Glucagon-Like Peptide-1 Gene Therapy in Obese Diabetic Mice Results in Long-Term Cure of Diabetes by Improving Insulin Sensitivity and Reducing Hepatic Gluconeogenesis

Heterozygous OB/ob mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred and maintained at the animal facility at Rosalind Franklin University of Medicine and Science. Animals were fed ad libitum on a standard rodent diet. Pair-fed diabetic ob/ob mice were given the same daily amount of food as that eaten by the corresponding rAd-GLP-1–treated group during the previous day. The wild, lean (OB/OB), and mutant obese (ob/ob) genotypes were screened by PCR. Mutant ob/ob mice were monitored for the development of hyperglycemia using a glucometer. All animal experiments were approved by the institutional animal care and use committee at the Rosalind Franklin University of Medicine and Science.

Recombinant adenoviral vectors expressing GLP-1 and β-galactosidase (recombinant adenovirus expressing β-galactosidase [rAd-βgal]), as a control, were constructed as previously described (10). The recombinant adenoviruses were produced and amplified in human embryonic kidney cell line (HEK-293). After purification of viruses by CsCl-gradient ultracentrifugation, viral titer was determined by optical particle units and 50% tissue culture infectious dose. Six- to 8-week-old diabetic male and female ob/ob mice (blood glucose levels >250 mg/dl for 3 consecutive days) were injected via the tail vein with rAd-GLP-1 or rAd-βgal (4 × 10⁹ plaque-forming units [pfu]), and body weights were measured weekly. Blood glucose levels were determined every 2 days using a glucometer.

Mice were not fed for 4 h, and a glucose solution (2 g/kg body wt) was injected intraperitoneally. Blood glucose levels were measured at 0, 30, 60, 90, 120, 150, and 180 min after glucose injection.

Mice were not fed for 4 h, and islets were isolated as described previously (11). Total RNA was isolated from the islets, and expression of insulin mRNA was analyzed by RT-PCR using the following primers: sense 5′-TCAGAGACCATCAGCAAGCAG-3′ and antisense 5′-GTCTGAAGGTCCCCGGGGCT-3′. Expression of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was analyzed as an internal control using the following primers: sense 5′-AGTGCCAGCCTCGTCCGTA-3′ and antisense 5′-TGAGCCCTTCCACAATGCAA-3′.

Total RNA was isolated from the liver, and cDNA was synthesized using a Superscript III Reverse Transcriptase Kit (Invitrogen, Carlsbad, CA). PCR was carried out in a LightCycler (Roche Applied Science, Indianapolis, IN) at 95°C for 15 min, followed by 45 cycles of 95°C for 15 s, 55°C for 20 s, and 72°C for 35 s. The primer sequences used were fatty acid synthase (FAS): sense 5′CAAGACTGACTCGGCT-3′, antisense 5′-GATGCAATCTATGTAGTAGGC-3′; phosphoenolpyruvate carboxykinase (PEPCK): sense 5′-CTCCGTAGCTGGTTCG-3′, antisense 5′-CGATCCGCAACGCAAA-3′; and glucose-6-phosphatase (G6Pase): sense 5′-GTGTTGACATCGGCCC-3′, antisense 5′-AACTGAAGCCGGTTAG-3′. As an internal control, GAPDH mRNA was amplified. Relative copy number was calculated using the threshold crossing point (C_t) as calculated by the LightCycler software combined with the ΔΔ C_t calculations.

Mice were not fed for 4 h, and blood samples were collected. The concentration of serum insulin was measured by a rat ultrasensitive insulin enzyme immunosorbent assay kit, which is cross-reactive with mouse insulin (Crystal Chem, Dowers Grove, IL). Serum triglycerides and free fatty acids (FFAs) were measured using a colorimetric kit (Sigma, St. Louis, MO) and FFA assay kit (Wako Chemicals, Neuss, Germany), respectively. Serum glucagon and GLP-1 levels were measured using a glucagon radioimmunoassay kit (Linco Research, St. Charles, MO) and GLP-1 radioimmunoassay kit (Phoenix Pharmaceuticals, Inc., Belmont, CA), respectively.

