Differential Role of Insulin/IGF-1 Receptor Signaling on Muscle Growth and Glucose Homeostasis

Summary

Insulin and IGF-1 are major regulators of muscle protein and glucose homeostasis. To determine how these pathways interact, we generated mice with muscle-specific knockout of IGF-1 receptor (IGF1R) and insulin receptor (IR). These MIGIRKO mice showed >60% decrease in muscle mass. Despite a complete lack of insulin/IGF-1 signaling in muscle, MIGIRKO mice displayed normal glucose and insulin tolerance. Indeed, MIGIRKO mice showed fasting hypoglycemia and increased basal glucose uptake. This was secondary to decreased TBC1D1 resulting in increased Glut4 and Glut1 membrane localization. Interestingly, overexpression of a dominant-negative IGF1R in muscle induced glucose intolerance in MIGIRKO animals. Thus, loss of insulin/IGF-1 receptor signaling impairs muscle growth, but not whole-body glucose tolerance due to increased membrane localization of glucose transporters. Nonetheless, presence of a dominant-negative receptor, even in the absence of functional IR/IGF1R, induces glucose intolerance, indicating that interactions between these receptors and other proteins in muscle can impair glucose homeostasis.

Introduction

Skeletal muscle insulin resistance is a prominent feature of type 2 diabetes, which precedes and predicts the development of disease in high-risk populations (Martin et al., 1992). In humans, up to 80% of the glucose infused during a hyperinsulinemic euglycemic clamp is disposed into muscle. However, genetic manipulation of insulin signaling specifically in muscle of mice has shown little effect on whole body glucose metabolism. For example, genetic deletion of the insulin receptor (IR) specifically in skeletal muscle of mice (MIRKO) did not cause dysglycemia or diabetes, although it did result in hypertriglyceridemia and mild obesity (Bruning et al., 1998). On the other hand, overexpression of a kinase-deficient insulin receptor (IR) in muscle of mice led to glucose intolerance with increased circulating insulin and triglyceride levels (Moller et al., 1996). Likewise, generation of a mouse that highly overexpressed a dominant negative, kinase-dead IGF-1 receptor (IGF1R) in muscle (MKR) develop severe glucose intolerance, insulin resistance, and diabetes (Fernandez et al., 2001). Further study revealed that expression of the MKR allele impairs both insulin and IGF-1 signaling in muscle due to hybrid receptor formation, suggesting that the normal glucose tolerance in MIRKO mice might be due to IGF1R compensating for loss of IR signaling in muscle.

In addition to glucose uptake and metabolism, insulin and IGF-1 signaling affect muscle growth and protein turnover (Schiaffino and Mammucari, 2011; Meek et al., 1998). Both insulin and IGF-1 have been shown to stimulate muscle protein synthesis (Fulks et al., 1975; Rommel et al., 2001) and inhibit protein degradation via the ubiquitin-proteasome and autophagy-lysosome pathways (Mammucari et al., 2007; Sandri et al., 2004). Indeed, IGF-1 treatment is sufficient to cause muscle hypertrophy via Akt activation of mTOR and inhibition of GSK3β (Rommel et al., 2001). On the other hand, deletion of IGF1R in muscle only modestly impairs muscle growth, suggesting alternative pathways can induce muscle growth possibly via IR (Mavalli et al., 2010).

At a molecular level, insulin and IGF-1 signal through highly homologous tyrosine kinase receptors, which are virtually ubiquitously expressed in mammals. IR and IGF1R specifically bind their respective ligands at physiological concentrations. However at high concentrations, each ligand can bind and initiate signaling with the opposite receptor. Both IR and IGF1R then initiate intracellular signaling via similar cascades beginning with tyrosine phosphorylation of insulin receptor substrates (IRS) which leads to activation of the phosphatidylinositol 3-kinase (PI3K)/Akt pathway, as well as other downstream signals (Taniguchi et al., 2006). In addition, IR and IGF1R interact with Src homology and collagen domain protein (Shc) to activate MAP kinase pathways. Signaling via these two pathways leads to a broad range of cellular effects on growth, proliferation, and metabolism.

In muscle, insulin-stimulated glucose uptake has been extensively studied. IR mediated activation of Akt leads to phosphorylation of AS160 and TBC1D1 to facilitate translocation of vesicles containing the Glut4 glucose transporter to the plasma membrane where they fuse leading to increased glucose uptake into the cell (Klip, 2009). Not surprisingly, knockout of Glut4 in muscle of mice leads to insulin resistance and hyperglycemia (Zisman et al., 2000). However, basal glucose uptake, i.e. that which occurs in the absence of insulin, has been ascribed to other glucose transporters, such as Glut1, which have higher constitutive association with the sarcolemma (Scheepers et al., 2004; Marette et al., 1992; Wang et al., 1996). Exercise is also an important factor in glucose transport inducing AMP dependent kinase (AMPK) activation and Glut4 vesicle translocation via phosphorylation of AS160 and TBC1D1, which is independent of insulin (Fujii et al., 2006; Koh et al., 2008).

To investigate to what extent insulin and/or IGF-1 receptor signaling pathways control glucose metabolism and protein homeostasis, we have deleted either IR, IGF1R or both in skeletal muscle using genetic recombination. While mice with single receptor deletions show little or no change in glucose homeostasis or muscle mass, mice with combined loss of IR and IGF1R in muscle (MIGIRKO) display dramatically decreased muscle mass and fiber size. Nonetheless, MIGIRKO mice show normal glucose and insulin tolerance, and even have fasting hypoglycemia due to enhanced basal glucose uptake into muscle secondary to increased expression and translocation of glucose transporters. Surprisingly, when MIGIRKO mice were crossed to mice carrying a dominant negative IGF1R, the resultant mice still developed glucose intolerance and dyslipidemia. Thus, combined loss of IR and IGF1R in muscle dramatically impairs muscle growth, but glucose tolerance is maintained by enhanced basal glucose transport. The induction of glucose intolerance in these mice by expression of a dominant negative IGF1R indicates that the dominant negative receptor can interact with other proteins on the cell to modify metabolic regulation.

