Denervation and High-Fat Diet Reduce Insulin Signaling in T-Tubules in Skeletal Muscle of Living Mice

All experiments were approved by the animal research committee of the Danish Ministry of Justice. Mice were always provided with water and studied in the fed state.

At 3 weeks of age, male Friend virus B-type (FVB) mice (Charles River, Sulzfeld, Germany) were given either normal standard diet or high-fat diet for 12 weeks. Standard diet consisted of 14% energy from fat, 66% from carbohydrate, and 20% from protein (Altromin, Lage, Germany). The high-fat diet was prepared by Harlan Teklad (TD93075), as previously described (5), and consisted of 55% of energy from fat, 24% from carbohydrate, and 21% from protein.

To study the effects of denervation, the right femoral nerve of 7- to 8-week-old FVB male mice (Charles River) was transsected 1 mm proximal to its branch, innervating the quadriceps muscle. In other experiments, 7- to 8-week-old FVB male mice had one extensor digitorum longus (EDL) and one soleus muscle denervated by resection of a 5-mm-long segment of the sciatic nerve in the thigh. After denervation, the skin was closed by sutures and confocal imaging analysis or glucose transport measurement was performed 10 days later.

The construction of GLUT4-GFP (8) and GFP-ADP–ribosylation factor nucleotide-binding site opener (ARNO) (12) has been described previously. The vastus lateralis part of the quadriceps muscles of 5-days–denervated or high fat–fed and control FVB mice (Charles River) were transfected with either two times 1 μg GLUT4-GFP or 1 μg ARNO-GFP cDNA/0.5 mg gold, as previously described (8).

Five days after transfection, mice were anesthetized subcutaneously with 0.8–1 ml/100 g body wt Hypnorm-Dormicum (25% Hypnorm [5 mg/ml] + 25% Dormicum [5 mg/ml] in water), and the skin covering the superficial part of the quadriceps muscle (86% type IIb fibers [13]) was opened. The mice were mounted on the confocal microscope, transfected muscle fibers with similar probe expression in the various groups were imaged, and insulin was stimulated during maintenance of euglycemia, as previously described (9). Sulforhodamine B was administrated and imaged, as previously described (9).

The TIFF (tagged image file format) images obtained with the Leica confocal software were imported into Metamorf Software (version 6.1; Universal Imaging). Image stacks were created and pixel values were corrected for noncellular fluorescence. Brightness and contrast were adjusted. For quantifications and visualization of GLUT4-GFP images, a Fourier noise filter was applied. The degree of GLUT4-GFP translocation to membrane surfaces was quantified by drawing a region of interest (ROI) surrounding the surface of interest and measuring its pixel values. For sarcolemma, measurements from both sarcolemma edges (excluding perinuclear areas) of the confocal picture were averaged. For T-tubules four ROIs, excluding big vesicular GLUT4-EGFP stores, were drawn and measurements from these were averaged. For GFP-ARNO quantification, ROI pixel values were measured at both sarcolemma edges (excluding perinuclear areas) and averaged. For GFP-ARNO at T-tubules, measurements were averaged from two 2-μm-wide ROIs along the length of the fiber at identical depths. Arbitrary units shown are the actual gray value of ROI fluorescence divided by ROI fluorescence value at t = 0. Uncompressed Quicktime movies were produced from the image stacks and subsequently compressed with Sorensen 3 codec in Quicktime Pro 6.

Quadriceps muscles were fixed and single fibers from the superficial white part of the muscle were teased to perform immunofluorescence staining, as previously described (14). A rabbit antibody against insulin receptor (no. ab10991-50; Abcam) was used and immunodetected with a secondary antibody conjugated to Alexa-488 (no. A11008; Invitrogen).

Seven- to 8-week-old FVB male mice (Charles River) were anesthetized subcutaneously with Hypnorm-Dormicum (0.8–1 ml/100 g body wt) and decapitated. EDL muscles (47% type IIa, 50% type IIb fibers [13]) were gently taken out and preincubated in test tubes at pH 7.4 and 29°C for 60 min in Krebs-Henseleit buffer gassed with 95% O₂ and 5% CO₂ and containing 8 mmol/l glucose, 1 mmol/l pyruvate, and 0.2% BSA. After preincubation, the muscles were incubated with 5 mol/l formamide (Sigma) in Krebs-Henseleit buffer for 60 min. The detubulation was induced by transferring muscles to Krebs-Henseleit buffer for 20 min, and then the muscles were washed twice in glucose-free Krebs-Henseleit buffer.

