GLUT4 Overexpression in db/db Mice Dose-Dependently Ameliorates Diabetes But Is Not a Lifelong Cure

TG mice carrying 11.8 kb (NotI fragment) of the human GLUT4 transporter genomic DNA were produced as described previously (7). Two lines of TG mice (hGLUT4db/db line 6 and line 8), ages 10–12 (young), 20–25 (middle-aged), and 28–30 weeks (old) and homozygous for the hGLUT4 transgene and obese NTG C57BL/KsJ-lepr^db/lepr^db of the same ages were used in this study; hGLUT4db/db line 6 (TG6) displays a two- to threefold overexpression of GLUT4 in skeletal muscle, cardiac muscle, and white adipose tissue, whereas hGLUT4db/db line 8 (TG8) displays a fivefold overexpression in these tissues. All mice used in this study were homozygous for the db gene and thus exhibited an obese, diabetic phenotype, with the exception that NTG C57BL/KsJ- lepr^db/+ mice were used as nondiabetic controls for direct comparison to NTG db/db mice in the experiments described in Fig. 5.

Mice in the postprandial state were anesthetized with 5 mg/100 g body weight of sodium pentobarbital. The extensor digitorum longus (EDL) muscles were used for all transport and photolabeling measurements. This muscle is composed of a fiber type (∼50% fast-twitch oxidative and 50% fast-twitch glycolytic) that is representative of the hind limb muscle fiber composition of the rat (18). The EDL muscles were dissected out, blotted on gauze, and transferred to 25-ml Erlenmeyer flasks containing 2 ml Krebs-Henseleit buffer (0.1% bovine serum albumin [BSA], 32 mmol/l mannitol, 8 mmol/l glucose, and ±13.3 nmol/l insulin). The flasks were incubated in a shaking water bath maintained at 35°C for 60 min and continuously gassed with 95% O₂/5% CO₂.

Glucose transport activity was measured using 2-deoxyglucose (2DG) as described in detail previously (8). After the incubations, the muscles were frozen between tongs cooled to the temperature of liquid nitrogen and stored at −70°C until they were processed for measurement of 2DG transport. Frozen muscle tissue was homogenized in 1 ml 0.6N perchloric acid. Homogenates were centrifuged at 5,000g, and aliquots of the supernatant were counted for radioactivity in a liquid scintillation counter.

Isolated muscles were photolabeled as described previously (10,19,20). After labeling, the muscles were trimmed of their tendons, blotted, and frozen between tongs cooled to the temperature of liquid nitrogen. Muscle tissue was stored at −70°C until it was processed.

Crude total membranes were prepared from the photolabeled muscles and solubilized, and GLUT4 was immunoprecipitated as described previously (10,19,20). Briefly, solubilized crude membranes were immunoprecipitated with an immunocomplex consisting of anti-GLUT4 antibody (courtesy of Hoffman–La Roche, Nutley, NJ) and protein A sepharose. Subsequent to immunoprecipitation, the immunocomplex was released from the protein A sepharose with SDS-urea sample buffer and then subjected to SDS-PAGE. Gels were stained with Coomassie blue and sliced by lane into 8-mm slices. The slices were dried and solubilized in 30% H₂O₂ and 2% ammonium hydroxide, and the radioactivity was quantitated in a liquid scintillation counter. Labeled GLUT4 protein was quantified by integrating the area under the ³H peak corresponding to the 45-kDa molecular weight marker and subtracting the average background radioactivity in the gel.