Materials and Methods

Animal care and use

Animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC). Male mice were used for all studies other than MKR-MIGIRKO studies which used both males and females. Muscle IR knockout (M-IR^−/−), muscle IGF1R knockout (M-IGF1R^−/−), and combined muscle IR/IGF1R knockout (MIGIRKO) mice were each generated by crossing mice carrying the Cre recombinase driven by a skeletal muscle actin promoter, ACTA1-Cre (Jackson Laboratories Stock Number: 006149), with mice carrying both floxed insulin and IGF1R receptor alleles (Boucher et al., 2012), i.e., IR^lox/loxIGF1R^lox/lox, then maintained as separate colonies. Since no differences were observed among the IR^lox/lox, IGF1R^lox/lox, IR^lox/+IGF1R^lox/lox, and IR^lox/loxIGF1R^lox/lox mice, the results on controls were pooled. MKR transgenic mice, which have the murine MCK promoter directing expression of the human IGF1R gene containing the K1003R mutation, were purchased from Jackson laboratories (Stock Number: 016618) and have been previously described (Fernandez et al., 2001). See Supplemental Experimental Procedures.

In vivo Glucose uptake

Glucose uptake into tissue was measured by intravenous injection with either saline or 1 mU/g insulin in combination with 0.33 μCi [¹⁴C]2-deoxyglucose/g administered via the retro-orbital sinus. After 45 minutes, [¹⁴C] levels in blood and tissue were determined. For full details, see Supplemental Experimental Procedures.

Ex Vivo Muscle Glucose Uptake

Glucose uptake was measured in extensor digitorum longus (EDL) and soleus strips as previously described (Hayashi et al., 1998). Briefly, mice were fasted starting at 22:00, and muscle harvested the next day between 10:00 and 13:00. EDL and isolated soleus strips were incubated with resting tension in the basal state or stimulated with 5 mU/ml of insulin for 40 minutes with addition of [³H]-2-deoxyglucose for the last 10 minutes.

Plasmid transfection and Intravital Microscopy

The construction of GLUT4-EGFP (Lauritzen et al., 2008; Lauritzen et al., 2002) and TBC1D1 (An et al., 2010) have been described previously. Mice were transfected using the Helios gene gun system (Bio-Rad) as previously described (Lauritzen et al., 2002; Lauritzen, 2010; Lauritzen and Schertzer, 2010). Briefly, mice were anaesthetized with 90 mg/kg pentobarbital; the skin was opened to expose the vastus lateralus to the bombardment of DNA/gold particles using the gene gun. Five days after transfection, vastus lateralus in random-fed mice were imaged. Quantification of Glut4-EGFP vesicular depots above 1 μm in size and measurement of average GLUT4-EGFP area were generated using MetaMorph software. See Supplemental Experimental Procedures.

Physiological and Analytical Measurements

Comprehensive Laboratory Animal Monitoring Systems (CLAMS, Columbus instruments) and DEXA measurements were performed at the Joslin Diabetes Research Center (DRC) core. Glucose tolerance tests and insulin tolerance tests were performed as previously described (Bruning et al., 1998). Insulin levels were measured using a mouse insulin ELISA kit (Crystal Chem), triglycerides were measured using a triglyceride assay kit (Abnova), and FGF21 serum levels were measured with a mouse/rat ELISA kit (R&D Systems, Cat# MF2100). In vivo insulin and IGF-1 signaling were performed in anesthetized, overnight-fasted mice by injecting either 5 U of regular insulin or 1 mg/kg of human IGF-1 (Sigma) via inferior vena cava (IVC) and 10-15 minutes later harvesting of tissues and snap freezing in liquid nitrogen. Lactate levels were measured in gastrocnemius at the Mayo Clinic Metabolomics Resource Core using TOF mass spectrometry.

Statistical Analyses

All data are presented as mean ± standard error of the mean (SEM). Student's t-test was performed for comparison of two groups, and ANOVA was performed for comparison of 3 or more groups to determine significance.

Results

Muscle specific deletion of IR and IGF1R decreases muscle growth and leads to early demise

To generate mice with skeletal muscle specific deletion of IR and IGF1R, we crossed mice which express the Cre recombinase under the control of human skeletal muscle actin promoter (ACTA1-Cre) with mice harboring floxed IR and IGF1R alleles. Previous attempts at deleting IR and IGF1R in muscle using Cre under the muscle creatine kinase (MCK) promoter allowed for expression in the heart, as well as skeletal muscle, and led to death within 21 days from cardiac failure (Laustsen et al., 2007). Cre expression in ACTA1-Cre mice is more restricted to skeletal muscle (Miniou et al., 1999), allowing successful generation of mice with muscle specific deletion of either IR (M-IR^−/−), IGF1R (M-IGF1R^−/−), or both IGF1R and IR (MIGIRKO) (Figure S1 and Table S1). We have named mice which harbor IR^lox/lox alleles and the ACTA1-Cre transgene as M-IR^−/− mice in order to distinguish them from the MIRKO mouse which was created using MCK-Cre (Bruning et al., 1998). Genomic DNA isolated from M-IR^−/− and MIGIRKO muscle (which also contains vascular cells, fibroblasts and satellite cells/myoblasts) showed a 50% recombination of IR^lox locus by quantitative RT-PCR (qPCR). IGF1R^lox locus was similarly recombined by 50% in M-IGF1R^−/− and MIGIRKO muscle without any change in liver DNA (Figure S1A-S1B). M-IR^−/− and MIGIRKO mice displayed a 90% decrease of IR mRNA expression by quantitative RT-PCR (qPCR) in muscle, while M-IGF1R^−/− and MIGIRKO mice showed a 50-60% reduction in IGF1R mRNA (Figure S1C). These changes in IR and IGF1R mRNA expression correlated well with decreases in protein levels corresponding to genotype (Figure 1A).

Figure 1. Deletion of IR and IGF1R in muscle dramatically decreases muscle size and survival.

