Subcellular Localization of GLUT4 in Nonstimulated and Insulin-Stimulated Soleus Muscle of Rat ANTJE BORNEMANN, THORKIL PLOUG, AND HENNING SCHMALBRUCH
Soleus muscles of fed rats were fixed by vascular perfusion with paraformaldehyde; individual fibers were teased and immunostained with a polyclonal antibody against the COOH-terminal of GLUT4. The binding sites were visualized by a horseradish peroxidase-coupled secondary antibody and diaminobenzidine. The fibers were embedded in epoxy resin and studied by electron microscopy. Strong immunoreactivity was found in subsarcolemmal clusters of vesicles and cisternae, Golgilike structures, and triadic junctions. Clusters of vesicles between myofibrils were occasionally stained. The plasma membrane was unlabeled. However, the plasma membrane was labeled when the rats had been injected with insulin (40 U/kg body wt) 15 min before perfusion fixation. In non-insulin-injected rats, the plasma membrane might show spotty staining close to clusters of intensely labeled subsarcolemmal vesicles. This may have been due to diffusion but may also indicate that there are domains of GLUT4 in the plasma membrane of nonstimulated fibers or that the endogenous insulin activity to some extent had translocated GLUT4 from the intracellular pool into the plasma membrane. Coated vesicles that were also labeled were found adjacent to subsarcolemmal vesicles and cisternae; it is possible that coated vesicles play a role during insulin- or contraction-induced translocation of GLUT4 between subsarcolemmal pool and plasma membrane. It has been proposed that glucose uptake into skeletal muscle fibers takes place across the t-tubule membrane rather than across the plasma membrane. This would explain the presence of GLUT4 at triadic
From the Department of Neuropathology, University of Mainz, Mainz, Germany; and the Department of Medical Physiology B and the Institute of Neurophysiology, The Panum Institute, University of Copenhagen, Copenhagen, Denmark. Address correspondence and reprint requests to Henning Schmalbruch, MD, Institute of Neurophysiology, The Panum Institute, Blegdamsvej 3 C, DK-2200 Copenhagen N, Denmark. Received for publication 10 July 1991 and accepted in revised form 1 November 1991.
DIABETES, VOL. 41, FEBRUARY 1992
junctions. Alternatively, we suggest that GLUT4 in t-tubules represents a second intracellular pool. Diabetes 41:215-21, 1992
G
lucose transporter molecules mediate the passage of glucose across the phospholipid bilayer of cell membranes. Five functional facilitative glucose transporter isoforms (GLUT1GLUT5) have been identified in different tissues (1). The isoform GLUT4 is supposed to facilitate glucose transport across the plasma membrane of skeletal muscle fibers and fat cells in response to stimulation by insulin (1) and, for skeletal muscle fibers, also by contractile activity (2). The subcellular localization of glucose transporter proteins was initially studied by biochemistry and more recently by immunohistochemistry and immunocytochemistry. Cushman and Wardzala (3) and Suzuki and Kono (4) found that the glucose transporter proteins in fat cells are predominantly localized in intracellular membranes and that they are translocated into the plasma membrane in response to stimulation with insulin. The translocation of GLUT4 in insulin-stimulated brown fat cells of rats was shown by immunocytochemistry by Slot et al. (5). Data for skeletal muscle are conflicting. Vilaro et al. (6) found GLUT4 in endothelial cells. Slot et al. (7) found no evidence for the presence of GLUT4 in endothelial cells and presented light micrographs of rat soleus muscles showing that the immunoreactivity for GLUT4 is confined to regions of the borders of the muscle fibers. In contrast, Friedman et al. (8) found GLUT4 in t-tubules but not in the periphery of the fibers. In this study, we demonstrate the intracellular localization of GLUT4 in fibers of the rat soleus muscle by a preembedding technique. We found GLUT4 in subsarcolemmal clusters of vesicles and cisternae, in Golgilike
215
complexes, at triadic junctions, and, only after stimulation with insulin, at the plasma membrane. RESEARCH DESIGN AND METHODS Animals. Male Wistar rats (250 g) bred by the Animal Department of the Panum Institute (Univ. of Copenhagen) r-••*/.-.£*£ were kept at 21-23°C on a 12-h light-dark cycle. They had free access to food and water. All experiments were performed at -1000. The rats were anesthetized with pentobarbital sodium (50 mg/kg body wt i.p.). "Stimulated" rats received insulin (Actrapid, Novo-Nordisk, Copenhagen; 40 U/kg body wt i.v.) 15 min before being killed. Immunohistochemistry. Rats were killed by cervical dislocation. The soleus muscle was rapidly frozen in a nitrogen slush and 8-|i,m cross sections were cut in a FIG. 1. Rat soleus muscle immunostained with anti-GLUT4 cryostat. Mounted sections were fixed for 20 min in (nonstimulated). Frozen section, horseradish peroxidase. The reaction product is granular and occupies mainly the periphery of muscle methanol and immunolabeled with a polyclonal antibody fibers. It is not possible at this magnification to determine whether against a peptide consisting of the 13 COOH-terminal staining is sarcolemmal or subsarcolemmal. Bar, 100 jim. amino acids of GLUT4 (9) according to standard procedures (10). The binding sites were visualized with either the peroxidase-antiperoxidase technique and diami- appearance (Fig. 1). This was seen after both fluoresnobenzidine (DAB) or a rhodamine-labeled secondary cence labeling and labeling with HRP-DAB. Sometimes the nuclei were surrounded by a rim of reaction product. antibody. The interior of the fibers contained few reactive sites. The Immunocytochemistry. The hindlimbs of the rats were sarcolemma itself and blood vessels and intramuscular perfused through the abdominal aorta, first with ice-cold nerve branches were not labeled. These observations phosphate buffer containing heparin (5000 U/L) and essentially confirmed the results reported by Slot et al. procaine hydrochloride (1 g/L) for 2 min and then with 2% (7). depolymerized paraformaldehyde in phosphate buffer for Semithin sections of epoxy-embedded muscle fibers 10 min. The soleus muscles remained in the fixing that had been stained with anti-GLUT4 by the