The triceps surae muscle group was removed and homogenized as described previously (4,10). The triceps surae comprises the same mixture of fiber types as the EDL and, therefore, should be representative of this muscle (18). Crude homogenate protein (100 μg sample protein) and plasma membrane standard (50 μg) along with molecular weight markers (Novex) were subjected to SDS-PAGE run under reducing conditions using a 10% resolving gel, as described previously (21). Samples from each group of mice were run in adjacent lanes. Resolved proteins were transferred to polyvinylidine fluoride (PVDF) sheets (Bio-Rad, Richmond, CA), as described previously (22), using a Bio-Rad SD semidry transfer unit. After transfer, the PVDF sheets were blocked in 20 mg/dl Tris and 500 mg/dl NaCl Tris-buffered saline (TBS; pH 7.5) at 25°C containing 5% nonfat dry milk. The sheets were then washed in TBS-containing 0.05% Tween-20 (TTBS) for 20 min and then incubated overnight with the polyclonal GLUT4 COOH-terminus antibody (8107P; kindly provided by Hoffman-La Roche, Nutley, NJ) in TTBS, 3% bovine serum albumin, and 0.02% sodium azide. The PVDF was then washed in TTBS and incubated with [¹²⁵I]protein A (DuPont–NEN, Boston, MA) at a concentration of 0.25 mCi/ml for 1 h. The sheets were then washed and air dried. Antibody binding was visualized by exposure to a Phosphorimager cassette (Molecular Dynamics, Sunnyvale, CA) for 72 h. Each band was corrected for background, and the density of each band was expressed as a percentage of a standard (50 μg rat skeletal muscle plasma membrane protein) run on each gel.

Nonanesthetized GLUT4 TG and age-matched NTG mice were fasted overnight and bled via the orbital sinus (0.025 ml) immediately before administration of an oral glucose load (1 g glucose per kg body wt) by gavage using a syringe equipped with a murine oral feeding needle (20-gauge; Popper & Sons, New Hyde Park, NY). Mice were subsequently bled after 30, 75, and 120 min. Plasma glucose was measured using the VP Super System Autoanalyzer (Abbott Laboratories, North Chicago, IL). Plasma insulin was determined using a radioimmunoassay (Linco Research, St. Charles, MO).

Studies were conducted in fasted (16 h), anesthetized female young (age 10–12 weeks) NTG and TG8 mice and old (age 28–30 weeks) NTG, TG6, and TG8 mice using clamp methodology as described previously (9). Briefly, mice were given an intravenous bolus injection (10 mU per mouse) and then continuously infused with insulin (U-100, R porcine/bovine insulin, Eli Lilly, Indianapolis, IN) at a rate of 1 mU/min (100 μl/h) for 2 h. Blood samples were obtained every 5 min from the tail for determination of glucose concentration; for this, the tip of the tail was cut and then clamped with a small hemostat to prevent bleeding between samples. Blood glucose concentration was measured using a glucose meter (One Touch II; Lifescan, Milipitas, CA). A 25% glucose solution was infused at a rate that was adjusted to maintain the blood glucose concentration at ∼17 mmol/l. At the end of the clamp, a 400-μl blood sample was obtained from the retro-orbital sinus for plasma insulin determination.

The data were analyzed by analysis of variance (ANOVA) to test the effects of group (control versus transgenic, young versus old) and treatment (basal, insulin) on muscle glucose uptake and glucose transporter distribution. When a significant F ratio was obtained, Fisher’s protected least significant difference (PLSD) post-hoc test was used to identify statistically significant differences (P < 0.05) between the means. Glucose clamp data for young and old mice were analyzed separately by ANOVA, followed by a PLSD test.

Levels of GLUT4 expression in NTG and TG mice are shown in Fig. 1. Young TG6 mice overexpressed skeletal muscle GLUT4 twofold as compared with NTG, and the overexpression level increased slightly in 20- to 25-week-old mice (threefold). Young TG8 mice exhibited a fivefold overexpression of GLUT4 in skeletal muscle, which declined slightly in 28- to 30-week-old mice to approximately fourfold.

