Fibroblast Growth Factor 21 (FGF21) and Glucagon-Like Peptide 1 Contribute to Diabetes Resistance in Glucagon Receptor–Deficient Mice

Glucagon is a 29-amino acid peptide hormone that is secreted by the α-cells in the pancreas and acts as a counter-regulatory hormone to insulin, such that glucagon is secreted during hypoglycemia, in contrast to insulin, which is secreted in response to hyperglycemia. Glucagon receptors are highly expressed on hepatocytes and stimulate hepatic glucose production in order to maintain euglycemia. Glucagon plays a role in the hyperglycemia observed in diabetes, as a deficiency of insulin leads to inadequate suppression of glucagon secretion from the α-cells (1). The excess glucagon secretion increases hepatic glucose production, exacerbating the existing hyperglycemia.

Glucagon receptor–deficient (Gcgr^−/−) mice have multiple phenotypic characteristics of interest in the study of diabetes. These mice display lower blood glucose levels throughout the day and have improved glucose tolerance (2). The Gcgr^−/− mice also have reduced adiposity, LDL cholesterol, and leptin levels, but normal body weight, food intake, and energy expenditure (2). Enlargement of the pancreas and postnatal hyperplasia of islets is observed, predominantly due to hyperplasia of α-cells up to 12-fold greater than Gcgr^+/+ mice, possibly as a result of a liver-derived circulating α-cell growth factor (3). An increase in circulating levels of glucagon-like peptide 1 (GLP-1) is also observed, most likely derived from the processing of elevated α-cell–produced proglucagon (2). The Gcgr^−/− mice have also been shown to be resistant to diet-induced obesity (4). Gcgr^−/− mice were also shown to be resistant to the development of insulin-deficient diabetes, as high doses of the diabetogen streptozotocin (STZ), which nearly completely destroyed all β-cells, did not cause hyperglycemia (4,5). The suggested mechanism behind the antidiabetic effect is lack of glucagon action (6), but the finding that the Gcgr^−/− mice not only do not develop diabetes upon treatment with STZ, but also still display rapid elimination of glucose, suggests that other factors besides insulin play a role in the postprandial regulation of blood glucose in these mice. In fact, Lee et al. (5) suggested that a hormone with insulinomimetic properties might be increased, causing the insulin-like responses to glucose in the insulin-deficient Gcgr^−/− mice; however, no increase in leptin or IGF-I levels could be observed. Fibroblast growth factor 21 (FGF21) is a 181-amino acid circulating protein that has been observed to increase in plasma upon long-term starvation (7,8), but also during high-fat feeding (9–11). FGF21 stimulates glucose uptake into adipocytes in an insulin-independent manner (12), and FGF21-mediated glucose uptake in skeletal muscle has been demonstrated (13). Furthermore, FGF21 has been shown to suppress hepatic glucose output in vivo and in vitro (14,15). Injection of recombinant FGF21 normalizes blood glucose levels, improves insulin sensitivity, and corrects dyslipidemia in several rodent and nonrodent animal models of type 2 diabetes (12,16,17). It is highly expressed in the liver and whole-pancreas extracts, with lower expression levels in muscle and adipose tissue (18,19). Given the pleiotropic actions of FGF21 on whole-body glucose regulation, we hypothesized that FGF21 may be a factor contributing to glucose regulation in insulin-deficient Gcgr^−/− mice. We therefore determined the contribution of FGF21 action to the regulation of glucose disposal in insulin-deficient Gcgr^−/− mice.

