Insulin stimulates glucose transport in skeletal muscle by glucose transporter GLUT4 translocation to sarcolemma and membrane invaginations, the t-tubules. Although muscle glucose uptake plays a key role in insulin resistance and type 2 diabetes, the dynamics of GLUT4 translocation and the signaling involved are not well described. We have now developed a confocal imaging technique to follow trafficking of green fluorescent protein–labeled proteins in living muscle fibers in situ in anesthetized mice. Using this technique, by imaging the dynamics of GLUT4 translocation and phosphatidylinositol 3,4,5 P3 (PIP3) production in response to insulin, here, for the first time, we delineate the temporal and spatial distribution of these processes in a living animal. We find a 10-min delay of maximal GLUT4 recruitment and translocation to t-tubules compared with sarcolemma. Time-lapse imaging of a fluorescent dye after intravenous injection shows that this delay is similar to the time needed for insulin diffusion into the t-tubule system. Correspondingly, immunostaining of muscle fibers shows that insulin receptors are present throughout the t-tubule system. Finally, PIP3 production, an early event in insulin signaling, progresses slowly along the t-tubules with a 10-min delay between maximal PIP3 production at sarcolemma compared with deep t-tubules following the appearance of dye-labeled insulin. Our findings in living mice indicate a major role of the t-tubules in insulin signaling in skeletal muscle and show a diffusion-associated delay in insulin action between sarcolemma and inner t-tubules.
Skeletal muscle is a key regulator of glucose homeostasis, and defects in glucose uptake in muscle are central in insulin resistance and type 2 diabetes. The GLUT4 glucose transporter in skeletal muscle fibers mediates glucose uptake by translocation from intracellular compartments to the plasma membrane upon stimulation with insulin (1,2). However, the intracellular trafficking of GLUT4 in muscle and the signaling regulating these processes are poorly understood. The lack of understanding reflects the weakness of current methods for analyzing insulin-mediated GLUT4 translocation in skeletal muscle. First, generally data can only be sampled at few time points, typically basal and 30 min after insulin stimulation (3–5), thereby missing intermediate GLUT4 trafficking. Second, in studies relying on subcellular fractionation or morphological GLUT4 translocation analysis, the effect of insulin can only be evaluated by comparisons between different muscles and not within single muscle fibers (4,6). Third, traditional methods are all invasive and may not accurately reflect in vivo conditions. To overcome these problems, we have developed confocal time-lapse imaging techniques for monitoring dynamic changes in localization of green fluorescent protein (GFP)-tagged proteins in skeletal muscle fibers in situ in living mice (7) (Fig. 1). Here, using these techniques, we reveal previously unnoticed signaling functions of the t-tubule system in muscle fibers, findings that call for changes in the concepts of t-tubules. Specifically, we report for the first time a delay in insulin-mediated GLUT4 translocation to t-tubules compared with sarcolemma, which reflects insulin diffusion to receptors in the t-tubules. Furthermore, we report that the majority of insulin-mediated phosphatidylinositol 3 kinase (PI3K) activity is present in the t-tubules and not in the sarcolemma.
RESEARCH DESIGN AND METHODS
Plasmid and transfection procedures.
The construction of GLUT4-GFP (7) and GFP–ADP ribosylation factor 1 nucleotide binding site opener (ARNO) (8) has been described previously. The quadriceps muscles of 7- to 8-week-old male C57BL/6 mice (Charles River, Germany) were transfected with either 1 μg GLUT4-GFP or ARNO-GFP cDNA/0.5 mg gold as previously described (7).
