%basal)] (means ± SE) is graphed. Data were analyzed by 2-way ANOVA. For male mice, data are statistically significantly different from WT for male AS160^−/−/Tbc1d1^−/− (P < 0.0001) and AS160^−/− mice (P = 0.0002), but not for Tbc1d1^−/− mice (P = 0.47). Data are statistically significantly different from Tbc1d1^−/− mice for AS160^−/− (P = 0.03) and AS160^−/−/Tbc1d1^−/− mice (P = 0.0004), but there was no difference between AS160^−/− and AS160^−/−/Tbc1d1^−/− mice (P = 0.18). For females, data are statistically significantly different from WT for AS160^−/−/Tbc1d1^−/− (P = 0.0007) and AS160^−/− (P = 0.008), but not for Tbc1d1^−/− mice (P = 0.33). AS160^−/−/Tbc1d1^−/− mice are statistically significantly different from Tbc1d1^−/− mice (P = 0.0003), but not from AS160^−/− mice (P = 0.23). Data for Tbc1d1^−/− and AS160^−/− mice are significantly different (P = 0.004). Bar graphs above curves show AUC. Similar results for glucose and insulin tolerance tests were obtained with a second group of male mice. C and D: Akt and Akt substrate phosphorylation in skeletal muscle, adipose tissues, and liver. Male WT (+/+) and AS160^−/−/Tbc1d1^−/− mice (−/−) at 15–17 wk of age were fasted for 6 h (8 AM to 2 PM) and then treated without (basal) or with insulin as described in the legend to Fig 2, E–G, before gastro, interscapular BAT and epididymal WAT, and liver were harvested. Homogenates (100 μg total protein/lane for gastro, BAT, and liver and 40 μg total protein/lane for WAT) were immunoblotted with antibodies against Akt p-Ser⁴⁷³ (p-Akt; top) and total Akt (Akt; bottom) (p-Akt and Akt immunoblots; C) and phosphorylated Akt substrates (p-AS; marked as p-AS160 in BAT and, since they are indistinguishable, as p-AS160/p-Tbc1d1 in gastro, p-AS immunoblots; D). Immunoblots representative of 2 mice for each genotype are shown. Immunoblots were repeated with another set of samples, and similar results were obtained. The signals for p-Akt and Akt were quantified and normalized to WT basal (assigned a value of 1), ratios for p-Akt/Akt were calculated, and means ± SE (n = 4) were obtained. Data for each of the tissues of WT and AS160^−/−/Tbc1d1^−/− mice were compared with unpaired 2-tailed t-tests; no significant differences were observed (data not shown). The vertical black line in the p-Akt and Akt immunoblots of WAT indicates the deletion of 2 empty lanes in the immunoblot image. E: insulin secretion during glucose tolerance tests. Glucose tolerance tests were performed with 12-wk-old male (left) and 13- to 14-wk-old female (right) WT (■ and solid black lines) and AS160^−/−/Tbc1d1^−/− mice (⧫ and dotted lines) as described above in A, except that glucose was given at 2 mg/g body wt, and blood for glucose and insulin measurements was taken only at 0, 10, and 30 min. Insulin levels at the different time points are shown (means ± SE; n = 7–8 for males, n = 5–9 for females). Data were statistically significantly different between AS160^−/−/Tbc1d1^−/− and WT mice for male and female mice (P = 0.0003 and P = 0.015, respectively, 2-way ANOVA). BG levels during the glucose tolerance tests at these early time points were similar between WT and AS160^−/−/Tbc1d1^−/− mice for male and female mice (P = 0.09 and 0.61, respectively, 2-way ANOVA). At the 0- and 10-min time points, insulin values were also significantly lower for male AS160^−/−/Tbc1d1^−/− compared with WT mice (P = 0.004 and 0.02, respectively, unpaired 2-tailed t-tests). For female AS160^−/−/Tbc1d1^−/− mice, 0- and 30-min insulin values were close to being statistically significantly lower (P = 0.06 and P = 0.097, respectively, unpaired 2-tailed t-tests). Note that these mice were fasted for 16 h, whereas mice for which insulin levels are shown in Fig. 3 were fasted for 6 h. *P < 0.05, comparing respective data with unpaired 2-tailed t-tests.

Fig. 6.

Energy homeostasis. Metabolic cage studies were performed with 12- to 14-wk-old male AS160^−/− (black bars, −/−; n = 6) and WT littermates (open bars, +/+; n = 7) (left) and 22- to 26-wk-old AS160^−/−/Tbc1d1^−/− (black bars, −/−; n = 6) and WT littermates (open bars, +/+; n = 6) (right). Data are presented as means ± SE. A: energy expenditure represented by oxygen consumption (V̇o₂)/h normalized to BW. B: food intake expressed in mg of chow consumed over 24 h during 2 dark or 2 light cycles and normalized to BW. C: respiratory exchange ratios (RER) between carbon dioxide production and oxygen consumption (V̇co₂/V̇o₂) during light and dark cycles are shown. D: total activity represents the average total no. of infrared beam breaks per 15 min during light or dark cycles while mice were in metabolic cages. *P < 0.05, unless indicated otherwise, between AS160^−/− or AS160^−/−/Tbc1d1^−/− and WT mice (unpaired 2-tailed t-tests). BW of AS160^−/− mice used in this study were significantly lower compared with WT littermates (23.9 ± 0.34 vs. 26.4 ± 0.93 g; n = 6–7, P = 0.035) and close to significantly lower for AS160^−/−/Tbc1d1^−/− mice compared with WT littermates (27.6 ± 1.02 vs. 30.6 ± 0.90 g; n = 6–7, P = 0.06). Body composition was determined by DEXA in the same mice. Data for AS160^−/−/Tbc1d1^−/− and WT littermates are shown in Fig. 1B. Similar relative data vs. WT were obtained for AS160^−/− mice, as presented in Fig. 1B.

Glucose uptake and GLUT4 expression and subcellular distribution.

Glucose uptake assays with isolated adipocytes and soleus and EDL muscles, subcellular fractionation of gastrocnemius and quadriceps muscles, and evaluation of GLUT4 levels by immunoblotting were performed as described (17).

Akt and Akt substrate phosphorylation in skeletal muscle and adipose tissues.

