Glucose Transporter Number, Function, and Subcellular Distribution in Rat Skeletal Muscle After Exercise Training LAURIE J. GOODYEAR, MICHAEL F. HIRSHMAN, PATRICIA M. VALYOU, AND EDWARD S. HORTON
Endurance exercise training can result in increased rates of insulin-stimulated glucose uptake in skeletal muscle; however, this effect may be lost rapidly once training ceases. To examine a mechanism for these changes, the skeletal-muscle glucose transport system of female rats exercise-trained in wheelcages for 6 wk were studied against a group of untrained female rats. The trained rats were studied immediately following and 2 and 5 days after removal from wheelcages; both trained and untrained rats were studied 30 min after insulin (90 nmol/rat, intraperitoneal) or saline injection. The total number of skeletal-muscle plasma-membrane glucose transporters (Ro), total muscle- homogenate and plasma-membrane GLUT4 protein, and rates of plasma-membrane vesicle D-facilitated glucose transport were higher in the exercise-trained rats immediately after exercise training and did not decrease significantly during the 5 days after cessation of training. On the other hand, exercise training did not alter microsomal-membrane total glucose-transporter number or GLUT4 protein, nor did training alter GLUT1 protein in total muscle homogenates nor either membrane fraction. The carrier-turnover number, an estimate of average functional activity of glucose transporters in the plasma membrane, was elevated slightly, but not significantly, in the trained muscle. In both the trained and untrained muscle, insulin administration resulted in translocation of glucose transporters from the microsomal-membrane fraction to the plasma membrane and an increase in the carrier-turnover
From the Metabolic Unit, Department of Medicine, University of Vermont, Burlington. Address correspondence and reprint requests to M.F. Hirshman, C-350 Given Bldg., Dept. of Medicine, University of Vermont, Burlington, VT 05405. Received for publication 16 September 1991 and accepted in revised form 30 March 1992. SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TBS, Tris-buffered saline; KpNPPase, K+-stimulated-p-nitrophenol phosphatase; TO, rat group studied immediately after removal from wheelcages; T2, rat group studied 2 days after removal from wheelcages; T5, rat groups studied 5 days after removal from wheelcages; Ro, total number of glucose transporters; ANOVA, analysis of variance.
DIABETES, VOL. 41, SEPTEMBER 1992
number. These data suggest that increased rates of glucose uptake in endurance-trained skeletal muscle results primarily from an increase in the number—and not an increase in the average functional activity—of glucose transporters present in the plasma membrane. Furthermore, these increases persist for several days after cessation of exercise training. The specific increase in the GLUT4, but not the GLUT1 glucose-transporter isoform, in response to training demonstrates that a common, chronic physiological stimulus can regulate the expression of the two glucose-transporter isoforms present in skeletal-muscle tissue differentially. Diabetes 41:1091-99,1992
G
lucose transport is the rate-limiting step for skeletal-muscle glucose use under normal physiological conditions (1). Rates of glucose transport can be altered by a change in the number of carrier proteins present in the plasma membrane and/or a change in the turnover number of the carrier proteins (glucose transporters). Recent studies in rat skeletal muscle have demonstrated that both glucosetransporter number and carrier-turnover number are increased by insulin stimulation (2-4) and a single bout of acute exercise (2,5,6). In addition, of the two genetically distinct glucose-transporter isoforms thought to be expressed in rat skeletal-muscle tissue (GLUT1, GLUT4), acute exercise and insulin stimulation have been shown to increase the plasma-membrane content of the GLUT4 isoform selectively (7-9). Skeletal muscle is the predominant tissue responsible for insulin-stimulated glucose disposal and a major site of insulin resistance in diabetes. While diabetes results in a decrease in skeletal-muscle insulin sensitivity, endurance exercise training increases the sensitivity and/or responsiveness of skeletal-muscle glucose uptake to insulin (10-12). However, some evidence shows that in both human and rat models, this adaptation to chronic
1091
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
exercise lasts only 24-60 h after the last training session (13-15). The mechanism for these adaptations to both training and detraining may be related to changes in the total number, activity, or subcellular distribution of glucose transporters, the ability of insulin or contractile activity to cause the translocation and fusion of transporters into the plasma membrane, and/or a change in the amount of a specific glucose-transporter isoform. Previous reports have shown that exercise training increases GLUT4 mRNA (16,17) and GLUT4 protein (9,17) in rat skeletal-muscle homogenates. However, these studies did not determine if exercise training alters the other components of the glucose-transport system. In this study, we measured the effects of 6 wk of wheelcage training on total glucose-transporter number and activity, and the distribution of GLUT4- and GLUT1-transporter isoforms. Furthermore, we evaluated these parameters at three time points after the cessation of training to determine if changes in transporter number, distribution, and/or activity parallel the reported changes in insulin sensitivity and responsiveness.
mM sucrose, pH 7.6; it then was polytroned at slow speed, homogenized, and diluted. The homogenate was centrifuged at 35,000 g for 20 min. The resulting pellet was used to prepare plasma membranes; the supernatant was used to prepare microsomal membranes (4). Cytochalasin B binding. Equilibrium D-glucose-inhibitable [3H]-cytochalasin B binding was measured, and the concentration of glucose transporters was determined (18). Scatchard plots were generated from binding studies in which membranes were incubated with cytochalasin B in the presence or absence of 500 mM D-glucose. Cytochalasin E (2000 nM) was present to decrease nonspecific binding. Bound cytochalasin B was separated from free cytochalasin B by centrifugation. Tracer amounts of [14C]urea were used to correct for [3H]cytochalasin B trapped in the pellet. The total number of glucose transporters {Ro) and the dissociation constant (K"d) were determined from a linear plot derived by subtraction along the radial axes of binding curves generated in the presence of D-glucose from those in the absence of D-glucose. Glucose-transport activity in plasma-membrane vesicles. D-[14C(U)]-glucose and i_-[3H(N)]-glucose uptake RESEARCH DESIGN AND METHODS in plasma-membrane vesicles were determined under Animal care and exercise training. Female Sprague- equilibrium exchange conditions using a rapid filtration Dawley rats were received at body wt ranging from technique at 25°C (6). Membranes were preequilibrated 100-150 g and maintained with ad libitum feeding (Pu- with a HEPES-buffered Krebs Ringer solution containing rina Laboratory Chow, Ralston-Purina, St. Louis, MO) for 1, 5, 10, 20, 40, or 60 mM i_- and D-glucose. Uptake was several days before placement into treatment groups. measured at four time points for each preparation (1-5 One group of rats (n = 52) was housed individually in sec). Transport was initiated by combining an aliquot of wheelcages where voluntary access to physical exercise the membranes with HEPES-buffered Krebs solution and was available at all times. The total number of wheelcage 1-60 mM D- and L-glucose with 6 jxCi of L-[ 3 H(N)]revolutions was recorded 3 times/wk, and the accumu- glucose and 1.6 ixCi of d-[14C(U)]-glucose. Transport lated distance was calculated at the end of each wk. was stopped by the addition of HEPES-buffered Krebs Age-matched control rats (n = 26) were housed in indi- containing 0.2 mM phloretin. The membranes were filvidual