Western blot of insulin receptor-β (IR-β) and IGF-1 receptor-β (IGF1R-β) was measured in quadriceps from mice with muscle-specific deletion of insulin receptor (M-IR^−/−), IGF-1 receptor (M-IGF1R^−/−), or both IGF-1 receptor and insulin receptor (MIGIRKO) (A). Representative profile and hindlimb dissection of control and MIGIRKO littermate mice (B). Body weight was measured weekly in control and MIGIRKO mice (C) (n=7-16). Survival curve of MIGIRKO mice compared to control, M-IR^−/−, and M-IGF1R^−/−, represented as “All other genotypes” (D) (n=14-20 per group). Body weight was measured at time of sacrifice of control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice (E) (n=5-8 knockout mice and pooled 20 controls). Representative muscle dissection from control and MIGIRKO mice (F). Dissected muscle weights measured from control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice (G) (n=5-9 knockout mice and pooled 22 controls). Representative cross section of TA muscle stained for SDH to demonstrate oxidative (purple) and glycolytic (gray/white) muscle fibers from control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice (H). Quantification of total number of muscle fibers normalized to cross sectional area of TA sections in mm² (I). Quantification of total oxidative and glycolytic fibers per TA section (J) (n=3-6 per group). (*-p<0.05, **-p<0.01 vs. control, ANOVA) All mice were 11-15 weeks old. Quad – quadriceps, TA – tibialis anterior, EDL – extensor digitorum longus, Sol – soleus, Gast – Gastrocnemius, SDH – Succinate Dehydrogenase. See also Figures S1-S2 and Table S2.

MIGIRKO mice showed an obvious growth phenotype with decreased body weight as early as three weeks of age (Figure 1B-1C). By 7 to 10 weeks of age, MIGIRKO mice exhibited severe muscle atrophy with spinal deformities and obvious kyphosis (Figure 1B). These mice progressed to have breathing difficulties and died between 15 and 25 weeks of age, most likely of respiratory failure (Figure 1D). By contrast M-IR^−/− or M-IGF1R^−/− mice had normal body weight and skeletal appearance and lived normally up to 52 weeks of age. By DEXA scanning and assessment of tissue weight at sacrifice, the decreased body weight in MIGIRKO mice could be attributed almost entirely to a loss of muscle mass with 59-68% reductions in individual muscle weights and a 32% decreased in total lean mass (Figure 1E-1G and Table S2). There was also a 9% loss of lean mass and a decrease in muscle weights in M-IR^−/−, whereas M-IGF1R^−/− had normal lean mass and muscle weights. In addition, M-IGF1R^−/− and MIGIRKO mice displayed a loss of fat mass (Figure S1D and Table S2). The cause of this loss of fat mass is unknown, but was not due to recombination of IR or IGF1R in fat or other tissues (Figure S1F-S1G), suggesting some form of communication between muscle and fat that is dependent on IGF-1 action. None of the changes in body weight and composition were attributable to dwarfism or any decrease in linear growth as assessed by femur length (Table S2).

Histologic analysis of tibialis anterior (TA) muscle using succinate dehydrogenase (SDH) staining revealed marked atrophy/lack of hypertrophy of MIGIRKO muscle fibers (Figure 1H). The total cross sectional area of the TA was mildly reduced in M-IR^−/− and markedly reduced in MIGIRKO mice (Figure S2). Quantification of fiber number normalized to the area of the TA cross section in square mm revealed that the decrease in muscle size in these two strains was due to atrophy and not a loss of muscle fibers (Figure 1I). Lastly, while glycolytic fiber number did not change, M-IGF1R^−/− and MIGIRKO mice showed increased numbers of oxidative fibers (Figure 1J and S2C).

MIGIRKO mice demonstrate normal glucose and insulin tolerance, but increased basal glucose uptake in muscle and fasting hypoglycemia

Glucose levels were unchanged in randomly fed animals, but after 16 hours of fasting, glucose levels were 35% lower in MIGIRKO mice compared to controls (Figure 2A). This occurred with no significant changes in either fasting or refed insulin and triglyceride levels (Figure 2B-2C). As expected, insulin and IGF-1 signaling was abolished in skeletal muscle from MIGIRKO animals injected with either 5 units of insulin or 1 mg/kg IGF-1 via inferior vena cava (Figure 2D and S3A). Insulin signaling was normal in M-IGF1R^−/− and blunted in M-IR^−/− (Figure S3B). However, upon western blotting of muscle extracts, MIGIRKO animals displayed an unexpected 3- to 10- fold increase in the protein levels of IRS-1, IRS-2, Akt1 and Akt2. This dramatic increase was not seen at the mRNA level, although mRNA for IRS-2 and Akt2 were increased in MIGIRKO by 2.2- and 1.6-fold, respectively (Figure S3C). This increase in protein levels was associated with a marked increase in the amount of phosphorylated Akt in the basal state, as well as an increase in downstream phosphorylation of GSK3β, FoxO1, and FoxO3a (Figure 2D and S3D-S3G). We hypothesized that other tyrosine kinases may constitutively activate the IRS-PI3K-Akt pathway in the absence of IR and IGF1R. Indeed, total EGFR levels are increased in MIGIRKO muscle (Figure S3H).

Figure 2. MIGIRKO mice display normal glucose tolerance, fasting hypoglycemia, and increased basal glucose uptake into muscle, despite abolished insulin signaling in muscle.

Blood glucose levels were measured in 8-10 week old MIGIRKO and control mice fasted overnight or randomly fed (A) (n=9-10). Insulin (B) and triglyceride (C) levels from 8-10 week-old MIGIRKO and control mice fasted overnight or refed for 4 hours (n=9-10). Insulin signaling was determined by western blot analysis in quadriceps muscle from 11-15 week-old MIGIRKO and control mice fasted overnight and treated with saline or insulin intravenously (D). Intraperitoneal glucose tolerance test (GTT) (E) and insulin tolerance test (ITT) (F) were performed in 8-10 week-old MIGIRKO and control mice (n=9-10). In vivo 2-deoxyglucose uptake was performed as described in Methods in control and MIGIRKO mice (G) (n=6-8 per group). (**-p<0.01 vs. control, student's t-test; #-p<0.05 vs. control with same treatment, †-p<0.05 vs. basal of same genotype, ANOVA) Quad – quadriceps, Gastroc – gastrocnemius, BAT – brown adipose tissue. See also Figure S3.