Mice were housed using a 12 h dark/light cycle and were fed standard rodent chow (R34; Lantmännen, Stockholm, Sweden). All experimental protocols were approved by the regional animal ethics committee in Lund, Sweden. Induction of diabetes by STZ was carried out as follows: 15-week-old female Gcgr^−/− and Gcgr^+/+ mice were anesthetized with an intraperitoneal injection of midazolam (12.5 mg/kg) (Dormicum; Roche, Basel, Switzerland), and a combination of fluanisone (25 mg/kg) and fentanyl (0.78 mg/kg) (Hypnorm; Janssen, Beerse, Belgium). STZ (Sigma-Aldrich, St. Louis, MO) or vehicle was injected into the tail vein at a dose of 150 mg/kg. The protocol was repeated 7 days later with a dose of 100 mg/kg according to the protocol described by Lee et al. (5). The dose response of STZ to circulating FGF21 was determined in male C57BL/6 mice given single intravenous injections of 0, 150, or 200 mg/kg. Blood sampling for FGF21 was performed 7 days after STZ treatment. For fasting plasma values, mice were fasted for 5 h and anesthetized, and blood was collected from the retrobulbar intraorbital sinus plexus. For the oral glucose tolerance test (OGTT), mice were fasted for 5 h prior to oral administration of d-glucose (75 mg/mouse) by gavage. Blood samples were collected before and at 15, 30, and 60 min after the administration of glucose. A neutralizing antibody (Ab) to FGF21 or an IgG control (AIS, Hong Kong) (10 μg/mouse) was given to Gcgr^−/− and STZ-treated Gcgr^−/− mice in a single intraperitoneal injection at the start of the fasting period, 5 h prior to the start of the OGTT. The use of the neutralizing Ab has been previously described (20). Exendin 9-39 (Ex9-39) (Sigma-Aldrich, St. Louis, MO) was injected intraperitoneally at a dose of 30 nmol/kg 10 min prior to the start of the OGTT.

Pancreata were removed and fixed in neutral buffered formalin prior to embedding in paraffin. The 5-µm-thick sections were mounted onto Superfrost Plus glass slides. Briefly, after deparaffinization, sections were blocked in 1% H₂O₂ and subsequently blocked with 0.5% Tris-NaCl blocking reagent (PerkinElmer, Waltham, MA). Sections were incubated with rabbit anti-FGF21 Ab (1:200; ab64857; Abcam, Cambridge, MA) overnight and developed with TSA-cy3 (tyramide signal amplification; PerkinElmer). On top of the primary antibodies, guinea pig anti-insulin (1:150; ab7842; Abcam) and mouse anti-glucagon antibodies were added: (1:100; Glu001; Novo BioLabs, Bagsvaerd, Denmark). Guinea pig Ab was visualized with donkey anti-guinea pig-cy5 and mouse Ab with donkey anti-mouse-cy2 (1:300; Jackson ImmunoResearch, West Grove, PA). DAPI (0.2 μg/mL) was used to stain nuclei. Images were obtained on an LSM510 Laser Scanning confocal microscope (Carl Zeiss, Oberkochen, Germany). False color images were generated with the LSM software. Absorption studies were conducted to verify the specificity of the anti-FGF21 antiserum.

Plasma glucose concentrations were determined using the glucose oxidase method as described previously (21). Plasma insulin was measured by ELISA (Mercodia, Uppsala, Sweden). Plasma FGF21 was measured using a mouse-/rat-specific FGF21 ELISA assay (Biovendor, Prague, Czech Republic). The detection limit was measured to be 7 pg/mL, and the intra-assay coefficient of variation was 4%. Plasma samples in which FGF21 was below the detection limit were set to 7 pg/mL, as this was the detection limit of the assay.

Liver and gonadal adipose tissue were excised and frozen for measurement of FGF21 protein levels. FGF21 expression in tissues was determined using a mouse-/rat-specific FGF21 ELISA (Biovendor). In brief, liver and fat tissues were homogenized and lysed in lysis buffer (Invitrogen, Carlsbad, CA) with a protein inhibitor cocktail mix (Invitrogen). Total protein was determined with a BCA protein assay from Pierce, using albumin as the standard.

Total RNA was extracted from the liver and white adipose tissue using Trizol (Invitrogen) and RNeasy mini-kit (Qiagen) according to the manufacturer’s instructions. cDNA was synthesized using iScript reverse-transcription kit (BioRad). Quantitative real-time PCR was performed on an ABI 7900 Sequence Detection System (Applied Biosystems) using a locked nucleic acid probe–based system from Roche. Primers were designed using Primer3 software (bioinfo.ut.ee/primer3). All samples were run in triplicate, and expression was calculated using the ΔΔC_T method. Samples were normalized to β-actin expression.

Data are presented as the mean ± SEM. The area under the curve (AUC) was calculated using the trapezoidal rule, and differences between groups were determined by one-way ANOVA or Student t test. Kruskal-Wallis test (one-way ANOVA) was performed on plasma FGF21 levels in mice given a single-dose injection of STZ. Differences between groups with regard to quantitative PCR were determined by Student t test. Significant differences were assumed for P values <0.05. All statistical analysis was performed using GraphPad Prism software, version 5.0 (San Diego, CA).