Four days after transfection, the overnight-fasted mice were anesthetized subcutaneously with Hypnorm-Dormicum (0.8–1 ml/100 g body wt), and the skin covering the quadriceps muscle was opened to expose the quadriceps. The mice were put to rest on the side and mounted and stabilized with Colténe President cement (Colténe/Whaledent, Mahwah, NJ) imbedding. The exposed muscle was covered by a coverglass and the mice mounted on the Leica TCS SP2 confocal microscope stage (Fig. 1). Confocal images were collected with a 63×, 1.2 NA Leica HCX PL APO water immersion objective. The 488-nm laser line with 40% of minimal laserpower was used for excitation of GFP while green emission light was collected in the fluorescein isothiocyanate channel. Selective photobleaching of a region of interest (ROI) was performed by scanning the ROI 72 times in 2 min at 100% of maximal 488 nm laserpower. Sulforhodamine B was imaged using 100% power of the 543-nm laser line and collecting emission light between 570 and 700 nm. Gain and offset settings in the Leica confocal software (version 2.61) were adjusted so that images of nontransfected fibers appeared essentially black. For insulin stimulation at t = 0 min, a 16.8-μl insulin bolus (656 mU, Actrapid; Novo Nordisk) (in 20% [wt/vol] glucose and 0.01% BSA) were injected in a tail vein, and this was followed by a glucose infusion, which in pilot studies had been shown to maintain euglycemia. For estimation of insulin kinetics, insulin was coinjected with 67 μg/μl sulforhodamine B. For insulin distribution studies, 17 μg/μl sulforhodamine B conjugated insulin (Princeton Biomolecules, Langhorne, PA) was coinjected with Actrapid. The fact that translocation of both GLUT4-GFP and GFP-ARNO was completely intact upon a second stimulation with insulin after 5 h recovery ensured that the procedures did not harm the preparation.
The quadriceps muscles of overnight-fasted anesthetized 7- to 8-week-old C57BL/6 mice were fixed in 2% paraformaldehyde and 0.15% picric acid in 0.1 mol/l Sorensens phosphate buffer (9). After an overnight wash in pH 7.3 PBS, single muscle fibers were isolated by dissection, immmunostained, and mounted as previously described (4). Primary antibody was against insulin receptor (no. ab10991; Abcam, Cambridge, U.K.), and secondary antibodies were conjugated to Alexa-488 (Molecular Probes).
The TIF 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 noizefilter was applied. The degree of GLUT4-GFP translocation to membrane surfaces was quantified by drawing a thin ROI, including sarcolemma or region of t-tubules and excluding vesicular structures. Differences in fluorescence between images at identical time points (e.g., in Fig. 2 in some t-tubule areas in A compared with D) reflect differences in magnification and variation in level of GLUT4-GFP expression within and between individual fibers. In image quantifications, values were expressed relative to values at t = 0 to compensate for such variation. Thus, arbitrary units shown are the actual gray value of ROI fluorescence divided by ROI fluorescence value at t = 0. For sulforhodamine B and GFP-ARNO quantification, pixel values in the raw image stacks were measured in a 2-μm-wide ROI along the length of the fiber. Uncompressed QuickTime movies were produced from the image stacks and subsequently compressed with Sorensen three codec in QuickTime Pro 6.
Dynamic visualization of muscle GLUT4-GFP translocation in anesthetized mice.
To study the dynamics of GLUT4 vesicle trafficking, skeletal muscle of living mice were transfected with GLUT4-GFP by gene gun, and 96 h later the transfected muscle fibers were imaged in their natural surroundings (Fig. 1). GLUT4-GFP has previously been shown to behave as endogenous GLUT4 (7,10). In the basal state GLUT4-GFP was distributed as reported earlier, i.e., in larger and smaller vesicular structures and in the perinuclear area, and with no continuous presence at the sarcolemma and no regular continuous striated pattern indicating uniform presence in the full length of all t-tubules (4) (Fig. 2A). Ten minutes after insulin administration, a clear GLUT4-GFP staining of the sarcolemma was seen, indicating translocation from the intracellular compartments (Fig. 2A). Maximal staining of sarcolemma was reached at 20 min after insulin with no further changes at 30 min. In contrast, after 10 min of insulin only a weakly striated GLUT4-GFP pattern was visible, indicating low t-tubular translocation (4,7). However, at 20 min a continuous and uniform striated pattern was seen, and this reached maximum at 30 min after insulin (Figs. 2A and online appendix movie 1 [available at http://diabetes.diabetesjournal.org]). The delay in maximal GLUT4-GFP translocation to the t-tubules was confirmed by quantification of the time-lapse series (Figs. 2B and C). Translocation resulting in insertion of GLUT4-GFP into the sarcolemma or t-tubule membrane cannot be distinguished from translocation of GLUT4-GFP containing vesicles to the juxtaposition of these membranes (within 0.2 μm).
