Lucy S. Jun,1 Rohn L. Millican,2 Eric D. Hawkins,1 Debra L. Konkol,1 Aaron D. Showalter,1

Michael E. Christe,1 M. Dodson Michael,1 and Kyle W. Sloop1

Absence of Glucagon andInsulin Action Reveals a Role forthe GLP-1 Receptor inEndogenous GlucoseProductionDiabetes 2015;64:819–827 | DOI: 10.2337/db14-1052

The absence of insulin results in oscillating hyperglycemiaand ketoacidosis in type 1 diabetes. Remarkably, micegenetically deficient in the glucagon receptor (Gcgr) arerefractory to the pathophysiological symptoms of insulindeficiency, and therefore, studies interrogating thisunique model may uncover metabolic regulatory mecha-nisms that are independent of insulin. A significant featureof Gcgr-null mice is the high circulating concentrations ofGLP-1. Hence, the objective of this report was to investi-gate potential noninsulinotropic roles of GLP-1 in micewhere GCGR signaling is inactivated. For these studies,pancreatic b-cells were chemically destroyed by strepto-zotocin (STZ) in Gcgr2/2:Glp-1r2/2 mice and in Glp-1r2/2

animals that were subsequently treated with a high-affinityGCGR antagonist antibody that recapitulates the physio-logical state of Gcgr ablation. Loss of GLP-1 action sub-stantially worsened nonfasting glucose concentrationsand glucose tolerance in mice deficient in, and undergoingpharmacological inhibition of, the GCGR. Further, lack ofthe Glp-1r in STZ-treated Gcgr2/2 mice elevated rates ofendogenous glucose production, likely accounting for thedifferences in glucose homeostasis. These results supportthe emerging hypothesis that non–b-cell actions of GLP-1analogs may improve metabolic control in patients withinsulinopenic diabetes.

Type 1 diabetes is a life-threatening syndrome resultingfrom unbridled hepatic glycogenolysis, gluconeogenesis,

and ketogenesis that perpetuate a state of chronic fuelutilization (1). Manifestation of this condition occurs due toloss of the anabolic effects of insulin on glucose and lipidmetabolism (1). A significant contributor to the dysregulatedphysiological state is the absence of the inhibitory effectinsulin imposes on glucagon secretion (2–7). The effect ofexcess glucagon is significant because of the broad stimu-latory actions glucagon has on catabolic processes in the liverand the hyperglucagonemia that is present in poorly con-trolled diabetes (8,9). Thus, due to its role in exacerbatingthe metabolic consequences of insulinopenia, glucagon wasproposed several years ago to be an essential component inthe pathogenesis of type 1 diabetes (10,11). Therefore, toimprove glucose homeostasis in the absence of insulin, ther-apies that suppress glucagon secretion or block its actionmay improve overall glycemia by reducing the effect of a pri-mary stimulus of endogenous glucose production (EGP).

Bolstering the now long-standing hypothesis thatsuppressing glucagon action would have antihyperglyce-mic effects are studies revealing the remarkable metabolicphenotype of glucagon receptor (Gcgr) knockout mice.Although these animals have extraordinarily high circulat-ing concentrations of glucagon and pancreatic a-cell hy-perplasia, Gcgr-deficient mice display lower blood glucoselevels, improved glucose tolerance, and have reduced ad-iposity (12,13). Gcgr2/2 animals also have slower gastricemptying and are resistant to diet-induced obesity(13,14). Compelling evidence exists supporting the

1Endocrine Discovery, Lilly Research Laboratories, Eli Lilly and Company,Indianapolis, IN2BioTechnology Discovery Research, Lilly Research Laboratories, Eli Lillyand Company, Indianapolis, IN

Corresponding author: Kyle W. Sloop, sloop_kyle_w@lilly.com.

Received 7 July 2014 and accepted 1 October 2014.

© 2015 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, andthe work is not altered.

See accompanying article, p. 715.

Diabetes Volume 64, March 2015 819

METABOLISM

concept that glucagon inhibition would improve glycemiccontrol in the state of insulin deficiency.

Studies have demonstrated that Gcgr2/2 mice admin-istered the diabetogenic chemical agent streptozotocin(STZ), a cytotoxic glucose analog that induces necroticdestruction of pancreatic b-cells (15), are resistant toSTZ-induced hyperglycemia (14). This discovery is provoc-ative and has many potential implications; however, theSTZ treatment failed to abolish plasma insulin concentra-tions and staining for insulin-positive cells in the pan-creata of STZ-treated Gcgr2/2 animals (14). To causemaximal b-cell loss, Unger and colleagues (16) establisheda high-dose STZ treatment regimen that resulted in nearlycomplete b-cell destruction in Gcgr2/2 mice. Importantly,these studies show the metabolic maladies that resultfrom insulinopenia, especially profound hyperglycemiaand hyperketonemia, do not develop in animals lackingthe GCGR (16). Together, data from these experimentssupport a substantial role of glucagon signaling in themetabolic phenotype of insulin deficiency.

Although experiments using the high-dose STZ pro-tocol to induce b-cell destruction in Gcgr-null mice werewell designed, protection from insulinopenic hyperglyce-mia in this model may result from a combination of fac-tors and not solely as a consequence of the direct loss ofGCGR function. The initial characterization of Gcgr2/2

mice demonstrated that these animals have high circulat-ing concentrations of the antidiabetogenic peptide GLP-1,likely generated by proteolytic processing of the high lev-els of proglucagon within the hyperplastic a-cells (12).Consistent with this compensatory a-cell hyperplasia inGcgr2/2 animals, concomitant elevation of plasma GLP-1also occurs with progressive hyperglucagonemia in studieswhere GCGR antagonism is pharmacologically achieved(17,18). Omar et al. (19) recently provided data support-ing a contributing role for GLP-1 in glucose homeostasisof STZ-treated Gcgr2/2 mice by showing animals admin-istered the GLP-1 receptor (GLP-1R) peptide antagonist,exendin-4(9-39), display poorer glucose tolerance comparedwith animals that had not received the antagonist. Thisreport also indicated circulating levels of fibroblast growthfactor 21 (FGF21) are increased in Gcgr2/2 mice, a findingparadoxical to other studies showing activation of theGCGR increases hepatic and circulating FGF21 levels(20,21). The combination of exendin-4(9-39) and FGF21antisera worsened the glucose excursion for animals un-dergoing an oral glucose tolerance test (OGTT) (19).These experiments exemplify the potential importanceof fundamentally understanding the mechanisms wherebyGcgr2/2 mice are refractory to insulin deficiency and mayenable new proposals of alternate treatment regimens fortype 1 diabetes.