cages and did not have access to a running wheel. tered rapidly (Millipore HA 0.45 |xm) and washed; the Food consumption and body wt were monitored twice filter and adhering membranes were analyzed by liquid weekly throughout the 6-wk study period. scintillation counting using quench correction for the dual The trained rats were subdivided into three groups; label. Facilitated transport was calculated by subtracting one group was studied immediately after removal from the initial rate of L-glucose influx from that of D-glucose. wheelcages (TO), a second group was studied 2 days Vmax and KV2 were determined using nonlinear least after removal from wheelcages (T2), and a third group squares fit of the data. Glucose-transporter turnover was studied 5 days after removal from wheelcages (T5). number was calculated by dividing Vmax by Ro. Rats were placed into groups after matching for body wt Immunoblotting. Plasma- and microsomal-membrane and number of wheelcage revolutions. On the day of the protein, along with Mr markers (Bio-Rad, Fullerton, CA), experiment, rats, in the fed state, were injected (intraper- were subjected to SDS-PAGE and run under nonreducitoneally) with either saline or 90 nmol pork insulin, and ing conditions using an 8% resolving gel, as described previously (4,19). Resolved proteins were transferred studied 30 min later. Membrane preparation. The rats were killed by decap- electrophoretically from the gel to a nitrocellulose memitation, and blood was collected in heparinized tubes for brane (20). Incubations were carried out in TBS 20 mM subsequent analysis of plasma glucose and insulin con- Tris, 500 mM NaCI, pH 7.5, at 22°C unless otherwise centrations (see below). Approximately 6.2 g of mixed indicated. The nitrocellulose transfer membrane was forelimb muscle (pectoralis, scapularis, triceps, biceps, blocked in TBS with 0.2% Tween 20 and 5% Carnation extensors) was removed from both forelimbs, cleaned nonfat dry milk. To identify GLUT4 proteins, the transfer free of fat and connective tissue, and weighed. A small membranes were incubated with the polyclonal antibody portion of pectoralis, subscapularis, and the white portion R820 at a dilution of 1/250. R820 was produced from a of the triceps muscle was removed for muscle glycogen synthetic peptide corresponding to a 12 amino acid determination (see below). Plasma- and microsomal- C-terminal sequence in rat GLUT4 (21) (East Acres, membrane fractions were isolated by a procedure de- Southbridge, MA). To identify GLUT1 proteins, memscribed previously (4). Briefly, muscle was minced in a branes were incubated with the polyclonal antibody A379 buffer containing 100 mM Tris, 0.2 mM EDTA, and 255 at a dilution of 1/200 (provided by S.W. Cushman,
1092
DIABETES, VOL. 41, SEPTEMBER 1992
L.J. GOODYEAR AND ASSOCIATES
TABLE 1 Body wt, citrate synthase activity, blood glucose, plasma insulin, and muscle-glycogen concentrations Study groups Insulin
Untrained 11
Body wt (g) Plasma glucose (mM) Plasma insulin (nM) Citrate synthase activity (nmol • mg protein' 1 • min~1) Muscle glycogen (|xg/mg muscle) White triceps Subscapularis Pectoralis
252 260 8.5 3.4 0.19 337 207 191
±4 ±5 ± 0.3 ±0.1* ±0.02 ± 58* ±9 ± 10
5.0 4.7 3.3 4.0 4.4 4.3
± 0.3 ± 0.3 ± 0.3 ± 0.3 ± 0.4 ± 0.3
TO
T2
T5
12
8 247 jt 5 239 dt 6 8.8 i t0.2 3.2 =t0.2* 0.19;t0.03 300 it65* 329 dt 19t 335 dt 19*
255 ± 11 243 ± 5 7.5 ± 0.6 3.3 ± 0.2* 0.29 ±0.12 413 ± 17*
243 ± 5 240 ± 6 8.7 ± 0.4 3.0 ±0.1* 0.20 ± 0.05 328 ± 27* 326 ± 15t 359 ± 27* 4.5 4.8 6.7 8.6
± 0.3 ± 0.4 ± 0.7t ± 0.6**
8.0 ± 0.7t 9.5 ± 0.6*
4.8 :t0.3 4.2:t0.4 4.3 dt0.5
8.3 ± 1.0** 6.2 ± 0.5 9.4 ± 0.5**
6
285 ± 11t 314 ± 2 2 * 4.2 4.6 4.1 5.8 5.3 8.3
± 0.2 ± 0.2 ± 0.2 ± 0.4 ± 0.4 ± 0.7**
Values are means ± SE. n = number of rats per group. *P < 0.05 vs. saline injected. t P < 0.05 vs. untrained saline injected. *P < 0.05 vs. untrained insulin injected.
National Institutes of Health, Bethesda, MD). A379 was produced from a synthetic peptide of a 12 amino acid C-terminus sequence from rat brain GLUT1 (22). Antibody binding to the transfer membranes was visualized by incubating with 125l-labeled protein A (Amersham, Arlington Heights, IL). Following incubation, the transfer membranes were exposed to Kodak X AR-5 film at -80°C for 16-48 h. Bands corresponding to specific glucose transporters were quantitated by video densitometry (Gel/Image Analysis Systems, Technology Resources, Nashville, TN). Serial dilution of GLUT1 or GLUT4 standards were run in each gel to compare samples run in separate gels and to determine linearity. GLUT1 and GLUT4 were expressed as the percent of standards in arbitrary units. Assays. Plasma-glucose concentrations were determined with a YSI model 27 glucose analyzer (Yellow Springs, OH). Insulin concentrations were measured by the radioimmunoassay procedure of Starr et al. (23) modified by using polyethylene glycol (Mr 8,000) to separate the bound from free insulin tracer. Muscleglycogen concentrations were determined by the procedure of Hultman (24). Homogenates were assayed for citrate synthase activity by the method of Srere (25). Homogenate and subcellular membrane protein was determined by the Coomassie Brilliant Blue method (BioRad) described by Bradford (26) using crystalline bovine serum albumin as the standard. KpNPPase-specific activity, a plasma-membrane marker, was assayed in the absence or presence of 20 mM K+ (27). UDP-galactoseA/-acetylglucosamine galactosyltransferase (galactosyltransferase), an enzyme marker associated with the Golgi apparatus, also was measured in each fraction according to the method of Fleischer et al. (28). Statistical analysis. Data were analyzed by A NOVA using the general linear models procedure with a priori
DIABETES, VOL. 41, SEPTEMBER 1992
comparisons tested using the appropriate contrasts (SAS). Data are reported as means ± SE. RESULTS
Body weight, food consumption, and wheelcage revolution data. From wk 2 until the end of the protocol, wheelcage-trained rats exercised an average of 12-17 km • rat"1 -day" 1 (data not shown). Food consumption was greater in wheelcage-exercised rats throughout the 6 wk of training (data not shown). No significant difference was found in body wt of trained rats studied immediately, or 2 or 5 days following removal from wheelcages (Table 1). Plasma-insulin and glucose concentrations. Exercise training had no effect on fed-state plasma-glucose or -insulin concentrations (Table 1). Insulin injection resulted in decreased plasma-glucose concentrations and plasma-insulin concentrations of >180 nM. Citrate synthase activity and glycogen concentrations. Citrate synthase activity, a commonly used marker of trained muscle, was increased significantly in all trained groups (Table 1). Glycogen concentrations were greater in pectoralis and subscapularis muscles from trained rats; however, in the non-insulin-stimulated muscles this effect was lost by 2 days after training. Insulin injection tended to increase pectoralis and subscapularis muscle glycogen concentrations, but only in the trained rats. Endurance training or insulin had no effect on the glycogen content of the white portion of the triceps muscle. Protein recovery. The total protein content of the homogenate and the plasma- and microsomal-membrane fractions are shown in Table 2. Total homogenate protein was similar for all experimental groups, indicating that training did not alter the protein content of the muscle
1093
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
TABLE 2 Total protein recoveries Study groups Insulin Total protein Homogenate (mg)
Untrained
TO
11
12
694 694 3.02 2.74 0.88 0.83
Plasma membrane (mg) Microsomal membrane (mg)
± ± ± ± ± ±
30 42 0.46 0.42 0.05 0.07
755 686 3.28 3.07 1.48 1.51
T5
T2
dt 21 dt 21 dt0.39 dt0.20 db0.21* dtO.19t
694 709 3.40 3.43 1.09 1.12
± 27 ± 27 ± 0.24 ± 0.22 ±0.10 ±0.16
755 714 3.48 3.78 1.51 1.25
± 48 ±51 ±0.15 ± 0.21 ±0.09* ±0.14
Values are means ± SE. n = number of rats per group. *P < 0.05 vs. untrained saline injected. t P < 0.05 vs. untrained insulin injected.