It has previously been shown that a special pool of small rapidly fusing GLUT4 vesicles is present just below the sarcolemma (4). To clarify whether the faster translocation to sarcolemma than to t-tubules might reflect the presence of these subsarcolemmal vesicles, a ROI around the sarcolemma and part of the t-tubular area was photobleached. Immediately after photobleaching, insulin was injected and the translocation of GLUT4 from surrounding nonbleached areas was followed. In the basal state, the bleached ROI appeared dark at the sarcolemma and the t-tubules, except for a few only partially bleached large GLUT4-GFP structures (Fig. 2D). However, 10 min after insulin a continuous GLUT4-GFP staining along the sarcolemma was visible in the bleached ROI. This staining reached a maximum at 20 min with no further increase 30 min after insulin. In contrast, striated GLUT4-GFP staining of the t-tubules was not visible 10 min after insulin but weakly visible at 20 min and clearly present at 30 min. The visual impression was confirmed by ROI quantification in the bleached area (Fig. 2E). Thus, the delay in GLUT4 translocation between sarcolemma and t-tubules was the same with and without prior bleaching. Accordingly, the delay was not due to a particularly high availability of GLUT4 stores in the vicinity of the sarcolemma, which is in line with the suggestion that rapidly recruitable GLUT4 vesicles also exist along the t-tubules (11).
Alternatively, the translocation delay might reflect differences between GLUT4 stores in either recruitment machinery or signaling pathway. To investigate this, we analyzed if GLUT4-GFP structures in close vicinity to sarcolemma were depleted faster compared with structures in the middle of the muscle fiber. Confocal z stack images were obtained of GLUT4-GFP in the two locations at different time points. Maximal and sum projection images were created from the z stacks. It is seen from maximal projections obtained just below the sarcolemma surface that after 10 min of insulin treatment, a marked depletion of larger GLUT4-GFP structures (i.e., reduction in size of 4- to 20-μm vesicles and spreading to smaller vesicular dots) was already achieved (Fig. 3A). Thirty minutes after insulin, the depletion was slightly further increased. The projection obtained 15 μm deeper inside the muscle fiber showed a delayed depletion pattern, with depletion not being evident until 20–30 min after insulin injection (Fig. 3B). The visual impression was confirmed by quantification (Fig. 3C). Thus, differences between sarcolemma and t-tubules were present in both recruitment (Fig. 3) and translocation (Fig. 2) of GLUT4-GFP. This finding indicates that the delayed response to insulin in t-tubules compared with sarcolemma does not reflect a slower GLUT4 translocation machinery in the interior of muscle fibers compared with surface near areas.
Role of the t-tubules in insulin signaling.
This leads us to speculate that recruitment of GLUT4 to t-tubules is elicited from these and not by insulin signaling from the sarcolemma as generally believed (12,13). While presence of insulin receptors in t-tubules has been indicated by a study using subcellular fractionation, the distribution of insulin receptors within the t-tubules has not been known (14). Accordingly, we immunostained muscle fibers for insulin receptors and found a striated pattern of uniform intensity indicating an even distribution of insulin receptors throughout the t-tubule system (Fig. 4A). To evaluate if insulin can in fact gain access to its receptors deep down the t-tubules, we coinjected insulin with a fluorescent dye, sulforhodamine B. This dye is nontoxic, is not taken up by muscle cells, and has previously been used to load t-tubules in vitro (15,16,17). Time-lapse images of muscle fibers in situ in anesthetized mice showed that sulforhodamine B arrived at the sarcolemma surface within 14 s after intravenous injection (Fig. 4B). Subsequently, the t-tubules were gradually loaded with the dye, resulting in a striated appearance of the fibers, which reached a maximum after ∼4 min (Figs. 4B and online appendix movie 2). Image quantification within a part of the t-tubules located 20 μm from the sarcolemma confirmed the visual impression of diffusion to the deep parts of the t-tubules (Fig. 4C). This is the first demonstration of exchange of molecules between the lumen of the t-tubules and the interstitial fluid in vivo. Diffusion of molecules scales with their hydrodynamic radius. The hydrodynamic radius of sulforhodamine B is not known. However, from direct determination of diffusion coefficients for various substances of different molecular weight, it appears that sulforhodamine B with a molecular weight of 558 D diffuses five times more rapidly in water than insulin with a molecular weight of 5,808 D (18). Accordingly, the 2 min needed for sulforhodamine B to reach the inner t-tubules (Fig. 4C) supports the view that the 10-min delay found for GLUT4 recruitment and translocation reflects the time necessary for insulin diffusion.