The significance of glucagon action during insulinopeniais further demonstrated in elegant studies showing thatreplenishment of hepatic GCGR using an adenovirussystem causes severe hyperglycemia in euglycemic,insulin-deficient Gcgr2/2 mice and that once expression

of GCGR wanes, euglycemia reappears (22). Althoughthese studies support the hypothesis of glucagon-inducedhyperglycemia as the main driver of diabetes, it remainsclear that there are multiple metabolic responses to Gcgrdeficiency, some of which may contribute to amelioratinghyperglycemia in this model. Therefore, this unique phe-notype should be fully interrogated to investigate theimportance of factors that appear dysregulated as a resultof the loss of Gcgr. Owing to the role of GLP-1 in glucosemetabolism (23) and its high circulating levels in Gcgr2/2

mice (12) as well as in animals administered GCGR antag-onists (17,18), we investigated the physiological conse-quences of Glp-1r ablation in insulin-deficient Gcgr2/2

mice and in insulinopenic wild-type and Glp-1r2/2 animalstreated with a high-affinity GCGR antagonist antibody.Results from these studies suggest that interventions toimprove glucose control in the absence of insulin may ben-efit greatly from adjunctive therapies that block glucagonaction and activate GLP-1R signaling.

RESEARCH DESIGN AND METHODS

AnimalsAll mouse models were generated by Eli Lilly andCompany in collaboration with Taconic (Hudson, NY).Glp-1r2/2 mice were previously described (24) and bredwith Gcgr+/2 mice licensed from Deltagen (San Mateo,CA) to produce double heterozygous Gcgr+/2:Glp-1r+/2

breeders that produced littermates with the followinggenotypes: Gcgr+/+:Glp-1r+/+ (wild-type); Gcgr+/+:Glp-1r2/2

(Glp-1r2/2); Gcgr2/2:Glp-1r+/+ (Gcgr2/2); and Gcgr2/2:Glp-1r2/2 double knockout (DKO) mice. PCR genotypingfor the Gcgr allele was performed with the following prim-ers (59-39): wild-type allele –T648S GTTGAGGAAACAGTAGAGAACAGCC and T648W ACCCTCATCCCTCTGCTGGGGGTCC; mutant allele –T648S and NEO3195 GGGCCAGCTCATTCCTCCCACTCAT. Genotyping for the Glp-1rallele has been described (24).

All mice were singly housed and fed a standard chowdiet (Teklad 2014; Harlan Laboratories, Indianapolis, IN)with bottled water ad libitum in a 12 h light/12 h darkcycle (lights off at 1800). Animals were maintained inaccordance with the Eli Lilly and Company InstitutionalAnimal Care and Use Committee and the NationalInstitutes of Health Guide for the Care and Use of Labora-tory Animals.

GCGR Models: Genetic Deletion and PharmacologicalBlockadeAnimals in the Gcgr ablation models were fasted for 4 h inthe morning and injected intraperitoneally with 130 mg/kgSTZ (Sigma-Aldrich, St. Louis, MO). This dose caused se-vere hyperglycemia in wild-type and Glp-1r2/2 mice,whereas the Gcgr2/2 and DKO animals remained healthy.Because the pharmacological GCGR blockade cohort ofwild-type and Glp-12/2 mice was a chronic study,a three-dose paradigm of lower STZ was used with aonce-weekly STZ dose of 110–115 mg/kg (or 10 mL/kg

820 GLP-1R Regulates EGP in STZ-Gcgr–Null Mice Diabetes Volume 64, March 2015

citrate vehicle) over 3 weeks. Diabetic animals (blood glu-cose $400 mg/dL) were randomized into groups forweekly anti-GCGR human IgG4 antibody (Ab-4) (25) orIgG4 isotype control subcutaneous injections of 10 mg/kg.The treatment dose was selected based on previous stud-ies in hyperglycemic ZDF rats (25). The concentration ofAb-4 (molecular weight = 143,965 g/mol) was determinedin solution by measuring ultraviolet light absorbance (280nm) and calculated using the molar extinction coefficientof the antibody amino acid sequence. Antibody activitywas measured by homogeneous time-resolved fluores-cence cAMP assays (Cisbio Assays, Bedford, MA) usingHEK293 cells expressing the mouse GCGR or GLP-1R.These assays demonstrated the ability of Ab-4 to blockglucagon action.

In Vivo Physiology and ImmunohistochemistryOGTTs were performed in mice fasted overnight for 16 hand orally gavaged with 2 g/kg glucose (24). For pancre-atic hormone extraction, the entire pancreas was removedand homogenized in acid ethanol. After centrifugation,supernatants were analyzed for hormone content. Plasma,serum, and pancreatic insulin, glucagon, and GLP-1 weremeasured using electrochemiluminescence assays (MesoScale Discovery, Rockville, MD). For pancreas immunohis-tochemistry, the whole pancreas was removed and fixed in10% neutral buffered formalin, followed by paraffin em-bedding. Pancreata were sectioned at 5 mm, and slideswere serially stained for insulin and glucagon with hema-toxylin counterstain. Gastric emptying tests were per-formed using mice that were fasted for 16 h and thenadministered a semiliquid diet containing acetaminophenvia oral gavage. The animals were then bled at 0, 30, 60,and 120 min via tail vein, and plasma acetaminophenconcentrations were measured using mass spectrometry.