FGF21 is highly expressed in the liver and in the pancreas; however, no data on the cellular expression or role of FGF21 in the pancreas have been published. In Fig. 1, an immunostaining of insulin, glucagon, and FGF21 is shown in wild-type littermates and in Gcgr^−/− mice. FGF21 is localized in both α- and β-cells in the control littermates (C57BL/6J) (Fig. 1); however, in the islet sections from the Gcgr^−/− mice FGF21 is highly expressed in the α-cells, and, as previously described, the Gcgr^−/− mice display α-cell hyperplasia (glucagon staining). To verify the specificity of the applied FGF21 Ab, a blocking experiment was performed with recombinant FGF21, in which the addition of recombinant FGF21 totally blocked the staining of FGF21 on islet sections (data not shown). FGF21 protein expression was also determined in liver and adipose tissue, two other FGF21-expressing tissues. As seen in Fig. 2A and B, FGF21 protein expression was also increased in liver and adipose tissue from Gcgr^−/− mice, although it was only statistically significant for adipose tissue. FGF21 mRNA as well as the mRNAs for the FGF receptors (FGFRs) FGFR1, FGFR2, FGFR3, FGFR4, and β-Klotho were also determined in these tissues. FGF21 and FGFR3 mRNAs were significantly increased in the livers of Gcgr^−/− mice (Table 1). FGFR1 mRNA was significantly decreased in adipose tissue from Gcgr^−/− mice (Table 1). β-Klotho mRNA expression was decreased in adipose tissue (P = 0.05) but was unchanged in the liver.

In agreement with the strong FGF21 immunosignal observed in the α-cells and the increased expression in liver and adipose tissue, plasma FGF21 was significantly increased in the Gcgr^−/− mice, reaching levels of 11,000 pg/mL (P = 0.0073 vs. wild-type) (Fig. 2C). In addition, there was a strong correlation between circulating FGF21 and 4 h fasting plasma glucose levels in both wild-type and Gcgr^−/− mice (Supplementary Table 1).

In order to determine whether the increased plasma level of FGF21 could be involved in protecting the Gcgr^−/− mice from STZ-induced diabetes, Gcgr^−/− and wild-type control mice were treated with multiple high doses of STZ, and the in vivo effects on FGF21 were examined. As expected, treatment with multiple high doses of STZ resulted in overt diabetes in wild-type control mice with fasting plasma glucose levels exceeding 27 mmol/L (500 mg/dL) (Fig. 3A). Consistent with previous findings, Gcgr^−/− mice have significantly lower fasting glucose levels. STZ treatment resulted in a minor increase in fasting glucose levels in Gcgr^−/− mice; however, they still had significantly lower fasting glucose levels than untreated wild-type controls (Fig. 3A).

Because the effect of STZ on circulating levels of FGF21 is unknown, plasma levels of FGF21 were measured in wild-type and Gcgr^−/− mice before and after STZ treatment. Wild-type mice treated with a single dose of STZ at 150 mg/kg showed no differences in plasma FGF-21 levels compared with untreated mice (Fig. 3B). However, a single dose of STZ at 200 mg/kg resulted in significantly reduced plasma FGF21 levels (Fig. 3B). The multiple high-dose STZ regimen (150 mg/kg followed by 100 mg/kg 7 days later) decreased circulating levels of FGF21 in wild-type and Gcgr^−/− mice (Fig. 3C and D), but this was only statistically significant for Gcgr^−/− mice (Fig. 3D).