To further test the hypothesis that upon administration of insulin, insulin diffusion into the t-tubules is accompanied by a progressing activation of insulin receptors in the t-tubule membranes and, in turn, GLUT4 translocation, we performed confocal time-lapse analysis of the early insulin signaling step, the activation of PI3K (19,20). For this we used the ARNO (21) fused to GFP (8,22). ARNO binds with exquisite specificity via its pleckstrin-homology domain to phosphatidylinositol-3,4,5 triphosphate (PIP3), the product of insulin-activated PI3K (23). By imaging GFP-ARNO in skeletal muscle fibers of living mice, we could dynamically follow the localization of active PI3K with time (Fig. 5A). As expected for a cytosolic protein (8), GFP-ARNO was distributed throughout the muscle cell in the basal state (Fig. 5A). Within ∼1 min after insulin injection, shortly after expected arrival of insulin at the sarcolemma, GFP-ARNO translocated to the sarcolemma. This indicates PIP3 production in response to insulin-mediated activation of PI3K at the sarcolemma. The GFP-ARNO signal had, surprisingly, already disappeared from the sarcolemma 4 min after insulin injection. This is in contrast to previous findings in 3T3-L1 adipocytes in vitro, where GFP-ARNO translocation to the plasma membrane was maintained for at least 10 min after insulin stimulation (22). However, in our in situ muscle setup, GFP-ARNO translocation spread from sarcolemma into the muscle fibers following their crosstriations down into the t-tubules, indicating a progressing PI3K activation and PIP3 production along the t-tubules (Fig. 5A). Because GFP-ARNO translocation also was short-lived in the t-tubules, PIP3 production moved in a wavelike fashion into the muscle fiber, and in the middle part of the fiber PIP3 production had faded ∼18 min after insulin and was completely absent at ∼32 min (Figs. 5A and online appendix movie 3). However, surprisingly, overall the majority of PI3K activity was found in the t-tubules and not in the sarcolemma. Quantification of GFP-ARNO at the sarcolemma and various depths down the t-tubule system confirmed the visually observed progression of GFP-ARNO translocation from sarcolemma to deep parts of t-tubules (Fig. 5B). The GFP-ARNO response strongly supports the view that in vivo insulin diffuses into the t-tubules and there activates receptors even in the deep parts. Interestingly, we have just been able to carry out experiments with fluorescent sulforhodamine B–labeled insulin. These experiments confirmed that insulin from plasma can reach the inner t-tubules (Fig. 6). Furthermore, the appearance of insulin within the t-tubules was closely matched with local GFP-ARNO translocation (Fig. 6).
In the present study, we, for the first time, followed insulin signaling and GLUT4 translocation in real time in skeletal muscle in situ in a living animal. A major new finding was that insulin-mediated GLUT4 translocation is delayed in the t-tubules compared with sarcolemma (Fig. 2). Photobleaching experiments showed that this is not due to differences in availability of GLUT4 storage vesicles between the interior of muscle fibers and surface near areas (Fig. 2D). Furthermore, analysis of GLUT4 depletion in the two locations showed that the delay is also not due to differences in the speed of translocation (Fig. 3). In fact, the delay in GLUT4 translocation between sarcolemma and t-tubules was identical to a delay in depletion of nearby GLUT4 stores. We speculated that the latter delay reflected a delay in insulin signaling. Another major finding of the study was in accordance with this view: insulin does not exclusively activate PI3K in the sarcolemma. Surprisingly, upon insulin stimulation, PI3K activation progressively spreads from the sarcolemma along the t-tubules (Fig. 5). Initially, at t = 2–4 min, the total fluorescence in the focal plane decreased ∼60%, probably reflecting diffusion of GFP-ARNO to sarcolemma areas outside the focal plane. At t = 8–12 min, total fluorescence peaked ∼20% higher than at t = 0, probably reflecting a surplus of GFP-ARNO arriving from outside the focal plane. The presented time course of PIP3 production fits well with previous biochemical measurements on whole white muscles of total insulin-stimulated PI3K activation (24). In contrast to previous studies, however, our temporal and spatial account of insulin signaling within individual muscle cells in situ in living mice surprisingly was able to reveal that t-tubules are, in fact, the major site of active PI3K. The spreading of GFP-ARNO fluorescence into the muscle fibers did not appear completely uniform. Possible explanations for this are differences in rate of insulin diffusion into t-tubules and, in turn, PI3K activation, differences in PI3K distribution, or differences in insulin deactivation within t-tubules.