Measurement of EGPOne week before EGP measurements, Gcgr2/2 and DKOmice were treated with 130 mg/kg STZ. Four days beforeEGP was assessed, catheters were placed in the left carotidartery and advanced to the aortic arch and in the rightjugular vein and advanced to the right atrium. Animalswere fasted overnight and allowed to acclimate to studycages for 2 h. All studies lasted 2 h. A bolus/continuousinfusion of 3-3H-glucose (6 mCi bolus and 0.1 mCi/min;PerkinElmer, Waltham, MA) was initiated and maintainedthroughout the test period. Arterial blood was obtainedevery 15 min to determine glucose concentrations. Bloodwas collected at the beginning and end of the 3-3H-

glucose infusion to monitor hematocrit and plasma insu-lin and to determine basal EGP.

Statistical AnalysesGraphing and statistical analyses were performed usingGraphPad Prism software. Data are presented as mean 6SEM and were compared using ANOVA, followed by theDunnett test. Repeated-measures ANOVA was used toassess significance between time courses. The statisticalsignificance threshold was set at P # 0.05.

RESULTS

The Glp-1r Is Required for Euglycemia in InsulinopenicGcgr2/2 MicePrevious studies show that Gcgr2/2 mice are resistant toSTZ-induced hyperglycemia (14,17). Blockade of GCGR,whether by genetic deletion or pharmacological antago-nism, leads to expansion of pancreatic a-cells due to theloss of an uncharacterized glucagon-mediated negativefeedback mechanism. As a result, plasma concentrationsof a-cell–derived glucagon and GLP-1 rise dramatically(Table 1) (17,26). Because GLP-1R activation helps regu-late glucose homeostasis (26–29), we investigatedwhether GLP-1R expression plays a role in maintainingeuglycemia in insulin-deficient Gcgr2/2 mice. For thesestudies, Glp-1r2/2 mice were crossed with Gcgr+/2 miceto produce double heterozygous breeders that enabled thegeneration of animals lacking both genes. To induce in-sulin deficiency, wild-type, Glp-1r2/2, Gcgr2/2, and DKOanimals were administered high-dose STZ (130 mg/kg)and evaluated over the next 7 days. Nonfasted glucoseconcentrations became pathologically high in wild-typeand Glp-1r2/2 mice, and therefore, these animals wereremoved from the study (Fig. 1A). As expected, Gcgr2/2

mice displayed unchanged glucose concentrations afterSTZ administration (Fig. 1A), even with a dramatic de-crease in pancreatic insulin content (Fig. 1D). Althoughglucose concentrations in DKO mice after STZ treatmentwere lower than in wild-type and Glp-1r2/2 animals, thesemice showed significantly higher glucose concentrationscompared with Gcgr2/2 mice, indicating the GLP-1R isneeded for post-STZ euglycemia in the Gcgr2/2 animal(Fig. 1A).

Before STZ treatment, Gcgr2/2 and DKO mice dis-played similar oral glucose tolerance and glucose-stimulated insulin secretion (GSIS) during OGTTs(Fig. 1B). Lack of the Glp-1r did not affect oral glucosetolerance in nondiabetic DKO animals, likely due to the

Table 1—Plasma concentrations of GLP-1 and glucagon in wild-type and loss-of-function models

Wild-type Glp-1r2/2 Gcgr2/2 DKO

Total GLP-1 (pg/mL) 32.2 6 4.8 30.9 6 7.7 95.9 6 13.1* 144.6 6 5.8*

Glucagon (pg/mL) 62.1 6 35.0 (6/8 ,LLOQ†) 50.4 6 10.4 (6/8 ,LLOQ†) 11,604 6 1,947 3,451 6 513

*P# 0.05 one-way ANOVA with Tukey post hoc test. †Six of eight animals had glucagon levels lower than the lower limit of quantitation(LLOQ) of 14 pg/mL; only the measurable values are reported. Statistical analysis was not performed because the data set wasincomplete.

diabetes.diabetesjournals.org Jun and Associates 821

compensatory upregulation of other incretin pathwaysupon Glp-1r ablation (30). In an OGTT 1 week after STZadministration, Gcgr2/2 mice continued to show rapidglucose clearance; however, STZ-treated DKO animals dis-played marked glucose intolerance (area under the curve[AUC] = 1,571 mg/dL/min) compared with Gcgr2/2 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 blockadeof the GLP-1R using exendin-4(9-39) worsens glucose tol-erance in STZ-treated Gcgr-null mice (19). An assessmentof gastric emptying between the two STZ-treated geno-types failed to show any difference (plasma acetamino-phen AUC(0–2 h): Gcgr

2/2 23.0 6 1.7, and DKO 25.4 61.8 mg $ h/mL). Importantly, there was no GSIS in theGcgr2/2 or DKO animals, further indicating both groupsbecame insulin-deficient (Fig. 1C).

To show near-maximum b-cell loss occurred in bothgenotypes, immunohistochemistry staining for insulinand glucagon was performed on pancreata from nondia-betic mice and 1-week after STZ administration (Fig. 1D).As previously shown, Gcgr ablation caused marked a-cellhyperplasia. STZ-treated animals exhibited profoundb-cell loss after STZ treatment, regardless of genotype(Fig. 1D). Body composition was also measured to

investigate evidence of differential insulin-mediated lipo-genesis between the groups, and there were no differencesin adipose or lean mass between Gcgr2/2 and DKO ani-mals before or after STZ administration (data not shown).Furthermore, plasma concentrations of FGF21 were notdifferent between these groups (data not shown). Finally,pancreatic insulin, GLP-1 (total), and glucagon levels weresimilar in Gcgr2/2 and DKO animals after STZ treatment(Table 2).

