Absence of Glucagon and Insulin Action Reveals a Role for the GLP-1 Receptor in Endogenous Glucose Production

Type 1 diabetes is a life-threatening syndrome resulting from unbridled hepatic glycogenolysis, gluconeogenesis, and ketogenesis that perpetuate a state of chronic fuel utilization (1). Manifestation of this condition occurs due to loss of the anabolic effects of insulin on glucose and lipid metabolism (1). A significant contributor to the dysregulated physiological state is the absence of the inhibitory effect insulin imposes on glucagon secretion (2–7). The effect of excess glucagon is significant because of the broad stimulatory actions glucagon has on catabolic processes in the liver and the hyperglucagonemia that is present in poorly controlled diabetes (8,9). Thus, due to its role in exacerbating the metabolic consequences of insulinopenia, glucagon was proposed several years ago to be an essential component in the pathogenesis of type 1 diabetes (10,11). Therefore, to improve glucose homeostasis in the absence of insulin, therapies that suppress glucagon secretion or block its action may improve overall glycemia by reducing the effect of a primary stimulus of endogenous glucose production (EGP).

Bolstering the now long-standing hypothesis that suppressing glucagon action would have antihyperglycemic effects are studies revealing the remarkable metabolic phenotype of glucagon receptor (Gcgr) knockout mice. Although these animals have extraordinarily high circulating concentrations of glucagon and pancreatic α-cell hyperplasia, Gcgr-deficient mice display lower blood glucose levels, improved glucose tolerance, and have reduced adiposity (12,13). Gcgr^−/− animals also have slower gastric emptying and are resistant to diet-induced obesity (13,14). Compelling evidence exists supporting the concept that glucagon inhibition would improve glycemic control in the state of insulin deficiency.

Studies have demonstrated that Gcgr^−/− mice administered the diabetogenic chemical agent streptozotocin (STZ), a cytotoxic glucose analog that induces necrotic destruction of pancreatic β-cells (15), are resistant to STZ-induced hyperglycemia (14). This discovery is provocative and has many potential implications; however, the STZ treatment failed to abolish plasma insulin concentrations and staining for insulin-positive cells in the pancreata of STZ-treated Gcgr^−/− animals (14). To cause maximal β-cell loss, Unger and colleagues (16) established a high-dose STZ treatment regimen that resulted in nearly complete β-cell destruction in Gcgr^−/− mice. Importantly, these studies show the metabolic maladies that result from insulinopenia, especially profound hyperglycemia and hyperketonemia, do not develop in animals lacking the GCGR (16). Together, data from these experiments support a substantial role of glucagon signaling in the metabolic phenotype of insulin deficiency.

Although experiments using the high-dose STZ protocol to induce β-cell destruction in Gcgr-null mice were well designed, protection from insulinopenic hyperglycemia in this model may result from a combination of factors and not solely as a consequence of the direct loss of GCGR function. The initial characterization of Gcgr^−/− mice demonstrated that these animals have high circulating concentrations of the antidiabetogenic peptide GLP-1, likely generated by proteolytic processing of the high levels of proglucagon within the hyperplastic α-cells (12). Consistent with this compensatory α-cell hyperplasia in Gcgr^−/− animals, concomitant elevation of plasma GLP-1 also occurs with progressive hyperglucagonemia in studies where GCGR antagonism is pharmacologically achieved (17,18). Omar et al. (19) recently provided data supporting a contributing role for GLP-1 in glucose homeostasis of STZ-treated Gcgr^−/− mice by showing animals administered the GLP-1 receptor (GLP-1R) peptide antagonist, exendin-4_(9-39), display poorer glucose tolerance compared with animals that had not received the antagonist. This report also indicated circulating levels of fibroblast growth factor 21 (FGF21) are increased in Gcgr^−/− mice, a finding paradoxical to other studies showing activation of the GCGR increases hepatic and circulating FGF21 levels (20,21). The combination of exendin-4_(9-39) and FGF21 antisera worsened the glucose excursion for animals undergoing an oral glucose tolerance test (OGTT) (19). These experiments exemplify the potential importance of fundamentally understanding the mechanisms whereby Gcgr^−/− mice are refractory to insulin deficiency and may enable new proposals of alternate treatment regimens for type 1 diabetes.