Mice were injected with insulin (2 units/kg body wt i.p.; Eli Lilly, Indianapolis, IN), and blood glucose levels were measured at 0, 30, 60, and 90 min after insulin injection.

Quantitative evaluation of the β-cell area was performed on insulin-stained sections using the UTHSCSA Image Tool program. More than 400 serial sections (5-μm thickness) were prepared from each mouse, and every 10th section was stained with anti-insulin antibody. The ratio of the β-cell area was calculated by dividing the area of all insulin-positive cells by the total area of the pancreas. The β-cell mass was calculated by multiplying the pancreas weight by the ratio of the β-cell area.

Insulin was extracted from the pancreas as previously described (12) and measured using an insulin enzyme immunosorbent assay kit (Crystal Chem).

Epididymal adipose tissue was removed, and adipocytes were prepared as previously described (13). Adipocytes (3–4 × 10⁵/ml) were suspended in Krebs-Ringer HEPES buffer and treated with 10 nmol/l insulin for 30 min. After incubation with 2-deoxy-d-[1-³H]glucose (0.5 μCi, 0.125 mmol/l; Amersham Pharmacia Biotech, Buckinghamshire, U.K.) for 3 min, glucose uptake was measured as previously described (14).

Mice were injected with insulin (2 units/kg, i.m.), and liver and muscle samples from the hind limb were harvested 15 min later. Tissue lysates (1.5 mg protein) were incubated overnight with anti-insulin receptor substrate (IRS)-1 (Upstate Biotech, Lake Placid, NY) at 4°C. Twenty microliters of agarose conjugate suspension (Santa Cruz Biotechnologies, Santa Cruz, CA) was added to the samples, and samples were incubated at 4°C on a rotating device for 3 h. After centrifugation, pellets were washed with radioimmunoprecipitation assay buffer (50 mmol/l Tris-HCl, pH 7.4; 1% NP-40; 0.25% Na-deoxycholate; 150 mmol/l NaCl; 1 mmol/l EDTA; 1 mmol/l PMSF; 1 μg/ml aprotinin, leupeptin, and pepstatin; 1 mmol/l Na₃VO₄; and 1 mmol/l NaF). Pellets were resuspended in 2× Laemmlie sample buffer and boiled for 5 min. These immunoprecipitated samples were used for SDS-PAGE and immunoblotting.

Immunoprecipated samples or whole lysates were prepared as described above and subjected to SDS-PAGE. Proteins were electrotransferred to nitrocellulose membranes. Membranes were incubated with anti–IRS-1, anti-phosphotyrosine (Upstate Biotech), anti-Akt, or anti–phospho-Akt (Ser 473) antibodies (Cell Signaling Technology, Danvers, MA) overnight at 4°C. After washing, membranes were incubated with a horseradish peroxidase–conjugated secondary antibody (goat anti-rabbit or goat anti-mouse IgG; Chemicon International, Temecula, CA) for 1 h at room temperature. Reactive bands were detected by enhanced chemiluminescence (Pierce Chemical, Rockford, IL) and quantified by scanning densitometry.

Protein kinase C (PKC) activity was measured using a PKC assay kit (Upstate Biotech). Lysates were incubated with the substrate (peptide sequence: QKRPSQRSKYL) and [γ-³²P]ATP, then phosphorylated peptide was separated using p81 phosphocellulose paper. Radioactivity on the disks was counted using a scintillation counter. Specific PKC activity was determined after subtraction of the activity without substrates.

Hepatic glucose production in vivo was determined as described previously (15). Briefly, mice were anesthetized, catheterized in the left jugular vein, and allowed 3–5 days of recovery. Mice were not fed for 18 h and infused with [3-³H]glucose (0.05 μCi/min; DuPont NEN Research Products, Boston, MA). Blood samples (35 μl) were taken from the tip of the tail at 90 and 120 min after the initiation of glucose infusion. Plasma glucose concentration was determined by a glucose assay kit (BioVision Research Products, Mountain View, CA). To determine [3-³H]glucose, 10 μl of plasma was mixed with an equal volume of 20% trichloroacetic acid to precipitate protein. The supernatant was heated at 65°C to dryness to evaporate [³H]H₂O, and the samples were reconstituted in 100 μl H₂O. Radioactivity was measured using a scintillation counter. Mean steady-state hepatic glucose production was calculated by dividing the [3-³H]glucose infusion rate by the mean plasma glucose specific activity at 90 and 120 min.