GCGR Antagonist-Mediated Glucose Normalization inSTZ-Induced Diabetes Requires the Glp-1rThe genetic results provide compelling evidence thata functional GLP-1R in Gcgr-null mice is needed to protectagainst hyperglycemia in the insulin-deficient state. Tofurther evaluate the mechanism by which blockade ofglucagon action improves glucose control, a high-affinityGCGR monoclonal antibody antagonist, Ab-4 (25), wasused to investigate whether therapeutic interventionwould reverse diabetes. Ab-4 shows potent antagonismof glucagon-induced cAMP accumulation in HEK293 cellsexpressing the mouse GCGR (Kb = 0.76 nmol/L) (Fig. 2A);these data are consistent with previously reported resultsfor Ab-4 in mouse GCGR-containing membrane-bindingassays (Ki = 0.83 nmol/L) (25). Ab-4 showed no antago-nism of the mouse GLP-1R (Fig. 2B). For the in vivo

Figure 1—GLP-1R is required for post-STZ euglycemia in Gcgr2/2 mice. A: Nonfasted blood glucose concentrations pre-STZ and 7 daysafter STZ administration (n = 9–10 per group). WT, wild-type. ****P # 0.0001 by two-way ANOVA, Gcgr2/2 vs. DKO 7 days post-STZ. B:Blood glucose concentrations during an OGTT (n = 6–10 per group). *P # 0.05 by two-way ANOVA, DKO pre-STZ vs. DKO 7 days post-STZ. C: GSIS during an OGTT (dotted line indicates assay lower limit of quantitation). D: Pancreata harvested from Gcgr2/2 and DKO micebefore and 7 days after STZ administration (blue, glucagon; brown, insulin; original magnification 320).

822 GLP-1R Regulates EGP in STZ-Gcgr–Null Mice Diabetes Volume 64, March 2015

studies, STZ-induced diabetes was established in wild-type and Glp-1r2/2 mice, and animals were administeredAb-4 once weekly for 7 weeks. To avoid the lethal hyper-glycemia observed for these genotypes (Fig. 1A), the STZ

treatment paradigm used here resulted in more restrainedhyperglycemia, thereby allowing the vehicle/control IgG4-treated mice to survive the study period. Using this ap-proach, in addition to investigating the therapeuticpotential of Ab-4 to improve hyperglycemia in diabeticwild-type animals, we pharmacologically recapitulatedGcgr ablation in a Glp-1r–null background. STZ-treatedwild-type and Glp-1r2/2 animals were randomized accord-ing to their nonfasted blood glucose concentrations intothree treatment groups: wild-type (STZ+IgG4), wild-type(STZ+Ab-4), and Glp-1r2/2 (STZ+Ab-4). Also included wasa nondiabetic wild-type control group that received onlycitrate injection initially, followed by weekly administra-tion of the isotype control IgG4 (Fig. 3A). Hyperglycemiawas quickly established in STZ-treated animals. Un-treated, diabetic wild-type (STZ+IgG4) mice remainedhyperglycemic throughout the study (Fig. 3A), whereasAb-4–treated insulinopenic wild-type (STZ+Ab-4) animalsshowed normalization of nonfasted blood glucose concen-trations after the first antibody dose and remained similarto nondiabetic wild-type controls (Fig. 3A). The Glp-1r2/2

(STZ+Ab-4) group, with pharmacologically blocked GCGRand genetically ablated Glp-1r, experienced markedglucose lowering compared with the wild-type animals(STZ+IgG4); however, their blood glucose levels remainedsignificantly 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 thatof nondiabetic wild-type controls (AUC = 1,096 mg/dL/min)(Fig. 3B). However, despite improvement in nonfastedblood glucose levels in Glp-1r2/2 (STZ+Ab-4) mice aftersix treatments (Fig. 3A), these animals showed profoundglucose intolerance (AUC = 3,180 mg/dL/min) similar tothe 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 comparedwith nondiabetic controls, similar to the diabetic wild-type and hyperglycemic Ab-4–treated Glp-1r2/2 mice(Fig. 3C). It appears that 6 weeks of Ab-4–mediatedGCGR inactivation had little effect on GSIS from anyresidual b-cells (Fig. 3C). These results are supported bythe fact that pancreatic insulin content was less than 10%of nondiabetic controls in all of the STZ-treated groups(Table 2).

Table 2—Pancreatic hormone levels in genetic and pharmacological Gcgr loss-of-function models

STZ-treated Wild-type Glp-1r2/2

Gcgr2/2 DKO Citrate+IgG4 STZ+IgG4 STZ+Ab-4 STZ+Ab-4

Insulin (ng)† 347 6 88* 278 6 37* 1,954 6 737 30 6 8* 169 6 28* 58 6 18*

Total GLP-1 (pg)† 15,155 6 895* 18,284 6 8,585* 332 6 62 333 6 64 2,224 6 215 4,879 6 583

Glucagon (pg)† 259,516 6 21,933* 371,341 6 12,702* 5,639 6 1,116 5,180 6 1,549 200,375 6 34,648* 243,613 6 28,626*

†Values are shown as the amount per grams of pancreas/grams of body weight. *P # 0.05 by one-way ANOVA with Tukey post hoctest.