The significance of glucagon action during insulinopenia is further demonstrated in elegant studies showing that replenishment of hepatic GCGR using an adenovirus system causes severe hyperglycemia in euglycemic, insulin-deficient Gcgr^−/− mice and that once expression of GCGR wanes, euglycemia reappears (22). Although these studies support the hypothesis of glucagon-induced hyperglycemia as the main driver of diabetes, it remains clear that there are multiple metabolic responses to Gcgr deficiency, some of which may contribute to ameliorating hyperglycemia in this model. Therefore, this unique phenotype should be fully interrogated to investigate the importance of factors that appear dysregulated as a result of the loss of Gcgr. Owing to the role of GLP-1 in glucose metabolism (23) and its high circulating levels in Gcgr^−/− mice (12) as well as in animals administered GCGR antagonists (17,18), we investigated the physiological consequences of Glp-1r ablation in insulin-deficient Gcgr^−/− mice and in insulinopenic wild-type and Glp-1r^−/− animals treated with a high-affinity GCGR antagonist antibody. Results from these studies suggest that interventions to improve glucose control in the absence of insulin may benefit greatly from adjunctive therapies that block glucagon action and activate GLP-1R signaling.

All mouse models were generated by Eli Lilly and Company in collaboration with Taconic (Hudson, NY). Glp-1r^−/− mice were previously described (24) and bred with Gcgr^+/− mice licensed from Deltagen (San Mateo, CA) to produce double heterozygous Gcgr^+/−:Glp-1r^+/− breeders that produced littermates with the following genotypes: Gcgr^+/+:Glp-1r^+/+ (wild-type); Gcgr^+/+:Glp-1r^−/− (Glp-1r^−/−); Gcgr^−/−:Glp-1r^+/+ (Gcgr^−/−); and Gcgr^−/−:Glp-1r^−/− double knockout (DKO) mice. PCR genotyping for the Gcgr allele was performed with the following primers (5′-3′): wild-type allele –T648S GTTGAGGAAACAGTAGAGAACAGCC and T648W ACCCTCATCCCTCTGCTGGGGGTCC; mutant allele –T648S and NEO3195 GGGCCAGCTCATTCCTCCCACTCAT. Genotyping for the Glp-1r allele has been described (24).

All mice were singly housed and fed a standard chow diet (Teklad 2014; Harlan Laboratories, Indianapolis, IN) with bottled water ad libitum in a 12 h light/12 h dark cycle (lights off at 1800). Animals were maintained in accordance with the Eli Lilly and Company Institutional Animal Care and Use Committee and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Animals in the Gcgr ablation models were fasted for 4 h in the morning and injected intraperitoneally with 130 mg/kg STZ (Sigma-Aldrich, St. Louis, MO). This dose caused severe hyperglycemia in wild-type and Glp-1r^−/− mice, whereas the Gcgr^−/− and DKO animals remained healthy. Because the pharmacological GCGR blockade cohort of wild-type and Glp-1^−/− mice was a chronic study, a three-dose paradigm of lower STZ was used with a once-weekly STZ dose of 110–115 mg/kg (or 10 mL/kg citrate vehicle) over 3 weeks. Diabetic animals (blood glucose ≥400 mg/dL) were randomized into groups for weekly anti-GCGR human IgG4 antibody (Ab-4) (25) or IgG4 isotype control subcutaneous injections of 10 mg/kg. The treatment dose was selected based on previous studies in hyperglycemic ZDF rats (25). The concentration of Ab-4 (molecular weight = 143,965 g/mol) was determined in solution by measuring ultraviolet light absorbance (280 nm) and calculated using the molar extinction coefficient of the antibody amino acid sequence. Antibody activity was measured by homogeneous time-resolved fluorescence cAMP assays (Cisbio Assays, Bedford, MA) using HEK293 cells expressing the mouse GCGR or GLP-1R. These assays demonstrated the ability of Ab-4 to block glucagon action.