EDL and soleus (87% type I fibers [13]) muscles were preincubated for at least 90 min in Krebs-Henseleit buffer gassed with 95% O₂ and 5% CO₂ and containing 8 mmol/l glucose, 1 mmol/l pyruvic acid, and 0.2% BSA at pH 7.4 and 29°C.

For glucose transport measurement, muscles were transferred to glucose-free Krebs-Henseleit buffer containing 2 mmol/l pyruvic acid and 0.2% BSA with or without insulin (100 mU/ml Actrapid; Novo Nordisk). Glucose transport was measured as a 2-deoxy-d-[³H]-glucose ([³H]2-DG) uptake with [¹⁴C]sucrose as the extracellular marker (15). Isotopes and unlabeled sugars were added to the incubation medium to yield final concentrations of 0.43 μCi [³H]2-DG and 0.32 μCi [¹⁴C]sucrose/ml and 1 mmol/l of both unlabeled 2-DG and sucrose. After 10 min of exposure to isotopes, muscles were briefly blotted on filter paper and immediately frozen in liquid nitrogen. Muscles were stored at −80°C until analyzed.

Blood was sampled at noon from the tail vein of anesthetized (subcutaneously with Hypnorm-Dormicum) 15-week-old FVB male mice that had either a high-fat diet or a standard diet for 12 weeks. Blood glucose concentrations were measured directly by EuroFlash glucose-monitoring system using the sensor strips of this system (Johnson & Johnson).

Data are presented as means ± SE. Statistical comparisons were carried out by two-way ANOVA and Student's t test, paired or unpaired as applicable. Post hoc analysis was done by the Tukey test. P < 0.05 was considered significant in two-tailed testing.

To study insulin signaling in insulin resistance, we looked at two conditions accompanied by insulin resistance: long-term muscle denervation and high-fat feeding. Long-term denervation of muscle (7–10 days) is known to result in an up to 60% decrease in insulin-induced, biochemically determined PI3-K activity and glucose transport (10,16). In agreement with this, we found a 40% decrease (P < 0.05) in insulin-mediated 2-DG uptake after 10 days of denervation of EDL (online appendix Fig. 1A [available at http://dx.doi.org/10.2337/db07-0516]) and soleus (data not shown) muscle. High-fat feeding of FVB mice has also been shown to induce insulin resistance and type 2 diabetes (5,11). In agreement with this, we found a 30% decrease (P < 0.05) in insulin-mediated 2-DG uptake in EDL (online appendix Fig. 1B) and soleus (data not shown) muscles after 12 weeks of high-fat diet feeding. Furthermore, in accordance with the development of whole-body insulin resistance, after 12 weeks of high-fat feeding the noon glucose concentration in tail vein blood had increased compared with values in control mice (8.9 ± 0.5 vs. 10.5 ± 0.5 mmol/l, n = 6–8; P < 0.05) (online appendix Fig. 1C).