Despite the lack of IR and IGF1R in muscle, glucose tolerance and insulin tolerance in MIGIRKO were unchanged compared to controls (Figure 2E-2F). To determine the fate of glucose in MIGIRKO animals, we performed an in vivo glucose uptake assay under basal or insulin-stimulated conditions as described in Methods. Interestingly, basal glucose uptake was increased in MIGIRKO muscles to the level of insulin-stimulated control muscle, and did not increase further with insulin treatment (Figure 2G). In other tissues, such as heart and brown adipose tissue (BAT), basal and insulin-stimulated glucose uptake in MIGIRKO were similar to controls, consistent with the absence of recombination of IR or IGFR in these tissues (Figure S1F-S1G).

Deletion of IR and IGF1R in muscle paradoxically increases glucose transporter expression and membrane localization

To determine the relative contributions of IR and IGF1R signaling to insulin-stimulated glucose uptake, we measured glucose uptake in extensor digitorum longus (EDL) and soleus muscles in vitro. Of note, insulin signaling in these two muscle groups was similar to quadriceps (Figure S4A). While glucose uptake in response to insulin in EDL from M-IR^−/− was blunted, glucose uptake in the soleus of M-IR^−/− and in EDL or soleus of M-IGF1R^−/− was unchanged compared to control (Figure 3A-3B). As found in vivo, in vitro basal glucose uptake was increased in MIGIRKO muscle, and this was unresponsive to insulin stimulation. This increase in basal glucose uptake in MIGIRKO was associated with increased protein levels of the glucose transporters Glut1 and Glut4 (Figure 3C-3D). These changes occurred with no changes in Glut1 expression at the mRNA level, and a decrease in Glut4 mRNA expression (Figure 3E).

Figure 3. Deletion of muscle IR and IGF1R paradoxically increases glucose transporter expression and membrane localization.

Ex vivo 2-deoxyglucose uptake was measured in EDL (A) and soleus (B) from 8-10 week-old control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice (n=5 knockouts and 12 pooled controls). Glut1 (C) and Glut4 (D) total protein measured by Western blot in control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO quadriceps (n=4). Glut1 and Glut4 mRNA levels were measured in quadriceps from control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice by qPCR (E) (n=5-8). Glut4-EGFP was transfected into vastus lateralus muscle and visualized 5 days later as described in Methods (F) Bar = 10 μm (n=2). Glut1 and Glut4 levels in plasma membrane (PM) isolates from mixed hindlimb muscle (G) (n=3). Phosphorylation of AMPK^T172 was measured in quadriceps (H) (n=4). (*-p<0.05, **-p<0.01 vs. control; †-p<0.05 vs. basal of same genotype, ANOVA) See also Figure S4 and Table S1 and S3.

Since basal glucose uptake in vivo and in vitro was increased in MIGIRKO muscle and signaling downstream of Akt was significantly enhanced in the basal state, we hypothesized that Glut4 translocation was enhanced in fasted or unstimulated MIGIRKO muscle. To evaluate Glut4 localization in vivo, we utilized a method of intravital imagining of a Glut4-EGFP protein transiently transfected into superficial muscle fibers of vastus lateralus utilizing a gene gun approach (Lauritzen et al., 2002; Lauritzen et al., 2006). Glut4-EGFP remained in larger intracellular depots with minimal surface localization in fibers from control mice, yet MIGIRKO fibers displayed a diffuse pattern with dispersed GLUT4-EGFP vesicle depots and increased surface localization (Figure 3F). This pattern of diffuse fluorescence with increased surface localization seen in MIGIRKO fibers is consistent with the pattern of Glut4-EGFP seen after stimulation with insulin (Figure S4C-S4D) or muscle contractions (Lauritzen et al., 2006; Lauritzen et al., 2010). Muscle fractionation experiments confirmed increased levels of Glut4 and Glut1 in plasma membrane isolates from MIGIRKO muscle compared to controls (Figure 3G). Increased Glut4 translocation is consistent with our observation that signaling downstream of Akt was increased in muscle from fasted MIGIRKO mice (Figure 2D). MIGIRKO muscles also displayed increased phosphorylation of AMPK in the fed state (Figure 3H), indicating activation of this pathway. AMPK phosphorylation remained elevated even when the mice were fasted with modest elevations in p-ACC, but phosphorylation of neither protein changed in response to insulin in either control or MIGIRKO mice (Figure S4B). Despite increased basal glucose uptake and Glut4 membrane localization, lactate levels were actually decreased in MIGIRKO muscle and glycogen content was unchanged in M-IR^−/−, M-IGF1R^−/−, and MIGIRKO muscle compared to controls (Table S3).

Deletion of IR and IGF1R in muscle leads to suppression of TBC1D1, and re-expression of TBC1D1 leads to re-internalization of Glut4

To gain insight into the mechanism for enhanced glucose transporter translocation, we investigated the phosphorylation status of AS160 and TBC1D1, both of which participate in Glut4 translocation and glucose uptake in muscle. Consistent with previous reports (Taylor et al., 2008), we found AS160 to be more abundant in oxidative soleus muscle and TBC1D1 more abundant in EDL, a more glycolytic muscle (Figure S4A). Phosphorylation of AS160 and TBC1D1 in response to insulin showed no differences between control, M-IR^−/− and M-IGF1R^−/−, but basal phosphorylation of AS160 was increased in EDL and soleus from fasted MIGIRKO mice (Figure S4A), consistent with increased Akt phosphorylation and activation. Interestingly, while phosphorylation of the 160 kDa band using a phospho-Akt substrate antibody (PAS 160) and AS160^T642 were increased, total levels of TBC1D1 were decreased compared to controls in EDL, soleus, and quadriceps muscle (Figure 4A and S4A). Quantitative PCR analysis revealed a significant ~20% decrease in TBC1D1 mRNA from M-IR^−/− mice and a dramatic 72% decrease in MIGIRKO muscle (Figure 4B). Conversely, AS160 mRNA levels were increased 2.1- and 2.6-fold in M-IR^−/− and MIGIRKO muscle, respectively.