preembedsolution for 2 h and then were stored in phosphate buffer ding method showed the same granular reaction prodat 4°C. After perfusion with paraformaldehyde, some muscles were perfused with buffer to interrupt fixation. uct, which was subsarcolemmal and preferentially The immunoreactivity for GLUT4 was stronger in these around myonuclei. In addition, longitudinal sections disshort-term-fixed muscles, but the structural preservation closed weak cross striations. High-power dark-field obwas better after 2-h fixation. The localization of reactive servations revealed that the label in some places was sites was the same after both procedures (see RESULTS). situated on both sides of the Z-line. This suggested Fragments of individual muscle fibers were isolated, staining of triadic junctions or t-tubules, which was contreated with 0.5% saponin, and incubated with the firmed by electron microscopy (see below). Muscles from nonstimulated rats and from insulinabove-mentioned antibody against GLUT4 for 2 days at 20°C. The binding sites of the primary antibody were stimulated rats did not differ. At the light-microscope visualized with a horseradish peroxidase (HRP)-labeled level, it was not possible to observe an effect of insulin in secondary antibody against rabbit Ig and DAB. For frozen or epoxy sections. Frozen sections or formaldehyde-fixed fibers incucontrols, the fibers were incubated with rabbit nonimmune serum rather than with the primary antibody. The bated with nonimmune serum, but otherwise processed stained fibers were osmicated and embedded in epoxy as usual, showed weak background activity but no resin. Sections 3 |xm thick were assessed with high- staining of particular structures. resolution light microscopy including dark-field illumina- Electron microscopy: nonstimulated fibers. Lowtion to identify well-preserved and stained fibers. The power micrographs of nonstimulated fibers showed blocks were then trimmed accordingly, and thin sections heavy labeling of groups of vesicles and flattened cisterfor electron microscopy were cut. Thin sections were nae in the subsarcolemmal region and sometimes in the stained with lead citrate. For technical details of the interior of the fiber between myofibrils (Fig. 2). The reactive sites measured up to 1-|j,m in diameter. They preembedding procedure, see ref. 11. occurred rarely in mitochondria-free areas of the fiber periphery (Fig. 3A) but were mostly at nuclear poles, often within subsarcolemmal clusters of mitochondria RESULTS Light microscopy. Frozen cross sections of nonstim- (Fig. 3, B and C). The distance between prominent ulated muscles showed reaction with anti-GLUT4 at the reactive sites localized below the sarcolemma, as seen in nuclear poles and between myonuclei and the sarco- longitudinal sections, ranged between 5 and 15 |xm. In lemma and, to a lesser extent, below the sarcolemma addition, there were small groups of reactive vesicles. between the nuclei. The reaction product had a granular This distribution explains the peripheral rim of reactivity *
:
•
•
.:.."•!:
216
••«fv
DIABETES, VOL. 41, FEBRUARY 1992
ANN. T. PLOUG.
FIG. 2. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). A: subsarcolemmal aggregation of mitochondria. A stack of flattened cistemae (open arrow) resembling Golgi apparatus is stained. Plasma membrane (curved arrow) is unstained. Note weak staining of several triadic junctions (small arrows). B-. a cluster of stained vesicles (open arrow) is seen between myofibrils In interior of a fiber. Several triadic junctions (small arrows) are distinctly stained. Bars, 0.5 jim.
that was seen in frozen cross sections (Fig. 1). Groups of labeled vesicles were squeezed into the narrow gap between myonucleus and plasma membrane. This gap was only 0.2- to 0.5-|im wide, and the reactive vesicles sometimes appeared to indent the nuclear membrane. Most triadic junctions were strongly labeled as well (Fig. 4). The reactive sites beneath the plasma membrane consisted of vesicles 50- to 100-nm diam and of narrow tubules or flattened cistemae. At nuclear poles (Fig. 3B), but also between mitochondria (Fig. 2/4), the reactive elements often resembled Golgi complexes. Many coated vesicles with reaction product were found. These vesicles were associated with Golgilike structures (Fig. 3, C and D) but also with clusters of small vesicles. They were never seen in contact with the plasma membrane or with t-tubules. The reaction product at triadic junctions was bound to
the membrane of the terminal cisterna facing the t-tubulus and also in the gap between t-tubule membrane and terminal cisterna (Fig. 4). Staining of the t-tubule membrane occasionally showed a periodicity of 20-25 nm. This corresponded roughly to the periodicity of the connecting feet of the triadic junctions (12). However, it was not possible to determine whether the stain was deposited at the base of the connecting feet or between them. Most of the plasma membrane of the muscle fibers was unstained (Fig. 2A\ Fig. 3, Sand C), but reaction product attached to the plasma membrane was regularly seen at sites close to stained clusters of subsarcolemmal vesicles (Fig. 3>4). Adjacent to heavily labeled groups of vesicles or flattened cistemae, the outer mitochondria membranes and the nuclear membrane showed spotty staining (Fig. 28; Fig. 3, Sand C). Electron microscopy: insulin-stimulated fibers. After insulin injection, the plasma membrane of the muscle
FIG. 3. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). A: subsarcolemmal cluster of reactive vesicles (open arrow). Plasma membrane close to reactive vesicles is also stained. No staining of capillary (C) endothelium. Bar, 0.5 y-m. ft group of mitochondria at nuclear pole. Several reactive sites (open arrows) consisting of small vesicles are seen between mitochondria. Nucleus (N), plasma membrane (curved arrows), and endothelial cell (below curved arrows) are unstained. Bar, 0.5 |im. C: small groups of stained vesicles and cistemae (open arrows) below plasma membrane (curved arrow). Spotty deposits of reaction product are on surface of mitochondria (small arrows). Bar, 0.2 p,m. D: high magnification of C. Arrows, several coated vesicles with reaction product in connection with flattened cistemae. Mitochondria (M) are unstained. Bar, 0.1 M-m.