We analyzed the spatial and temporal changes in insulin signaling in quadriceps muscle of denervated and high fat–fed mice by time-lapse analysis of the early insulin signaling step, the activation of PI3-K (17,18). For this we used the ARNO (19) fused to GFP (12,20). ARNO binds with exquisite specificity via its pleckstrin homology domain to phosphatidylinositol 3,4,5 P₃ (PIP3), the product of insulin-activated PI3-K (21). The production of PIP3 has been shown to be a direct measure of PI3-K activation, and GFP-ARNO has been shown to track and indicate local concentrations of PIP3 (20,22). By imaging GFP-ARNO in skeletal muscle fibers of living mice, we could dynamically follow the local PI3-K activity with time (Fig. 1A). Skeletal muscle of living mice were denervated 10 days before imaging, and GFP-ARNO–transfected muscle fibers were imaged in their natural surroundings. In the basal state, GFP-ARNO was distributed throughout the muscle cell in both control (Fig. 1A, t = 0 min) and denervated (Fig. 1B, t = 0 min) muscle as expected for a cytosolic protein (12). Furthermore, within ∼2 min after insulin injection, GFP-ARNO translocated to the sarcolemma in both groups (Fig. 1A and B, t = 2 min, arrows). However, the PIP3 production persisted for a longer time and at a higher level at the sarcolemma in denervated compared with control muscle (Fig. 1A and B, t = 4 min). In fact, image quantification of ROIs, including sarcolemma (Fig. 2 E and F), revealed a 2.6-fold higher PIP3 production at the sarcolemma over the 20 min of observation in denervated muscle compared with control (Fig. 2A and G). In contrast to findings in sarcolemma, the wave-like spreading of the PIP3 production into the T-tubules seen in normal muscle (Fig. 1A, t = 4 and 14-min arrows; online appendix movie 1) was completely absent in denervated muscle (Fig. 1B; online appendix movie 2). Instead, in the latter muscle only a few fragmented spots of PIP3 production emerged in the T-tubules (Fig. 1B, t = 7 min, arrow), and overall PIP3 production was markedly reduced in the T-tubules (Fig. 1B, t = 4–20 min). Image quantification of ROIs at various distances from the sarcolemma (Fig. 2E and F, t = 0 min) also showed that the wave of PIP3 production in the T-tubules in response to insulin was almost completely abolished (ΔAUC, P < 0.05) (Fig. 2B–D, H). So, not only did denervation alter the progression pattern of PIP3 production, but the overall PIP3 production was also strongly reduced.

To see if these changes in PIP3 production in T-tubules were secondary to the rearrangement of the T-tubule system, which has been shown by electron microscopy to take place during long-term denervation (23), sulforhodamine B (24–26) was coinjected with insulin and glucose. This dye is nontoxic, is not taken up by muscle cells, and has previously been used to load T-tubules both in vitro (24–26) and in vivo (9). As shown in Fig. 1C, a striated pattern emerged as previously seen in normal innervated muscle (9). Thus, no T-tubular rearrangement was visible at the level of light microscopy and no changes in accessibility to the inner T-tubules had taken place, a fact excluding that such rearrangement explained the denervation-induced change in appearance of PIP3 production. Similarly, immunofluorescence staining of insulin receptors delineated the sarcolemma and, in addition, showed a striated pattern of uniform intensity indicating an even distribution of insulin receptors throughout the T-tubule system, as previously described in normal muscle (9) (Fig. 3A and B).

After 12 weeks of high-fat diet, quadriceps muscles were transfected with GFP-ARNO and subjected to confocal image analysis in vivo. During basal conditions, GFP-ARNO was distributed similarly throughout the muscle cell in both control (Fig. 4A, t = 0 min) and fat-fed muscle (Fig. 4B, t = 0 min). Within ∼2 min after insulin injection, GFP-ARNO translocated to the sarcolemma in both groups, but translocation was slightly reduced in the high fat–fed mice (Fig. 4A and B, t = 2 min, arrows). Quantification showed the increase in PIP3 production at the sarcolemma to be reduced by 42% over the time of PI3-K activation in the high fat–fed mice compared with controls (Fig. 5 A and E). Moreover, while in muscle of mice fed normal diet a strong and regular wave-like spreading of the PIP3 production along the T-tubules was seen (Fig. 4A, t = 9 min, arrows), in mice fed a high-fat diet a less strong GFP-ARNO signal was found (Fig. 4B, t = 9 min, arrows), indicating reduced PIP3 production with only a few strong “spikes” running into the T-tubules (Fig. 4A and B). However, overall, the wave-like pattern was still present, although weaker and more irregular than seen in normal mice (Fig. 4A and B). Quantification at various distances from the sarcolemma revealed a marked reduction of ∼65% of T-tubular PIP3 production over the period of observation (ΔAUC, P < 0.05) (Fig. 5B–D and F). Sulforhodamine B staining of T-tubules (results not shown), as well as immunofluorescence staining of insulin receptors in sarcolemma and T-tubules (Fig. 3C and D), appeared as in normal muscle.