To investigate the role of FGF21 in preventing the development of overt hyperglycemia in the STZ-treated Gcgr^−/− mice, an FGF21 neutralizing Ab was used. The FGF21 neutralizing Ab decreased FGF21-induced signaling in vitro (Supplementary Fig. 1) and has previously been shown to neutralize FGF21 in vivo (20). As shown in Fig. 4A, the plasma level of FGF21 is decreased after injection of the neutralizing FGF21 Ab, and the level of FGF21 after Ab injection is not significantly different than that in control mice. The neutralizing Ab was injected into mice 5 h prior to the evaluation of glucose tolerance with an OGTT. As seen in Fig. 4B, and in agreement with published data (2), under normal conditions the Gcgr^−/− mice display lower fasting plasma glucose and lower glycemic excursions during the OGTT than wild-type controls. As seen in Fig. 4B, early insulin release in the Gcgr^−/− mice is increased compared with the controls. Upon STZ treatment, the Gcgr^−/− mice have slightly increased plasma glucose levels during the OGTT, but still significantly better glycemic control than untreated wild-type mice (Fig. 4A) despite a near total lack of insulin (Fig. 4B). Neutralization of FGF21 with the FGF21 Ab in STZ-treated Gcgr^−/− mice significantly increased the glycemic excursion equaling that of untreated wild-type controls, without significant changes in insulin. As shown in Fig. 4C and D, the incremental AUC for glucose was significantly increased as a result of FGF21 neutralization in STZ-treated Gcgr^−/− mice, whereas that for insulin was unchanged. In agreement with previous reports (5), the Gcgr^−/− mice do not become diabetic upon STZ treatment.

Despite the fact that immunoneutralization of FGF21 significantly increased glycemia in the STZ-treated Gcgr^−/−mice, they were not hyperglycemic and had relatively normal glucose elimination. This was despite being near totally insulin-deficient. This led us to believe that another factor could be working in addition to FGF21 to aid in glucose elimination and prevent hyperglycemia in the absence of insulin. Gcgr^−/− mice have been previously shown to have dramatically elevated levels of GLP-1 (2). Insulin- and glucagon-independent effects of GLP-1 on glucose homeostasis have been recently demonstrated (22,23). We thus determined whether GLP-1 contributes to the improved glycemic profile of Gcgr^−/− mice, even after β-cell destruction with STZ. GLP-1 receptor signaling was blocked with the GLP-1 receptor antagonist Ex9-39 prior to the OGTT. In STZ-treated Gcgr^−/− mice given Ex9-39, the glycemic excursion was significantly increased, by 100% compared with STZ treatment alone, without changes in the insulin response (Fig. 5A and B). The resulting incremental AUC for glucose in the STZ Ex9-39–treated Gcgr^−/− mice was not significantly different than that of wild-type mice (Fig. 5C), despite the Gcgr^−/− mice not having a measurable insulin response (Fig. 5D). STZ-treated Gcgr^−/− mice were pretreated with the neutralizing Ab to FGF21 and then given Ex9-39 prior to the OGTT. The combination of FGF21 neutralization and GLP-1 receptor antagonism resulted in a significant increase in glucose excursion during the OGTT, by 159% compared with STZ treatment alone, and significantly greater than that of wild-type controls (Fig. 6A and C), without significant changes in insulin levels (Fig. 6B and D).

Insulin and glucagon are the two most important hormones for controlling blood glucose levels. Nearly 40 years ago, Unger and colleagues (24,25) and Gerich et al. (26) published a series of seminal studies demonstrating the importance of glucagon action in the pathophysiology of diabetes and the potential of glucagon-suppressing agents for the treatment of diabetes. Recently, this same group revisited the physiological consequences of suppression of glucagon action in nearly total insulin-deficient states using mice with glucagon receptor deficiency (5,27). A key finding in these studies was a lack of the lethal metabolic consequences of insulin deficiency in the absence of glucagon action. Another key finding was completely normal postprandial glucose elimination during an OGTT in these insulin-deficient mice. It was proposed that another hormone with insulinomimetic properties could be contributing to the regulation of postprandial glycemia in these mice. Leptin and IGF-I were investigated but were found not to be altered (5). However, the involvement of other, as yet undetermined, factors in the diabetes resistance seen in this unique model could not be ruled out.

FGF21 is one such factor. Recent studies have demonstrated clear, insulin-independent, glucose-lowering effects of FGF21 in vitro and in vivo (15,28). Circulating levels of FGF21 were dramatically elevated in Gcgr^−/− mice, and there was abundant FGF21 in β-cells and the hyperplastic α-cells of Gcgr^−/− mice. FGF21 protein expression was also increased in adipose tissue, and mRNA expression was increased in the liver of Gcgr^−/− mice. It was clear in our study that β-cells express FGF21 protein. In support of our finding, plasma FGF21 was decreased in both wild-type and Gcgr^−/− mice after β-cell destruction with high-dose STZ treatment. This finding suggests that β-cells themselves secrete FGF21, that insulin induces FGF21 secretion from insulin-responsive tissues, or both. In support of this concept, significant decreases in serum FGF21 levels in human subjects with recent-onset type 1 diabetes have been reported (29). Although plasma FGF21 levels decreased significantly after multiple high-dose STZ treatments, they were still more than 10-fold greater in Gcgr^−/− mice than in wild-type mice and thus were fully capable of mediating effects on glycemia. This led us to investigate whether they might be mediating glucose tolerance in the model even after β-cell destruction with STZ.