Mice were fasted for 6 h (8 AM-2 PM) and then anesthetized with Nembutal (50 μg/g body wt). Insulin (21 mU/g body wt, Humulin R; Eli Lilly) or no insulin was injected intraperitoneally, and 30 min later mice were euthanized by cervical dislocation. Epididymal white adipose tissue, interscapular brown adipose tissue, gastrocnemius, and liver were dissected and immediately frozen in liquid nitrogen. Tissues were stored at −80°C until further processing. Tissue homogenates were prepared in homogenization buffer (25 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM Na₃VO₄, 20 mM sodium pyrophosphate, 3 mM N-ethylmaleimide, 1 μg/ml pepstatin, 10 μM EP475, 10 μg/ml aprotinin, 0.2 mM PMSF, 10 mM β-glycerophosphate, and 10 mM sodium fluoride) as described (17). SDS samples containing equal amounts of protein were separated by SDS-PAGE and immunoblotted with phospho-Akt substrate (p-AS) antibodies (cat. no. 9611; Cell Signaling Technology), p-Akt Ser⁴⁷³-specific antibodies (cat. no. 9271; Cell Signaling Technology), and total Akt antibodies (cat. No. 9272; Cell Signaling Technology) (17). Akt phosphorylation and Akt expression in isolated adipocytes, soleus, and EDL were determined as described (17).

Blood glucose and plasma insulin, glucose and insulin tolerance, and glucose disposal.

Blood glucose and plasma insulin, glucose and insulin tolerance, and glucose disposal in vivo were analyzed as described previously (14, 17), with the exception that mice were fasted for 7 h (6 AM to 1 PM) before the determination of glucose disposal in vivo.

Body composition, energy expenditure, and respiratory exchange ratio.

Body composition was determined using dual energy X-ray absorptiometry (14). Oxygen consumption (V̇o₂), carbon dioxide production (V̇co₂), food intake, and ambulatory activity were determined in an Oxymax metabolic chamber system (Comprehensive Laboratory Animal Monitoring System from Columbus Instruments, Columbus, OH) (14). Respiratory exchange ratios (RER) were calculated as the ratio of carbon dioxide production to oxygen consumption (V̇co₂/V̇o₂). Heat production was calculated by Oxymax software using gas exchange data with the following formula: heat (kcal·kg body wt⁻¹·h⁻¹) = (3.815 + 1.232 × RER) × V̇o₂.

Statistical analysis.

Statistical analysis was performed with GraphPad Prism software (GraphPad Software Incorporation). One-sample t-tests with wild-type basal assigned values of 100 or 1, unpaired two-tailed t-tests, and two-way ANOVA were used to compare data (means ± SE), as specified in the figure legends. Differences were considered statistically significant for P values <0.05. P values between 0.05 and 0.1 are given in the figures to indicate strong trends.

RESULTS

Growth and development of mice with AS160 and/or Tb1d1 deletions.

AS160^+/−/Tbc1d1^+/− breeder pairs produced close to the expected 6.25% of AS160^−/−/Tbc1d1^−/− (6.5%), AS160^−/− (6.9%), and Tbc1d1^−/− (6.9%) mice but fewer than the expected 6.25% of wild-type mice (4.5%) among 741 offspring. Thus, deletion of AS160 and/or Tbc1d1 does not increase embryonic lethality but may provide a competitive advantage during early development. However, AS160 and/or Tbc1d1 deficiencies cause abnormalities in growth and body composition. When measuring body weights at different ages, we observed that male AS160^−/−/Tbc1d1^−/− mice weighed significantly less than wild-type mice at 9, 12, and 15 wk of age and trended lower at 18 (P = 0.07) and 21 wk (P = 0.12) of age (Fig. 1A, top). Male AS160^−/− and Tbc1d1^−/− mice weighed consistently less than wild-type (P = 0.007 and P = 0.02, respectively, 2-way ANOVA across all ages) but were not significantly different from AS160^−/−/Tbc1d1^−/− mice. Body weights for AS160^−/− and Tbc1d1^−/− mice were not significantly different from each other. For females, body weights of AS160^−/−/Tbc1d1^−/− and AS160^−/− mice were similar to wild-type mice, but Tbc1d1^−/− mice weighed significantly less than wild-type mice at 9, 12, and 15 but not at 18 and 21 wk of age (Fig. 1A, bottom). Tbc1d1^−/− mice were also significantly lower in weight than AS160^−/− mice at all ages. Compared with AS160^−/−/Tbc1d1^−/− mice, Tbc1d1^−/− mice weighed less when data from all ages were included in the analysis (P = 0.0025, 2-way ANOVA). When determining body composition, we found that for males, AS160^−/− mice had significantly decreased and AS160^−/−/Tbc1d1^−/− mice showed a trend to lower fat mass (P = 0.13) (Fig. 1B, top). In contrast, female AS160^−/− and AS160^−/−/Tbc1d1^−/− mice had significantly increased fat mass (Fig. 1B, bottom). Body composition was normal for male and female Tbc1d1^−/− mice (Fig. 1B, top and bottom). In summary, deletions of AS160 and/or Tbc1d1 lead to sex-specific changes in body weight and body composition. Body weight changes of AS160^−/− and Tbc1d1^−/− mice coalesce in AS160^−/−/Tbc1d1^−/− mice.

Glucose uptake in isolated AS160^−/−/Tbc1d1^−/− adipocytes and skeletal muscles.

Previously, we described that in AS160^−/− mice, adipocyte glucose uptake is increased by 1.7-fold under basal conditions and decreased by 35% after insulin stimulation (17). In contrast, we found that Tbc1d1^−/− adipocytes showed normal basal and insulin-stimulated glucose uptake (14). As shown in Fig. 2A, upon deletion of both AS160 and Tbc1d1, basal glucose uptake is increased 1.9-fold, and insulin-stimulated glucose uptake is normal in isolated adipocytes. Basal glucose uptake in AS160^−/−/Tbc1d1^−/− adipocytes is thus increased to a similar extent as in AS160^−/− adipocytes, suggesting that AS160 controls basal glucose uptake in adipocytes with no contribution from Tbc1d1.

For isolated skeletal muscles, we previously observed ∼30% decreased insulin-stimulated glucose uptake in AS160^−/− soleus (17) and Tbc1d1^−/− EDL muscles (14). But insulin-stimulated glucose uptake was normal in AS160^−/− EDL (17) and Tbc1d1^−/− soleus (14). Basal glucose uptake was normal in soleus and EDL of both AS160^−/− and Tbc1d1^−/− mice (14, 17). As shown in Fig. 2, B and C, for AS160^−/−/Tbc1d1^−/− mice, insulin-stimulated glucose uptake is decreased by 42% in soleus and by 44% in EDL, whereas basal glucose uptake is normal in both soleus and EDL. The impaired insulin-stimulated glucose uptake was not due to impaired signaling; insulin-induced phosphorylation of Akt was normal in isolated AS160^−/−/Tbc1d1^−/− soleus and EDL (data not shown). Thus, combined deletion of AS160 and Tbc1d1 leads to defects in insulin-stimulated glucose uptake in both soleus and EDL. The magnitudes of the defects in soleus and EDL are similar to the ones observed previously in AS160^−/− and Tbc1d1^−/− mice, respectively.

GLUT4 expression in AS160^−/−/Tbc1d1^−/− adipose tissues and skeletal muscles.