OGTTs were performed in mice fasted overnight for 16 h and orally gavaged with 2 g/kg glucose (24). For pancreatic hormone extraction, the entire pancreas was removed and homogenized in acid ethanol. After centrifugation, supernatants were analyzed for hormone content. Plasma, serum, and pancreatic insulin, glucagon, and GLP-1 were measured using electrochemiluminescence assays (Meso Scale Discovery, Rockville, MD). For pancreas immunohistochemistry, the whole pancreas was removed and fixed in 10% neutral buffered formalin, followed by paraffin embedding. Pancreata were sectioned at 5 µm, and slides were serially stained for insulin and glucagon with hematoxylin counterstain. Gastric emptying tests were performed using mice that were fasted for 16 h and then administered a semiliquid diet containing acetaminophen via oral gavage. The animals were then bled at 0, 30, 60, and 120 min via tail vein, and plasma acetaminophen concentrations were measured using mass spectrometry.

One week before EGP measurements, Gcgr^−/− and DKO mice were treated with 130 mg/kg STZ. Four days before EGP was assessed, catheters were placed in the left carotid artery and advanced to the aortic arch and in the right jugular vein and advanced to the right atrium. Animals were fasted overnight and allowed to acclimate to study cages for 2 h. All studies lasted 2 h. A bolus/continuous infusion of 3-³H-glucose (6 μCi bolus and 0.1 μCi/min; PerkinElmer, Waltham, MA) was initiated and maintained throughout the test period. Arterial blood was obtained every 15 min to determine glucose concentrations. Blood was collected at the beginning and end of the 3-³H-glucose infusion to monitor hematocrit and plasma insulin and to determine basal EGP.

Graphing and statistical analyses were performed using GraphPad Prism software. Data are presented as mean ± SEM and were compared using ANOVA, followed by the Dunnett test. Repeated-measures ANOVA was used to assess significance between time courses. The statistical significance threshold was set at P ≤ 0.05.

Previous studies show that Gcgr^−/− mice are resistant to STZ-induced hyperglycemia (14,17). Blockade of GCGR, whether by genetic deletion or pharmacological antagonism, leads to expansion of pancreatic α-cells due to the loss of an uncharacterized glucagon-mediated negative feedback mechanism. As a result, plasma concentrations of α-cell–derived glucagon and GLP-1 rise dramatically (Table 1) (17,26). Because GLP-1R activation helps regulate glucose homeostasis (26–29), we investigated whether GLP-1R expression plays a role in maintaining euglycemia in insulin-deficient Gcgr^−/− mice. For these studies, Glp-1r^−/− mice were crossed with Gcgr^+/− mice to produce double heterozygous breeders that enabled the generation of animals lacking both genes. To induce insulin deficiency, wild-type, Glp-1r^−/−, Gcgr^−/−, and DKO animals were administered high-dose STZ (130 mg/kg) and evaluated over the next 7 days. Nonfasted glucose concentrations became pathologically high in wild-type and Glp-1r^−/− mice, and therefore, these animals were removed from the study (Fig. 1A). As expected, Gcgr^−/− mice displayed unchanged glucose concentrations after STZ administration (Fig. 1A), even with a dramatic decrease in pancreatic insulin content (Fig. 1D). Although glucose concentrations in DKO mice after STZ treatment were lower than in wild-type and Glp-1r^−/− animals, these mice showed significantly higher glucose concentrations compared with Gcgr^−/− mice, indicating the GLP-1R is needed for post-STZ euglycemia in the Gcgr^−/− animal (Fig. 1A).