We investigated if the observed changes in PI3-K activity after denervation and high-fat diet had any functional consequences at the level of GLUT4 translocation. In control muscles, insulin induced a threefold increase in GLUT4-GFP at the sarcolemma and a sixfold increase in the T-tubules in agreement with previous findings (Fig. 6A, t = 30 min, arrows; C and D) (8,9,14). In basal denervated muscles, GLUT4 localization in the T-tubule region was changed compared with control, now having a diffuse rather than dotted distribution, as also reported earlier (27). After insulin stimulation of denervated muscle, in sarcolemma the increased PIP3 signal (Figs. 1 and 2) was accompanied by an increased GLUT4-GFP translocation compared with findings in control muscle (Fig. 6B, t = 30 min, arrows). Image quantification showed the increase in GLUT4-GFP translocation in denervated muscle to be 2.3-fold above control (Fig. 6C). Furthermore, corresponding with the finding of an abolished PIP3 signal in the T-tubules in denervated muscle, insulin stimulation did not induce any visible GLUT4-GFP translocation, as no enhancement of the striated pattern in the T-tubule area emerged (Fig. 6B). Image quantification also showed a complete abolition of GLUT4-GFP translocation in the T-tubules in denervated muscle compared with control (Fig. 6D). Thus, in denervated insulin-resistant muscle both insulin signaling and GLUT4 translocation are markedly impaired in the T-tubules while being increased at the sarcolemma.

Image analysis in the high-fat fed mice showed that at the sarcolemma the diminished PIP3 signal in response to insulin was accompanied by a slightly weaker staining in GLUT4-GFP transfected compared with normal muscle, indicating reduced sarcolemmal translocation (Fig. 7B vs. a). Quantification supported this observation showing a reduction in GLUT4-GFP translocation to sarcolemma of 16% in high fat–fed compared with control mice (Fig. 7C). Furthermore, corresponding with the decrease in PIP3 signal in the T-tubules (Fig. 4A and B), the insulin-induced GLUT4-GFP translocation was also reduced in fat-fed compared with control mice, as indicated by emergence of only weak striations (Fig. 7A vs. B). Image quantification supported this observation showing a reduction in GLUT4-GFP translocation to the T-tubules of 68% (Fig. 7D).

The above findings pointed at an essential role of impaired insulin signaling and GLUT4 translocation in T-tubules for insulin resistance. Therefore, in order to directly evaluate the importance of the T-tubules in normal mouse muscle, we measured the effect of detubulation on insulin-mediated 2-DG transport in incubated muscles. EDL muscles were subjected to osmotic shock by which the T-tubules lose their connection to the sarcolemma and, accordingly, have no access to insulin and glucose from the extracellular fluid (28–30). Detubulation reduced basal glucose transport by 50% (0.04 ± 0.006 vs. 0.08 ± 0.02 μmol · g⁻¹ · 5 min⁻¹; n = 5–6; P < 0.05) and completely abolished the insulin-induced increase in glucose transport (to 0.04 ± 0.06 vs. 0.18 ± 0.03 μmol · g⁻¹ · 5 min⁻¹; n = 5–6; P < 0.05) (Fig. 8).

In the present study, the spatial and temporal distribution of insulin signaling and GLUT4 translocation was, for the first time, followed in insulin-resistant skeletal muscle in situ in living animals. A major new finding is that, whether elicited by muscle denervation or high-fat diet, insulin resistance is associated with a marked reduction of insulin-mediated PIP3 production and GLUT4 translocation in the T-tubules, while these variables are less diminished or even increased at the sarcolemma. Emphasizing the role of T-tubules relative to that of sarcolemma for glucose transport in muscle, insulin-mediated glucose transport was completely abolished by osmotic disruption of T-tubules in normal muscle (Fig. 8). We are the first to report changes in the opposite direction of both PIP3 production and GLUT4 translocation in T-tubules compared with sarcolemma, as seen in denervated muscle. The finding shows that the defects in insulin action are markedly compartmentalized in insulin-resistant muscle and further supports that GLUT4 translocation is controlled by local insulin signaling (9).