The effect of increased GLUT4 expression on glucose homeostasis in vivo was assessed by performing oral glucose tolerance tests on TG and NTG db/db mice of various ages. Both lines of young (age 10–12 weeks) TG mice showed a similar plasma glucose profile (Fig. 2A); fasting plasma glucose was significantly reduced in TG mice compared with the severely hyperglycemic NTG db/db mice. Furthermore, the glucose rise was substantially blunted compared with NTG, which confirms our previously published results (7). Surprisingly, the improved glycemic control observed in TG6 mice at age 10 to 12 weeks was lost in mice at age 21 weeks (Fig. 2B), as both fasting plasma glucose and the glucose excursion were identical to that observed in age-matched NTG db/db controls. In marked contrast, TG8, which expresses higher GLUT4 levels, maintained the same level of improved glycemic control observed in 10- to 12-week-old mice to at least age 21 weeks. However, by age ∼30 weeks, the improvement in glycemic control observed in TG8 was substantially reduced, as only the fasting plasma glucose and 2-h time point after the glucose challenge were significantly lower than that observed in NTG db/db mice (Fig. 2C).

The time course of plasma glucose and glucose infusion rate (GIR) for 10- to 12-week-old and 28- to 30-week-old NTG and TG mice are shown in Figs. 3 and 4. Relatively similar blood glucose concentrations of ∼17 mmol/l and GIRs were achieved by 90 min of infusion; data for GIRs and plasma glucose values averaged over 90–120 min are summarized in Table 1. Mean plasma insulin concentrations in blood samples taken at the end of the clamp were >13 nmol/l in all of the groups. In general, the NTG mice had higher insulin concentrations (∼24 nmol/l) than the TG mice. Lower plasma insulin levels in the TG mice at the end of the clamp likely reflected increased insulin binding and clearance that is to be expected in the more insulin-responsive TG mice. In young TG8 mice (age 10–12 weeks), the GIR required to maintain steady-state plasma glucose was elevated 1.8-fold compared with NTG db/db mice of the same age (102 vs. 56 μmol · min⁻¹ · kg⁻¹, respectively; Table 1 and Fig. 3B). In old NTG mice (28–30 weeks of age), the GIR required to maintain plasma glucose at ∼17 mmol/l decreased 70% relative to the young NTG mice (18 vs. 56 μmol · min^–1 · kg^–1 for old and young NTG, respectively; Table 1). The low GIR in both the young and old NTG mice (Figs. 3B and 4B) indicates that these mice are extremely insulin resistant in vivo. The GIR in old TG8 and TG6 mice was increased >5.5-fold relative to the age-matched NTG mice (Table 1 and Fig. 4B). GIR during most of the clamp was elevated in the TG8 relative to the TG6; however, during the last 30 min of the clamp, when the glucose values were relatively steady, the GIR was only slightly increased and this difference was not statistically significant (Figs. 4A and B, Table 1). These data suggest that both lines of TG mice remain responsive to high levels of exogenous insulin with increasing age.

Pancreases from some mice were excised and weighed; the total pancreatic weight in TG8 mice at age 10 and 30 weeks was significantly greater than in NTG or TG6 mice (Table 2), which is consistent with the diminished pancreatic atrophy in the more insulin-sensitive TG8 mice.

Cell surface GLUT4 labeling and 2DG uptake are shown in Figs. 5A and B, respectively. Insulin stimulation increased 2DG transport approximately threefold in EDL muscles from both NTG db/+ and db/db mice. Among the db/db mice, both basal and insulin-stimulated 2DG was greater in EDL muscles from old NTG db/db than young NTG db/db mice. In young and old TG6 and TG8 mice, both basal and insulin-stimulated 2DG transport was elevated above that in NTG mice, with the exception of basal transport in old TG6 mice. In young and old mice, both basal and insulin-stimulated 2DG transport was greater in TG8 mice than TG6 mice, with the exception of basal 2DG transport in young mice, but not in proportion to the greater levels of GLUT4 expression. In old TG6 and TG8 mice, both basal and insulin-stimulated 2DG transport was increased as compared with young mice. In general, the changes in 2DG transport were paralleled by similar changes in cell surface GLUT4 photolabeling. However, a discrepancy was noted between insulin-stimulated 2DG transport and cell surface GLUT4 photolabeling in the old TG6 and TG8 mice, in that whereas 2DG transport tended to increase with aging, cell surface GLUT4 photolabeling either remained unchanged (TG6 mice) or declined (TG8 mice).