Figure 4. Deletion of IR and IGF1R in muscle leads to suppression of TBC1D1, and re-expression of TBC1D1 normalizes Glut4 localization.

AS160 phospho- and total protein, phospho-Akt substrate 160 kDa band (PAS 160), as well as TBC1D1 total protein were measured by Western blot in control and MIGIRKO quadriceps (A) (n=4). TBC1D1 and AS160 mRNA levels were measured in TA muscle from control, M-IR^−/−, M-IGF1R^−/−, and MIGIRKO mice by qPCR (C) (n=4-8). Glut4-EGFP was transfected into vastus lateralus muscle along with empty vector (EV) or with TBC1D1 and visualized 5 days later in the fed state (C) Bar = 10 μm. Quantification of average area of all Glut4 depots >1 μm from control and MIGIRKO mice transfected with Glut4-EGFP + EV and MIGIRKO transfected with Glut4-EGFP + TBC1D1. (D and E) (n=2 control and 3 MIGIRKO mice per group with 3-7 fibers each). (*-p<0.05, **-p<0.01 vs. control ANOVA; §-p<0.05 vs. MIGIRKO + EV, student's t-test) See also Figure S4 and Table S1.

We hypothesized that the observed decrease of total TBC1D1 levels along with increased AS160 phosphorylation in MIGIRKO muscle contributed to the re-localization of Glut4 to the sarcolemma. To test this hypothesis directly, we transiently re-expressed TBC1D1 in MIGIRKO muscle (Figure S4E) and determined the localization of Glut4-EGFP using intravital imaging of muscle fibers. Transient expression of Glut4-EGFP with an empty vector (EV) in control muscle fibers again showed large depots of Glut4, whereas a diffuse pattern with increased membrane localization was seen in MIGIRKO fibers (Figure 4C). Co-expression of TBC1D1 with Glut4-EGFP in MIGIRKO fibers normalized the pattern of Glut4 localization back to large intracellular depots. Quantification of the Glut4 depot area using MetaMorph software was performed as described in Methods. Total area of GLUT4-EGFP vesicle depots above 1 μm in size was lower in MIGIRKO animals compared to controls, reflecting re-localization of the depots to t-tubules and sarcolemma (Figure 4D-4E). Re-expression of TBC1D1 in MIGIRKO fibers significantly increased Glut4 depot area compared to MIGIRKO + EV in intramyofibrillar compartments (Figure 4D-4E), where 90% of Glut4 vesicles reside (Wang et al., 1996).

Increased energy expenditure in MIGIRKO mice is correlated with increased browning of subcutaneous white fat and increased glucose uptake into fat

MIGIRKO mice display fasting hypoglycemia and increased basal muscle glucose uptake even in the fasted state, but glucose tolerance and insulin tolerance were normal, indicating that whole body metabolic adaptations are likely to occur when insulin signaling is abolished in skeletal muscle. To better investigate these metabolic changes, we assessed metabolic actions of MIGIRKO mice using the Comprehensive Laboratory Animal Monitoring System (CLAMS). This revealed that MIGIRKO mice ate about 20% less food and drank less water per mouse than their controls, but when normalized to lean body mass, the water intake was unchanged and the food intake was actually increased (Figure 5A). Likewise, oxygen consumption and CO₂ production normalized to lean body mass were significantly increased in MIGIRKO mice during both day and night cycles (Figure 5B), while respiratory exchange ratio (RER) was unchanged (Figure 5C). This occurred despite a significant decrease in activity of the MIGIRKO mice as measured by number of times a horizontal axis was crossed (Figure 5D).

Figure 5. Increased energy expenditure in MIGIRKO mice is associated with browning of subcutaneous fat.

Daily food and water intake were measured in control and MIGIRKO animals and normalized per mouse or per milligram of lean body weight (LBW) (A) (n=9-10). Oxygen consumption (VO₂) and carbon dioxide production (VCO₂) per kg of LBW were measured using CLAMS metabolic cages (B) (n=9-10). Respiratory exchange ratio (RER) was measured in control and MIGIRKO mice(C) (n=9-10). Activity was measured as the number of times an animal crossed a horizontal laser (D) (n=9-10). Hematoxylin and Eosin staining was performed on inguinal subcutaneous white adipose tissue (sWAT) from control and MIGIRGO animals (E). Markers of brown adipose tissue (BAT) were measured by QPCR in sWAT and epididymal WAT (eWAT) (F). In vivo 2-deoxyglucose uptake under basal or insulin-stimulated conditions was measured in sWAT and eWAT from control and MIGIRKO mice (G). Fgf21 mRNA levels from quadriceps of MIGIRKO and control mice either randomly fed or fasted for 16 hours were measured by qPCR (H). Serum FGF21 levels in randomly fed control and MIGIRKO mice (I). (*-p<0.05, **-p<0.01 vs. control, student's t-test; #-p<0.05 vs. control with same treatment, †-p<0.05 vs. basal of same genotype, ANOVA). See also Table S1 and S4.