DIABETES, VOL. 41, FEBRUARY 1992
217
GLUT4 IN SKELETAL MUSCLE FIBERS
FIG. 3. Legend on previous page.
A. BORNEMANN, T. PLOUG. AND H. SCHMALBRUCH
fibers was heavily labeled, and unlabeled stretches were rare (Fig. 5/4). The labeling of the Golgilike structures and of the subsarcolemmal vesicles did not differ from that in nonstimulated fibers. Triadic junctions appeared stained less often. However, in view of the inconsistencies inherent in immunocytochemical procedures, it was not possible to determine whether there was decrease compared with nonstimulated fibers. Fibers incubated with nonimmune serum rather than the primary antibody were unstained in control rats (Fig. 58). Unspecific DAB precipitates were found in the interstitial space, sometimes attached to basal laminae, but not in muscle fibers. The subcellular structure of muscle fibers treated with nonimmune serum was better preserved than that of fibers immunostained with antiGLUT4 because the DAB reaction damages the ultrastructure because gaseous O2 may form. DISCUSSION Our results demonstrate that anti-GLUT4 in nonstimulated fibers of the rat soleus muscle labels subsarcolemmal groups of vesicles, Golgilike structures at the nuclear poles and between mitochondria (Figs. 2 and 3), and triadic junctions (Fig. 4). Occasionally, clusters of vesicles localized between myofibrils in the interior of the fiber are also labeled (Fig. 28). The plasma membrane of nonstimulated fibers does not react, but when it runs close to clusters of strongly reactive vesicles, it is stained as well (Fig. 3, A and C). The plasma membrane becomes distinctly labeled after injection of insulin (Fig. 5A). Endothelial cells do not react with anti-GLUT4 (Fig. 3A). The product of the DAB reaction becomes attached to membrane surfaces close to reactive sites. This limits the spatial resolution of the preembedding method. Diffusion of the reaction product probably explains the spotty staining of mitochondria and of the nuclear membrane close to reactive sites (Fig. 3, 8 and C). The staining of the plasma membrane above reactive vesicles that was observed in nonstimulated fibers (Fig. 3/4 and C) may be an artifact as well. Alternatively, we suggest that part of the GLUT4 content of skeletal muscle fibers is contained in domains of the plasma membrane or that the endogenous insulin activity of fed rats to some extent had translocated GLUT4 into the plasma membrane. Slot and colleagues (5,7) demonstrated in brown fat cells of rats that the immunoreactivity for GLUT4 is translocated from intracellular membrane systems into the plasma membrane when the rats are given insulin. The same obviously applies to skeletal muscle fibers (Fig. 5/4). The results of biochemical studies also indicate that insulin induces the translocation of GLUT4 from an intracellular pool into the plasma membrane (2,13). Contractile activity obviously has a similar effect as insulin (2), and the spotty staining of the plasma membrane in muscles that had not been stimulated by insulin might
FIG. 4. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). Strong reactivity of triadic junctions. Reaction product is in gap between membrane of t-tubules (solid arrows) and terminal cisternae (open arrows) or attached to terminal cisternae. Bars, 0.2 |im (A) and 0.1 jim (B).
DIABETES, VOL. 41, FEBRUARY 1992
219
GLUT4 IN SKELETAL MUSCLE FIBERS
B
FIG. 5. Rat soieus muscle after insulin stimulation immunostained with anti-GLUT4 (A) and with nonimmune serum (B). A: prominent deposits of reaction product along plasma membrane (curved arrows). (For comparison, see Fig. 2A). Partial staining of outer face of mitochondria (small arrows) is presumably due to diffusion of reaction product. Note reactive vesicles below sarcolemma, some of which are overstained (open arrows). B: micrograph shows sarcolemma (curved arrows) and mitochondria at nuclear pole of muscle fiber treated with nonimmune serum instead of anti-GLUT4. Note complete absence of reaction product. N, nucleus. Bar, 1 n-m.
even be attributed to inadvertent contractile activity during the perfusion procedure. Slot et al. (7) studied the localization of GLUT4 by light microscopy of immunostained frozen sections of the rat soieus muscle. They found the reaction confined to regions of the cell border; the staining pattern disclosed no cross striations corresponding to labeling of t-tubules or triadic junctions. In electron micrographs of cardiac muscle cells, Slot et al. (7) found the label for GLUT4 at tubulovesicular structures beneath the plasma membrane, in association with Golgi complexes, and close to t-tubules. The plasma membrane was not labeled. The observations in cardiac muscle resemble our findings in skeletal muscle. The presence of GLUT4 at triadic junctions of skeletal muscle fibers is in agreement with previous biochemical results (14). However, in a later study (13), the same workers (14) did not find markers for t-tubule membranes in a membrane fraction containing GLUT4. Our material did not disclose whether the antigen was in the membranes of the t-tubule or of the terminal cistemae. The staining intensity of the triadic junctions varied in
220
different areas of the same section, between different preparations, and also between nonstimulated and insulin-stimulated muscles. Because of inconsistencies of all immunocytochemical methods and the technical problems specific for preembedding methods (fixation with loss of antigenicity, poor preservation of the ultrastructure, diffusion gradients), we do not wish to speculate on these differences. Friedman et al. (8) studied human muscle fibers with immunocytochemistry and reported that the label for GLUT4 was bound to triadic junctions and to elements of the sarcoplasmic reticulum; they did not mention the subsarcolemmal membrane complexes that in this study contained the bulk of reactivity. The discrepancy with this study may be due to the fact that different muscles were used but may also reflect technical differences. Friedman et al. stained thin sections of paraformaldehyde-glutaral.dehyde fixed and osmicated tissue that had been embedded in LR white; osmium was not dissolved before the antibody was applied. In our hands, formaldehydefixed and nonosmicated muscle fibers after embedding