Previously, we showed that in AS160^−/− and Tbc1d1^−/− mice there is decreased GLUT4 protein expression that follows an adipose- and skeletal muscle-specific pattern consistent with the differential tissue distributions of AS160 and Tbc1d1 (14, 17). Specifically, in AS160^−/− mice we found that GLUT4 protein was reduced in white (58%) and brown adipose tissues (19%) and in soleus (35%) and gastrocnemius muscles (13%) (17). In Tbc1d1^−/− mice we observed decreased GLUT4 protein in EDL (58%), tibialis anterior (57%), and gastrocnemius muscles (37%) (14). As shown in Fig. 2D, in AS160^−/−/Tbc1d1^−/− mice GLUT4 protein levels are decreased in all adipose and skeletal muscle tissues analyzed: white adipose tissue (44%), brown adipose tissue (13%, P = 0.11), soleus (41%), EDL (50%), gastrocnemius (60%), and tibialis anterior (70%). Thus, AS160^−/−/Tbc1d1^−/− mice show reductions in GLUT4 that combine the decreases in GLUT4 expression observed previously in tissues of single AS160- and Tbc1d1-knockout mice.

GLUT4 subcellular distribution in AS160^−/−/Tbc1d1^−/− skeletal muscles.

Our previous analysis of GLUT4 subcellular distribution in pooled gastrocnemius/quadriceps (gastro/quad) showed that GLUT4 plasma membrane localization under basal conditions was increased in AS160^−/− and Tbc1d1^−/− mice, but to different extents (14, 17). In AS160^−/− gastro/quad plasma membranes, relative GLUT4 amounts under basal conditions were elevated to levels found in wild-type plasma membrane after insulin exposure, and no further increase in response to insulin was observed. In Tbc1d1^−/− gastro/quad plasma membranes, relative GLUT4 levels were increased under basal conditions, but not to levels found in wild-type plasma membranes after insulin stimulation. GLUT4 increased further in response to insulin in Tbc1d1^−/− gastro/quad plasma membranes to reach the same relative levels as in wild-type plasma membranes. Since AS160 and Tbc1d1 are both relatively well expressed in gastrocnemius and quadriceps (14, 17, 30, 33), we examined whether AS160^−/−/Tbc1d1^−/− mice combined the defects of the single-knockout mice. As shown in Fig. 2E, under basal conditions, GLUT4 amounts in AS160^−/−/Tbc1d1^−/− gastro/quad plasma membranes were the same as in wild-type mice. In response to insulin, GLUT4 plasma membrane levels increased 1.7-fold in wild-type but remained at basal levels in AS160^−/−/Tbc1d1^−/− plasma membranes. When taking into account that total GLUT4 expression was decreased by 60% in AS160^−/−/Tbc1d1^−/− gastro/quad membranes (Fig. 2F), the relative amount of GLUT4 in basal AS160^−/−/Tbc1d1^−/− plasma membranes was increased by 2.7-fold compared with wild-type basal and 1.5-fold compared with wild type after insulin stimulation (Fig. 2G). There was no significant change in the relative amounts of GLUT4 in AS160^−/−/Tbc1d1^−/− plasma membranes in response to insulin, with relative GLUT4 levels still 1.2-fold higher compared with wild type after insulin treatment (Fig. 2G). Increased relative GLUT4 plasma membrane localization under basal conditions and the lack of an insulin-elicited increase was not due to abnormalities in Akt activation. Akt phosphorylation in gastrocnemius (determined in vivo in the same setting used for subcellular fractionation) was similar for wild-type and AS160^−/−/Tbc1d1^−/− mice under basal and insulin-stimulated conditions (Fig. 4C); but as expected, in AS160^−/−/Tbc1d1^−/− mice, insulin-elicited phosphorylation of Akt substrates of 160 kDa, the size of both AS160 and Tbc1d1, was abolished in response to insulin (Fig. 4D).

In summary, AS160^−/−/Tbc1d1^−/− gastro/quad show increased normalized GLUT4 plasma membrane association under basal and insulin-stimulated conditions, and the relative increases are higher than we observed previously in AS160^−/− or Tbc1d1^−/− gastro/quad plasma membranes.

Whole body glucose homeostasis and insulin sensitivity in mice with AS160 and/or Tbc1d1 deletions.

Our previous studies demonstrated normal (male) or impaired (female) glucose tolerance, impaired insulin tolerance (male and female), and lower fasting blood glucose (male and female) in AS160^−/− mice (17). In Tbc1d1^−/− mice we found normal (male) or mildly impaired (female) insulin and glucose tolerance and increased fasting blood glucose (male) (14). To directly compare the effects of deletions of AS160 and/or Tbc1d1 on glucose homeostasis, we analyzed male and female AS160^−/−/Tbc1d1^−/− mice together with AS160^−/−, Tbc1d1^−/−, and wild-type littermates, measured fed and fasting blood glucose and insulin levels, and performed glucose and insulin tolerance tests (Figs. 3 and 4).

Random-fed and fasting blood glucose levels were significantly lower or trended lower in male and female AS160^−/− and AS160^−/−/Tbc1d1^−/− mice but were normal in male and female Tbc1d1^−/− compared with wild-type mice (Fig. 3, A and C). In random-fed males, plasma insulin levels were significantly lower in AS160^−/− and AS160^−/−/Tbc1d1^−/− mice but were normal in Tbc1d1^−/− compared with wild-type mice (Fig. 3B, top). After 6 h of fasting, plasma insulin levels were significantly lower in male AS160^−/− and Tbc1d1^−/− mice but not in AS160^−/−/Tbc1d1^−/− mice compared with wild-type (Fig. 3D, top). In female AS160^−/−, Tbc1d1^−/−, and AS160^−/−/Tbc1d1^−/− mice, plasma insulin levels were similar to wild-type mice under random-fed and fasting conditions (Fig. 3, B and D, bottom).

Glucose tolerance was impaired in male AS160^−/− and AS160^−/−/Tbc1d1^−/− mice but not Tbc1d1^−/− compared with wild-type mice (Fig. 4A, left). For females, glucose tolerance was mildly impaired in AS160^−/− mice but was normal in Tbc1d1^−/− and AS160^−/−/Tbc1d1^−/− compared with wild-type mice (Fig. 4A, right). Glucose tolerance is determined by insulin signaling and action and insulin secretion. To obtain a measurement for insulin action, we performed insulin tolerance tests. Male and female AS160^−/− and AS160^−/−/Tbc1d1^−/− mice showed a diminished response to lower blood glucose levels after insulin injection, whereas Tbc1d1^−/− mice responded like wild type mice (Fig. 4B). The impaired insulin response was not due to impaired insulin signaling upstream of AS160 and Tbc1d1 in major insulin target tissues responsible for regulating glucose disposal and output. As shown in Fig. 4C, insulin-stimulated phosphorylation of Akt was intact in AS160^−/−/Tbc1d1^−/− gastrocnemius, white and brown adipose tissues, and liver, but as expected with deletion of both AS160 and Tbc1d1, insulin-induced phospho-Akt substrates of 160 kDa were missing (Fig. 4D). We also did not previously observe any abnormalities in insulin-elicited Akt phosphorylation in AS160^−/− mice (17). To test for defects in insulin secretion, we measured insulin levels after a glucose challenge in AS160^−/−/Tbc1d1^−/− mice. We found significantly decreased insulin levels in male and female AS160^−/−/Tbc1d1^−/− compared with wild-type mice (Fig. 4E). However, insulin levels in AS160^−/− mice after a glucose challenge were not significantly different from wild-type mice (17).