Before STZ treatment, Gcgr^−/− and DKO mice displayed similar oral glucose tolerance and glucose-stimulated insulin secretion (GSIS) during OGTTs (Fig. 1B). Lack of the Glp-1r did not affect oral glucose tolerance in nondiabetic DKO animals, likely due to the compensatory upregulation of other incretin pathways upon Glp-1r ablation (30). In an OGTT 1 week after STZ administration, Gcgr^−/− mice continued to show rapid glucose clearance; however, STZ-treated DKO animals displayed marked glucose intolerance (area under the curve [AUC] = 1,571 mg/dL/min) compared with Gcgr^−/− mice (AUC = 647 mg/dL/min) and compared with their pre-STZ treatment profile (AUC = 919 mg/dL/min; Fig. 1B). These results are similar to those showing acute blockade of the GLP-1R using exendin-4_(9-39) worsens glucose tolerance in STZ-treated Gcgr-null mice (19). An assessment of gastric emptying between the two STZ-treated genotypes failed to show any difference (plasma acetaminophen AUC_(0–2 h): Gcgr^−/− 23.0 ± 1.7, and DKO 25.4 ± 1.8 μg ⋅ h/mL). Importantly, there was no GSIS in the Gcgr^−/− or DKO animals, further indicating both groups became insulin-deficient (Fig. 1C).

To show near-maximum β-cell loss occurred in both genotypes, immunohistochemistry staining for insulin and glucagon was performed on pancreata from nondiabetic mice and 1-week after STZ administration (Fig. 1D). As previously shown, Gcgr ablation caused marked α-cell hyperplasia. STZ-treated animals exhibited profound β-cell loss after STZ treatment, regardless of genotype (Fig. 1D). Body composition was also measured to investigate evidence of differential insulin-mediated lipogenesis between the groups, and there were no differences in adipose or lean mass between Gcgr^−/− and DKO animals before or after STZ administration (data not shown). Furthermore, plasma concentrations of FGF21 were not different between these groups (data not shown). Finally, pancreatic insulin, GLP-1 (total), and glucagon levels were similar in Gcgr^−/− and DKO animals after STZ treatment (Table 2).

The genetic results provide compelling evidence that a functional GLP-1R in Gcgr-null mice is needed to protect against hyperglycemia in the insulin-deficient state. To further evaluate the mechanism by which blockade of glucagon action improves glucose control, a high-affinity GCGR monoclonal antibody antagonist, Ab-4 (25), was used to investigate whether therapeutic intervention would reverse diabetes. Ab-4 shows potent antagonism of glucagon-induced cAMP accumulation in HEK293 cells expressing the mouse GCGR (K_b = 0.76 nmol/L) (Fig. 2A); these data are consistent with previously reported results for Ab-4 in mouse GCGR-containing membrane-binding assays (K_i = 0.83 nmol/L) (25). Ab-4 showed no antagonism of the mouse GLP-1R (Fig. 2B). For the in vivo studies, STZ-induced diabetes was established in wild-type and Glp-1r^−/− mice, and animals were administered Ab-4 once weekly for 7 weeks. To avoid the lethal hyperglycemia observed for these genotypes (Fig. 1A), the STZ treatment paradigm used here resulted in more restrained hyperglycemia, thereby allowing the vehicle/control IgG4-treated mice to survive the study period. Using this approach, in addition to investigating the therapeutic potential of Ab-4 to improve hyperglycemia in diabetic wild-type animals, we pharmacologically recapitulated Gcgr ablation in a Glp-1r–null background. STZ-treated wild-type and Glp-1r^−/− animals were randomized according to their nonfasted blood glucose concentrations into three treatment groups: wild-type (STZ+IgG4), wild-type (STZ+Ab-4), and Glp-1r^−/− (STZ+Ab-4). Also included was a nondiabetic wild-type control group that received only citrate injection initially, followed by weekly administration of the isotype control IgG4 (Fig. 3A). Hyperglycemia was quickly established in STZ-treated animals. Untreated, diabetic wild-type (STZ+IgG4) mice remained hyperglycemic throughout the study (Fig. 3A), whereas Ab-4–treated insulinopenic wild-type (STZ+Ab-4) animals showed normalization of nonfasted blood glucose concentrations after the first antibody dose and remained similar to nondiabetic wild-type controls (Fig. 3A). The Glp-1r^−/− (STZ+Ab-4) group, with pharmacologically blocked GCGR and genetically ablated Glp-1r, experienced marked glucose lowering compared with the wild-type animals (STZ+IgG4); however, their blood glucose levels remained significantly elevated compared with the wild-type (STZ+Ab-4) mice similarly treated with the antibody (Fig. 3A).