DIABETES, VOL 41, FEBRUARY 1992
A. BORNEMANN, T. PLOUG, AND H. SCHMALBRUCH
in LR white had almost completely lost reactivity for GLUT4 (unpublished observations). We observed labeling of many coated vesicles that were close to clusters of smooth vesicles and to Golgilike structures (Fig. 3D) but never in contact with the plasma membrane. Slot et al. (5) found in brown fat cells that anti-GLUT4 labeling of coated vesicles increases after insulin stimulation; they concluded that coated vesicles are involved in retrieval of GLUT4 from the plasma membrane, i.e., in endocytosis. Coated vesicles in myotubes ferry newly synthesized acetylcholine receptor molecules from the Golgi apparatus into the plasma membrane (14). This indicates that coated vesicles in muscle cells may participate in exocytosis. It is unknown whether and how coated vesicles participate in the translocation of GLUT4 in muscle fibers. Friedman et al. (8) suggested that glucose transport takes place across t-tubule membranes rather than across the plasma membrane and that this explains the presence of GLUT4 at these membranes. This hypothesis appears attractive: t-tubules run close to the deposits of intramyofibrillar glycogen (12), and many glycolytic enzymes have been located in the I-band of the myofibrils (16). However, the length and narrowness of the t-tubules may represent an obstacle for the diffusion of glucose. The strong immunoreactivity of the plasma membrane after insulin stimulation confirms biochemical evidence that GLUT4 from an intracellular pool is translocated into this membrane (2,13). The clusters of subsarcolemmal vesicles and the Golgilike structures may represent the source of GLUT4, but the strong reactivity in nonstimulated fibers makes it unlikely that a possible reduction after insulin at these sites can be detected with our staining method. GLUT4 in the membrane of the transverse tubules may be involved in glucose uptake of nonstimulated fibers. Alternatively, the t-tubule membranes may represent a second intracellular GLUT4 pool in addition to the subsarcolemmal vesicles. It is possible that insulin and contractile activity recruit GLUT4 from different sources within the muscle fiber (2). This would agree with the observation that the effects of both stimulation regimes are strictly additive (17). ACKNOWLEDGMENTS This work was supported by the Danish Medical Research Council, the Novo Foundation, the Danish Diabe-
DIABETES, VOL. 41, FEBRUARY 1992
tes Foundation, the Nordic Insulin Foundation, Benthine Lund Fond, and Neuropathology and Applied Neurobiology Journal Bursary. We thank M. Bjaerg for technical assistance.
REFERENCES 1. Bell Gl, Kayano T, Buse JB, Burant CF, Takeda J, Lin D, Fukumoto H, Seino S: Molecular biology of mammalian glucose transporters. Diabetes Care 13:198-208, 1990 2. Douen AG, Ramlal T, Rastogi S, Bilan PJ, Cartee GD, Vranic M, Holloszy JO, Klip A: Exercise induces recruitment of the "insulinresponsive glucose transporter": evidence for distinct intracellular insulin- and exercise-recruitable transporter pools in skeletal muscle. J Biol Chem 265:13427-30, 1990 3. Cushman SW, Wardzala LJ: Potential mechanism of insulin action on glucose transport in the isolated rat adipose cell. J Biol Chem 255:4758-62, 1980 4. Suzuki K, Kono T: Evidence that insulin causes translocation of glucose transport activity to the plasma membrane from an intracellular storage site. Proc Natl Acad Sci USA 77:2542-45, 1980 5. Slot JW, Geuze HJ, Gigengack S, Lienhard GE, James DE: Immunolocalization of the insulin regulatable glucose transporter in brown adipose tissue of the rat. J Cell-Biol 113:123-35, 1991 6. Vilaro S, Palacin M, Pilch PF, Testar X, Zorzano A: Expression of an insulin-regulatable glucose carrier in muscle and fat endothelial cells. Nature (Lond) 342:798-800, 1989 7. Slot JW, Moxley R, Geuze HJ, James DE: No evidence for expression of the insulin-regulatable glucose transporter in endothelial cells. Nature (Lond) 346:369-71, 1990 8. Friedman JE, Dudek RW, Whitehead DS, Downes DL, Frisell WR, Caro JF, Dohm GL: Immunolocalization of glucose transporter GLUT4 within human skeletal muscle. Diabetes 40:150-54, 1991 9. Ploug T, Stallknecht BM, Pedersen O, Kahn BB, Ohkuwa T, Vinten J, Galbo H: Effect of endurance training on glucose transport capacity and glucose transporter expression in rat skeletal muscle. Am J Physiol 259:E778-86, 1990 10. Polak JM, Van Noorden S: An Introduction to Immunocytochemistry: Current Techniques and Problems. Oxford, UK, Oxford Univ. Press, 1987 11. Bornemann A, Schmalbruch H: Desmin and vimentin in regenerating muscles. Muscle Nerve. In press 12. Schmalbruch H: Skeletal Muscle. New York, Springer-Verlag, 1985 13. Douen AG, Burdett E, Ramlal T, Rastogi S, Vranic M, Klip A: Characterization of glucose transporter-enriched membranes from rat skeletal muscle: assessment of endothelial cell contamination and presence of sarcoplasmic reticulum and transverse tubules. Endocrinology 128:611-16,1991 14. Burdett E, Beeler T, Klip A: Distribution of glucose transporters and insulin receptors in the plasma membrane and transverse tubules of skeletal muscle. Arch Biochem Biophys 253:279-86, 1987 15. Bursztajn S, Fischbach GD: Evidence that coated vesicles transport acetylcholine receptors to the surface membrane of chick myotubes. J Cell Biol 98:498-506, 1984 16. Sigel P, Pette D: Intracellular localization of glycogenolytic and glycolytic enzymes in white and red rabbit skeletal muscle: a gel film method for coupled enzyme reactions in histochemistry. J Histochem Cytochem 17:225-37, 1969 17. Ploug T, Galbo H, Vinten J, Jorgensen M, Richter EA: Kinetics of glucose transport in rat muscle: effects of insulin and contractions. Am J Physiol 253:E12-20, 1987
221