Electron microscopy studies have shown that the arrangement of T-tubules changes within 10 days of muscle denervation (23). So, it might be speculated that reduced accessibility of insulin to its receptors might explain the reduced insulin signaling in T-tubules seen after denervation in the present study. However, the total T-tubule area increases with denervation (23). Furthermore, in the present study the dye sulforhodamine B, which has previously been used to locate T-tubules (9,24–26), diffused into the T-tubules of denervated muscle with no delay and caused a striated pattern, as previously seen (9) in normal muscle (Fig. 1). Reduced insulin signaling in the T-tubules was also not due to reduced presence of insulin receptors in the T-tubules (Fig. 3). Thus, we show, for the first time, that plasma molecule access to T-tubules and insulin receptor distribution are not affected in insulin-resistant denervated muscle.

In previous studies of the mechanism of denervation-induced insulin resistance, biochemical methods have been applied on crude muscle homogenate. In line with the present findings, it was shown that after 7–10 days of denervation insulin receptor content was normal, whereas insulin-stimulated tyrosin phosphorylation of the receptor and of its downstream signaling molecule, insulin receptor substrate-1, as well as the association of insulin receptor substrate-1 with PI3-K, insulin receptor substrate-1–associated PI3-K activity, PI3-K activity, insulin-mediated protein kinase B/Akt activation, and total amount of GLUT4 mRNA and protein were diminished (6,10). However, those studies were able to reveal neither the dynamics nor the location of the changes. In contrast, the present study has indicated that after a similar denervation protocol, insulin signaling and GLUT4-GFP translocation are exclusively reduced in T-tubules and that in these, the wave-like spreading of PIP3 production seen in normal muscle, as well as the GLUT4-GFP translocation, are absent in their entire length (Figs. 1, 2, and 6).

These findings indicate that after long-term denervation, diminished insulin-mediated glucose transport in skeletal muscle reflects impaired GLUT4 translocation to T-tubules rather than to sarcolemma. However, because denervation reduces the total endogenous GLUT4 content of muscle by 25–50% (6,31,32), at similar GLUT4-GFP expression the labeled fraction of GLUT4 (specific activity) will be higher in denervated compared with control muscle. Accordingly, in spite of enhanced translocation of GLUT4-GFP to sarcolemma, after denervation, translocation of endogenous GLUT4 might nevertheless be reduced and so account for the diminished insulin-mediated glucose transport. Still, this was hardly the case because the increase in GLUT4-GFP translocation to sarcolemma in denervated compared with control muscle was 2.3-fold, while increases in specific activity calculated from reported decreases in endogenous GLU4 would be only 1.3- to 2-fold, allowing for a 1.15- to 1.73-fold higher endogenous GLUT4 translocation to sarcolemma in denervated muscle.

The redistributed amounts of both GFP-ARNO and GLUT4-GFP in absolute terms also depend on the total amount of the probes expressed in the studied cells. However, this did not vary consistently between groups, and, furthermore, any random variation was corrected for by expressing fluorescence upon insulin stimulation relative to local basal values.

In unstimulated denervated muscle we found, in agreement with others (27), in the T-tubule region a diffuse distribution of GLUT4-GFP rather than a dotted distribution as seen in control muscle (Fig. 6). At the level of light microscopy resolution, it is not possible to distinguish between GLUT4-GFP in vesicles in the juxtaposition of the T-tubules and GLUT4-GFP within these membranes. So, we cannot completely exclude that in denervated muscle part of the diffusively located GLUT4-GFP is locally translocated to the T-tubule membranes. However, judged from the insulin-induced PIP3 production, any GLUT4 translocation would be minimal. Because we and others find no increased basal glucose uptake after long-term denervation, it is also not likely that GLUT4 within T-tubule membranes was high in the basal state in denervated muscle.

Our preparation is set up for imaging of muscle fibers in the superficial layers of the quadriceps muscle. This part of the quadriceps muscle consists of white type IIb fibers (13). However, in accordance with previous findings (6,31–33), we found no differences in changes in insulin sensitivity after denervation in red soleus muscle, consisting predominantly of type 1 fibers, compared with EDL muscle, consisting predominantly of type 2 fibers (13). Accordingly, our findings in the superficial quadriceps would be expected to apply, as well, to other types of muscle (6,31–33).