used for the preparation of membranes. Plasma-membrane protein recovered was not changed by training nor by insulin injection. However, the recovery of microsomal-membrane protein was greater in some trained groups compared with the untrained (Table 2). Insulin injection did not alter microsomal-membrane protein recovery. Marker enzymes. Marker enzyme data from the homogenate and membrane fractions are shown in Tables 3 and 4. Activities of the plasma-membrane marker enzyme KpNPPase were enriched by —30- to 40-fold in the plasma-membrane fraction compared with the original homogenate, whereas the microsomal fraction was enriched by - 7 - to 11-fold (Table 3). KpNPPase-specific activities, recoveries, and enrichments were elevated somewhat in plasma membranes from the trained groups, however, no significant differences were ob-
served among the homogenates, plasma membranes, or microsomal membranes of any experimental group. Galactosyltransferase, an enzyme associated with the Golgi apparatus, was enriched by -35- to 50-fold in the microsomal fraction, whereas the plasma-membrane fraction was enriched by - 1 1 - to 13-fold (Table 4). Neither exercise training nor acute insulin treatment significantly altered galactosyltransferase-specific activities, recoveries, and enrichments of the membrane fractions or the homogenates. Although it was apparent from the protein and marker enzyme data that the membrane fractions produced were relatively distinct and were not altered by exercise training or insulin treatment, note that recoveries of marker enzymes may not have reflected glucose-transporter recovery accurately, and extrapolation of results to intact tissue provides only relative estimates of plasma-
TABLE 3 KpNPPase-specific activities, percent recoveries, and enrichments of subcellular membrane fractions from skeletal muscle of basal and insulin-stimulated rat forelimb Study groups Insulin n KpNPPase Homogenate Specific activity (nmol • mg • 30 min" Plasma membrane Specific activity (nmol • mg • 30 min" Recovery (%) Fold enrichment Microsomal membrane Specific activity (nmol • mg • 30 min" Recovery (%) Fold enrichment
Untrained
TO
T2
T5
157 ± 21 143 ± 18
168 ± 15 185 ± 9
11 157 ± 18 160 ± 17
148 ± 2 0 154 ± 19
4522 4507 13.7 12.3 33.3 28.6
± ± ± ± ± ±
533 574 1.3 1.6 5.8 3.0
5353 5612 17.4 18.2 41.3 40.2
± 577 ±736 dt 1.4 dt2.6 dt4.6 dt 4.4
5527 ± 528 5789 ± 663 19.6 ±3.2 20.3 ± 1.8 40.9 ± 6.8 41.8 ±2.6
6393 6130 18.7 17.6 39.5 33.8
± 340 ±388 ± 2.7 ± 1.3 ± 4.0 ± 1.4
1010 1078 0.9 0.8 6.9 7.1
±100 ± 167 ±0.1 ±0.1 ± 0.9 ± 1.1
991 857 1.4 1.3 7.5 7.1
dt 117 dt 8 8 dt0.2 dt0.2 ± 1.1 ± 1.5
1092 1192 1.4 1.2 8.8 11.2
1109 1217 1.4 1.1 6.9 6.6
±73 ±66 ±0.2 ±0.1 ± 0.7 ± 0.3
± 107 ± 170 ±0.3 ±0.1 ± 2.2 ± 3.8
Values are means ± SE. n = number of rats in group. NS among groups.
1094
DIABETES, VOL. 41, SEPTEMBER 1992
L.J. GOODYEAR AND ASSOCIATES
TABLE 4 UDP-galactose. A/-acetyglucosamine galactosyltransferase-specific activities, percent recoveries, and enrichments of subcellular membrane fractions from skeletal muscle of basal and insulin-stimulated rat forelimb Study groups Insulin Galactosyltransferase Homogenate Specific activity (nmol • mg~ f • 2 h~1) Plasma membrane Specific activity (nmol • mg~ f • 2 h~1) Recovery (%) Fold enrichment Microsomal membrane Specific activity (nmol • mg • 2 h~1) Recovery (%) Fold enrichment
Untrained
TO
11
12
T2
T5
2.5 ± 0.2 2.5 ±0.1
2.7 ± 0.2 2.8 ± 0.2
2.9 ± 0.2 2.7 ±0.3
2.7 ±0.1 2.9 ±0.1
28.4 ± 2.0 32.0 ± 1.8 5.0 2t0.6 5.0 2t0.8 12.0 2t 1.3 13.3 2t 1.3
31.7 ±2.8 29.9 ± 3.2 5.2 ± 0.6 4.9 ± 0.6 12.1 ±0.9 11.3 ± 1.5
33.3 ± 4.7 32.3 ± 3.1 5.5 ± 0.7 6.0 ± 0.5 11.9 ± 1.6 12.5 ± 1.2
31.3 ±2.8 32.9 ± 2.9 5.5 ± 0.5 6.2 ± 0.7 11.7 ± 1.0 11.5 ± 1.2
113db8.2 130 2t8.6 6.0 2t0.3 6.2 2t0.7 45.7 2t2.2 53.7 2t5.2
100 ± 9.9 105 ± 13.1 6.8 ± 0.7 7.3 ± 0.7 37.7 ±3.1 37.7 ± 4.2
103 ±7.0 134 ± 29.2 5.6 ±0.5 5.9 ± 0.5 35.8 ± 1.5 49.7 ± 15.0
94 ± 7.4 107 ± 10.0 7.2 ± 0.9 6.2 ± 0.4 35.5 ± 3.3 37.2 ± 3.8
Values are means ± SE. n = number of rats per group. NS among groups.