Glucose disposal in skeletal muscles, adipose tissues, and liver contributes to whole body glucose homeostasis (18). As described above and in our previous studies, AS160 and/or Tbc1d1 deletion results in abnormal basal or insulin-stimulated glucose uptake in isolated adipocytes and skeletal muscles (Fig. 2, A–C) (14, 17). However, our previous studies with AS160^−/− and Tbc1d1^−/− mice also demonstrated that in vivo and ex vivo glucose uptake do not always match (14, 17). To determine whether defects in glucose disposal in individual tissues contributed to impaired insulin and glucose tolerance in male AS160^−/−/Tbc1d1^−/− mice, we injected radioactive 2-deoxyglucose in either saline (for basal) or glucose (to trigger insulin release) and evaluated glucose uptake in different skeletal muscles and adipose tissues and glucose incorporation into glycogen in liver. We observed significant abnormalities in glucose uptake in brown adipose tissue, gastrocnemius, and tibialis anterior muscles but not white adipose tissue of AS160^−/−/Tbc1d1^−/− mice (Fig. 5, A–D). In AS160^−/−/Tbc1d1^−/− brown adipose tissue basal glucose uptake was increased by 1.7-fold vs. wild-type basal and was just 25% lower than for glucose-injected wild-type brown adipose tissue (Fig. 5A). Insulin stimulation caused a similar 2.1- and 2.3-fold increase in glucose uptake in AS160^−/−/Tbc1d1^−/− and wild-type brown adipose tissue, respectively, resulting in 1.6-fold higher glucose uptake in AS160^−/−/Tbc1d1^−/− than in wild-type mice (Fig. 5A). In gastrocnemius muscle, basal glucose uptake was decreased by 32% in AS160^−/−/Tbc1d1^−/− compared with wild-type mice (Fig. 5C). Insulin increased glucose uptake to similar extents in AS160^−/−/Tbc1d1^−/− and wild-type gastrocnemius (1.4- vs. 1.3-fold, respectively). Thus, insulin-stimulated glucose uptake in AS160^−/−/Tbc1d1^−/− gastrocnemius was similar to wild-type basal and 28% lower than in insulin-stimulated wild type mice (Fig. 5C). In tibialis anterior muscle, basal glucose uptake in AS160^−/−/Tbc1d1^−/− mice was increased 1.8-fold over wild-type and matched glucose uptake in insulin-stimulated wild-type mice (Fig. 5D). An additional 1.4-fold increase in glucose uptake in response to insulin in AS160^−/−/Tbc1d1^−/− tibialis anterior resulted in a 1.5-fold higher glucose uptake compared with insulin-stimulated wild-type mice (Fig. 5D). Basal glucose incorporation into glycogen in AS160^−/−/Tbc1d1^−/− liver was decreased by 30% compared with wild-type liver. But with 2.5- vs. twofold increases in AS160^−/−/Tbc1d1^−/− and wild-type livers, respectively, similar levels of glucose incorporation were achieved after glucose injection (Fig. 5E). Integrated glucose levels during these tests, presented as areas under the curve, were similar for saline- and glucose-injected AS160^−/−/Tbc1d1^−/− and wild-type mice (Fig. 5F). The discrepancy between normal (Fig. 5F) and increased (Fig. 4A, left) areas under the curve in glucose tolerance tests for AS160^−/−/Tbc1d1^−/− mice can be explained by different fasting periods before the tests, 7 vs. 16 h, respectively. After 7 h of fasting, AS160^−/−/Tbc1d1^−/− mice, compared with wild-type mice, exhibit significantly lower blood glucose levels (140.3 ± 6.3 vs. 159.1 ± 2.8 mg/dl, P = 0.009, n = 9–11), thus making glucose areas under the curve similar to wild types. When glucose levels after the injection of glucose are normalized to basal glucose, 7 h-fasted AS160^−/−/Tbc1d1^−/− mice also show relatively higher glucose excursions than wild-type, which is consistent with impaired glucose tolerance observed after 16 h of fasting. We used two different fasting times to match previously published results for glucose tolerance tests (3, 14, 17, 30, 32, 36) and measurements of glucose disposal (14, 17, 32, 36).

Fig. 5.

Glucose (Gluc) disposal in adipose tissues, skeletal muscles, and liver in vivo. GU and glucose incorporation into glycogen were determined in vivo in male 13- to 14-wk-old WT (open bars) and AS160^−/−/Tbc1d1^−/− mice (black bars) fasted for 7 h (6 AM to 1 PM) and then injected with 2-deoxy-d-[1,2-³H]glucose in either saline (Sal) or Gluc (2 mg/g body wt). After 120 min, mice were euthanized and skeletal muscles, adipose tissues, and liver dissected. A–D: GU in vivo in skeletal muscles and adipose tissues. The amount of radioactive 2-deoxy-glucose phosphate, representing 2-deoxy-glucose taken up into cells, was determined and GU expressed in nmol·g tissue⁻¹·min⁻¹. Graphs show means ± SE for BAT (n = 6–9; A), epididymal WAT (n = 4–6; B), gastro (n = 6–9; C), and TA (n = 5–9; D). E: glucose incorporation into glycogen in liver was determined by measuring deoxy-d-[1,2-³H]glucose incorporated into liver glycogen, and data (means ± SE; n = 4–6) expressed as μg·g liver⁻¹·min⁻¹ were plotted. Data were compared with unpaired 2-tailed t-tests, and significant differences (P < 0.05, unless indicated otherwise) are denoted by * comparing data with WT Sal; ^comparing data for AS160^−/−/Tbc1d1^−/− Sal and AS160^−/−/Tbc1d1^−/− Gluc to WT Gluc; and #comparing data between AS160^−/−/Tbc1d1^−/− Sal and AS160^−/−/Tbc1d1^−/− Gluc. F: integrated BG levels derived from measurements at 0, 15, 30, 45, 60, and 120 min after Gluc or Sal injection are shown as AUC [means ± SE for Sal (n = 7) and Gluc (n = 8–10)] for AS160^−/−/Tbc1d1^−/− (black bars) and WT mice (open bars). No significant differences were observed between saline- and glucose-injected AS160^−/−/Tbc1d1^−/− and WT mice used in these tests (P = 0.18 and P = 0.72, respectively, unpaired 2-tailed t-tests).