Linear regression analyses of GLUT4 ATB-BMPA labeling and 2DG transport are shown in Fig. 6. ATB-BMPA surface-labeled GLUT4 was significantly correlated with 2DG transport (r² = 0.66) for all samples. This correlation improved when the data are subdivided into young (r² = 0.92) and old (r² = 0.76) mice. However, the slopes of the regression lines differ between young and old mice, such that less of an increase in cell surface GLUT4 is observed in old mice compared with young mice for the same incremental increase in 2DG transport.

In the present study, we investigated the effects of different levels of GLUT4 overexpression on insulin resistance in two lines of db/db transgenic mice: hGLUT4 line 6 (two- to threefold overexpression) and hGLUT4 line 8 (four- to fivefold overexpression). In young hGLUT4db/db of both lines, glucose disposal after an oral glucose challenge was markedly improved, similar to previous results (7). Despite the significantly higher levels of GLUT4 in the muscles of the TG8 mice, the improved ability to dispose an oral glucose load was comparable in young TG6 and TG8 mice. However, in older TG6 mice, the enhanced glycemic control observed in the young mice was no longer evident. In contrast, TG8 mice maintained their enhanced glucose tolerance completely until at least age 21 weeks and to a slightly lesser extent until age 30 weeks. Thus, although it seems that the greater level of GLUT4 overexpression in the TG8 mice does not cause a proportional improvement in glucose tolerance in young mice, it does dramatically reduce the decline in glucose tolerance with aging observed in the TG6 mice.

To further assess the effects of GLUT4 overexpression in vivo, we conducted hyperglycemic-hyperinsulinemic clamp studies in mice of age ∼10 and ∼30 weeks. The severe insulin resistance in the NTG db/db mice forced us to conduct these experiments at a blood glucose level of 17 mmol/l. At both ages, TG8 mice had increased GIR compared with NTG db/db mice. Glucose disposal was also increased in 30-week-old TG6 mice relative to NTG mice. Although TG6 mice have a lower level of GLUT4 overexpression relative to TG8 mice, during the last 30 min of the clamp, their GIR was not significantly lower than the GIR of the TG8 mice. These data directly demonstrate that overexpression of GLUT4 on an insulin-resistant background results in improved glucose homeostasis as a result of enhanced peripheral glucose utilization. However, the level of GLUT4 overexpression and insulin-stimulated glucose disposal in vivo at high insulin concentrations does not show a one-to-one correspondence. In addition, the response to insulin in TG mice in the glucose clamp studies did not diminish. This finding differs from the results obtained in NTG mice, in which the GIRs required to maintain a blood glucose level of 17 mmol/l decreased ∼70% between age 10 and 30 weeks. The glucose clamp data suggest that despite the in vivo impairment in glucose tolerance demonstrated by the results of the oral glucose challenge, skeletal muscle of both TG6 and TG8 mice maintain their maximal capacity to respond to insulin during aging. The apparent discrepancy between the oral glucose tolerance test and glucose clamp data might be explained if the secretion of endogenous insulin from the pancreas during the oral glucose challenge did not reach the high levels infused during the clamps. Therefore, this could indicate that the higher level of expression of GLUT4 in skeletal muscle may preserve pancreatic function in the TG8 mice through chronically lowered blood glucose. This hypothesis is supported by our findings that pancreatic atrophy was reduced in TG8 mice (Table 2).