Data are presented as means ± SD. Statistical significance of differences was analyzed by unpaired Student's t test for comparison of two groups or ANOVA followed by Tukey honestly significant difference test for multiple comparisons. P < 0.05 was accepted as significant.

To examine whether rAd-GLP-1 treatment efficiently produces GLP-1 in vivo, we injected rAd-GLP-1 (4× 10⁹ pfu) into diabetic ob/ob mice and measured serum GLP-1 levels after 4 h without food at 1, 2, and 4 weeks after rAd-GLP-1 treatment. Serum GLP-1 levels in rAd-GLP-1–treated mice were high at 1 week and gradually decreased at 2 and 4 weeks after treatment, whereas serum GLP-1 levels were very low in wild-type mice and rAd-βgal–treated diabetic mice (Fig. 1A). Blood glucose levels gradually decreased in rAd-GLP-1–treated diabetic ob/ob mice and reached normoglycemia within 4 days after treatment. Mice then became slightly hypoglycemic for a week and then returned to normoglycemia, which was maintained for 60 days when the experiment was terminated. In contrast, rAd-βgal–treated mice remained hyperglycemic, as did untreated diabetic ob/ob mice (Fig. 1B). Food intake in rAd-GLP-1–treated mice rapidly decreased over the first 4 days after treatment, continued to decrease over days 6–8, and then gradually increased to an amount similar to that ingested by rAd-βgal–treated and untreated control groups by 20 days after treatment (Fig. 1C). Changes in food intake appeared to be inversely correlated with the amount of circulating GLP-1. Gain of body weight was significantly lower in rAd-GLP-1–treated as compared with rAd-βgal–treated mice over the 8 weeks of the experiment (Fig. 1D).

To determine whether rAd-GLP-1 treatment affects triglyceride and FFA production, we measured serum triglyceride and FFA levels after 4 h without food at 2 weeks after rAd-GLP-1 treatment. Both serum triglyceride (Fig. 2A) and FFA levels (Fig. 2B) were significantly decreased compared with rAd-βgal–treated mice. Serum triglyceride and FFA levels in the pair-fed group were not significantly different from those of untreated or rAd-βgal–treated mice.

To determine whether blood glucose levels are properly controlled in rAd-GLP-1–treated ob/ob mice, we performed intraperitoneal glucose tolerance tests in normoglycemic ob/ob mice at 2 weeks after rAd-GLP-1 treatment. Blood glucose levels in rAd-GLP-1–treated mice were significantly lower at all time points following glucose injection compared with rAd-βgal–treated mice, and the kinetics of exogenous glucose clearance were similar to that in lean wild-type mice. In contrast, untreated and rAd-βgal–treated ob/ob mice did not reach normal glucose levels after glucose loading. Blood glucose levels in the pair-fed group were not significantly different at all time points compared with the untreated and rAd-βgal–treated groups, except at 60 min after glucose loading (Fig. 3).

As GLP-1 is known to have proliferative, differentiating, and antiapoptotic effects on β-cells (2), we measured β-cell mass. The β-cell mass was significantly increased in rAd-GLP-1–treated mice compared with rAd-βgal–treated mice (Fig. 4A). We then measured the pancreatic insulin content and insulin mRNA expression in islets and found that pancreatic insulin content was significantly decreased in rAd-GLP-1–treated mice (Fig. 4B), which is consistent with the decrease in insulin mRNA expression in pancreatic islets compared with rAd-βgal–treated mice (Fig. 4C). When we examined serum insulin levels before and at 30 min after glucose injection, we found that basal serum insulin levels were significantly reduced in rAd-GLP-1–treated ob/ob mice compared with rAd-βgal–treated mice. However, glucose-stimulated insulin secretion was increased over basal levels in rAd-GLP-1–treated mice, whereas there was no increase in rAd-βgal–treated mice (Fig. 4D). These results indicate that rAd-GLP-1 treatment improved β-cell function.