At sacrifice, subcutaneous white adipose tissue (sWAT) from the inguinal region of MIGIRKO mice was noted to be more brown in color than in normal mice, and hematoxalin and eosin staining of sWAT revealed large patches of adipocytes with multiloculated lipid droplets and abundant capillaries indicating browning of the white fat in MIGIRKO mice (Figure 5E). Consistent with browning, mRNA expression of brown adipose tissue (BAT) markers such as Ucp1, Dio2, and Elovl3 were increased by 3- to 5- fold in sWAT of MIGIRKO mice, but not in eWAT (Figure 5F), and basal glucose uptake was increased into sWAT, and insulin-stimulated glucose uptake was increased in both sWAT and eWAT (Figure 5G). Recent studies have implicated a circulating protein called irisin, which is derived from FNDC5 and secreted from muscle, in browning of WAT (Bostrom et al., 2012), however, levels of Fndc5 mRNA were decreased in quadriceps muscle from MIGIRKO (Table S4). FGF21 is another circulating hormone that can lead to browning of sWAT and has recently been implicated in metabolic adaptations to autophagy inhibition in muscle (Kim et al., 2012), as well as other stresses such as ER stress. Interestingly, Fgf21 mRNA levels in quadriceps were modestly increased, especially in the fasted state (Figure 5H), with increases in mRNA levels of macrophage markers but not ER stress markers (Table S4). However, when we tested circulating levels of FGF21, these were not changed in randomly fed MIGIRKO mice (Figure 5I).

Deletion of Insulin and IGF-1 receptors in muscle does not predispose mice to diabetes even after a high fat diet

To determine if deletion of either IR, IGF1R or both in muscle would predispose mice to metabolic derangements or diabetes, all genotypes were challenged with high fat diet (HFD). MIGIRKO mice were 25% smaller when dietary challenge was initiated, but mice of each genotype gained a similar percent of weight (15%) on HFD after 8 weeks compared to controls on chow diet (CD) (Figure 6A). In control mice, metabolic derangements were present as early as 4 weeks on HFD as indicated by increased insulin levels and increased serum triglycerides (Figure 6B). Interestingly, although serum triglycerides were equally elevated in MIGIRKO mice as in controls on HFD, insulin levels did not increase in MIGIRKO animals on HFD.

Figure 6. MIGIRKO mice are not predisposed to diabetes even after high fat diet feeding.

Body weights of control and MIGIRKO mice were measured weekly while on chow diet (CD) or high fat diet (HFD) beginning at 6 weeks of age (A) (n=4-8). Serum insulin levels from randomly fed mice and triglycerides from overnight fasted control and MIGIRKO mice on CD or HFD for 4 weeks (B) (n=4-8). Intraperitoneal glucose tolerance test (GTT) (C) was performed and area under the curve (AUC) (D) was calculated for MIGIRKO and control mice on CD or HFD for 9 weeks (n=3-9). Intraperitoneal insulin tolerance test (ITT) (E) was performed and area above the curve (AAC) (F) was calculated for mice on CD or HFD for 8 weeks (n=3-9). VO₂ (G), VCO₂ (H), and RER (I) were measured in animals on CD or HFD during both light and dark cycles using CLAMS metabolic cages (n=3-9). (*-p<0.05 vs. control with same diet, †-p<0.05 vs. CD of same genotype, student's t-test). See also Figure S5.

Glucose tolerance tests again revealed fasting hypoglycemia in MIGIRKO compared to controls on the same diet (Figure 6C). However, both control and MIGIRKO mice on HFD exhibited impaired glucose tolerance, as measured by increased area under the curve (AUC), with no differences between genotypes on the same diet (Figure 6D). M-IR^−/− mice did show modest impairment of glucose tolerance on CD, but similar to MIGIRKO animals, M-IR^−/− and M-IGF1R^−/− mice became glucose intolerant with increased AUC on HFD, and no differences were observed when compared to IR^lox/lox or IGF1R^lox/lox controls on the same diet (Figure S5B-S5C). Insulin tolerance tests at a dose of 1.0 mU/g body weight also remained similar between control and MIGIRKO mice, regardless of diet (Figure 6E-6F). Finally CLAMS analysis of both control and MIGIRKO animals on HFD revealed increases in VO₂ and VCO₂ compared to CD mice, with increases in both VO₂ and VCO₂ in MIGIRKO compared to controls, regardless of the diet (Figure 6G-6H). Both control and MIGIRKO mice showed the expected suppression of RER on HFD indicating increased fat utilization, with no differences between genotypes on the same diet (Figure 6I).

Overexpression of a dominant negative IGF1R in muscle of MIGIRKO mice leads to metabolic derangements despite absence of IR and IGF1R

Previous studies have shown that mice overexpressing a dominant negative, kinase inactive IGF1R in muscle (MKR) develop overt diabetes, presumably through inhibition of endogenous IR and IGF1R function (Fernandez et al., 2001). To further explore this hypothesis, we expressed the mutant IGF1R in MIGIRKO muscle by crossing MKR and MIGIRKO mice (MKR-MIGIRKO). As has been previously reported (Fernandez et al., 2002), we observed that MKR mice have reduced body weight compared to controls (Figure 7A). Unlike MKR mice on a FVB background (Fernandez et al., 2001), MKR mice on this mixed genetic background exhibited no differences in glucose levels upon fasting or refeeding (Figure 7B-7C), however, they did develop significant glucose intolerance when compared to controls (Figure 7D-7E). Expression of the MKR allele in MKR-MIGIRKO mice did not further reduce the body weight of MIGIRKO mice (Figure 7A), nor did it affect the development of mild hypoglycemia upon fasting when compared to control and MKR mice (Figure 7B). However, upon refeeding for only 4 hours, MKR-MIGIRKO mice displayed an exaggerated rebound in glucose levels.

Figure 7. Overexpression of a dominant negative, kinase inactive IGF1R in muscle of MIGIRKO mice induces glucose intolerance and impaired glucose uptake in heart.

Body weight was measured in 8-10 week-old control mice, mice with overexpression of a kinase inactive IGF1R in muscle (MKR), MIGIRKO mice, and MKR-MIGIRKO mice (A) (n=3-7). Blood glucose was measured in 8-10 week-old control, MKR, MIGIRKO, and MKR-MIGIRKO mice after an overnight fast and after 4 hours of refeeding (B) (n=3-7). Glut1 and Glut4 were measured in quadriceps (C) (n=8-10). Intraperitoneal glucose tolerance test (GTT) (D) was performed and area under the curve (AUC) (E) was calculated for 7-15 week-old mice (n=4-7). In vivo 2-deoxyglucose uptake was performed during an IV GTT in control, MKR, MIGIRKO, and MKR-MIGIRKO mice (F) (n=6-11). Western blots for IGF1R expression and insulin signaling were performed on quadriceps and heart from control, MKR, MIGIRKO, and MKR-MIGIRKO mice after a 5 U of insulin via IVC. (*-p<0.05, **-p<0.01 vs. control ANOVA, #-p<0.05 vs. MIGIRKO, student's t-test). See also Figure S7-S8.