When comparing the above-described glucose disposal results with our previous measurements in AS160^−/− and Tbc1d1^−/− mice, we found that in tissues in which only AS160 or Tbc1d1 is predominantly expressed, such as brown adipose tissue or tibialis anterior muscles, respectively, changes in glucose uptake in AS160^−/−/Tbc1d1^−/− mice were similar to the ones observed previously in AS160^−/− (17) and Tbc1d1^−/− mice (14). In gastrocnemius muscle, which expresses both Rab GAPs well (14, 17, 33), the changes in AS160^−/−/Tbc1d1^−/− mice mix the findings obtained with single-knockout mice. Glucose uptake in AS160^−/− gastrocnemius is decreased under basal conditions and does not increase in response to insulin (17). Tbc1d1^−/− gastrocnemius shows normal glucose uptake under both basal and insulin-stimulated conditions (14). Thus, basal glucose in AS160^−/−/Tbc1d1^−/− gastrocnemius is similarly impaired as in AS160^−/− gastrocnemius, but insulin is able to stimulate glucose uptake in AS160^−/−/Tbc1d1^−/− as in Tbc1d1^−/− gastrocnemius. In liver that expresses little AS160 or Tbc1d1, we also observe a mixed phenotype. Glucose incorporation into glycogen is decreased under basal conditions and does not further increase in response to insulin in AS160^−/− mice (17), but it is normal in Tbc1d1^−/− mice under basal and insulin-stimulated conditions (14). In AS160^−/−/Tbc1d1^−/− mice, liver glucose incorporation into glycogen is decreased under basal conditions as in AS160^−/− liver but normal after insulin stimulation as in Tbc1d1^−/− liver.

In summary, AS160^−/−/Tbc1d1^−/− mice show abnormalities in glucose and insulin tolerance and fed and fasting glucose and insulin levels that are more similar to defects in AS160^−/− than Tbc1d1^−/− mice. Impaired insulin and glucose tolerance in AS160^−/−/Tbc1d1^−/− mice may be explained by decreased glucose uptake in large skeletal muscles and deficient insulin secretion. Decreased fasting glucose levels in AS160^−/−/Tbc1d1^−/− mice may be due to increased basal glucose uptake in brown adipose tissue and some skeletal muscles (such as tibialis anterior). Consistent with this, basal glucose incorporation into glycogen in liver is decreased in AS160^−/−/Tbc1d1^−/− (Fig. 5E). Glucose uptake in different skeletal muscles and adipose tissues of AS160^−/−/Tbc1d1^−/− mice is differentially affected, and the defects represent a combination of abnormalities observed previously in AS160^−/− (17) and Tbc1d1^−/− mice (14). Changes in glucose uptake in vivo are not associated with abnormalities in insulin signaling upstream of AS160 and Tbc1d1 (Fig. 4C), and they are not always consistent with decreased GLUT4 protein levels (Fig. 2D) and GLUT4 amounts in the plasma membrane (Fig. 2E). Discrepancies between GLUT4 expression, GLUT4 levels in the plasma membrane, and glucose uptake were also observed in AS160^−/− and Tbc1d1^−/− mice (14, 17). In skeletal muscles of AS160- and Tbc1d1-deficient mice, basal intracellular GLUT4 retention is defective, but GLUT4 cell surface exposure is still subject to regulation by insulin (14, 17). Thus, changes in plasma membrane GLUT4 levels do not match changes in glucose uptake. Furthermore, additional currently unknown factors affect glucose uptake in vivo in AS160- and Tbc1d1-deficient skeletal muscles (14, 17).

Energy homeostasis in mice with AS160 and/or Tbc1d1 deletions.

As described above, body weights of male AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice were consistently lower than for wild-type mice. To evaluate whether decreased body weights were a consequence of changes in energy homeostasis, we performed metabolic cage studies. Previously, we described our observations with Tbc1d1^−/− mice (14), and here we show the analysis of AS160^−/− and AS160^−/−/Tbc1d1^−/− mice (Fig. 6). We found that, during the dark cycle, oxygen consumption, a measure for energy expenditure, is elevated by 10% in AS160^−/−/Tbc1d1^−/− and AS160^−/− mice compared with matching wild types (Fig. 6A). Similarly, Tbc1d1^−/− mice also show a 10% increase in oxygen consumption (14). During the light cycle, there were no significant changes in oxygen consumption for AS160^−/−/Tbc1d1^−/− or AS160^−/− (Fig. 6A) and Tbc1d1^−/− mice (14), although AS160^−/−/Tbc1d1^−/− and Tbc1d1^−/− mice showed trends toward higher rates (Fig. 6A) (14). Food intake trended 25, 17, and 11% higher in AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice, respectively, under dark conditions compared with wild-type mice (Fig. 6B) (14). Respiratory exchange ratios were similar to wild type mice for AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice during light and dark cycles (Fig. 6C) (14). Total activity was similar to wild type for AS160^−/−/Tbc1d1^−/− (Fig. 6D) and Tbc1d1^−/− mice (14). However, AS160^−/− mice had decreased activity during the light and dark cycles compared with wild-type mice (Fig. 6D). There were no abnormalities in heat production in AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice (data not shown). In summary, AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice show abnormalities in energy homeostasis that are compatible with lower body weights. The changes in food intake, which are more pronounced in AS160^−/−/Tbc1d1^−/− mice than in AS160^−/− and Tbc1d1^−/− mice, suggest nonredundant functions for each of the two Rab GAPs in controlling feeding. Whether the similar changes in energy expenditure in AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice signify redundancy or blend different complex functions of AS160 and Tbc1d1 in the regulation of energy homeostasis cannot be distinguished.