After 6 weeks of treatment with Ab-4, the wild-type (STZ+Ab-4) group showed superior glucose tolerance (AUC = 1,618 mg/dL/min) in an OGTT, similar to that of nondiabetic wild-type controls (AUC = 1,096 mg/dL/min) (Fig. 3B). However, despite improvement in nonfasted blood glucose levels in Glp-1r^−/− (STZ+Ab-4) mice after six treatments (Fig. 3A), these animals showed profound glucose intolerance (AUC = 3,180 mg/dL/min) similar to the untreated diabetic controls (AUC = 3,032 mg/dL/min) during an OGTT (Fig. 3B). In addition, wild-type (STZ+Ab-4) animals displayed significantly impaired GSIS compared with nondiabetic controls, similar to the diabetic wild-type and hyperglycemic Ab-4–treated Glp-1r^−/− mice (Fig. 3C). It appears that 6 weeks of Ab-4–mediated GCGR inactivation had little effect on GSIS from any residual β-cells (Fig. 3C). These results are supported by the fact that pancreatic insulin content was less than 10% of nondiabetic controls in all of the STZ-treated groups (Table 2).

As with Gcgr^−/− mice, circulating GLP-1 concentrations were elevated in both Ab-4–treated groups compared with wild-type (citrate+IgG4) and wild-type (STZ+IgG4) animals (Fig. 3D). Fasting insulin levels in all STZ-treated animals were similarly depleted regardless of antibody treatment, indicating β-cell recovery was unlikely to be occurring despite the high levels of GLP-1 (Fig. 3C). Finally, pancreatic GLP-1 and glucagon content were elevated in the Ab-4–treated groups compared with nondiabetic and diabetic wild-type controls (Table 2). Therefore, the elevated ambient glucose concentrations and impaired oral glucose tolerance in Glp-1r^−/− (STZ+Ab-4) compared with the wild-type (STZ+Ab-4) animals cannot be explained by differences in insulin, glucagon, or GLP-1 levels. Because GLP-1R action inhibits gastric transit and food intake, we performed semiliquid gastric emptying tests in all groups and found no differences (plasma acetaminophen AUC_[0–2 h]: wild-type [citrate+IgG4] 18.5 ± 1.7; wild-type [STZ+IgG4] 26.0 ± 2.8; wild-type [STZ+Ab-4] 25.0 ± 2.0; Glp-1r^−/− [STZ+Ab-4] 21.0 ± 5.6 μg ⋅ h/mL). Food intake in the normoglycemic groups was not different; but as expected, the two hyperglycemic groups ate more, which is consistent with a catabolic state (average food intake: wild-type [citrate+IgG4] 126.5 ± 6.3; wild-type [STZ+IgG4] 263.6 ± 28.6; wild-type [STZ+Ab-4] 142.6 ± 9.4; Glp-1r^−/− [STZ+Ab-4] 225.8 ± 25.3 mg food/g body weight/day).

Although there is little meaningful evidence indicating expression of the GLP-1R in the liver (24,31), there are several reports of GLP-1–mediated regulation of hepatic glucose production in a manner separate from its ability to suppress glucagon release and, intriguingly, independent of insulin (32–35). We therefore measured EGP rates to investigate whether a GLP-1R–dependent mechanism accounts for regulating glycemia in Gcgr^−/− animals. Although the Gcgr^−/− and DKO mice experienced similar degrees of insulin deficiency in response to high-dose STZ administration (Fig. 4A), the insulinopenic DKO animals showed significant fasting hyperglycemia (241 ± 29 vs. 100 ± 6 mg/dL in Gcgr^−/−). The elevated fasting glucose in the DKO correlated to a 46% higher EGP rate compared with insulin-deficient Gcgr^−/− mice (Fig. 4B). EGP in the Gcgr-null animals was similar to normal mice (non-STZ, satellite animals; EGP = 12.8 ± 0.7 mg/kg/min). Basal glycolytic rates and FFA levels were similar in both groups (Fig. 4C and D). Although modeling suggests that the differences in basal EGP cannot account for all of the difference in fasting hyperglycemia (data not shown), these results confirm an insulin-independent and GLP-1R–dependent increase of EGP in the DKO mice compared with Gcgr^−/− animals after STZ-induced β-cell destruction.