Another characteristic of Gcgr^−/− mice is a compensatory increase in proglucagon gene transcription and a substantial increase in circulating GLP-1 levels (2,30,31). Although the insulin-independent effects of FGF21 on glycemia are generally well accepted, insulin-independent effects of GLP-1 on glucose lowering have only recently begun to be appreciated. In recent years, several studies in rodents, canines, and humans have demonstrated insulin-independent effects of GLP-1 on circulating glucose levels (23,32–34). Thus, it stood to reason that one or both of these factors could contribute to the lack of postprandial hyperglycemia seen in insulin-deficient Gcgr^−/− mice. The Gcgr^−/− mice exhibited far lower glucose excursions than wild-type controls. This was still the case after induction of insulin deficiency with STZ. When neutralizing FGF21 with a neutralizing Ab, glucose tolerance was significantly worsened in insulin-deficient Gcgr^−/− mice, supporting the notion that FGF21 contributes to the diabetes resistance of Gcgr^−/− mice. Interestingly, blockade of the GLP-1 receptor alone also worsened glucose tolerance in insulin-deficient Gcgr^−/− mice. Despite the impairment of glucose tolerance induced in insulin-deficient Gcgr^−/− mice by both FGF21 neutralization and GLP-1 receptor antagonism individually, insulin-deficient Gcgr^−/− mice did not have hyperglycemia or diabetes in either case. Only the combination of both FGF21 neutralization and concurrent GLP-1 receptor blockade resulted in significant hyperglycemia in insulin-deficient Gcgr^−/− mice. This strongly suggests that GLP-1 and FGF21 work in a complementary way to prevent postprandial hyperglycemia in mice lacking insulin and glucagon action. That said, these Gcgr^−/− mice had normal fasting glucose, even with nearly total insulin deficiency and this was unaffected by FGF21 neutralization or GLP-1 receptor blockade. This supports the idea that glucagon is the sole mediator of fasting hyperglycemia in diabetes, and that chronic blockade of FGF21 and GLP-1 action in insulin-deficient Gcgr^−/− mice would likely not result in overt diabetes due to the lack of glucagon action. Thus, the total elimination of glucagon action does prevent the lethal catabolic consequences of total insulin deficiency, and FGF21 and GLP-1, and possibly other factors, contribute to the regulation of postprandial glycemia in insulin-deficient states.

In conclusion, genetic ablation of the glucagon receptor results in a dramatically increased circulating level of FGF21, which promotes glucose tolerance in an insulin-independent manner. In addition, FGF21 acts together with GLP-1 to prevent hyperglycemia in insulin-deficient Gcgr^−/− mice. Our results support the idea that pharmacological antagonism of glucagon action could be clinically beneficial for individuals with diabetes.

Acknowledgments. The authors thank Kristina Andersson, Catarina Blennow, and Maria Anderberg of Lund University. The authors thank Karen Arevad, Kirsten Haugegaard, and Anette Bjerregaard of Novo Nordisk A/S for excellent technical assistance. The authors also thank Rick Gelling of Abbott Nutrition for fruitful discussion of the Gcgr^−/− mice phenotype.

Funding. This work was supported by grants from the Swedish Research Council, Region Skåne, and the Lund University Faculty of Medicine.

Duality of Interest. B.An., J.H., K.R., and E.N. are employees of Novo Nordisk A/S. B.Ah. has received speaking fees and research grants from Novo Nordisk A/S. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. B.A.O. and B.An. designed the study, performed experiments and data analysis, and wrote the manuscript. J.H. and K.R. performed experiments and contributed to the writing of the manuscript. E.N. helped with writing the manuscript. B.Ah. designed the study, performed data analysis, and wrote the manuscript. B.Ah. 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. Parts of this study were presented in abstract form at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, Pennsylvania, 8–12 June 2012.