and microsomal-membrane glucose-transporter number. groups pooled, respectively). In addition, the KV2 for In addition, skeletal-muscle fractionation techniques re- glucose transport was not altered by exercise training covered only about 15-20% of plasma membranes. (26.8 ± 2.3 vs. 26.1 ±1.3 mM glucose, control vs. Thus, although it is unlikely in light of the extensive trained groups pooled, respectively). Glucose-transcharacterization of these fractions (2,4), it is possible that porter turnover number was calculated only from prepatransporters with different properties were not recovered rations where both the Ro and Vmax were determined by this technique. (n = 5-8/group). When the data were analyzed by conCytochalasin B binding. The number of glucose trans- trast comparisons of each individual group, neither exerporters in the plasma membranes was significantly cise training nor insulin stimulation had any significant greater in trained skeletal muscle compared with un- effects on transporter turnover number. However, in trained muscle (Fig. 1). This increase persisted for 5 days measuring the overall effect of insulin stimulation on after removal of rats from wheelcages. When the plasma- turnover number (insulin-stimulated groups pooled), inmembrane cytochalasin B binding data were normalized sulin significantly increased glucose-transporter turnover for KpNPPase activity, increases still were associated number (P < 0.01). The absolute increase, or the A value with exercise training (data not shown). Exercise training for the change in Ro, Vmax, and turnover number with had no effect on the number of glucose transporters insulin stimulation was not significantly different between associated with the microsomal membranes. trained and untrained preparations (data not shown). Insulin caused a significant increase in the number of Western blot analysis. Video densitometry analysis was plasma-membrane glucose transporters in all groups, performed on autoradiographs from immunoblots, and except for T2, in which the increase was not significant the data expressed in arbitrary units was standardized to (Fig. 1). Concomitant to the increase in the plasma- GLUT4 and GLUT1 control standards prepared from membrane fractions, glucose transporters in the mi- several pooled microsomal membranes. Exercise traincrosomal fractions decreased with insulin stimulation. ing resulted in a significant increase in GLUT4 protein in This decrease was not significant in T2 or T5. the whole muscle homogenates (basal and insulin-stimGlucose transport in plasma-membrane vesicles. Fig- ulated data pooled) (Table 6). GLUT1 protein was not ure 2 shows vesicle glucose-transport Vmax, transporter altered by exercise training or insulin injection in either number Ro, and the calculated transporter turnover num- the plasma- or microsomal-membrane fractions (Table ber from the untrained group and the trained groups 6). pooled. Table 5 reports Vmax and turnover number for Compared with untrained muscle, the relative abuneach group individually. Vesicle glucose-transport Vmax dance of GLUT4 protein in the plasma membranes was was significantly greater in plasma membranes from significantly greater in trained skeletal muscle when trained rats (P< 0.01). Insulin stimulation resulted in an measured immediately after training (Fig. 3). Although increase in Vmax in both untrained and trained rats. The under both basal and insulin-stimulated conditions we KV2 was not altered significantly by insulin stimulation found no significant decrease in plasma-membrane (25.0 ± 1.4 vs. 27.8 ± 1.8 mM glucose, basal vs. insulin GLUT4 in T2 and T5 (compared with TO), the increase in
DIABETES, VOL. 41, SEPTEMBER 1992
1095
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
A. PLASMA MEMBRANE
CIUCOM Transport ( V m a , )
I.. B. MICROSOMAL MEMBRANE
i
Transporter Numbsr ( Do )
»i
FIG. 2. Effects of exercise training and Insulin stimulation on skeletal-muscle plasma-membrane glucose-transporter turnover number (A), glucose transport Vmax (fl), and glucose-transporter number Ro (C). Results are means ± SE, n = 5-17/group. *P < 0.05 vs. corresponding saline injected; t P < 0.05 vs. untrained/saline injected; tP < 0.05 vs. untrained/insulin injected.
Insulin
-
+
FIG. 1. Total number of glucose transporters (Ro) determined by cytochalasin B binding in skeletal-muscle (A) plasma- and (6) microsomal-membranes from exercise-trained and untrained rats. Cytochalasin B binding studies were performed as described in METHODS. Results are means ± SE, n = 6-12/group. *P < 0.05 vs. corresponding saline injected; IP < 0.05 vs. untrained/saline injected; %P < 0.05 vs. untrained/insulin injected.