DISCUSSION

Studies with cultured fat and muscle cells suggest similar roles for AS160 and Tbc1d1 in regulating intracellular GLUT4 retention and release to the cell surface and glucose uptake (26). However, the distinct tissue distribution of the two Rab GAPs (5, 14, 17, 30, 33) and their differential phosphorylation in response to different stimuli (6, 34) support distinct physiological roles. Consistent with these notions, AS160- and Tbc1d1-deficient mice exhibit similar changes in glucose uptake and GLUT4 expression, but in different skeletal muscles, and adipose tissues are affected only in AS160-knockout mice (3, 9, 14, 17, 30, 32, 36). As described here and by Chadt et al. (3), AS160^−/−/Tbc1d1^−/− mice combine these defects. The phenotypes of AS160^−/−, Tbc1d1^−/−, and AS160^−/−/Tbc1d1^−/− mice published by the different groups are summarized in Table 1. Specifically, AS160^−/− mice have abnormalities in glucose uptake in adipocytes and soleus muscle (17, 36) and Tbc1d1^−/− mice in EDL muscle (4, 9, 14, 30, 32), whereas AS160^−/−/Tbc1d1^−/− mice exhibit changes in glucose uptake in adipocytes, soleus, and EDL muscles (Fig. 2, A–C) (3). GLUT4 expression is decreased in adipose tissues and in soleus and gastrocnemius muscles of AS160^−/− mice (17, 36), in all skeletal muscles except soleus of Tbc1d1^−/− mice (3, 14, 30), and in all examined skeletal muscles and adipose tissues of AS160^−/−/Tbc1d1^−/− mice (Fig. 2D) (3). With regard to in vivo glucose uptake, increased basal and insulin-stimulated glucose uptake in AS160^−/−/Tbc1d1^−/− brown adipose tissue and tibialis anterior muscles (Figs. 5, A and D) combine similar abnormalities in AS160^−/− brown adipose tissue (17) and Tbc1d1^−/− tibialis anterior (14, 32). In gastrocnemius muscle that normally expresses both AS160 and Tbc1d1 relatively well (14, 17, 33), glucose uptake and GLUT4 expression abnormalities of AS160^−/− or Tbc1d1^−/− mice mix in AS160^−/−/Tbc1d1^−/− mice. Glucose uptake in AS160^−/−/Tbc1d1^−/− gastrocnemius is decreased under basal and insulin-stimulated conditions (Fig. 5C) similarly to AS160^−/− gastrocnemius (17). But in AS160^−/−/Tbc1d1^−/− gastrocnemius, glucose uptake normally responds to insulin (Fig. 5C) like in Tbc1d1^−/− (14) but unlike in AS160^−/− gastrocnemius (17). GLUT4 is decreased by ∼20 and 40% in male AS160^−/− (17) and Tbc1d1^−/− gastrocnemius (14), respectively, and 60% in AS160^−/−/Tbc1d1^−/− gastrocnemius (Fig. 2D). Abnormalities in GLUT4 plasma membrane localization observed in single AS160^−/− and Tbc1d1^−/− mice also add up in large skeletal muscles. Subcellular fractionation of pooled gastrocnemius and quadriceps revealed increased relative plasma membrane association of GLUT4 under basal conditions in AS160^−/− (17), Tbc1d1^−/− (14), and AS160^−/−/Tbc1d1^−/− mice (Fig. 2G), but the extent of the relative increases differed. In AS160^−/− muscles it was increased to insulin-stimulated wild-type levels (17), in Tbc1d1^−/− muscles it was increased compared with basal wild type but was lower than in insulin-stimulated wild type (14), and in AS160^−/−/Tbc1d1^−/− muscles it was higher than in insulin-stimulated wild type (Fig. 2G). Whereas relative plasma membrane GLUT4 did not change further in AS160^−/− (17) or AS160^−/−/Tbc1d1^−/− muscles in response to insulin (Fig. 2G), it increased further to insulin-stimulated wild-type levels in Tbc1d1^−/− muscles (14). All of these observations support nonredundant functions for the two Rab GAPs, with a unique function for AS160 in adipocytes and split functions for AS160 and Tbc1d1 in skeletal muscles. In some skeletal muscles, such as soleus, and tibialis anterior and EDL muscles, AS160 or Tbc1d1 are key. In the large skeletal muscles, both gastrocnemius and quadriceps play roles. Whether in the large mixed fiber type muscles each Rab GAP functions in distinct muscle fibers or controls different intracellular GLUT4 compartments that may be sensitive to different stimuli needs to be determined.

Table 1.

Comparison of single- and double-AS160 (Tbc1d4)- and Tbc1d1-deficient mice

			AS160−/−	Tbc1d1−/−	AS160−/−/Tbc1d1−/−
Glucose uptake ex vivo	Adipocytes		B↑ (17, 36); B→ (3)	B→ (3, 14)	B↑ (TS); B→ (3)
			I↓ (3, 17, 36)	I→ (3, 14)	I→ (TS); I↓ (3)
	Soleus		B→ (3, 17, 36)	B→ (3, 14, 30, 32)	B→ (3, TS)
			I↓ (3, 17, 36)	I→ (3, 14, 30, 32)	I↓ (3, TS)
	EDL		B→ (3, 17, 36)	B→ (3, 9, 14, 30, 32)	B→ (3, TS)
			I→ (3, 17, 36)	I↓ (3, 9, 14, 30, 32)	I↓ (3, TS)
GLUT4 expression	WAT		↓ (3, 17, 36)	→ (3, 14, 30)	↓ (3, TS)
	BAT		↓ (17)	→ (14)	↓ (TS)
	Gastro		↓ (17, 36); → (3)	↓ (14); → (3)	↓ (3, TS)
	Quad		↓ (17, 36); → (3)	↓ (3, 14, 30)	↓ (3)
	TA		→ (3, 36)	↓ (3, 9, 14, 30)	↓ (3, TS)
	Soleus		↓ (3, 17, 36)	→ (3, 14, 30, 32)	↓ (3, TS)
	EDL		→ (3, 17, 36)	↓ (3, 9, 14, 30, 32)	↓ (3,TS)
GLUT4 subcellular location relative to total	PM WAT		B↑ (17); I↓ (17); no IR (17); reg IR (36)	ND	ND
	PM gastro/quad		B↑ (17); I→ (17); no IR (17)	B↑ (14); I→ (14); IR↓ (14)	B↑ (TS); I↑ (TS); no IR (TS)
	CS adipocytes		B↑ (17); I↑ (17); reg IR (17)	ND	ND
	CS soleus		B↓ (17); I→ (17); reg IR (17)	B→ (3); I→ (3); reg IR (3)	ND
	CS EDL		ND	B→ (14); B↑ (3); I→ (3, 14); reg IR (14); no IR (3)	ND
Whole body glucose homeostasis	Glucose tolerance		→ (3, 17, 36); ↓ (TS)	→ (3, 14, 30, 32, TS)	↓ (3, TS)
	Insulin tolerance		↓ (17, 36, TS); → (3)	→ (3, 14, 30, 32, TS)	↓ (3, TS)
	Glucose fast		↓ (17, 36, TS) → (3)	→ (3, 14, 32, TS)	↓ (TS); → (3)
	Insulin fast		↓ (17, TS) → (3, 36)	→ (3, 14, 30); ↑ (32); ↓ (TS)	→ (3, TS)
	Glucose disposal in vivo	Total	B→ (36); I↓ (36)	B→ (32); I→ (32)	ND
		Gastro/quad	B→ (17); I↓ (17, 36)	B→ (14); I→ (14, 32)	B↓ (TS); I↓ (TS)
		TA	B→ (17); I↓ (17)	B↑ (14); I↑ (14, 32)	B↑ (TS); I↑ (TS)
		BAT	B↑ (17); I↑ (17)	B↓ (14); I→ (14, 32)	B↑ (TS); I↑ (TS)
		Liver	B→ (17); I↓ (17)	B→ (14); I→ (14, 32)	B↓ (TS); I→ (TS)
Energy homeostasis	Body weight#		→ (17, 36, TS); ↓ (3)	→ (14, 30, 32, TS); ↓ (3, 9)	↓ (3, TS)
	Total fat mass		↓ (TS)→ (3)	→ (3, 9, 14, 30, TS)	→ (3, TS)
	EE		↑ (TS)↓ (36)	↑ (14)	↑ (TS)
	RER		↓ (3); → (TS)	↓ (3); → (14)	↓ (3); → (TS)
	Food intake		↑ (TS); → (36)	↑ (14)	↑ (TS)
	Activity		↓ (TS); → (36)	→ (14)	→ (TS)