Gcgr^−/− mice have reduced plasma glucose concentrations in the fed and fasted states as well as improved glucose tolerance compared with wild-type littermates (12,13). These data provide proof-of-concept that eliminating glucagon action improves glycemic control. The rodent loss-of-function data are consistent with seminal studies performed in humans showing inhibition of glucagon secretion by somatostatin infusion reduces plasma glucose (36). Similarly, models of experimental diabetes in which insulin deficiency is achieved by STZ demonstrated the pathogenic potential of unbridled glucagon secretion (37). These reports, together with more recent findings that Gcgr^−/− mice remain euglycemic in the state of insulin deficiency (16,22), emphasize the importance of glucagon and support the notion that suppression of glucagon action may be sufficient to reduce many of the debilitating symptoms of type 1 diabetes (38).

Intriguingly, however, the profound physiology of Gcgr^−/− mice has not been observed in other experimental models and has therefore inspired investigations exploring the unique consequences of Gcgr deficiency. This model has been used to study glucagon and α-cell homeostasis, especially in relation to glucose metabolism in the liver. For example, the discovery that selective deletion of hepatic Gcgr also results in animals with reduced blood glucose levels, improved glucose tolerance, hyperglucagonemia, and α-cell hyperplasia suggests that some control of α-cell function occurs via an unknown circulating factor (39). This concept is supported by experiments indicating α-cell area in wild-type islets increased when transplanted under the kidney capsule of Gcgr^−/− mice (39).

The goal of our studies using the Gcgr knockout model was to determine whether the GLP-1R plays a significant role in maintaining normal glycemia when these animals are made insulin deficient. Use of the Gcgr^−/−, Glp-1r^−/−, and Gcgr^−/−:Glp-1r^−/− genetic models here builds on previous work in Gcgr^−/− mice showing that administration of the GLP-1R peptide antagonist, exendin-4_(9-39), worsens glucose tolerance in these animals (19). Further, use of the high-affinity monoclonal GCGR antagonist antibody enabled complimentary studies that pharmacologically recapitulated the genetically driven loss of Gcgr in insulinopenic wild-type and Glp-1r^−/− animals. In total, the data presented in this report indicate that Gcgr ablation per se is not sufficient to confer protection from STZ-induced defects on glucose metabolism and that GLP-1R expression is needed for euglycemia in insulin-deficient Gcgr loss-of-function models.

Compared with reports characterizing single deletion of the Glp-1r on the C57BL/6 genetic background using various recombination approaches (24,40,41), our studies reveal the physiological consequences of Glp-1r ablation are more pronounced in an insulinopenic state when combined on a background of Gcgr deficiency or in a setting in which GCGR function is pharmacologically inhibited. In the short-term studies presented here, the STZ treatment paradigm caused wild-type and Glp-1r^−/− mice to reach blood glucose levels above 700 mg/dL within 1 week, preventing further experiments with these animals. Similar to previous studies (16), STZ treatment of Gcgr^−/− mice reduced the plasma and pancreatic insulin level due to near-maximal β-cell destruction, yet insulin deficiency had no effect on ambient glucose concentrations or glucose tolerance in the OGTT experiments. However, loss of Glp-1r in these animals resulted in significantly elevated glucose levels and impaired glucose tolerance. Importantly, comparison of GSIS during the OGTTs showed no differences in the single Gcgr^−/− mice versus DKO animals. These studies indicate that in the absence of insulin, a GLP-1R–mediated mechanism helps regulate glucose homeostasis in Gcgr^−/− mice.