Soleus muscles of fed rats were fixed by vascular perfusion with paraformaldehyde; individual fibers were teased and immunostained with a polyclonal antibody against the COOH-terminal of GLUT4. The binding sites were visualized by a horseradish peroxidase-coupled secondary antibody and diaminobenzidine. The fibers were embedded in epoxy resin and studied by electron microscopy. Strong immunoreactivity was found in subsarcolemmal clusters of vesicles and cisternae, Golgilike structures, and triadic junctions. Clusters of vesicles between myofibrils were occasionally stained. The plasma membrane was unlabeled. However, the plasma membrane was labeled when the rats had been injected with insulin (40 U/kg body wt) 15 min before perfusion fixation. In non-insulin-injected rats, the plasma membrane might show spotty staining close to clusters of intensely labeled subsarcolemmal vesicles. This may have been due to diffusion but may also indicate that there are domains of GLUT4 in the plasma membrane of nonstimulated fibers or that the endogenous insulin activity to some extent had translocated GLUT4 from the intracellular pool into the plasma membrane. Coated vesicles that were also labeled were found adjacent to subsarcolemmal vesicles and cisternae; it is possible that coated vesicles play a role during insulin- or contraction-induced translocation of GLUT4 between subsarcolemmal pool and plasma membrane. It has been proposed that glucose uptake into skeletal muscle fibers takes place across the t-tubule membrane rather than across the plasma membrane. This would explain the presence of GLUT4 at triadic
From the Department of Neuropathology, University of Mainz, Mainz, Germany; and the Department of Medical Physiology B and the Institute of Neurophysiology, The Panum Institute, University of Copenhagen, Copenhagen, Denmark. Address correspondence and reprint requests to Henning Schmalbruch, MD, Institute of Neurophysiology, The Panum Institute, Blegdamsvej 3 C, DK-2200 Copenhagen N, Denmark. Received for publication 10 July 1991 and accepted in revised form 1 November 1991.
DIABETES, VOL. 41, FEBRUARY 1992
junctions. Alternatively, we suggest that GLUT4 in t-tubules represents a second intracellular pool. Diabetes 41:215-21, 1992
G
lucose transporter molecules mediate the passage of glucose across the phospholipid bilayer of cell membranes. Five functional facilitative glucose transporter isoforms (GLUT1GLUT5) have been identified in different tissues (1). The isoform GLUT4 is supposed to facilitate glucose transport across the plasma membrane of skeletal muscle fibers and fat cells in response to stimulation by insulin (1) and, for skeletal muscle fibers, also by contractile activity (2). The subcellular localization of glucose transporter proteins was initially studied by biochemistry and more recently by immunohistochemistry and immunocytochemistry. Cushman and Wardzala (3) and Suzuki and Kono (4) found that the glucose transporter proteins in fat cells are predominantly localized in intracellular membranes and that they are translocated into the plasma membrane in response to stimulation with insulin. The translocation of GLUT4 in insulin-stimulated brown fat cells of rats was shown by immunocytochemistry by Slot et al. (5). Data for skeletal muscle are conflicting. Vilaro et al. (6) found GLUT4 in endothelial cells. Slot et al. (7) found no evidence for the presence of GLUT4 in endothelial cells and presented light micrographs of rat soleus muscles showing that the immunoreactivity for GLUT4 is confined to regions of the borders of the muscle fibers. In contrast, Friedman et al. (8) found GLUT4 in t-tubules but not in the periphery of the fibers. In this study, we demonstrate the intracellular localization of GLUT4 in fibers of the rat soleus muscle by a preembedding technique. We found GLUT4 in subsarcolemmal clusters of vesicles and cisternae, in Golgilike
215
complexes, at triadic junctions, and, only after stimulation with insulin, at the plasma membrane. RESEARCH DESIGN AND METHODS Animals. Male Wistar rats (250 g) bred by the Animal Department of the Panum Institute (Univ. of Copenhagen) r-••*/.-.£*£ were kept at 21-23°C on a 12-h light-dark cycle. They had free access to food and water. All experiments were performed at -1000. The rats were anesthetized with pentobarbital sodium (50 mg/kg body wt i.p.). "Stimulated" rats received insulin (Actrapid, Novo-Nordisk, Copenhagen; 40 U/kg body wt i.v.) 15 min before being killed. Immunohistochemistry. Rats were killed by cervical dislocation. The soleus muscle was rapidly frozen in a nitrogen slush and 8-|i,m cross sections were cut in a FIG. 1. Rat soleus muscle immunostained with anti-GLUT4 cryostat. Mounted sections were fixed for 20 min in (nonstimulated). Frozen section, horseradish peroxidase. The reaction product is granular and occupies mainly the periphery of muscle methanol and immunolabeled with a polyclonal antibody fibers. It is not possible at this magnification to determine whether against a peptide consisting of the 13 COOH-terminal staining is sarcolemmal or subsarcolemmal. Bar, 100 jim. amino acids of GLUT4 (9) according to standard procedures (10). The binding sites were visualized with either the peroxidase-antiperoxidase technique and diami- appearance (Fig. 1). This was seen after both fluoresnobenzidine (DAB) or a rhodamine-labeled secondary cence labeling and labeling with HRP-DAB. Sometimes the nuclei were surrounded by a rim of reaction product. antibody. The interior of the fibers contained few reactive sites. The Immunocytochemistry. The hindlimbs of the rats were sarcolemma itself and blood vessels and intramuscular perfused through the abdominal aorta, first with ice-cold nerve branches were not labeled. These observations phosphate buffer containing heparin (5000 U/L) and essentially confirmed the results reported by Slot et al. procaine hydrochloride (1 g/L) for 2 min and then with 2% (7). depolymerized paraformaldehyde in phosphate buffer for Semithin sections of epoxy-embedded muscle fibers 10 min. The soleus muscles remained in the fixing that had been stained with anti-GLUT4 by the preembedsolution for 2 h and then were stored in phosphate buffer ding method showed the same granular reaction prodat 4°C. After perfusion with paraformaldehyde, some muscles were perfused