Major characteristics of single- and double-AS160 (Akt substrate of 160 kDa; Tbc1d4)- and Tbc1d1-deficient mice described in this study (TS) and in previous publications (3, 9, 14, 17, 30, 32, 36) are listed. EDL, extensor digitorum longus; gastro, gastrocnemius; quad, quadriceps; PM, plasma membrane; CS, cell surface; TA, tibialis anterior; WAT, white adipose tissue; BAT, brown adipose tissue; EE, energy expenditure; RER, respiratory exchange ratio; B, basal; I, insulin stimulated; IR, insulin response; no IR, no insulin response; reg IR, normal insulin response; ND, not determined. Numbers inside parentheses following given changes refer to references in which measurements were reported. Only data for male mice fed a standard chow diet were included. Changes in parameters are indicated as follows: ↑increase, ↓decrease, →no change when compared with wild type. #Only body weight measurements compared at single time points were included.

This and previous studies agree on the location of glucose uptake defects in isolated adipocytes and skeletal muscles in single- and double-AS160 and Tbc1d1-knockout mice. However, the nature and magnitude of the defects differ between studies, leading to different conclusions on AS160 and Tbc1d1 functions. In isolated adipocytes, normal (3) and increased (Fig. 2A) (17, 36) basal glucose uptake and no change (Fig. 2A), a small decrease (17), or a large decrease (3, 36) in insulin-stimulated glucose uptake were observed in AS160^−/− and AS160^−/−/Tbc1d1^−/− mice. In isolated AS160^−/−, Tbc1d1^−/− or AS160^−/−/Tbc1d1^−/− soleus and/or EDL muscle basal glucose uptake were normal (Fig. 2, B and C) (3, 14, 17, 36) or decreased (9, 30, 32), and insulin increased glucose uptake significantly, albeit to a lesser extent than in wild-type mice (Fig. 2, B and C) (9, 14, 17, 32). However, two previous studies observed no significant stimulation of glucose uptake in response to insulin in soleus and/or EDL of AS160^−/−, Tbc1d1^−/−, and/or AS160^−/−/Tbc1d1^−/− mice (3, 36). The different abnormalities in glucose uptake occur in the presence of similar decreases in GLUT4 levels and normal GLUT1 amounts as well as normal insulin-elicited Akt phosphorylation (3, 36). It is very likely that discrepancies are due to differences in strain backgrounds, but they clearly affect conclusions on the roles of AS160 and Tbc1d1. The similar decreases in insulin-stimulated glucose uptake we observed in isolated soleus and EDL muscles of single- and double-AS160 and Tbc1d1-knockout mice on the same C57BL/6N background (Fig. 2, B and C) (14, 17) corroborate nonredundant roles for AS160 and Tbc1d1 in different skeletal muscles. The lack of an insulin response previously observed in skeletal muscles of single- and double-AS160 and Tbc1d1-knockout mice on a C57BL/6J background did not allow such discrimination (3, 36), but it suggested that glucose uptake regulation in soleus and EDL is dependent entirely on AS160 and/or Tbc1d1. GLUT4 cell surface labeling of Tbc1d1^−/− EDL further supported this conclusion (3). In contrast, our earlier GLUT4 cell surface labeling studies suggest that impaired insulin-stimulated glucose uptake in isolated AS160^−/− soleus and Tbc1d1^−/− EDL muscles is due to decreased total GLUT4 levels in the presence of normal insulin-stimulated GLUT4 cell surface exposure (14, 17). But as this and our earlier studies clearly demonstrate, AS160 and Tbc1d1 function in intracellular GLUT4 retention in large skeletal muscles (Fig. 2, E–G) (14, 17).

Similar changes in glucose uptake in isolated adipocytes, EDL, and soleus muscles of single- and double-AS160 and Tbc1d1-knockout mice suggest comparable roles for AS160 and Tbc1d1 in different cell types and muscles (3). Although we agree that AS160 and Tbc1d1 play similar, although not identical, roles in skeletal muscles, and AS160 controls intracellular GLUT4 retention in both adipocytes and skeletal muscle, our earlier study with AS160^−/− mice implied different roles for AS160 in the regulation of glucose uptake and GLUT4 cell surface localization between adipocytes and skeletal muscles (17). Control of GLUT4 cell surface exposure is AS160 independent in skeletal muscles but at least partially AS160 dependent in adipocytes (14, 17). The differential changes in glucose uptake in isolated AS160^−/−/Tbc1d1^−/− adipocytes and soleus muscle in this study (Fig. 2) further corroborate differential AS160 functions in the two cell types. Further analysis of the molecular functions of AS160 and Tbc1d1 and differences thereof between skeletal muscles and adipocytes will be necessary to resolve current discrepancies.

Our in vivo studies assessing whole body glucose homeostasis and insulin tolerance demonstrate similar phenotypes for AS160^−/−/Tbc1d1^−/− and AS160^−/− mice (Figs. 3 and 4). This applies to reduced blood glucose levels under random-fed and fasting conditions for males and females and to plasma insulin levels under random-fed conditions for males. Also, male AS160^−/− and AS160^−/−/Tbc1d1^−/− mice match for impaired glucose tolerance, and male and female AS160^−/− and AS160^−/−/Tbc1d1^−/− mice match for impaired insulin tolerance. Our observations are at least in part different from the previous study with single- and double-AS160 and Tbc1d1-knockout mice in that impaired glucose and insulin tolerance were observed only for AS160^−/−/Tbc1d1^−/− mice (3). The earlier study further describes normal glucose and insulin levels in AS160- and/or Tbc1d1-deficient mice under fasting and fed conditions but decreased blood glucose levels in AS160^−/−/Tbc1d1^−/− under fed conditions (3). Whereas we observed impaired insulin secretion during the glucose tolerance test in AS160^−/−/Tbc1d1^−/− mice (Fig. 4E), normal insulin levels were reported in the earlier study (3). Most likely differences in strains (C57BL/6J vs. C57BL/6N) (15) and in environmental conditions account for discrepant observations (see below for further details). But our findings suggest that AS160 predominantly determines the in vivo phenotype of AS160^−/−/Tbc1d1^−/− mice as it relates to glucose homeostasis and insulin tolerance. Note that this occurs in the presence of different abnormalities in glucose disposal in tibialis anterior and liver but similarly impaired glucose uptake in gastrocnemius of AS160^−/− and AS160^−/−/Tbc1d1^−/− mice (Fig. 5) (17). A key role for AS160 in the regulation of glucose homeostasis is consistent with studies in humans with AS160 truncation mutations (8, 22). Human carriers of such mutations exhibit strikingly similar defects to those of AS160-deficient mice: impaired glucose tolerance, peripheral insulin resistance consistent with impaired glucose uptake in skeletal muscle, decreased fasting blood glucose, and decreased GLUT4 expression in skeletal muscle concomitant with drastically increased incidence of type 2 diabetes (8, 22).