To address whether rAd-GLP-1 treatment improves insulin sensitivity, we performed insulin tolerance tests. rAd-GLP-1–treated mice showed an enhanced reduction in glucose levels in response to exogenous insulin at 30 min following insulin injection compared with untreated and rAd-βgal–treated ob/ob mice, and this reduction was comparable with lean wild-type mice. Glucose reduction in the pair-fed group was not different from untreated and rAd-βgal–treated mice (Fig. 5A). Because insulin induces glucose uptake in peripheral tissues, resulting in reduction of glucose levels, we measured the insulin-stimulated glucose transport in adipocytes of rAd-GLP-1–treated ob/ob mice. Glucose transport was significantly lower in untreated and rAd-βgal–treated diabetic ob/ob mice, whereas rAd-GLP-1 treatment restored glucose transport to the levels seen in lean wild-type mice (Fig. 5B). These results indicate that improved insulin sensitivity contributes to increased glucose uptake, resulting in improved glucose homeostasis.

To determine whether the improvement of insulin sensitivity in rAd-GLP-1–treated ob/ob mice is accompanied by restoration of insulin signal transduction, we first examined the levels of total and phosphorylated IRS-1 protein in muscle of rAd-GLP-1–treated mice. The total amount of IRS-1 protein was not changed, but insulin-stimulated phosphorylation of IRS-1 was upregulated in the muscle of rAd-GLP-1–treated mice compared with untreated and rAd-βgal–treated mice (Fig. 6A). We then examined the activation of downstream molecules of insulin signaling in the muscle of rAd-GLP-1–treated ob/ob mice. Insulin-stimulated phosphorylation of Akt was significantly increased in rAd-GLP-1–treated mice compared with untreated and rAd-βgal–treated ob/ob mice, whereas the total Akt and basal level of phosphorylated Akt were not altered (Fig. 6B). Insulin-stimulated PKC activity was also increased in the muscle of rAd-GLP-1–treated ob/ob mice compared with untreated and rAd-βgal–treated mice (Fig. 6C).

We then examined the insulin-stimulated signaling molecules in the liver of rAd-GLP-1–treated ob/ob mice. The basal level of IRS-1 protein was decreased in diabetic ob/ob mice, and rAd-GLP-1 treatment significantly increased this level comparable with that of wild-type lean mice. In addition, insulin-stimulated phosphorylation of IRS-1 was significantly increased in rAd-GLP-1–treated mice compared with untreated and rAd-βgal–treated mice (Fig. 7A). However, rAd-GLP-1 treatment did not affect either insulin-stimulated activation of Akt or total Akt protein levels in the liver (Fig. 7B). PKC activity was significantly increased after stimulation with insulin, but the basal level of PKC activity was unaffected compared with untreated and rAd-βgal–treated ob/ob mice (Fig. 7C).

To determine whether rAd-GLP-1 treatment affects glucose production, we measured hepatic glucose production by clamp studies. rAd-GLP-1 treatment significantly decreased basal hepatic glucose production compared with rAd-βgal–treated ob/ob mice (Fig. 8A). Glucagon increases gluconeogenesis in the liver (16), and GLP-1 is known to downregulate glucagon secretion (17). Thus, we examined whether decreased hepatic glucose production is due to the reduction of glucagon. We found that 4-h fasting serum glucagon levels at 2 weeks after treatment were not significantly different between rAd-GLP-1–and rAd-βgal–treated groups (Fig. 8B).

To determine whether rAd-GLP-1 treatment affects expression of genes involved in glucose metabolism in the liver, we examined the expression of G6Pase and PEPCK mRNA, which are involved in gluconeogenesis. The expression of both G6Pase and PEPCK mRNA was significantly decreased in the liver of rAd-GLP-1–treated ob/ob mice compared with that in untreated and rAd-βgal–treated diabetic ob/ob mice (Figs. 8C and D). Because we found that serum triglyceride levels were reduced by rAd-GLP-1 treatment, we examined the expression of FAS mRNA, which is involved in lipogenesis, and found that the expression of FAS mRNA was also significantly decreased in the liver after rAd-GLP-1 treatment (Fig. 8E).