with buffer to interrupt fixation. uct, which was subsarcolemmal and preferentially The immunoreactivity for GLUT4 was stronger in these around myonuclei. In addition, longitudinal sections disshort-term-fixed muscles, but the structural preservation closed weak cross striations. High-power dark-field obwas better after 2-h fixation. The localization of reactive servations revealed that the label in some places was sites was the same after both procedures (see RESULTS). situated on both sides of the Z-line. This suggested Fragments of individual muscle fibers were isolated, staining of triadic junctions or t-tubules, which was contreated with 0.5% saponin, and incubated with the firmed by electron microscopy (see below). Muscles from nonstimulated rats and from insulinabove-mentioned antibody against GLUT4 for 2 days at 20°C. The binding sites of the primary antibody were stimulated rats did not differ. At the light-microscope visualized with a horseradish peroxidase (HRP)-labeled level, it was not possible to observe an effect of insulin in secondary antibody against rabbit Ig and DAB. For frozen or epoxy sections. Frozen sections or formaldehyde-fixed fibers incucontrols, the fibers were incubated with rabbit nonimmune serum rather than with the primary antibody. The bated with nonimmune serum, but otherwise processed stained fibers were osmicated and embedded in epoxy as usual, showed weak background activity but no resin. Sections 3 |xm thick were assessed with high- staining of particular structures. resolution light microscopy including dark-field illumina- Electron microscopy: nonstimulated fibers. Lowtion to identify well-preserved and stained fibers. The power micrographs of nonstimulated fibers showed blocks were then trimmed accordingly, and thin sections heavy labeling of groups of vesicles and flattened cisterfor electron microscopy were cut. Thin sections were nae in the subsarcolemmal region and sometimes in the stained with lead citrate. For technical details of the interior of the fiber between myofibrils (Fig. 2). The reactive sites measured up to 1-|j,m in diameter. They preembedding procedure, see ref. 11. occurred rarely in mitochondria-free areas of the fiber periphery (Fig. 3A) but were mostly at nuclear poles, often within subsarcolemmal clusters of mitochondria RESULTS Light microscopy. Frozen cross sections of nonstim- (Fig. 3, B and C). The distance between prominent ulated muscles showed reaction with anti-GLUT4 at the reactive sites localized below the sarcolemma, as seen in nuclear poles and between myonuclei and the sarco- longitudinal sections, ranged between 5 and 15 |xm. In lemma and, to a lesser extent, below the sarcolemma addition, there were small groups of reactive vesicles. between the nuclei. The reaction product had a granular This distribution explains the peripheral rim of reactivity *
:
•
•
.:.."•!:
216
••«fv
DIABETES, VOL. 41, FEBRUARY 1992
ANN. T. PLOUG.
FIG. 2. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). A: subsarcolemmal aggregation of mitochondria. A stack of flattened cistemae (open arrow) resembling Golgi apparatus is stained. Plasma membrane (curved arrow) is unstained. Note weak staining of several triadic junctions (small arrows). B-. a cluster of stained vesicles (open arrow) is seen between myofibrils In interior of a fiber. Several triadic junctions (small arrows) are distinctly stained. Bars, 0.5 jim.
that was seen in frozen cross sections (Fig. 1). Groups of labeled vesicles were squeezed into the narrow gap between myonucleus and plasma membrane. This gap was only 0.2- to 0.5-|im wide, and the reactive vesicles sometimes appeared to indent the nuclear membrane. Most triadic junctions were strongly labeled as well (Fig. 4). The reactive sites beneath the plasma membrane consisted of vesicles 50- to 100-nm diam and of narrow tubules or flattened cistemae. At nuclear poles (Fig. 3B), but also between mitochondria (Fig. 2/4), the reactive elements often resembled Golgi complexes. Many coated vesicles with reaction product were found. These vesicles were associated with Golgilike structures (Fig. 3, C and D) but also with clusters of small vesicles. They were never seen in contact with the plasma membrane or with t-tubules. The reaction product at triadic junctions was bound to
the membrane of the terminal cisterna facing the t-tubulus and also in the gap between t-tubule membrane and terminal cisterna (Fig. 4). Staining of the t-tubule membrane occasionally showed a periodicity of 20-25 nm. This corresponded roughly to the periodicity of the connecting feet of the triadic junctions (12). However, it was not possible to determine whether the stain was deposited at the base of the connecting feet or between them. Most of the plasma membrane of the muscle fibers was unstained (Fig. 2A\ Fig. 3, Sand C), but reaction product attached to the plasma membrane was regularly seen at sites close to stained clusters of subsarcolemmal vesicles (Fig. 3>4). Adjacent to heavily labeled groups of vesicles or flattened cistemae, the outer mitochondria membranes and the nuclear membrane showed spotty staining (Fig. 28; Fig. 3, Sand C). Electron microscopy: insulin-stimulated fibers. After insulin injection, the plasma membrane of the muscle
FIG. 3. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). A: subsarcolemmal cluster of reactive vesicles (open arrow). Plasma membrane close to reactive vesicles is also stained. No staining of capillary (C) endothelium. Bar, 0.5 y-m. ft group of mitochondria at nuclear pole. Several reactive sites (open arrows) consisting of small vesicles are seen between mitochondria. Nucleus (N), plasma membrane (curved arrows), and endothelial cell (below curved arrows) are unstained. Bar, 0.5 |im. C: small groups of stained vesicles and cistemae (open arrows) below plasma membrane (curved arrow). Spotty deposits of reaction product are on surface of mitochondria (small arrows). Bar, 0.2 p,m. D: high magnification of C. Arrows, several coated vesicles with reaction product in connection with flattened cistemae. Mitochondria (M) are unstained. Bar, 0.1 M-m.