basal plasma-membrane GLUT4 in these groups compared with the untrained groups was not statistically significant. When the plasma-membrane GLUT4 data were normalized for KpNPPase activity, increases still were associated with exercise training (data not shown). Exercise training had no effect on GLUT4 associated with the microsomal membranes. Insulin caused a significant increase in the relative amount of plasma-membrane GLUT4 in all groups, except for T5, in which the increase was not statistically significant. Concomitant with the increase in the plasmamembrane fractions, GLUT4 in the microsomal fractions decreased with insulin stimulation. DISCUSSION
This study demonstrates that exercise training increases the number of glucose transporters in rat skeletal muscle, and that this increase is specific for the GLUT4 transporter isoform. Under basal, fed conditions, this increase is manifested in the plasma-membrane fraction, as exercise training had no effect on the number of glucose transporters in the intracellular microsomal fraction. Although the number of glucose transporters was increased by exercise training, the functional activity of
1096
plasma-membrane glucose transporters was not altered as the average glucose-transporter turnover number was not increased significantly. Thus, increased rates of glucose uptake reported in previous studies (10-12,17) and the increased rates of vesicular glucose transport seen in this study must be attributable primarily to an increase in the number of glucose transporters rather than an increase in the functional activity of the transporter protein. When skeletal muscle is exposed to an acute insulin stimulus, the number of glucose transporters in the plasma membrane and the average carrier-turnover number of these transporters are increased (2-4). In this study, a maximal insulin stimulus increased plasmamembrane glucose-transporter number and turnover number to the same degree in trained and untrained skeletal muscle. This finding demonstrates that differences between trained and untrained muscles are caused by adaptations already present in the basal, fed state. In addition, and perhaps more importantly, it suggests that the mechanism of increased insulin-stimulated skeletal-muscle glucose uptake in this altered physiological state is not attributable to a change in the translocation phenomena. The effect of exercise training on several components of the glucose-transport system, including transporter number and transport Vmax, persisted for 5 days after cessation of training. Because the effect of training was still present 2-5 days after exercise and because an acute bout of exercise results in only short-term changes in the number and activity of plasma-membrane glucose transporters (
Endurance exercise training can result in increased rates of insulin-stimulated glucose uptake in skeletal muscle; however, this effect may be lost rapidly once training ceases. To examine a mechanism for these changes, the skeletal-muscle glucose transport system of female rats exercise-trained in wheelcages for 6 wk were studied against a group of untrained female rats. The trained rats were studied immediately following and 2 and 5 days after removal from wheelcages; both trained and untrained rats were studied 30 min after insulin (90 nmol/rat, intraperitoneal) or saline injection. The total number of skeletal-muscle plasma-membrane glucose transporters (Ro), total muscle- homogenate and plasma-membrane GLUT4 protein, and rates of plasma-membrane vesicle D-facilitated glucose transport were higher in the exercise-trained rats immediately after exercise training and did not decrease significantly during the 5 days after cessation of training. On the other hand, exercise training did not alter microsomal-membrane total glucose-transporter number or GLUT4 protein, nor did training alter GLUT1 protein in total muscle homogenates nor either membrane fraction. The carrier-turnover number, an estimate of average functional activity of glucose transporters in the plasma membrane, was elevated slightly, but not significantly, in the trained muscle. In both the trained and untrained muscle, insulin administration resulted in translocation of glucose transporters from the microsomal-membrane fraction to the plasma membrane and an increase in the carrier-turnover
From the Metabolic Unit, Department of Medicine, University of Vermont, Burlington. Address correspondence and reprint requests to M.F. Hirshman, C-350 Given Bldg., Dept. of Medicine, University of Vermont, Burlington, VT 05405. Received for publication 16 September 1991 and accepted in revised form 30 March 1992. SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; TBS, Tris-buffered saline; KpNPPase, K+-stimulated-p-nitrophenol phosphatase; TO, rat group studied immediately after removal from wheelcages; T2, rat group studied 2 days after removal from wheelcages; T5, rat groups studied 5 days after removal from wheelcages; Ro, total number of glucose transporters; ANOVA, analysis of variance.
DIABETES, VOL. 41, SEPTEMBER 1992
number. These data suggest that increased rates of glucose uptake in endurance-trained skeletal muscle results primarily from an increase in the number—and not an increase in the average functional activity—of glucose transporters present in the plasma membrane. Furthermore, these increases persist for several days after cessation of exercise training. The specific increase in the GLUT4, but not the GLUT1 glucose-transporter isoform, in response to training demonstrates that a common, chronic physiological stimulus can regulate the expression of the two glucose-transporter isoforms present in skeletal-muscle tissue differentially. Diabetes 41:1091-99,1992
G
lucose transport is the rate-limiting step for skeletal-muscle glucose use under normal physiological conditions (1). Rates of glucose transport can be altered by a change in the number of carrier proteins present in the plasma membrane and/or a change in the turnover number of the carrier proteins (glucose transporters). Recent studies in rat skeletal muscle have demonstrated that both glucosetransporter number and carrier-turnover number are increased by insulin stimulation (2-4) and a single bout of acute exercise (2,5,6). In addition, of the two genetically distinct glucose-transporter isoforms thought to be expressed in rat skeletal-muscle tissue (GLUT1, GLUT4), acute exercise and insulin stimulation have been shown to increase the plasma-membrane content of the GLUT4 isoform selectively (7-9). Skeletal muscle is the predominant tissue responsible for insulin-stimulated glucose disposal and a major site of insulin resistance in diabetes. While diabetes results in a decrease in skeletal-muscle insulin sensitivity, endurance exercise training increases the sensitivity and/or responsiveness of skeletal-muscle glucose uptake to insulin (10-12). However, some evidence shows that in both human and rat models, this adaptation to chronic
1091
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
exercise lasts only 24-60 h after the last training session (13-15). The mechanism for these adaptations to both training and detraining may be related to changes in the total number, activity, or subcellular distribution of glucose transporters, the ability of insulin or contractile activity to cause the translocation and fusion of transporters into the plasma membrane, and/or a change in the amount of a specific glucose-transporter isoform. Previous reports have shown that exercise training increases GLUT4 mRNA (16,17) and GLUT4 protein (9,17) in rat skeletal-muscle homogenates. However, these studies did not determine if exercise training alters the other components of the glucose-transport system. In this study, we measured the effects of 6 wk of wheelcage training on total glucose-transporter number and activity, and the distribution of GLUT4- and GLUT1-transporter isoforms. Furthermore, we evaluated these parameters at three time points after the cessation of training to determine if changes in transporter number, distribution, and/or activity parallel the reported changes in insulin sensitivity and responsiveness.