Interestingly, expression of the dominant negative IGF1R in MKR-MIGIRKO mice also resulted in development of glucose intolerance in MIGIRKO mice similar to that observed when the MKR transgene was expressed in control mice (Figure 7D-7E). The impaired glucose tolerance in MKR and MKR-MIGIRKO mice was associated with elevated circulating triglycerides, similar to that observed in HFD fed animals (Figure S6A), but with no significant change in insulin levels (Figure S6B). Surprisingly, MKR-MIGIRKO show decreased Glut1 levels compared to MIGIRKO (Figure 7C), but the total level was similar to that observed in MKR. By contrast, MKR-MIGIRKO muscle shows markedly increased levels of Glut4 protein compared to control and MKR mice, which were similar to what was observed in MIGIRKO muscle.

We determined in vivo glucose uptake during an intravenous glucose tolerance test (IV GTT) to see which tissues may account for the changes in glucose tolerance in MKR-MIGIRKO mice. Glucose values during the IV GTT again demonstrated mild but significant glucose intolerance in MKR compared to controls, and glucose intolerance in MKR-MIGIRKO compared to MIGIRKOs (Figure S6C). Glucose uptake into quadriceps and gastrocnemius muscle during IV GTT was unchanged in MKR mice compared to controls (Figure 7F). MIGIRKO and MKR-MIGIRKO mice showed increased glucose uptake in skeletal muscle compared to WT with no changes between MIGIRKO and MKR-MIGIRKO muscle. Interestingly, glucose uptake into heart was significantly decreased in MKR and MKR-MIGIRKO mice compared to controls (Figure 7F), whereas no change was observed in other insulin sensitive tissues (Figure S6D). As previously observed for other genes on the MCK promoter (Bruning et al., 1998), expression of MKR allele was very high in both skeletal muscle and heart as shown by IGF1R western blot analysis (Figure 7G). Total levels of Akt isoforms, as well as phosphorylation of Akt, in MKR-MIGIRKO skeletal muscle were the same as MIGIRKO, but these levels were elevated compared to that observed in control and MKR skeletal muscle, and did not respond to insulin or IGF-1 treatment (Figure 7G and S7). However, insulin signaling in heart as measured by phosphorylation of Akt, GSK3β, and FoxO isoforms was unchanged (Figure 7G), indicating that changes in Akt signaling in MKR and MKR-MIGIRKO hearts are unlikely to account for the impaired glucose uptake when compared to Controls.

Discussion

Skeletal muscle insulin resistance is an important component in the pathogenesis of type 2 diabetes and metabolic syndrome, and may occur years prior to onset of disease (Martin et al., 1992). However, while deletion of IR in skeletal muscle of MIRKO mice causes some features of the metabolic syndrome, it alone does not cause diabetes or hyperglycemia (Bruning et al., 1998). Two potential explanations for this discrepancy are the possibility that there is residual insulin signaling in muscle via the IGF1 receptor or that exercise-induced glucose uptake compensates for this insulin resistance and maintains glucose uptake. Consistent with the first hypothesis, Fernandez et. al. (Fernandez et al., 2001) have reported that MKR mice which overexpress a dominant-negative form of the human IGF1R in muscle develop diabetes at a young age, suggesting that this receptor can hybridize with IR and IGF1R to block insulin and IGF-1 signaling and induce hyperglycemia.

To test the first hypothesis directly, in the present study, we specifically deleted IR and IGF1R in muscle to create the MIGIRKO mouse. Indeed, we find that IR and IGF1R compensate for each other to maintain muscle growth, such that when both are deleted, the mice display no insulin or IGF-1 signaling in skeletal muscle and have a marked decrease in muscle mass and fiber size. Despite this, these mice display normal whole body glucose tolerance, indicating that, in mice, neither of these receptors alone or in combination is required in muscle to maintain normal glucose tolerance. Furthermore, loss of IR and IGF1R in muscle does not lead to diabetes, even when the mice are challenged with a high fat diet. That is not to say that these MIGIRKO mice do not display perturbations in muscle glucose metabolism. To the contrary, MIGIRKO mice show fasting hypoglycemia, which is mediated by increased glucose transporter protein levels and translocation leading to increased basal glucose uptake in muscle. Furthermore, energy expenditure was increased and associated with increased glucose uptake in BAT, WAT, and increased markers of browning of sWAT, possibly contributing to the normal glucose tolerance in MIGIRKO mice. Surprisingly, despite the lack of effect of IR/IGF1R knockout in muscle on whole body glucose tolerance, expression of the dominant-negative IGF1R in muscle does lead to glucose intolerance and some of the metabolic derangements associated with metabolic syndrome. Thus, insulin signaling via the IGF1R in muscle is not a compensatory mechanism by which glucose tolerance is maintained when the insulin receptor is deleted in muscle.

Mechanistically, this study defines a critical role for IR/IGF1R in muscle to suppress basal glucose transport, especially in the fasted state (low insulin). Normally, muscle utilizes glucose for energy production primarily in the fed state and transitions to fatty acids as a primary fuel source upon fasting. Upon refeeding, the muscle rapidly switches back to glucose utilization in a paradigm termed metabolic flexibility (Storlien et al., 2004). One component of muscle metabolic flexibility is the activation of glucose transport by insulin, which occurs via enhanced Glut4 translocation. Somewhat unexpectedly, we find that total absence of IR and IGF1R signaling leads to a paradoxical increase in Glut1 and Glut4 proteins and increased localization of these to the plasma membrane, even in the fasted state. Thus the MIGIRKO mouse develops mild fasting hypoglycemia, rather than hyperglycemia.