DIABETES, VOL. 41, FEBRUARY 1992
217
GLUT4 IN SKELETAL MUSCLE FIBERS
FIG. 3. Legend on previous page.
A. BORNEMANN, T. PLOUG. AND H. SCHMALBRUCH
fibers was heavily labeled, and unlabeled stretches were rare (Fig. 5/4). The labeling of the Golgilike structures and of the subsarcolemmal vesicles did not differ from that in nonstimulated fibers. Triadic junctions appeared stained less often. However, in view of the inconsistencies inherent in immunocytochemical procedures, it was not possible to determine whether there was decrease compared with nonstimulated fibers. Fibers incubated with nonimmune serum rather than the primary antibody were unstained in control rats (Fig. 58). Unspecific DAB precipitates were found in the interstitial space, sometimes attached to basal laminae, but not in muscle fibers. The subcellular structure of muscle fibers treated with nonimmune serum was better preserved than that of fibers immunostained with antiGLUT4 because the DAB reaction damages the ultrastructure because gaseous O2 may form. DISCUSSION Our results demonstrate that anti-GLUT4 in nonstimulated fibers of the rat soleus muscle labels subsarcolemmal groups of vesicles, Golgilike structures at the nuclear poles and between mitochondria (Figs. 2 and 3), and triadic junctions (Fig. 4). Occasionally, clusters of vesicles localized between myofibrils in the interior of the fiber are also labeled (Fig. 28). The plasma membrane of nonstimulated fibers does not react, but when it runs close to clusters of strongly reactive vesicles, it is stained as well (Fig. 3, A and C). The plasma membrane becomes distinctly labeled after injection of insulin (Fig. 5A). Endothelial cells do not react with anti-GLUT4 (Fig. 3A). The product of the DAB reaction becomes attached to membrane surfaces close to reactive sites. This limits the spatial resolution of the preembedding method. Diffusion of the reaction product probably explains the spotty staining of mitochondria and of the nuclear membrane close to reactive sites (Fig. 3, 8 and C). The staining of the plasma membrane above reactive vesicles that was observed in nonstimulated fibers (Fig. 3/4 and C) may be an artifact as well. Alternatively, we suggest that part of the GLUT4 content of skeletal muscle fibers is contained in domains of the plasma membrane or that the endogenous insulin activity of fed rats to some extent had translocated GLUT4 into the plasma membrane. Slot and colleagues (5,7) demonstrated in brown fat cells of rats that the immunoreactivity for GLUT4 is translocated from intracellular membrane systems into the plasma membrane when the rats are given insulin. The same obviously applies to skeletal muscle fibers (Fig. 5/4). The results of biochemical studies also indicate that insulin induces the translocation of GLUT4 from an intracellular pool into the plasma membrane (2,13). Contractile activity obviously has a similar effect as insulin (2), and the spotty staining of the plasma membrane in muscles that had not been stimulated by insulin might
FIG. 4. Rat soleus muscle immunostained with anti-GLUT4 (nonstimulated). Strong reactivity of triadic junctions. Reaction product is in gap between membrane of t-tubules (solid arrows) and terminal cisternae (open arrows) or attached to terminal cisternae. Bars, 0.2 |im (A) and 0.1 jim (B).
DIABETES, VOL. 41, FEBRUARY 1992
219
GLUT4 IN SKELETAL MUSCLE FIBERS
B
FIG. 5. Rat soieus muscle after insulin stimulation immunostained with anti-GLUT4 (A) and with nonimmune serum (B). A: prominent deposits of reaction product along plasma membrane (curved arrows). (For comparison, see Fig. 2A). Partial staining of outer face of mitochondria (small arrows) is presumably due to diffusion of reaction product. Note reactive vesicles below sarcolemma, some of which are overstained (open arrows). B: micrograph shows sarcolemma (curved arrows) and mitochondria at nuclear pole of muscle fiber treated with nonimmune serum instead of anti-GLUT4. Note complete absence of reaction product. N, nucleus. Bar, 1 n-m.
even be attributed to inadvertent contractile activity during the perfusion procedure. Slot et al. (7) studied the localization of GLUT4 by light microscopy of immunostained frozen sections of the rat soieus muscle. They found the reaction confined to regions of the cell border; the staining pattern disclosed no cross striations corresponding to labeling of t-tubules or triadic junctions. In electron micrographs of cardiac muscle cells, Slot et al. (7) found the label for GLUT4 at tubulovesicular structures beneath the plasma membrane, in association with Golgi complexes, and close to t-tubules. The plasma membrane was not labeled. The observations in cardiac muscle resemble our findings in skeletal muscle. The presence of GLUT4 at triadic junctions of skeletal muscle fibers is in agreement with previous biochemical results (14). However, in a later study (13), the same workers (14) did not find markers for t-tubule membranes in a membrane fraction containing GLUT4. Our material did not disclose whether the antigen was in the membranes of the t-tubule or of the terminal cistemae. The staining intensity of the triadic junctions varied in
220
different areas of the same section, between different preparations, and also between nonstimulated and insulin-stimulated muscles. Because of inconsistencies of all immunocytochemical methods and the technical problems specific for preembedding methods (fixation with loss of antigenicity, poor preservation of the ultrastructure, diffusion gradients), we do not wish to speculate on these differences. Friedman et al. (8) studied human muscle fibers with immunocytochemistry and reported that the label for GLUT4 was bound to triadic junctions and to elements of the sarcoplasmic reticulum; they did not mention the subsarcolemmal membrane complexes that in this study contained the bulk of reactivity. The discrepancy with this study may be due to the fact that different muscles were used but may also reflect technical differences. Friedman et al. stained thin sections of paraformaldehyde-glutaral.dehyde fixed and osmicated tissue that had been embedded in LR white; osmium was not dissolved before the antibody was applied. In our hands, formaldehydefixed and nonosmicated muscle fibers after embedding