Figure 2—The GCGR monoclonal antibody Ab-4 blocks glucagonactivity at the mouse GCGR. A: Glucagon-induced cAMP accumu-lation in HEK293 cells expressing the mouse GGCR is inhibited byAb-4 (Kb = 0.76 nmol/L). B: Ab-4 does not inhibit GLP-1-inducedcAMP accumulation in HEK293 cells expressing the mouse GLP-1R.

diabetes.diabetesjournals.org Jun and Associates 823

As with Gcgr2/2 mice, circulating GLP-1 concentrationswere elevated in both Ab-4–treated groups compared withwild-type (citrate+IgG4) and wild-type (STZ+IgG4) animals(Fig. 3D). Fasting insulin levels in all STZ-treated animalswere similarly depleted regardless of antibody treatment,indicating b-cell recovery was unlikely to be occurringdespite the high levels of GLP-1 (Fig. 3C). Finally, pancre-atic GLP-1 and glucagon content were elevated in theAb-4–treated groups compared with nondiabetic anddiabetic wild-type controls (Table 2). Therefore, the ele-vated ambient glucose concentrations and impaired oralglucose tolerance in Glp-1r2/2 (STZ+Ab-4) compared withthe wild-type (STZ+Ab-4) animals cannot be explained bydifferences in insulin, glucagon, or GLP-1 levels. BecauseGLP-1R action inhibits gastric transit and food intake,we performed semiliquid gastric emptying tests in allgroups and found no differences (plasma acetaminophenAUC[0–2 h]: wild-type [citrate+IgG4] 18.5 6 1.7; wild-type[STZ+IgG4] 26.0 6 2.8; wild-type [STZ+Ab-4] 25.0 6 2.0;Glp-1r2/2 [STZ+Ab-4] 21.0 6 5.6 mg $ h/mL). Food in-take in the normoglycemic groups was not different; butas expected, the two hyperglycemic groups ate more,which is consistent with a catabolic state (average foodintake: wild-type [citrate+IgG4] 126.5 6 6.3; wild-type [STZ+IgG4] 263.6 6 28.6; wild-type [STZ+Ab-4]142.6 6 9.4; Glp-1r2/2 [STZ+Ab-4] 225.8 6 25.3 mgfood/g body weight/day).

Glp-1r Ablation Increases EGP in InsulinopenicGcgr-Null MiceAlthough there is little meaningful evidence indicatingexpression of the GLP-1R in the liver (24,31), there areseveral reports of GLP-1–mediated regulation of hepaticglucose production in a manner separate from its abilityto suppress glucagon release and, intriguingly, indepen-dent of insulin (32–35). We therefore measured EGP ratesto investigate whether a GLP-1R–dependent mechanismaccounts for regulating glycemia in Gcgr2/2 animals. Al-though the Gcgr2/2 and DKO mice experienced similardegrees of insulin deficiency in response to high-doseSTZ administration (Fig. 4A), the insulinopenic DKO ani-mals showed significant fasting hyperglycemia (241 6 29vs. 100 6 6 mg/dL in Gcgr2/2). The elevated fastingglucose in the DKO correlated to a 46% higher EGP ratecompared with insulin-deficient Gcgr2/2 mice (Fig. 4B).EGP in the Gcgr-null animals was similar to normal mice(non-STZ, satellite animals; EGP = 12.86 0.7 mg/kg/min).Basal glycolytic rates and FFA levels were similar in bothgroups (Fig. 4C and D). Although modeling suggests thatthe differences in basal EGP cannot account for all of thedifference in fasting hyperglycemia (data not shown),these results confirm an insulin-independent andGLP-1R–dependent increase of EGP in the DKO micecompared with Gcgr2/2 animals after STZ-induced b-celldestruction.

Figure 3—GCGR blockade improves glucose tolerance in diabetic wild-type (WT) but not Glp-1r2/2 mice. A: Weekly nonfasted bloodglucose levels over a 7-week treatment period of the GCGR antibody Ab-4 (n = 11–18 per group). *P # 0.001 by two-way ANOVA vs. WTcitrate+IgG4; #P # 0.05 by two-way ANOVA vs. WT STZ+Ab-4. B: OGTT performed on the sixth week of antibody treatment (n = 4–5 pergroup). *P # 0.05 for WT STZ+IgG4 vs. WT citrate+IgG4; #P # 0.05–0.0001for Glp-1r2/2+STZ+Ab-4 vs. WT citrate+IgG4. C: GSIS duringan OGTT performed on the sixth week of antibody treatment (n = 4–5 per group). *P # 0.0001 for WT citrate+IgG4 vs. STZ-treated groups;#P # 0.01 for WT citrate+IgG4 vs. WT STZ+IgG4 and Glp-1r2/2 STZ+Ab-4. D: Plasma total GLP-1 and glucagon levels in response toGCGR blockade at the end of the 7-week treatment period (n = 4–9 per group). **P # 0.001; ***P # 0.001; ****P # 0.0001 by two-wayANOVA.

824 GLP-1R Regulates EGP in STZ-Gcgr–Null Mice Diabetes Volume 64, March 2015

DISCUSSION

Gcgr2/2 mice have reduced plasma glucose concentrationsin the fed and fasted states as well as improved glucosetolerance compared with wild-type littermates (12,13).These data provide proof-of-concept that eliminatingglucagon action improves glycemic control. The rodentloss-of-function data are consistent with seminal studiesperformed in humans showing inhibition of glucagon se-cretion by somatostatin infusion reduces plasma glucose(36). Similarly, models of experimental diabetes in whichinsulin deficiency is achieved by STZ demonstrated thepathogenic potential of unbridled glucagon secretion (37).These reports, together with more recent findings thatGcgr2/2 mice remain euglycemic in the state of insulindeficiency (16,22), emphasize the importance of glucagonand support the notion that suppression of glucagon ac-tion may be sufficient to reduce many of the debilitatingsymptoms of type 1 diabetes (38).

Intriguingly, however, the profound physiology ofGcgr2/2 mice has not been observed in other experimen-tal models and has therefore inspired investigationsexploring the unique consequences of Gcgr deficiency.This model has been used to study glucagon and a-cellhomeostasis, especially in relation to glucose metabolismin the liver. For example, the discovery that selectivedeletion of hepatic Gcgr also results in animals with re-duced blood glucose levels, improved glucose tolerance,hyperglucagonemia, and a-cell hyperplasia suggests thatsome control of a-cell function occurs via an unknowncirculating factor (39). This concept is supported by

experiments indicating a-cell area in wild-type islets in-creased when transplanted under the kidney capsule ofGcgr2/2 mice (39).