One potential explanation for the age-dependent waning of the improved glucose disposal after an oral glucose challenge was that the expression of the hGLUT4 transgene diminished as the mice aged. However, we demonstrated that total cellular GLUT4 protein expression actually increased slightly in 20- to 25-week-old TG6 mice compared with 10-week-old TG6 mice (Fig. 1). Although there was a small decrease (14%; P < 0.05) in GLUT4 expression in TG8, GLUT4 remained more than fourfold overexpressed in this transgenic line of mice. Furthermore, continued functional expression of the GLUT4 transgene in vivo was confirmed in 30-week-old TG8 mice by quantitating cardiac glycogen content; these mice maintained a similar fourfold increase in cardiac glycogen content compared with NTG mice, as was observed in 10-week-old mice (W.J. Zavadoski, S.C. McCoid, and E.M. Gibbs, unpublished observations). The minimal changes in GLUT4 expression during the period in which there was a dramatic decline in glycemic control (most notably in TG6 mice) strongly suggests that a factor independent of the level of total cellular GLUT4 expression was responsible for the diabetic progression. Along this line, pancreatic atrophy is known to occur in NTG db/db mice as they age (13,14), and here we demonstrated that the atrophy is diminished in TG8 but not TG6 mice at age ∼30 weeks. This finding offers a potential explanation for the more prolonged improvement in glucose disposal observed in TG8 mice compared with TG6 mice, in that secretion of endogenous insulin after an oral glucose challenge seems to remain sufficient to promote enhanced glucose disposal in TG8 mice until a later age than in TG6 mice. In turn, the smaller effect on glucose disposal after a glucose challenge in TG6 mice compared with TG8 may be attributable to the higher ambient plasma glucose levels and induction of a more severe insulin resistance (4,22,23).

In our initial test of insulin action in vitro using EDL muscles, we were surprised to discover that neither basal nor insulin-stimulated glucose transport was abnormal in muscles isolated from NTG db/db mice compared with NTG db/+ mice (10) (Fig. 5). Moreover, the glucose transport activities were roughly paralleled by the levels of cell surface GLUT4. Thus, in contrast to the expected result, isolated EDL muscles from db/db mice do not exhibit an insulin-resistant glucose transport response to a maximally stimulating concentration of insulin. Furthermore, the lack of insulin resistance was also observed in epitrochlearis and soleus muscles isolated from db/db mice (G.J. Etgen, Jr., and E.M. Gibbs, unpublished observations). Compared with NTG db/db mice, progressively increasing GLUT4 overexpression results in progressively increasing basal and insulin-stimulated transport activities and cell surface GLUT4 levels. Moreover, the cell surface GLUT4 labeling data are highly correlated with the 2DG transport rates (r² = 0.92) (Fig. 6), indicating that the elevations in basal and insulin-stimulated glucose transport can be fully explained by increased appearance of GLUT4 at the cell surface. This conclusion agrees with results of our earlier study in transgenic GLUT db/+ mice (10). A similar correlation between cell surface GLUT4 and glucose transport activity was also observed in older mice, although when the surface labeling data were plotted as a function of transport activity, the slope of the regression line was less steep and the correlation was not as good (r² = 0.76) as observed in younger mice. The reduction in the slope occurs because of the discrepancy between insulin-stimulated 2DG transport activity and cell surface GLUT4 in the older mice (i.e., there is a greater increase in 2DG transport activity than in cell surface labeling).Whatever the reason for this discrepancy, it is apparently not due to an alteration in GLUT1 levels, because previous studies have shown that GLUT1 levels are unaltered in transgenic mice in which GLUT4 has been ablated (24). Furthermore, the amount of GLUT1 in skeletal muscle is small in comparison with GLUT4 and functions only in regulation of basal transport (25), whereas the discrepancy in the present work occurred under insulin-stimulated conditions. One potential explanation for this finding is that in the older mice, GLUT4 may exist in an occluded state and may transport glucose but may not be accessible to photolabel (26), although a previous report indicated that occluded transporters do not function in glucose transport (26). Alternatively, there may be some unknown alteration in either GLUT4 or some factor that affects GLUT4 activity with aging that makes it more intrinsically active. Although previous studies have indicated that there is a one-to-one relationship between transport activity and cell surface labeling, this hypothesis has not been tested previously in older animals (10,19,20,27). However, an explanation for this apparent change in the relationship between cell surface GLUT4 and glucose transport remains to be determined by further study.