mM sucrose, pH 7.6; it then was polytroned at slow speed, homogenized, and diluted. The homogenate was centrifuged at 35,000 g for 20 min. The resulting pellet was used to prepare plasma membranes; the supernatant was used to prepare microsomal membranes (4). Cytochalasin B binding. Equilibrium D-glucose-inhibitable [3H]-cytochalasin B binding was measured, and the concentration of glucose transporters was determined (18). Scatchard plots were generated from binding studies in which membranes were incubated with cytochalasin B in the presence or absence of 500 mM D-glucose. Cytochalasin E (2000 nM) was present to decrease nonspecific binding. Bound cytochalasin B was separated from free cytochalasin B by centrifugation. Tracer amounts of [14C]urea were used to correct for [3H]cytochalasin B trapped in the pellet. The total number of glucose transporters {Ro) and the dissociation constant (K"d) were determined from a linear plot derived by subtraction along the radial axes of binding curves generated in the presence of D-glucose from those in the absence of D-glucose. Glucose-transport activity in plasma-membrane vesicles. D-[14C(U)]-glucose and i_-[3H(N)]-glucose uptake RESEARCH DESIGN AND METHODS in plasma-membrane vesicles were determined under Animal care and exercise training. Female Sprague- equilibrium exchange conditions using a rapid filtration Dawley rats were received at body wt ranging from technique at 25°C (6). Membranes were preequilibrated 100-150 g and maintained with ad libitum feeding (Pu- with a HEPES-buffered Krebs Ringer solution containing rina Laboratory Chow, Ralston-Purina, St. Louis, MO) for 1, 5, 10, 20, 40, or 60 mM i_- and D-glucose. Uptake was several days before placement into treatment groups. measured at four time points for each preparation (1-5 One group of rats (n = 52) was housed individually in sec). Transport was initiated by combining an aliquot of wheelcages where voluntary access to physical exercise the membranes with HEPES-buffered Krebs solution and was available at all times. The total number of wheelcage 1-60 mM D- and L-glucose with 6 jxCi of L-[ 3 H(N)]revolutions was recorded 3 times/wk, and the accumu- glucose and 1.6 ixCi of d-[14C(U)]-glucose. Transport lated distance was calculated at the end of each wk. was stopped by the addition of HEPES-buffered Krebs Age-matched control rats (n = 26) were housed in indi- containing 0.2 mM phloretin. The membranes were filvidual cages and did not have access to a running wheel. tered rapidly (Millipore HA 0.45 |xm) and washed; the Food consumption and body wt were monitored twice filter and adhering membranes were analyzed by liquid weekly throughout the 6-wk study period. scintillation counting using quench correction for the dual The trained rats were subdivided into three groups; label. Facilitated transport was calculated by subtracting one group was studied immediately after removal from the initial rate of L-glucose influx from that of D-glucose. wheelcages (TO), a second group was studied 2 days Vmax and KV2 were determined using nonlinear least after removal from wheelcages (T2), and a third group squares fit of the data. Glucose-transporter turnover was studied 5 days after removal from wheelcages (T5). number was calculated by dividing Vmax by Ro. Rats were placed into groups after matching for body wt Immunoblotting. Plasma- and microsomal-membrane and number of wheelcage revolutions. On the day of the protein, along with Mr markers (Bio-Rad, Fullerton, CA), experiment, rats, in the fed state, were injected (intraper- were subjected to SDS-PAGE and run under nonreducitoneally) with either saline or 90 nmol pork insulin, and ing conditions using an 8% resolving gel, as described previously (4,19). Resolved proteins were transferred studied 30 min later. Membrane preparation. The rats were killed by decap- electrophoretically from the gel to a nitrocellulose memitation, and blood was collected in heparinized tubes for brane (20). Incubations were carried out in TBS 20 mM subsequent analysis of plasma glucose and insulin con- Tris, 500 mM NaCI, pH 7.5, at 22°C unless otherwise centrations (see below). Approximately 6.2 g of mixed indicated. The nitrocellulose transfer membrane was forelimb muscle (pectoralis, scapularis, triceps, biceps, blocked in TBS with 0.2% Tween 20 and 5% Carnation extensors) was removed from both forelimbs, cleaned nonfat dry milk. To identify GLUT4 proteins, the transfer free of fat and connective tissue, and weighed. A small membranes were incubated with the polyclonal antibody portion of pectoralis, subscapularis, and the white portion R820 at a dilution of 1/250. R820 was produced from a of the triceps muscle was removed for muscle glycogen synthetic peptide corresponding to a 12 amino acid determination (see below). Plasma- and microsomal- C-terminal sequence in rat GLUT4 (21) (East Acres, membrane fractions were isolated by a procedure de- Southbridge, MA). To identify GLUT1 proteins, memscribed previously (4). Briefly, muscle was minced in a branes were incubated with the polyclonal antibody A379 buffer containing 100 mM Tris, 0.2 mM EDTA, and 255 at a dilution of 1/200 (provided by S.W. Cushman,
1092
DIABETES, VOL. 41, SEPTEMBER 1992
L.J. GOODYEAR AND ASSOCIATES
TABLE 1 Body wt, citrate synthase activity, blood glucose, plasma insulin, and muscle-glycogen concentrations Study groups Insulin
Untrained 11
Body wt (g) Plasma glucose (mM) Plasma insulin (nM) Citrate synthase activity (nmol • mg protein' 1 • min~1) Muscle glycogen (|xg/mg muscle) White triceps Subscapularis Pectoralis
252 260 8.5 3.4 0.19 337 207 191
±4 ±5 ± 0.3 ±0.1* ±0.02 ± 58* ±9 ± 10
5.0 4.7 3.3 4.0 4.4 4.3
± 0.3 ± 0.3 ± 0.3 ± 0.3 ± 0.4 ± 0.3
TO
T2
T5
12
8 247 jt 5 239 dt 6 8.8 i t0.2 3.2 =t0.2* 0.19;t0.03 300 it65* 329 dt 19t 335 dt 19*
255 ± 11 243 ± 5 7.5 ± 0.6 3.3 ± 0.2* 0.29 ±0.12 413 ± 17*
243 ± 5 240 ± 6 8.7 ± 0.4 3.0 ±0.1* 0.20 ± 0.05 328 ± 27* 326 ± 15t 359 ± 27* 4.5 4.8 6.7 8.6
± 0.3 ± 0.4 ± 0.7t ± 0.6**
8.0 ± 0.7t 9.5 ± 0.6*
4.8 :t0.3 4.2:t0.4 4.3 dt0.5
8.3 ± 1.0** 6.2 ± 0.5 9.4 ± 0.5**
6
285 ± 11t 314 ± 2 2 * 4.2 4.6 4.1 5.8 5.3 8.3
± 0.2 ± 0.2 ± 0.2 ± 0.4 ± 0.4 ± 0.7**
Values are means ± SE. n = number of rats per group. *P < 0.05 vs. saline injected. t P < 0.05 vs. untrained saline injected. *P < 0.05 vs. untrained insulin injected.
National Institutes of Health, Bethesda, MD). A379 was produced from a synthetic peptide of a 12 amino acid C-terminus sequence from rat brain GLUT1 (22). Antibody binding to the transfer membranes was visualized by incubating with 125l-labeled protein A (Amersham, Arlington Heights, IL). Following incubation, the transfer membranes were exposed to Kodak X AR-5 film at -80°C for 16-48 h. Bands corresponding to specific glucose transporters were quantitated by video densitometry (Gel/Image Analysis Systems, Technology Resources, Nashville, TN). Serial dilution of GLUT1 or GLUT4 standards were run in each gel to compare samples run in separate gels and to determine linearity. GLUT1 and GLUT4 were expressed as the percent of standards in arbitrary units. Assays. Plasma-glucose concentrations were determined with a YSI model 27 glucose analyzer (Yellow Springs, OH). Insulin concentrations were measured by the radioimmunoassay procedure of Starr et al. (23) modified by using polyethylene glycol (Mr 8,000) to separate the bound from free insulin tracer. Muscleglycogen concentrations were determined by the procedure of Hultman (24). Homogenates were assayed for citrate synthase activity by the method of Srere (25). Homogenate and subcellular membrane protein was determined by the Coomassie Brilliant Blue method (BioRad) described by Bradford (26) using crystalline bovine serum albumin as the standard. KpNPPase-specific activity, a plasma-membrane marker, was assayed in the absence or presence of 20 mM K+ (27). UDP-galactoseA/-acetylglucosamine galactosyltransferase (galactosyltransferase), an enzyme marker associated with the Golgi apparatus, also was measured in each fraction according to the method of Fleischer et al. (28). Statistical analysis. Data were analyzed by A NOVA using the general linear models procedure with a priori
DIABETES, VOL. 41, SEPTEMBER 1992
comparisons tested using the appropriate contrasts (SAS). Data are reported as means ± SE. RESULTS
Body weight, food consumption, and wheelcage revolution data. From wk 2 until the end of the protocol, wheelcage-trained rats exercised an average of 12-17 km • rat"1 -day" 1 (data not shown). Food consumption was greater in wheelcage-exercised rats throughout the 6 wk of training (data not shown). No significant difference was found in body wt of trained rats studied immediately, or 2 or 5 days following removal from wheelcages (Table 1). Plasma-insulin and glucose concentrations. Exercise training had no effect on fed-state plasma-glucose or -insulin concentrations (Table 1). Insulin injection resulted in decreased plasma-glucose concentrations and plasma-insulin concentrations of >180 nM. Citrate synthase activity and glycogen concentrations. Citrate synthase activity, a commonly used marker of trained muscle, was increased significantly in all trained groups (Table 1). Glycogen concentrations were greater in pectoralis and subscapularis muscles from trained rats; however, in the non-insulin-stimulated muscles this effect was lost by 2 days after training. Insulin injection tended to increase pectoralis and subscapularis muscle glycogen concentrations, but only in the trained rats. Endurance training or insulin had no effect on the glycogen content of the white portion of the triceps muscle. Protein recovery. The total protein content of the homogenate and the plasma- and microsomal-membrane fractions are shown in Table 2. Total homogenate protein was similar for all experimental groups, indicating that training did not alter the protein content of the muscle
1093
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
TABLE 2 Total protein recoveries Study groups Insulin Total protein Homogenate (mg)
Untrained
TO
11
12
694 694 3.02 2.74 0.88 0.83
Plasma membrane (mg) Microsomal membrane (mg)
± ± ± ± ± ±
30 42 0.46 0.42 0.05 0.07
755 686 3.28 3.07 1.48 1.51
T5
T2
dt 21 dt 21 dt0.39 dt0.20 db0.21* dtO.19t
694 709 3.40 3.43 1.09 1.12
± 27 ± 27 ± 0.24 ± 0.22 ±0.10 ±0.16
755 714 3.48 3.78 1.51 1.25
± 48 ±51 ±0.15 ± 0.21 ±0.09* ±0.14
Values are means ± SE. n = number of rats per group. *P < 0.05 vs. untrained saline injected. t P < 0.05 vs. untrained insulin injected.