After insulin injection, maximal PIP3 production was delayed ∼10 min in the deep parts of the t-tubules compared with the sarcolemma (Fig. 5). Interestingly, using the dye sulforhodamine B, we were able to show for the first time in vivo in living mice that molecules from the interstitial space can gain access to the interior of muscle cells through the t-tubules (Figs. 4B and C). From the sulforhodamine B diffusion time adjusted for differences in molecular weights (18), the time needed for insulin to diffuse from the sarcolemma to the deep parts of the t-tubules can be estimated. This calculation showed that the 10-min delay in both PI3K activation and GLUT4 translocation in t-tubules compared with sarcolemma may reflect the time needed for insulin diffusion through the t-tubules. In agreement with the view that insulin signaling may be elicited from the t-tubules, we found insulin receptors uniformly distributed throughout the t-tubule system (Fig. 4A). Furthermore, experiments with sulforhodamine B–labeled insulin confirmed that insulin from plasma can reach the inner t-tubules and that its rate of diffusion into the t-tubules corresponds with values estimated from our sulforhodamine B studies and can account for PI3K activation in the t-tubules (Fig. 6). However, the differences in both time of maximal PIP3 production and arrival of sulforhodamine B dye or dye-labeled insulin between sarcolemma and deep parts of t-tubules (∼20–30 μm from sarcolemma) were significantly longer than predicted by diffusion coefficients in water of insulin (6 μm2/s) and sulforhodamine B (30 μm2/s), respectively (18). This indicates that diffusion into the t-tubules must be quite restricted, possibly due to the biochemical composition of the intraluminal space, which includes macromolecules acting as gels and increasing viscosity in the 30 × 100-nm-narrow t-tubules (25,26). Importantly, in contrast to the present view (13,27,12), our findings indicate that all GLUT4 transporters in muscle cells are controlled by neighboring insulin receptors.
Based on the findings presented, we suggest that the t-tubule system plays a major role in the spreading of insulin stimulation deeply into the muscle fiber stuffed with myofibrillar proteins, which may impede signaling from the sarcolemma. Since the t-tubule system compromises two to three times the surface area of the sarcolemma (26), this system may have a major impact on the overall insulin signaling and GLUT4 recruitment and, in turn, probably also on glucose transport, because the smaller glucose molecule expectedly traverses the t-tubules more easily than sulforhodamine B and insulin. This view is supported by our finding that t-tubule GLUT4 translocation reached its maximal value 30 min after insulin injection, which corresponds well with the time gap of 30–40 min between insulin arrival in the interstitial space in muscle and maximal muscle glucose uptake found by others in humans and animals (28,29). Using immunogold labeling of muscle sections in transgenic mice, insulin-mediated GLUT4 translocation to t-tubules has previously been found (30). In accordance with the proposed role of t-tubules, the same study showed that in incubated normal rat muscle disruption of t-tubules abolished insulin-mediated glucose transport (30).
Thus, it may be speculated that a decrease in the number and accessibility of t-tubules or changes in their luminal matrix impairing diffusion might be primary defects in muscular insulin resistance as seen in, for example, type 2 diabetes. The study also points to the possibility that the t-tubules, in addition to their well-established role in excitation-contraction coupling, play a significant role for hormonal and metabolic events in general in muscle. So, the concept of t-tubules serving only in excitation-contraction coupling may have to be revised and their role in physiology and pathophysiology, e.g., diabetes, reevaluated.
Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Faculty of Health Sciences, University of Copenhagen; the Danish Diabetes Association; The Velux Foundation; the Novo Nordic Research Foundation; and the Danish National Research Foundation.
We thank Dr. J.N. Andersen, Cold Spring Harbor Laboratory (Cold Spring Harbor, NY), for advice and encouragement; L. Kall for bioanalytical assistance; and K. Juel Sørensen and S. Lohmann for engineering assistance.