It is notable that despite the marked in vivo insulin resistance observed for the NTG db/db mice during hyperinsulinemic clamps, their muscles are completely insulin responsive in vitro. This finding suggests that the factor that causes impaired in vivo insulin action in NTG db/db mice is not intrinsic to the muscle itself but rather is a humoral factor that may be related to the fasting hyperglycemia observed in db/db NTG mice and could, in fact, be glucose itself. Recent evidence has shown that chronic exposure of muscle to high circulating levels of glucose can downregulate the glucose transport system (28,29). It has also been shown that incubating muscles from patients with diabetes in a medium containing a lower glucose concentration rapidly reverses the impaired insulin-stimulated glucose transport to normal levels (23). Furthermore, we have recently observed a similar phenomenon in ob/ob mice, which like db/db mice, are hyperglycemic and fail to show insulin resistance in isolated skeletal muscle preparations (G.J. Etgen and E.M Gibbs, unpublished observations). Importantly, this result differs from the situation with the obese, insulin-resistant Zucker diabetic fatty (ZDF) rat, which is not hyperglycemic and whose muscles remain insulin resistant in vitro (27,30). The speculation that hyperglycemia per se causes the in vivo skeletal muscle insulin resistance that can be reversed in vitro is strengthened by the very recent observation by Dolan et al. (27), who found that skeletal muscle isolated from ZDF rats, which are hyperglycemic, is not insulin resistant in vivo. Thus, it is possible that the impaired in vivo insulin action in all three of the hyperglycemic rodent models (db/db, ob/ob, and ZDF) is attributable to the chronically elevated blood glucose that occurs in these animals and that incubation in normoglycemic buffer in vitro restores insulin sensitivity.

Old TG6 mice displayed an abnormal glucose tolerance that was identical to that of the NTG db/db mice. However, during hyperinsulinemic-hyperglycemic clamps, the GIRs of these mice were significantly higher than that of NTG mice. This observation indicates that the muscles of these mice were still able to respond to insulin, because under the clamp conditions used, most of the infused glucose is taken up by skeletal muscle, and the high circulating levels of insulin should fully suppress hepatic glucose production (31). Therefore, it is possible that the impaired glucose tolerance in the old TG6 mice was attributable to a relative reduction of circulating insulin or to a change in hepatic insulin sensitivity. Because insulin levels were markedly lower during the glucose tolerance test than the clamp, it is likely that hepatic glucose production would not be suppressed in the old animals. Furthermore, as mentioned above, the pancreatic weights of TG6 mice were significantly smaller than those of old TG8 mice, which supports this hypothesis. However, it is possible that a decrement in insulin action at a lower blood insulin concentration (i.e., a change in sensitivity) could contribute to the impairment in glucose tolerance in old TG mice. In other words, if the hyperglycemic clamps were performed at a lower insulin concentration, the GIR for old TG6 mice may not differ from that of old NTG mice. Which of these possibilities is correct remains to be determined.

In summary, the present study demonstrates that overexpression of GLUT4 in an insulin-resistant animal model improves both in vivo and in vitro skeletal muscle glucose transport in db/db mice. Moderate (twofold) overexpression resulted in proportional increases in 2DG transport, but overexpression to a higher level (fivefold) did not increase 2DG transport proportionally. Despite the loss of in vivo glucose tolerance in the TG6 mice, GLUT4 overexpression prevented the decline in skeletal muscle insulin responsiveness observed in the NTG mice. Thus, these data suggest that increasing GLUT4 at the plasma membrane may be an effective therapy for type 2 diabetes, even in the face of severe insulin resistance.

We thank Geoffrey D. Holman for the gracious donation of the ATB-BMPA photolabel.