GLP-1 has been studied as a potential therapy for type 2 diabetes; however, its short duration of action, owing to its short half-life, limits the maintenance of therapeutic levels by administration of GLP-1 peptide (1,2). Various methods have been tried to overcome this problem, including development of a long-acting analog of GLP-1 and an inhibitor of the GLP-1 degrading enzyme (5). Clinical trials using Exenatide, a long-acting, synthetic version of exendin-4, found that it improved glycemic control by stimulating glucose-dependent insulin secretion, suppressing glucagon secretion, slowing gastric emptying, and enhancing β-cell function; however, twice daily injections were required (18). A plasmid construct containing a modified cDNA of GLP-1 (7–37) injected into Zucker diabetic fatty rats lowered blood glucose levels, but not to the normal range (19), probably due to the low expression level of GLP-1.

In this study, we sought to deliver sustained, therapeutic levels of GLP-1 by injecting diabetic ob/ob mice with a recombinant adenovirus expressing GLP-1 under the control of the cytomegalovirus promoter (rAd-GLP-1). Circulating GLP-1 was significantly increased for at least 4 weeks compared with rAd-βgal–treated diabetic and untreated normal mice, indicating that a substantial amount of circulating GLP-1 is exogenously produced by rAd-GLP-1 therapy. A recent report showed that intramuscular injection of a plasmid expressing a GLP-1/IgG-Fc fusion construct significantly lowered fasting blood glucose levels in db/db mice but not until 3 months after injection (20). In our study, diabetic ob/ob mice given a single injection of rAd-GLP-1 showed lowered blood glucose levels within 4 days, and normoglycemia was maintained. The rapid remission of diabetes in our study relative to the recent report is probably because we used ob/ob mice, which are less severely hyperglycemic than db/db mice, and an adenoviral vector, which permits high transgene expression due to high transduction efficiency.

Previous reports showed that GLP-1 or exendin-4 treatment increased β-cell mass (21,22). Consistent with this, our rAd-GLP-1 therapy also increased β-cell mass in ob/ob mice. Despite this increase in β-cell mass, we found that pancreatic insulin content was significantly decreased in rAd-GLP-1–treated ob/ob mice compared with rAd-βgal–treated mice, which is supported by decreased insulin mRNA expression in the islets and decreased fasting serum insulin levels in rAd-GLP-1–treated mice. Similarly, treatment with exendin-4 or GLP-1 reduced fasting serum insulin levels in ob/ob and db/db mice (4,21). Our interpretation is that rAd-GLP-1 increases β-cell mass due to its proliferative and antiapoptotic effects on β-cells, but basal insulin content is reduced, probably due to the improvement of insulin sensitivity. Also, the increase in insulin secretion after glucose loading suggests that β-cell function is also improved by rAd-GLP-1 treatment.

It has been reported that long-term treatment with GLP-1 or its analog improved insulin sensitivity in both animal models and human type 2 diabetic patients (4,6,23). Improvement of insulin sensitivity by exendin-4 treatment was found to be independent of body weight (23). However, the precise mechanisms are not fully understood. Insulin tolerance tests showed that exogenous insulin appropriately cleared blood glucose in rAd-GLP-1–treated ob/ob mice. In addition, hepatic glucose production and serum FFA and triglyceride levels were significantly decreased compared with rAd-βgal–treated mice. There was no significant difference in insulin tolerance and serum FFA and triglyceride levels between the pair-fed group and the untreated or rAd-βgal–treated groups, suggesting that these effects in rAd-GLP-1–treated mice are not likely to be the result of reduced food intake and body weight. Although all these results showed that rAd-GLP-1 treatment clearly improved insulin sensitivity, hyperinsulinemic-euglycemic clamp studies will be required to directly quantify the improvement of insulin sensitivity.