The measurements of body weight and body composition also demonstrate that AS160^−/−/Tbc1d1^−/− mice match AS160^−/− mice better. But body weights of male AS160^−/−/Tbc1d1^−/− mice track with reduced weights for both male AS160^−/− and Tbc1d1^−/− mice (Fig. 1A) (3). The lower body weights are consistent with increased energy expenditure of AS160^−/−/Tbc1d1^−/−, AS160^−/−, and Tbc1d1^−/− mice (Fig. 6A) (14), which may only partially be compensated by increased food intake (Fig. 6B) (14) and decreased activity (AS160^−/− mice only; Fig. 6D). The earlier study with single- and double-AS160 and Tbc1d1-knockout mice on a chow diet did not evaluate energy expenditure, food intake, or activity (3). Previous studies showed increased resting metabolic rates in high-fat diet-fed Tbc1d1^−/− mice (9) but decreased oxygen consumption and heat production in chow-fed AS160^−/− mice (36). We did not observe any abnormalities in RER in male single- or double-AS160 and Tbc1d1-knockout mice (Fig. 6) (14). But earlier, we observed decreased RER in female Tbc1d1^−/− mice and increased fatty acid oxidation in male and female Tbc1d1^−/− mice (14). Decreased RER together with increased fatty acid oxidation in skeletal muscle was reported in male AS160^−/−, Tbc1d1^−/−, and AS160^−/−/Tbc1d1^−/− mice (3). Thus, deletion of AS160 and Tbc1d1 may again have different effects on whole body energy homeostasis, depending on strain background. How AS160 and Tbc1d1 control body weight, energy expenditure, and fatty acid oxidation is currently not known. But Tbc1d1 mutations and single nucleotide polymorphisms are associated with body weight and body composition changes in animals (4, 12, 37, 38) and with obesity in humans (16, 20, 23, 31).

While studying male and female single- and double-AS160 and Tbc1d1-knockout mice, we observed sex-specific changes in in vivo phenotypes. Body weights of male mice with single and double deletions of AS160 and Tbc1d1 were decreased (Fig. 1A) (3). In contrast for females, only Tbc1d1^−/− mice had lower body weights (Fig. 1A). Body composition was changed in opposite directions for AS160^−/− and AS160^−/−/Tbc1d1^−/− male and female mice; fat mass was decreased in male and increased in female mice (Fig. 1B). In males, glucose tolerance was impaired in AS160^−/− and AS160^−/−/Tbc1d1^−/− mice; in females, glucose tolerance was impaired only in AS160^−/− mice (Fig. 4A). Insulin levels did not decrease concomitantly with lower glucose levels in female AS160^−/− and AS160^−/−/Tbc1d1^−/− mice but did so in male mice (Fig. 3). The reason for the sex-specific differences is currently not known, but previously, we made similar observations when we characterized male and female AS160- and Tbc1d1-deficient mice (14, 17). As we speculated then, inherent differences in fat and muscle mass between male and female mice together with adipose tissue and skeletal muscle-specific expression patterns of AS160 and Tbc1d1 most likely contribute to this (14, 17). So far, we have not found any differences in abnormalities at the cellular level between male and female mice (14, 17). Similarly, subtle differences in body composition due to dissimilar genetic backgrounds and differences in environmental conditions (food composition, light-dark cycles, temperature) may also account for differences in phenotypes for male AS160 and/or Tbc1d1-deficient mice between different studies. The different genotypes of parents, single heterozygous for AS160 deletion and double heterozygous for AS160 and Tbc1d1 deletions in the previous (17) and current report, respectively, may also cause subtle differences in body composition of male AS160^−/− offspring, leading to the observed discrepancy in glucose tolerance. Glucose tolerance was impaired in AS160^−/− mice in this study (Fig. 4A) but not in the previous study (17).

In conclusion, our study provides evidence that double-knockout mice for AS160 and Tbc1d1 combine defects of single-knockout mice, supporting distinct roles for the two Rab GAPs. AS160 and Tbc1d1 clearly regulate GLUT4 trafficking and expression and thus glucose uptake mainly in ways consistent with their differential expression in different cells. At the cellular level they most likely share similar roles but may respond to different stimuli. However, AS160 and Tbc1d1 functions are not limited to regulating GLUT4. Although GLUT4 is reduced similarly in AS160^−/−/Tbc1d1^−/− mice and in mice heterozygous for GLUT4 deletion at the whole body level (GLUT4^+/− mice) (29), they show different abnormalities. Male GLUT4^+/− mice develop fed hyperglycemia and hyperinsulinemia and have normal body weights and fasting glucose and insulin, a phenotype clearly different from AS160^−/−/Tbc1d1^−/− mice (Ref. 3 and the present study). Furthermore, AS160 and Tbc1d1 are not only expressed in muscle and adipose tissues, where GLUT4 is abundant, but are also found in other tissues (2, 4, 17, 25, 33). Indeed, different reports have demonstrated roles for AS160 in the regulation of the trafficking of various membrane proteins in muscle and fat cells as well as other cell types, including the insulin-regulated aminopeptidase (11), the epithelial sodium channel (19), the sodium-potassium adenosine trisphospatase Na⁺/K⁺-ATPase (1), and the fatty acid translocase FAT/CD36 (27). The definition of the exact molecular functions of AS160 and Tbc1d1 in the regulation of GLUT4 and other membrane proteins and their unique roles in cellular and whole body homeostasis will be key topics for future research. This research will further determine whether selective targeting of AS160 and Tbc1d1 may hold promise for future treatments of obesity, diabetes, and other diseases.

GRANTS

This study was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant R01-DK-81471 to S. R. Keller.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.R.H., N.N.W., and S.R.K. conception and design of research; S.R.H. and N.N.W. performed experiments; S.R.H., N.N.W., and S.R.K. analyzed data; S.R.H., N.N.W., and S.R.K. approved final version of manuscript; S.R.K. interpreted results of experiments; S.R.K. prepared figures; S.R.K. drafted manuscript; S.R.K. edited and revised manuscript.