The goal of our studies using the Gcgr knockout modelwas to determine whether the GLP-1R plays a significantrole in maintaining normal glycemia when these animalsare made insulin deficient. Use of the Gcgr2/2, Glp-1r2/2,and Gcgr2/2:Glp-1r2/2 genetic models here builds on pre-vious work in Gcgr2/2 mice showing that administrationof the GLP-1R peptide antagonist, exendin-4(9-39), wors-ens glucose tolerance in these animals (19). Further, useof the high-affinity monoclonal GCGR antagonist anti-body enabled complimentary studies that pharmacologi-cally recapitulated the genetically driven loss of Gcgr ininsulinopenic wild-type and Glp-1r2/2 animals. In total,the data presented in this report indicate that Gcgr abla-tion per se is not sufficient to confer protection from STZ-induced defects on glucose metabolism and that GLP-1Rexpression is needed for euglycemia in insulin-deficientGcgr loss-of-function models.

Compared with reports characterizing single deletionof the Glp-1r on the C57BL/6 genetic background usingvarious recombination approaches (24,40,41), our studiesreveal the physiological consequences of Glp-1r ablationare more pronounced in an insulinopenic state when com-bined on a background of Gcgr deficiency or in a setting inwhich GCGR function is pharmacologically inhibited. Inthe short-term studies presented here, the STZ treatmentparadigm caused wild-type and Glp-1r2/2 mice to reachblood glucose levels above 700 mg/dL within 1 week,

Figure 4—DKO mice have increased EGP compared with Gcgr2/2 mice after STZ treatment. A: Fasted C-peptide levels in STZ-treatedGcgr2/2 and DKO mice before EGP measurement. B: EGP in STZ-treated Gcgr2/2 and DKO mice under noninsulin stimulated conditions(n = 6–7 per group). ****P # 0.0001 by two-tailed t test. C: Basal whole-body glycolytic rates. D: Plasma free fatty acids (FFA) levels in bothgroups during basal EGP measurements.

diabetes.diabetesjournals.org Jun and Associates 825

preventing further experiments with these animals. Sim-ilar to previous studies (16), STZ treatment of Gcgr2/2

mice reduced the plasma and pancreatic insulin leveldue to near-maximal b-cell destruction, yet insulin defi-ciency had no effect on ambient glucose concentrations orglucose tolerance in the OGTT experiments. However, lossof Glp-1r in these animals resulted in significantly ele-vated glucose levels and impaired glucose tolerance. Im-portantly, comparison of GSIS during the OGTTs showedno differences in the single Gcgr2/2 mice versus DKOanimals. These studies indicate that in the absence ofinsulin, a GLP-1R–mediated mechanism helps regulateglucose homeostasis in Gcgr2/2 mice.

Pharmacologically inhibiting the GCGR can result ineffects resembling those of Gcgr ablation (17,18), al-though studies with such agents in insulinopenic modelshave not been reported. In the 6-week studies reportedhere, STZ-induced diabetic wild-type or Glp-1r2/2 micewere administered once-weekly doses of an anti-GCGRantibody to determine whether the GLP-1R helps regulateglycemia upon GCGR inhibition by a therapeutic agent.Strikingly, 1 week after the first dose of antibody, thediabetic wild-type mice exhibited complete normalizationof ambient glucose levels, which continued for the dura-tion of the treatment period; this is the first report ofa GCGR antagonist ameliorating hyperglycemia in insuli-nopenic diabetes. STZ-treated Glp-1r2/2 mice showedsome improvement but remained significantly hypergly-cemic during the study period. After six treatment cycles,OGTT experiments demonstrated that the antibody-treated Glp-1r2/2 mice exhibited markedly impaired glucoseclearance, similar to untreated, diabetic control animals.In contrast, previously diabetic wild-type animals admin-istered the GCGR antibody showed glucose excursionscomparable to nondiabetic wild-type mice. These resultsare consistent with our DKO experiments and furtherhighlight the importance of the GLP-1R in glucose metab-olism when the GCGR is inhibited.

Owing to nearly maximal b-cell loss in STZ-treatedGcgr2/2 mice, the glucoregulatory effects mediated bythe GLP-1R that help maintain euglycemia in theseanimals likely occurs via a mechanism independent ofenhancing GSIS. Experiments measuring EGP in STZ-treated Gcgr2/2 and DKO mice showed DKO animalshave a higher rate of EGP compared with Gcgr2/2 mice(basal glycolytic rates were similar between the groups).These results agree with data showing GLP-1 infusion infasted humans decreases EGP (42) and with studies inGlp-1r2/2 mice displaying impaired suppression of EGPand liver glycogen accumulation during hyperinsulinemic-euglycemic clamp experiments (32). On the basis of datafrom our experiments shown here, we propose the loss ofGLP-1R–dependent EGP suppression in insulin-deficientDKO mice likely accounts for differences in glycemiccontrol observed in the DKO animals versus the Gcgr single-knockout mice. Further work to determine the keyGLP-1R–expressing tissue(s) that regulates whole body

EGP is warranted. Other studies have demonstrated thatactivation of the GLP-1R in the central nervous systemreduces hepatic glucose output (43). Whether GLP-1R–expressing vagal afferents from the enteric system initiatecentral nervous system–driven effects on EGP remainsunclear. Conditional deletion of the Glp-1r in the periph-eral nervous system and in regions such as the hypotha-lamic arcuate nucleus should be explored to betterunderstand the suppressive role of the GLP-1R on EGP.