Pharmacologically inhibiting the GCGR can result in effects resembling those of Gcgr ablation (17,18), although studies with such agents in insulinopenic models have not been reported. In the 6-week studies reported here, STZ-induced diabetic wild-type or Glp-1r^−/− mice were administered once-weekly doses of an anti-GCGR antibody to determine whether the GLP-1R helps regulate glycemia upon GCGR inhibition by a therapeutic agent. Strikingly, 1 week after the first dose of antibody, the diabetic wild-type mice exhibited complete normalization of ambient glucose levels, which continued for the duration of the treatment period; this is the first report of a GCGR antagonist ameliorating hyperglycemia in insulinopenic diabetes. STZ-treated Glp-1r^−/− mice showed some improvement but remained significantly hyperglycemic during the study period. After six treatment cycles, OGTT experiments demonstrated that the antibody-treated Glp-1r^−/− mice exhibited markedly impaired glucose clearance, similar to untreated, diabetic control animals. In contrast, previously diabetic wild-type animals administered the GCGR antibody showed glucose excursions comparable to nondiabetic wild-type mice. These results are consistent with our DKO experiments and further highlight the importance of the GLP-1R in glucose metabolism when the GCGR is inhibited.

Owing to nearly maximal β-cell loss in STZ-treated Gcgr^−/− mice, the glucoregulatory effects mediated by the GLP-1R that help maintain euglycemia in these animals likely occurs via a mechanism independent of enhancing GSIS. Experiments measuring EGP in STZ-treated Gcgr^−/− and DKO mice showed DKO animals have a higher rate of EGP compared with Gcgr^−/− mice (basal glycolytic rates were similar between the groups). These results agree with data showing GLP-1 infusion in fasted humans decreases EGP (42) and with studies in Glp-1r^−/− mice displaying impaired suppression of EGP and liver glycogen accumulation during hyperinsulinemic-euglycemic clamp experiments (32). On the basis of data from our experiments shown here, we propose the loss of GLP-1R–dependent EGP suppression in insulin-deficient DKO mice likely accounts for differences in glycemic control observed in the DKO animals versus the Gcgr single-knockout mice. Further work to determine the key GLP-1R–expressing tissue(s) that regulates whole body EGP is warranted. Other studies have demonstrated that activation of the GLP-1R in the central nervous system reduces hepatic glucose output (43). Whether GLP-1R–expressing vagal afferents from the enteric system initiate central nervous system–driven effects on EGP remains unclear. Conditional deletion of the Glp-1r in the peripheral nervous system and in regions such as the hypothalamic arcuate nucleus should be explored to better understand the suppressive role of the GLP-1R on EGP.

Our studies using the Gcgr^−/− model demonstrate a noninsulin-dependent role of the GLP-1R in controlling EGP. These results suggest that, in addition to the established ability of GLP-1 analogs to suppress glucagon secretion, GLP-1R activation in the state of insulin deficiency may improve overall glucose homeostasis. Early proof-of-concept trials show adjunctive treatment with the GLP-1 mimetics, liraglutide and exenatide, confers some efficacy in type 1 diabetic patients (44–46). Although it is unclear whether a high-affinity GCGR antagonist antibody will emerge as a therapeutic option, small-molecule GCGR antagonists that block glucagon action, but do not lead to profound compensatory hyperglucagonemia, may offer a viable path forward (47,48). Directly blocking glucagon action while further reducing EGP by GLP-1R activation may improve the metabolic symptoms of insulinopenia.

Duality of Interest. At the time of this work, all authors were employees of Eli Lilly and Company and may own company stock or possess stock options. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.S.J. designed the study, performed experiments and data analysis, and wrote the manuscript. R.L.M. performed data analysis and contributed to the writing of the manuscript. E.D.H., D.L.K., A.D.S., and M.E.C. designed the study, performed data analyses, and contributed to the writing of the manuscript. M.D.M. and K.W.S. designed the study, performed data analysis, and wrote the manuscript. K.W.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Portions of this study were presented during an oral presentation at the 74th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 13-17 June 2014.