used for the preparation of membranes. Plasma-membrane protein recovered was not changed by training nor by insulin injection. However, the recovery of microsomal-membrane protein was greater in some trained groups compared with the untrained (Table 2). Insulin injection did not alter microsomal-membrane protein recovery. Marker enzymes. Marker enzyme data from the homogenate and membrane fractions are shown in Tables 3 and 4. Activities of the plasma-membrane marker enzyme KpNPPase were enriched by —30- to 40-fold in the plasma-membrane fraction compared with the original homogenate, whereas the microsomal fraction was enriched by - 7 - to 11-fold (Table 3). KpNPPase-specific activities, recoveries, and enrichments were elevated somewhat in plasma membranes from the trained groups, however, no significant differences were ob-
served among the homogenates, plasma membranes, or microsomal membranes of any experimental group. Galactosyltransferase, an enzyme associated with the Golgi apparatus, was enriched by -35- to 50-fold in the microsomal fraction, whereas the plasma-membrane fraction was enriched by - 1 1 - to 13-fold (Table 4). Neither exercise training nor acute insulin treatment significantly altered galactosyltransferase-specific activities, recoveries, and enrichments of the membrane fractions or the homogenates. Although it was apparent from the protein and marker enzyme data that the membrane fractions produced were relatively distinct and were not altered by exercise training or insulin treatment, note that recoveries of marker enzymes may not have reflected glucose-transporter recovery accurately, and extrapolation of results to intact tissue provides only relative estimates of plasma-
TABLE 3 KpNPPase-specific activities, percent recoveries, and enrichments of subcellular membrane fractions from skeletal muscle of basal and insulin-stimulated rat forelimb Study groups Insulin n KpNPPase Homogenate Specific activity (nmol • mg • 30 min" Plasma membrane Specific activity (nmol • mg • 30 min" Recovery (%) Fold enrichment Microsomal membrane Specific activity (nmol • mg • 30 min" Recovery (%) Fold enrichment
Untrained
TO
T2
T5
157 ± 21 143 ± 18
168 ± 15 185 ± 9
11 157 ± 18 160 ± 17
148 ± 2 0 154 ± 19
4522 4507 13.7 12.3 33.3 28.6
± ± ± ± ± ±
533 574 1.3 1.6 5.8 3.0
5353 5612 17.4 18.2 41.3 40.2
± 577 ±736 dt 1.4 dt2.6 dt4.6 dt 4.4
5527 ± 528 5789 ± 663 19.6 ±3.2 20.3 ± 1.8 40.9 ± 6.8 41.8 ±2.6
6393 6130 18.7 17.6 39.5 33.8
± 340 ±388 ± 2.7 ± 1.3 ± 4.0 ± 1.4
1010 1078 0.9 0.8 6.9 7.1
±100 ± 167 ±0.1 ±0.1 ± 0.9 ± 1.1
991 857 1.4 1.3 7.5 7.1
dt 117 dt 8 8 dt0.2 dt0.2 ± 1.1 ± 1.5
1092 1192 1.4 1.2 8.8 11.2
1109 1217 1.4 1.1 6.9 6.6
±73 ±66 ±0.2 ±0.1 ± 0.7 ± 0.3
± 107 ± 170 ±0.3 ±0.1 ± 2.2 ± 3.8
Values are means ± SE. n = number of rats in group. NS among groups.
1094
DIABETES, VOL. 41, SEPTEMBER 1992
L.J. GOODYEAR AND ASSOCIATES
TABLE 4 UDP-galactose. A/-acetyglucosamine galactosyltransferase-specific activities, percent recoveries, and enrichments of subcellular membrane fractions from skeletal muscle of basal and insulin-stimulated rat forelimb Study groups Insulin Galactosyltransferase Homogenate Specific activity (nmol • mg~ f • 2 h~1) Plasma membrane Specific activity (nmol • mg~ f • 2 h~1) Recovery (%) Fold enrichment Microsomal membrane Specific activity (nmol • mg • 2 h~1) Recovery (%) Fold enrichment
Untrained
TO
11
12
T2
T5
2.5 ± 0.2 2.5 ±0.1
2.7 ± 0.2 2.8 ± 0.2
2.9 ± 0.2 2.7 ±0.3
2.7 ±0.1 2.9 ±0.1
28.4 ± 2.0 32.0 ± 1.8 5.0 2t0.6 5.0 2t0.8 12.0 2t 1.3 13.3 2t 1.3
31.7 ±2.8 29.9 ± 3.2 5.2 ± 0.6 4.9 ± 0.6 12.1 ±0.9 11.3 ± 1.5
33.3 ± 4.7 32.3 ± 3.1 5.5 ± 0.7 6.0 ± 0.5 11.9 ± 1.6 12.5 ± 1.2
31.3 ±2.8 32.9 ± 2.9 5.5 ± 0.5 6.2 ± 0.7 11.7 ± 1.0 11.5 ± 1.2
113db8.2 130 2t8.6 6.0 2t0.3 6.2 2t0.7 45.7 2t2.2 53.7 2t5.2
100 ± 9.9 105 ± 13.1 6.8 ± 0.7 7.3 ± 0.7 37.7 ±3.1 37.7 ± 4.2
103 ±7.0 134 ± 29.2 5.6 ±0.5 5.9 ± 0.5 35.8 ± 1.5 49.7 ± 15.0
94 ± 7.4 107 ± 10.0 7.2 ± 0.9 6.2 ± 0.4 35.5 ± 3.3 37.2 ± 3.8
Values are means ± SE. n = number of rats per group. NS among groups.