Our studies using the Gcgr2/2 model demonstratea noninsulin-dependent role of the GLP-1R in controllingEGP. These results suggest that, in addition to the estab-lished ability of GLP-1 analogs to suppress glucagon se-cretion, GLP-1R activation in the state of insulin deficiencymay improve overall glucose homeostasis. Early proof-of-concept trials show adjunctive treatment with the GLP-1mimetics, liraglutide and exenatide, confers some efficacyin type 1 diabetic patients (44–46). Although it is unclearwhether a high-affinity GCGR antagonist antibody willemerge as a therapeutic option, small-molecule GCGRantagonists that block glucagon action, but do not leadto profound compensatory hyperglucagonemia, may offera viable path forward (47,48). Directly blocking glucagonaction while further reducing EGP by GLP-1R activationmay improve the metabolic symptoms of insulinopenia.

Duality of Interest. At the time of this work, all authors were employeesof 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 experi-ments and data analysis, and wrote the manuscript. R.L.M. performed dataanalysis 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 thewriting of the manuscript. M.D.M. and K.W.S. designed the study, performed dataanalysis, and wrote the manuscript. K.W.S. is the guarantor of this work and, assuch, had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis.Prior Presentation. Portions of this study were presented during an oralpresentation at the 74th Scientific Sessions of the American Diabetes Associa-tion, San Francisco, CA, 13-17 June 2014.

References1. Ennis ED, Kreisberg RA. Diabetic ketoacidosis and the hyperglycemic hy-perosmolar syndrome. In Diabetes Mellitus: a Fundamental and Clinical Text.LeRoith D, Taylor SI, Olefsky JM, Eds. Philadelphia, Lippincott Williams & Wilkins,2000, p. 336–3472. Maruyama H, Hisatomi A, Orci L, Grodsky GM, Unger RH. Insulin within isletsis a physiologic glucagon release inhibitor. J Clin Invest 1984;74:2296–22993. Greenbaum CJ, Havel PJ, Taborsky GJ Jr, Klaff LJ. Intra-islet insulin permitsglucose to directly suppress pancreatic A cell function. J Clin Invest 1991;88:767–7734. Hope KM, Tran PO, Zhou H, Oseid E, Leroy E, Robertson RP. Regulation ofalpha-cell function by the beta-cell in isolated human and rat islets deprived ofglucose: the “switch-off” hypothesis. Diabetes 2004;53:1488–14955. Banarer S, McGregor VP, Cryer PE. Intraislet hyperinsulinemia prevents theglucagon response to hypoglycemia despite an intact autonomic response. Di-abetes 2002;51:958–9656. Xu E, Kumar M, Zhang Y, et al. Intra-islet insulin suppresses glucagonrelease via GABA-GABAA receptor system. Cell Metab 2006;3:47–58

826 GLP-1R Regulates EGP in STZ-Gcgr–Null Mice Diabetes Volume 64, March 2015

7. Meier JJ, Kjems LL, Veldhuis JD, Lefèbvre P, Butler PC. Postprandial sup-pression of glucagon secretion depends on intact pulsatile insulin secretion: furtherevidence for the intraislet insulin hypothesis. Diabetes 2006;55:1051–10568. Raskin P, Unger RH. Hyperglucagonemia and its suppression—importancein the metabolic control of diabetes. N Engl J Med 1978;299:433–4369. Dinneen S, Alzaid A, Turk D, Rizza R. Failure of glucagon suppression con-tributes to postprandial hyperglycaemia in IDDM. Diabetologia 1995;38:337–34310. Unger RH. The Banting Memorial Lecture 1975. Diabetes and the alpha cell.Diabetes 1976;25:136–15111. Unger RH, Orci L. The essential role of glucagon in the pathogenesis ofdiabetes mellitus. Lancet 1975;1:14–1612. Gelling RW, Du XQ, Dichmann DS, et al. Lower blood glucose, hyper-glucagonemia, and pancreatic alpha cell hyperplasia in glucagon receptorknockout mice. Proc Natl Acad Sci U S A 2003;100:1438–144313. Parker JC, Andrews KM, Allen MR, Stock JL, McNeish JD. Glycemic controlin mice with targeted disruption of the glucagon receptor gene. Biochem BiophysRes Commun 2002;290:839–84314. Conarello SL, Jiang G, Mu J, et al. Glucagon receptor knockout mice areresistant to diet-induced obesity and streptozotocin-mediated beta cell loss andhyperglycaemia. Diabetologia 2007;50:142–15015. Eleazu CO, Eleazu KC, Chukwuma S, Essien UN. Review of the mechanismof cell death resulting from streptozotocin challenge in experimental animals, itspractical use and potential risk to humans. J Diabetes Metab Disord 2013;12:6016. Lee Y, Wang MY, Du XQ, Charron MJ, Unger RH. Glucagon receptor knockoutprevents insulin-deficient type 1 diabetes in mice. Diabetes 2011;60:391–39717. Sloop KW, Cao JX, Siesky AM, et al. Hepatic and glucagon-like peptide-1-mediated reversal of diabetes by glucagon receptor antisense oligonucleotideinhibitors. J Clin Invest 2004;113:1571–158118. Gu W, Yan H, Winters KA, et al. Long-term inhibition of the glucagon re-ceptor with a monoclonal antibody in mice causes sustained improvement inglycemic control, with reversible alpha-cell hyperplasia and hyperglucagonemia.J Pharmacol Exp Ther 2009;331:871–88119. Omar BA, Andersen B, Hald J, Raun K, Nishimura E, Ahrén B. Fibroblastgrowth factor 21 (FGF21) and glucagon-like peptide 1 contribute to diabetesresistance in glucagon receptor-deficient mice. Diabetes 2014;63:101–11020. Habegger KM, Stemmer K, Cheng C, et al. Fibroblast growth factor 21mediates specific glucagon actions. Diabetes 2013;62:1453–146321. Arafat AM, Kaczmarek P, Skrzypski M, et al. Glucagon increases circulatingfibroblast growth factor 21 independently of endogenous insulin levels: a novelmechanism of glucagon-stimulated lipolysis? Diabetologia 2013;56:588–59722. Lee Y, Berglund ED, Wang MY, et al. Metabolic manifestations of insulindeficiency do not occur without glucagon action. Proc Natl Acad Sci U S A 2012;109:14972–1497623. Campbell JE, Drucker DJ. Pharmacology, physiology, and mechanisms ofincretin hormone action. Cell Metab 2013;17:819–83724. Jun LS, Showalter AD, Ali N, et al. A novel humanized GLP-1 receptor modelenables both affinity purification and Cre-LoxP deletion of the receptor. PLoS ONE2014;9:e9374625. Korytko AI, Millican RL. Glucagon receptor antagonists. U.S. patentUS7968686 B2. Eli Lilly and Company, 28 June 201126. Perfetti R, Hui H. The role of GLP-1 in the life and death of pancreatic betacells. Horm Metab Res 2004;36:804–81027. Doyle ME, Egan JM. Mechanisms of action of glucagon-like peptide 1 in thepancreas. Pharmacol Ther 2007;113:546–59328. Li Y, Hansotia T, Yusta B, Ris F, Halban PA, Drucker DJ. Glucagon-likepeptide-1 receptor signaling modulates beta cell apoptosis. J Biol Chem 2003;278:471–47829. Rondas D, Bugliani M, D’Hertog W, et al. Glucagon-like peptide-1 protectshuman islets against cytokine-mediated b-cell dysfunction and death: a proteo-mic study of the pathways involved. J Proteome Res 2013;12:4193–4206