DIABETES, VOL 41, FEBRUARY 1992
A. BORNEMANN, T. PLOUG, AND H. SCHMALBRUCH
in LR white had almost completely lost reactivity for GLUT4 (unpublished observations). We observed labeling of many coated vesicles that were close to clusters of smooth vesicles and to Golgilike structures (Fig. 3D) but never in contact with the plasma membrane. Slot et al. (5) found in brown fat cells that anti-GLUT4 labeling of coated vesicles increases after insulin stimulation; they concluded that coated vesicles are involved in retrieval of GLUT4 from the plasma membrane, i.e., in endocytosis. Coated vesicles in myotubes ferry newly synthesized acetylcholine receptor molecules from the Golgi apparatus into the plasma membrane (14). This indicates that coated vesicles in muscle cells may participate in exocytosis. It is unknown whether and how coated vesicles participate in the translocation of GLUT4 in muscle fibers. Friedman et al. (8) suggested that glucose transport takes place across t-tubule membranes rather than across the plasma membrane and that this explains the presence of GLUT4 at these membranes. This hypothesis appears attractive: t-tubules run close to the deposits of intramyofibrillar glycogen (12), and many glycolytic enzymes have been located in the I-band of the myofibrils (16). However, the length and narrowness of the t-tubules may represent an obstacle for the diffusion of glucose. The strong immunoreactivity of the plasma membrane after insulin stimulation confirms biochemical evidence that GLUT4 from an intracellular pool is translocated into this membrane (2,13). The clusters of subsarcolemmal vesicles and the Golgilike structures may represent the source of GLUT4, but the strong reactivity in nonstimulated fibers makes it unlikely that a possible reduction after insulin at these sites can be detected with our staining method. GLUT4 in the membrane of the transverse tubules may be involved in glucose uptake of nonstimulated fibers. Alternatively, the t-tubule membranes may represent a second intracellular GLUT4 pool in addition to the subsarcolemmal vesicles. It is possible that insulin and contractile activity recruit GLUT4 from different sources within the muscle fiber (2). This would agree with the observation that the effects of both stimulation regimes are strictly additive (17). ACKNOWLEDGMENTS This work was supported by the Danish Medical Research Council, the Novo Foundation, the Danish Diabe-
DIABETES, VOL. 41, FEBRUARY 1992
tes Foundation, the Nordic Insulin Foundation, Benthine Lund Fond, and Neuropathology and Applied Neurobiology Journal Bursary. We thank M. Bjaerg for technical assistance.
REFERENCES 1. Bell Gl, Kayano T, Buse JB, Burant CF, Takeda J, Lin D, Fukumoto H, Seino S: Molecular biology of mammalian glucose transporters. Diabetes Care 13:198-208, 1990 2. Douen AG, Ramlal T, Rastogi S, Bilan PJ, Cartee GD, Vranic M, Holloszy JO, Klip A: Exercise induces recruitment of the "insulinresponsive glucose transporter": evidence for distinct intracellular insulin- and exercise-recruitable transporter pools in skeletal muscle. J Biol Chem 265:13427-30, 1990 3. Cushman SW, Wardzala LJ: Potential mechanism of insulin action on glucose transport in the isolated rat adipose cell. J Biol Chem 255:4758-62, 1980 4. Suzuki K, Kono T: Evidence that insulin causes translocation of glucose transport activity to the plasma membrane from an intracellular storage site. Proc Natl Acad Sci USA 77:2542-45, 1980 5. Slot JW, Geuze HJ, Gigengack S, Lienhard GE, James DE: Immunolocalization of the insulin regulatable glucose transporter in brown adipose tissue of the rat. J Cell-Biol 113:123-35, 1991 6. Vilaro S, Palacin M, Pilch PF, Testar X, Zorzano A: Expression of an insulin-regulatable glucose carrier in muscle and fat endothelial cells. Nature (Lond) 342:798-800, 1989 7. Slot JW, Moxley R, Geuze HJ, James DE: No evidence for expression of the insulin-regulatable glucose transporter in endothelial cells. Nature (Lond) 346:369-71, 1990 8. Friedman JE, Dudek RW, Whitehead DS, Downes DL, Frisell WR, Caro JF, Dohm GL: Immunolocalization of glucose transporter GLUT4 within human skeletal muscle. Diabetes 40:150-54, 1991 9. Ploug T, Stallknecht BM, Pedersen O, Kahn BB, Ohkuwa T, Vinten J, Galbo H: Effect of endurance training on glucose transport capacity and glucose transporter expression in rat skeletal muscle. Am J Physiol 259:E778-86, 1990 10. Polak JM, Van Noorden S: An Introduction to Immunocytochemistry: Current Techniques and Problems. Oxford, UK, Oxford Univ. Press, 1987 11. Bornemann A, Schmalbruch H: Desmin and vimentin in regenerating muscles. Muscle Nerve. In press 12. Schmalbruch H: Skeletal Muscle. New York, Springer-Verlag, 1985 13. Douen AG, Burdett E, Ramlal T, Rastogi S, Vranic M, Klip A: Characterization of glucose transporter-enriched membranes from rat skeletal muscle: assessment of endothelial cell contamination and presence of sarcoplasmic reticulum and transverse tubules. Endocrinology 128:611-16,1991 14. Burdett E, Beeler T, Klip A: Distribution of glucose transporters and insulin receptors in the plasma membrane and transverse tubules of skeletal muscle. Arch Biochem Biophys 253:279-86, 1987 15. Bursztajn S, Fischbach GD: Evidence that coated vesicles transport acetylcholine receptors to the surface membrane of chick myotubes. J Cell Biol 98:498-506, 1984 16. Sigel P, Pette D: Intracellular localization of glycogenolytic and glycolytic enzymes in white and red rabbit skeletal muscle: a gel film method for coupled enzyme reactions in histochemistry. J Histochem Cytochem 17:225-37, 1969 17. Ploug T, Galbo H, Vinten J, Jorgensen M, Richter EA: Kinetics of glucose transport in rat muscle: effects of insulin and contractions. Am J Physiol 253:E12-20, 1987
221