Others have found that insulin-mediated glucose uptake in muscle is diminished after only 24 h of denervation (16,34). In the present study, PIP3 production and GLUT4 translocation were also analyzed after 24 h of denervation. We found no changes compared with controls (data not shown). This is in agreement with the fact that insulin-stimulated receptor phosphorylation, biochemically determined PI3-K activity, and GLUT4 translocation determined by subcellular fractionation have also been shown to be preserved (16,34,35). So, apparently the insulin resistance associated with short-term denervation must be attributed to downregulation of the metabolic machinery (e.g., hexokinase).

In previous studies of the mechanism of high-fat diet–induced insulin resistance, the insulin binding in muscle was normal (36), a finding agreeing with the observation that the number of insulin receptors was reduced neither in sarcolemma nor in T-tubules in high-fat–compared with standard diet–fed mice (Fig. 3). Based on biochemical measurements, some have argued that defects in insulin signaling in muscle are late events seen after 30 but not after 8 weeks of high-fat diet (7). In contrast, other studies have reported reduced insulin-stimulated PI3-K and Akt/protein kinase B activities after only 4 weeks of high-fat diet (5,37) and reduced Akt/protein kinase B signaling after 7 weeks of high-fat diet (38). Our study, for the first time, shows that 12 weeks of high-fat diet reduce PI3-K activity in the entire length of the T-tubules and, to a lesser extent, in sarcolemma (Figs. 4 and 5) and, furthermore, that this compartmentalized reduction in insulin signaling is closely coupled with reduced GLUT4 translocation (Fig. 7). Previous biochemical studies have found either normal (4,5) or reduced (6,37,39) overall GLUT4 content in muscle from high fat–fed animals, while GLUT4 translocation was reduced (4,5,7). One study using subcellular fractionation found completely abrogated insulin-stimulated GLUT4 translocation to both sarcolemma and T-tubules (38).

In studies with the isolated perfused rat hindquarter, 7 days of muscle denervation has been shown to diminish insulin-stimulated glucose transport in the three types of skeletal muscle more than streptozotocin-induced diabetes (6). Compatible with impairments of the same mechanisms in the two disease states, the effect of combined denervation and diabetes was not more marked than that of denervation alone (6). These findings are in accordance with our findings on GLUT4 translocation and PI3-K activity in quadriceps muscle from muscle-denervated or high fat–fed mice. However, the effect of denervation on insulin-mediated glucose transport in incubated EDL and soleus muscles was not markedly higher than the effect of high-fat feeding. EDL and soleus muscles were used instead of quadriceps muscle because quadriceps muscle cannot be easily isolated. It is possible that the more moderate effect of denervation in the present experiments on incubated muscle compared with the findings in the rat hindquarter reflects that the effect of impaired T-tubule function is blurred in incubated muscle because diffusion of insulin and glucose into the T-tubules may be hindered when muscle is not stretched.

In conclusion, the present study demonstrates new in vivo evidence of a marked compartmentalization of insulin signaling in skeletal muscle and that impaired insulin signaling in T-tubules is an essential part of development of insulin resistance. Moreover, the study has, for the first time, shown that insulin signaling may be changed in opposite directions in T-tubules compared with sarcolemma. Future studies must clarify the mechanism for the latter finding and for the effects of denervation and high-fat diet on insulin signaling in muscle. In this context, it may be recalled that although the T-tubules extend from the sarcolemma, the membrane composition of these two cellular compartments vary markedly regarding cholesterol, phospholipids, proteins, receptors, and enzymes (40,41).

This work was supported by the Faculty of Health Sciences, University of Copenhagen; the Danish Diabetes Association; the Weimann Foundation; the Novo Nordic Research Foundation; the Beckett Foundation; and the Danish National Research Foundation. H.A. was partly supported by the National Natural Science Foundation of China (grant no. 30270636).

We thank L. Kall for bioanalytical assistance and S. Lohmann for engineering assistance.