30. Pederson RA, Satkunarajah M, McIntosh CH, et al. Enhanced glucose-dependent insulinotropic polypeptide secretion and insulinotropic action inglucagon-like peptide 1 receptor -/- mice. Diabetes 1998;47:1046–105231. Richards P, Parker HE, Adriaenssens AE, et al. Identification and charac-terization of GLP-1 receptor-expressing cells using a new transgenic mousemodel. Diabetes 2014;63:1224–123332. Ayala JE, Bracy DP, James FD, Julien BM, Wasserman DH, Drucker DJ. Theglucagon-like peptide-1 receptor regulates endogenous glucose production andmuscle glucose uptake independent of its incretin action. Endocrinology 2009;150:1155–116433. D’Alessio D, Vahl T, Prigeon R. Effects of glucagon-like peptide 1 on thehepatic glucose metabolism. Horm Metab Res 2004;36:837–84134. Dardevet D, Moore MC, DiCostanzo CA, et al. Insulin secretion-independenteffects of GLP-1 on canine liver glucose metabolism do not involve portal veinGLP-1 receptors. Am J Physiol Gastrointest Liver Physiol 2005;289:G806–G81435. Dhanesha N, Joharapurkar A, Shah G, et al. Exendin-4 reduces glycemia byincreasing liver glucokinase activity: an insulin independent effect. PharmacolRep 2012;64:140–14936. Gerich JE, Lorenzi M, Bier DM, et al. Prevention of human diabetic ke-toacidosis by somatostatin. Evidence for an essential role of glucagon. N Engl JMed 1975;292:985–98937. Cherrington AD, Lacy WW, Chiasson JL. Effect of glucagon on glucoseproduction during insulin deficiency in the dog. J Clin Invest 1978;62:664–67738. Unger RH, Cherrington AD. Glucagonocentric restructuring of diabetes:a pathophysiologic and therapeutic makeover. J Clin Invest 2012;122:4–1239. Longuet C, Robledo AM, Dean ED, et al. Liver-specific disruption of themurine glucagon receptor produces a-cell hyperplasia: evidence for a circulatinga-cell growth factor. Diabetes 2013;62:1196–120540. Wilson-Pérez HE, Chambers AP, Ryan KK, et al. Vertical sleeve gastrectomyis effective in two genetic mouse models of glucagon-like Peptide 1 receptordeficiency. Diabetes 2013;62:2380–238541. Hansotia T, Baggio LL, Delmeire D, et al. Double incretin receptorknockout (DIRKO) mice reveal an essential role for the enteroinsular axis intransducing the glucoregulatory actions of DPP-IV inhibitors. Diabetes 2004;53:1326–133542. Prigeon RL, Quddusi S, Paty B, D’Alessio DA. Suppression of glucoseproduction by GLP-1 independent of islet hormones: a novel extrapancreaticeffect. Am J Physiol Endocrinol Metab 2003;285:E701–E70743. Burmeister MA, Ferre T, Ayala JE, King EM, Holt RM, Ayala JE. Acute ac-tivation of central GLP-1 receptors enhances hepatic insulin action and insulinsecretion in high-fat-fed, insulin resistant mice. Am J Physiol Endocrinol Metab2012;302:E334–E34344. Kielgast U, Krarup T, Holst JJ, Madsbad S. Four weeks of treatment withliraglutide reduces insulin dose without loss of glycemic control in type 1 diabeticpatients with and without residual beta-cell function. Diabetes Care 2011;34:1463–146845. Varanasi A, Bellini N, Rawal D, et al. Liraglutide as additional treatment fortype 1 diabetes. Eur J Endocrinol 2011;165:77–8446. Ghazi T, Rink L, Sherr JL, Herold KC. Acute metabolic effects of exenatide inpatients with type 1 diabetes with and without residual insulin to oral and in-travenous glucose challenges. Diabetes Care 2014;37:210–21647. Engel SS, Xu L, Andryuk PJ, et al. Efficacy and tolerability of MK-0893,a glucagon receptor antagonist (GRA), in patients with type 2 diabetes (T2DM)[abstract 0309-OR]. Presented at the 71st Scientific Sessions of the AmericanDiabetes Association, 24–28 June 2011, San Diego, CA48. Kazda C, Headlee S, Ding Y, et al. The glucagon receptor antagonistLY2409021 significantly lowers HbA1c and is well tolerated in patients withT2DM—a 24-week phase 2 study [abstract 112-OR/0112]. Presented at the73rd Scientific Sessions of the American Diabetes Association, 21-25 June 2013,Chicago, IL

diabetes.diabetesjournals.org Jun and Associates 827