and microsomal-membrane glucose-transporter number. groups pooled, respectively). In addition, the KV2 for In addition, skeletal-muscle fractionation techniques re- glucose transport was not altered by exercise training covered only about 15-20% of plasma membranes. (26.8 ± 2.3 vs. 26.1 ±1.3 mM glucose, control vs. Thus, although it is unlikely in light of the extensive trained groups pooled, respectively). Glucose-transcharacterization of these fractions (2,4), it is possible that porter turnover number was calculated only from prepatransporters with different properties were not recovered rations where both the Ro and Vmax were determined by this technique. (n = 5-8/group). When the data were analyzed by conCytochalasin B binding. The number of glucose trans- trast comparisons of each individual group, neither exerporters in the plasma membranes was significantly cise training nor insulin stimulation had any significant greater in trained skeletal muscle compared with un- effects on transporter turnover number. However, in trained muscle (Fig. 1). This increase persisted for 5 days measuring the overall effect of insulin stimulation on after removal of rats from wheelcages. When the plasma- turnover number (insulin-stimulated groups pooled), inmembrane cytochalasin B binding data were normalized sulin significantly increased glucose-transporter turnover for KpNPPase activity, increases still were associated number (P < 0.01). The absolute increase, or the A value with exercise training (data not shown). Exercise training for the change in Ro, Vmax, and turnover number with had no effect on the number of glucose transporters insulin stimulation was not significantly different between associated with the microsomal membranes. trained and untrained preparations (data not shown). Insulin caused a significant increase in the number of Western blot analysis. Video densitometry analysis was plasma-membrane glucose transporters in all groups, performed on autoradiographs from immunoblots, and except for T2, in which the increase was not significant the data expressed in arbitrary units was standardized to (Fig. 1). Concomitant to the increase in the plasma- GLUT4 and GLUT1 control standards prepared from membrane fractions, glucose transporters in the mi- several pooled microsomal membranes. Exercise traincrosomal fractions decreased with insulin stimulation. ing resulted in a significant increase in GLUT4 protein in This decrease was not significant in T2 or T5. the whole muscle homogenates (basal and insulin-stimGlucose transport in plasma-membrane vesicles. Fig- ulated data pooled) (Table 6). GLUT1 protein was not ure 2 shows vesicle glucose-transport Vmax, transporter altered by exercise training or insulin injection in either number Ro, and the calculated transporter turnover num- the plasma- or microsomal-membrane fractions (Table ber from the untrained group and the trained groups 6). pooled. Table 5 reports Vmax and turnover number for Compared with untrained muscle, the relative abuneach group individually. Vesicle glucose-transport Vmax dance of GLUT4 protein in the plasma membranes was was significantly greater in plasma membranes from significantly greater in trained skeletal muscle when trained rats (P< 0.01). Insulin stimulation resulted in an measured immediately after training (Fig. 3). Although increase in Vmax in both untrained and trained rats. The under both basal and insulin-stimulated conditions we KV2 was not altered significantly by insulin stimulation found no significant decrease in plasma-membrane (25.0 ± 1.4 vs. 27.8 ± 1.8 mM glucose, basal vs. insulin GLUT4 in T2 and T5 (compared with TO), the increase in
DIABETES, VOL. 41, SEPTEMBER 1992
1095
EXERCISE TRAINING AND GLUCOSE TRANSPORTERS
A. PLASMA MEMBRANE
CIUCOM Transport ( V m a , )
I.. B. MICROSOMAL MEMBRANE
i
Transporter Numbsr ( Do )
»i
FIG. 2. Effects of exercise training and Insulin stimulation on skeletal-muscle plasma-membrane glucose-transporter turnover number (A), glucose transport Vmax (fl), and glucose-transporter number Ro (C). Results are means ± SE, n = 5-17/group. *P < 0.05 vs. corresponding saline injected; t P < 0.05 vs. untrained/saline injected; tP < 0.05 vs. untrained/insulin injected.
Insulin
-
+
FIG. 1. Total number of glucose transporters (Ro) determined by cytochalasin B binding in skeletal-muscle (A) plasma- and (6) microsomal-membranes from exercise-trained and untrained rats. Cytochalasin B binding studies were performed as described in METHODS. Results are means ± SE, n = 6-12/group. *P < 0.05 vs. corresponding saline injected; IP < 0.05 vs. untrained/saline injected; %P < 0.05 vs. untrained/insulin injected.
basal plasma-membrane GLUT4 in these groups compared with the untrained groups was not statistically significant. When the plasma-membrane GLUT4 data were normalized for KpNPPase activity, increases still were associated with exercise training (data not shown). Exercise training had no effect on GLUT4 associated with the microsomal membranes. Insulin caused a significant increase in the relative amount of plasma-membrane GLUT4 in all groups, except for T5, in which the increase was not statistically significant. Concomitant with the increase in the plasmamembrane fractions, GLUT4 in the microsomal fractions decreased with insulin stimulation. DISCUSSION
This study demonstrates that exercise training increases the number of glucose transporters in rat skeletal muscle, and that this increase is specific for the GLUT4 transporter isoform. Under basal, fed conditions, this increase is manifested in the plasma-membrane fraction, as exercise training had no effect on the number of glucose transporters in the intracellular microsomal fraction. Although the number of glucose transporters was increased by exercise training, the functional activity of
1096
plasma-membrane glucose transporters was not altered as the average glucose-transporter turnover number was not increased significantly. Thus, increased rates of glucose uptake reported in previous studies (10-12,17) and the increased rates of vesicular glucose transport seen in this study must be attributable primarily to an increase in the number of glucose transporters rather than an increase in the functional activity of the transporter protein. When skeletal muscle is exposed to an acute insulin stimulus, the number of glucose transporters in the plasma membrane and the average carrier-turnover number of these transporters are increased (2-4). In this study, a maximal insulin stimulus increased plasmamembrane glucose-transporter number and turnover number to the same degree in trained and untrained skeletal muscle. This finding demonstrates that differences between trained and untrained muscles are caused by adaptations already present in the basal, fed state. In addition, and perhaps more importantly, it suggests that the mechanism of increased insulin-stimulated skeletal-muscle glucose uptake in this altered physiological state is not attributable to a change in the translocation phenomena. The effect of exercise training on several components of the glucose-transport system, including transporter number and transport Vmax, persisted for 5 days after cessation of training. Because the effect of training was still present 2-5 days after exercise and because an acute bout of exercise results in only short-term changes in the number and activity of plasma-membrane glucose transporters (