Several mechanisms contribute to the increased basal glucose uptake in muscle following loss of IR and IGF1R. First, deletion of IR/IGF1R in muscle leads to decreased levels of TBC1D1. TBC1D1 is a Rab-GAP protein expressed in primarily in glycolytic muscle, and is homologous to the Rab-GAP AS160 (also known as TBC1D4) (Taylor et al., 2008). Both AS160 and TBC1D1 are inhibited by Akt- or AMPK-mediated phosphorylation to promote GLUT4 translocation to the plasma membrane (Taylor et al., 2008). We observe that re-expression of TBC1D1 in MIGIRKO muscle in vivo is able to reverse the abnormal Glut4 localization. These data are further supported by in vitro studies which show that silencing of TBC1D1 in L6 myotubes or adipocytes results in increased basal Glut4 and Glut1 translocation, respectively (Zhou et al., 2008; Ishikura and Klip, 2008). Interestingly, germline deletion of TBC1D1 results in decreased glucose uptake in muscle, which was consistently associated with decreased Glut4 levels (Dokas et al., 2013; Szekeres et al., 2012). In MIGIRKO mice, we find decreased levels of TBC1D1 protein, but increased levels of Glut1 and Glut4, which are related to decreased protein turnover as mRNA levels were unchanged or decreased. In addition, Akt and AMPK are chronically activated in MIGIRKO muscle, which was unexpected and may relate to energy stress, changes in protein turnover, or unmasking of a feedback loop. While, little is known about the control of TBC1D1 expression, our data indicate that IR/IGF1R signaling plays an important role in regulation of TBC1D1 and AS160 levels in muscle.

It has been known for some time that IGF-1 treatment can induce muscle hypertrophy via the Akt-mTOR pathways, yet previous studies have indicated that deletion of IGF1R alone in muscle only modestly changes myocyte size and morphology (Schiaffino and Mammucari, 2011; Mavalli et al., 2010). The present study indicates that signaling via either IR or IGF1R is sufficient to maintain muscle mass. This indicates that physiologic levels of insulin or IGF-1 ligand are sufficient to promote proteins synthesis and suppress protein degradation as long as either the IR or the IGF1R are present. Recent work has identified FoxO transcription factors, which are known targets for IR/IGF1R signaling, as critical mediators of muscle protein degradation and atrophy (Mammucari et al., 2007; Sandri et al., 2004).

The browning of the sWAT in MIGIRKO mice also contributes to the metabolic phenotype. Previous reports confirm that changes in the autophagy pathway in muscle can lead to increases energy expenditure via FGF21 induced browning of WAT (Kim et al., 2012). Although circulating FGF21 levels were not increased, local FGF21 production by the muscle remains a possible mechanism for the browning of sWAT observed in these mice. These mice were raised at room temperature (25°C), which is not thermo-neutral, and the reduced body size may contribute to increased need for thermogenic capacity.

Our work also identifies a distinction between lack of IR/IGF1R signaling and insulin resistance in which the receptors are present, but activation by ligand is decreased. Thus, while combined deletion of muscle IR and IGF1R does not alter glucose or insulin tolerance in mice, expression of a dominant negative IGF1 receptor in skeletal muscle (and heart) in both control and MIGIRKO mice can produce mild glucose intolerance and lipid abnormalities. This is similar to previous observations demonstrating that MKR mice develop diabetes associated with dyslipidemia, hepatic steatosis, and insulin resistance (Kim et al., 2003; Vaitheesvaran et al., 2010), but in our study this occurs even in mice lacking normal insulin and IGF-1 receptors in muscle. We speculate that at least two possibilities contribute to this phenomenon. First, the dominant-negative IGF1R may bind to a receptor other than IR/IGF1R, such as Met (Fafalios et al., 2011), or to one or more downstream signaling proteins, such as IRS-1 and Shc, to transmit a signal to the myocyte that actively perturbs lipid homeostasis and interrupts tissue cross-talk, whereas deletion of the receptors does not transmit such a signal. Secondly, cardiac insulin/IGF-1 resistance, when combined with skeletal muscle resistance, may contribute to more glucose intolerance and lipid abnormalities than seen with skeletal muscle insulin resistance alone.

Our lab has previously shown that deletion of IR/IGF1R in preadipocytes protects them from apoptosis, and reintroduction of a non-functional IR transmitted a signal which conferred susceptibility to apoptosis (Boucher et al., 2010). This suggests that the unoccupied insulin and IGF1 receptors can generate a signal that is different from that normally mediated by the occupied receptor. The current study likewise indicates that deletion of IR or IGF1R is fundamentally different from loss of insulin or IGF-1 signaling in which the receptors are present, but the ligand or ligands are missing. Further work will be needed to fully characterize the nature of the signals coming from unoccupied insulin and IGF-1 receptors and the specific IR/IGF1R receptor-protein interactions that contribute to metabolic disease.

In summary, our study demonstrates that IR or IGF1R signaling is critical for normal muscle growth, but that deletion of both receptors in muscle does not lead to impaired glucose tolerance due to underlying feedback loops which maintain a high level of glucose uptake even in the fasted/unstimulated state. In addition, deletion of IR and IGF1R in muscle is unable to induce diabetes or worsen metabolic parameters in mice challenged with a high fat diet. On the other hand, loss of IR and IGF1R in muscle leads to increased basal glucose uptake due to increases in levels of Glut1 and Glut4 transporters, chronic activation of Akt and AMPK signaling, and a loss of TBC1D1 expression. Finally, we find that the presence of a non-functional IGF1R in muscle of animals lacking IR and IGF1R can induce glucose intolerance and metabolic derangements, indicating a novel mechanism of altered signaling by this receptor mutant. These data add a new layer of understanding, such that the metabolic changes that occur in insulin resistant states may be a consequence of signals transmitted from a poorly functional IR or IGF1R in muscle. Further investigation of the protein interactions of IR and IGF1R in the insulin resistant state will potentially provide new targets for the treatment of type 2 diabetes and its complications.