Impaired glucose transport is one of the major factors contributing to insulin resistance, and insulin-mediated glucose disposal is greatly reduced in type 2 diabetes (24). rAd-GLP-1 treatment significantly improved insulin-stimulated glucose uptake in adipocytes compared with rAd-βgal–treated adipocytes. Insulin stimulates glucose transport through activation of the insulin signaling pathway. The binding of insulin to its receptor results in the activation of the receptor tyrosine kinase, which subsequently phosphorylates IRS-1/2 (25). The phosphorylated IRS-1/2 activates phosphoinositol-3-kinase, and activation of the phosphoinositol-3-kinase signaling pathway activated downstream protein kinases such as protien kinase B (PKB)/Akt and PKC (26,27). Treatment of diabetic ob/ob mice with rAd-GLP-1 increased insulin-stimulated tyrosine phosphorylation of IRS-1 in both muscle and liver. PKB/Akt is required for insulin-stimulated glucose transport and glycogen synthesis in muscle and adipocytes (28–30) and is also required for glycogen synthesis and inhibition of gluconeogenesis in the liver (31–33). It is known that ob/ob mice have defects in the activation of PKB/Akt in both liver and muscle (34); thus, we examined whether rAd-GLP-1 treatment would reverse this defect. There was a marked increase in the level of insulin-stimulated Akt-Ser 473 phosphorylation in the muscle but not in the liver. This difference might be due to the differential regulation of signaling molecules induced by insulin in different tissues in rAd-GLP-1–treated ob/ob mice. PKCs are also involved in insulin-stimulated glucose transport in muscle and adipocytes (35), and atypical PKC activation is impaired in the muscle of ob/ob mice (36). We found that insulin-induced PKC activity significantly increased in both muscle and liver after rAd-GLP-1 treatment. Taken together, these results indicate that rAd-GLP-1 therapy restored insulin signaling by increasing the activation of IRS-1, PKB/Akt, and PKC, subsequently improving glucose transport in peripheral tissues such as adipose tissue.

Gluconeogenesis plays an important role in glucose homeostasis (37). Elevated hepatic glucose production, due to increased gluconeogenesis, glycogenolysis, or both, is associated with the pathogenesis of type 2 diabetes (38,39). We found that basal hepatic glucose production was significantly decreased by rAd-GLP-1 treatment. The expression of G6Pase, which is involved in gluconeogenesis and glycogenolysis (40), and PEPCK, which is involved in gluconeogenesis (41), was significantly decreased in the liver of rAd-GLP-1–treated ob/ob mice, suggesting that the blood glucose–lowering effect of rAd-GLP-1 might be due in part to the reduction of hepatic glucose output. As elevated glucagon levels are correlated with increased hepatic glucose production (42), we examined whether rAd-GLP-1 treatment reduced glucagon secretion. However, we found no significant changes in fasting serum glucagon levels between rAd-GLP-1–and rAd-βgal–treated mice, suggesting that the decrease of hepatic gluconeogenesis may not be a result of reduced glucagon levels. The effect of GLP-1 on inhibition of glucagon secretion is glucose dependent (17). As blood glucose levels are already normalized at 2 weeks after treatment, GLP-1 may not affect fasting glucagon levels at this time. This result is consistent with previous findings, which showed that continuous infusion of GLP-1 or treatment with GLP-1 analog had no effect on fasting glucagon levels (6,43,44). An increase in hepatic triglyceride synthesis and secretion is observed in type 2 diabetes (45). FAS plays a central role in de novo lipogenesis (46,47). In rAd-GLP-1–treated mice, FAS mRNA expression in liver and serum triglycerides were decreased, suggesting that GLP-1 reduces lipid production. Consistent with this result, it was reported that treatment of ob/ob mice with exendin-4, an analog of GLP-1, reversed hepatic steatosis (4).

The presence of GLP-1 receptors in peripheral tissues such as muscle, fat, and liver is controversial (4,48,49); therefore, it is unclear whether the improvement in insulin sensitivity in extrapancreatic tissues by rAd-GLP-1 treatment is due to direct effects on these tissues or indirect effects through the regulation of other molecules. Regardless, it is clear that a single administration of rAd-GLP-1 resulted in long-term remission of diabetes in ob/ob mice. The effect of rAd-GLP-1 treatment on the long-term remission of diabetes might result from the improvement of β-cell function, reduction of gluconeogenesis, and the improvement of insulin sensitivity.

This research was sponsored, in part, by a grant from the American Diabetes Association.

We thank Dr. Daniel Drucker for providing GLP-1 cDNA, Dr. Pastor Conceyro for technical help with the catheterization for the clamp studies, Keith Philibert and Adam Starkey for animal care, and Dr. Ann Kyle for editorial assistance.