Glucagon-like peptide 1 (GLP-1) is known to suppress glucagon secretion, but the mechanism by which GLP-1 exerts this effect is unclear. In this study, we demonstrated GLP-1 receptor (GLP-1R) expression in α-cells using both antibody-dependent and antibody-independent strategies. A novel α-cell–specific GLP-1R knockout (αGLP-1R−/−) mouse model was created and used to investigate its effects on glucagon secretion and glucose metabolism. Male and female αGLP-1R−/− mice both showed higher nonfasting glucagon levels than their wild-type littermates, whereas insulin and GLP-1 levels remained similar. Female αGLP-1R−/− mice exhibited mild glucose intolerance after an intraperitoneal glucose administration and showed increased glucagon secretion in response to a glucose injection compared with the wild-type animals. Furthermore, using isolated islets, we confirmed that αGLP-1R deletion did not interfere with β-cell function but affected glucagon secretion in a glucose-dependent bidirectional manner: the αGLP-1R−/− islets failed to inhibit glucagon secretion at high glucose and failed to stimulate glucagon secretion at very low glucose condition. More interestingly, the same phenomenon was recapitulated in vivo under hypoglycemic and postprandial (fed) conditions. Taken together, this study demonstrates that GLP-1 (via GLP-1R in α-cells) plays a bidirectional role, either stimulatory or inhibitory, in glucagon secretion depending on glucose levels.
Introduction
Glucagon-like peptide 1 (GLP-1) is one of the hormones encoded with glucagon by the proglucagon gene. In the pancreas, GLP-1 suppresses glucagon secretion from α-cells and stimulates insulin secretion from β-cells in a blood-glucose dependent manner (1,2). The effects of GLP-1 on β-cells are mediated by the GLP-1 receptor (GLP-1R), which activates various pathways to promote insulin secretion and β-cell survival (3,4). The action of GLP-1 on α-cells, however, is not clear. Given the importance of GLP-1–based medicines in the treatment of type 2 diabetes, it is essential to fully understand the actions of GLP-1R and its roles not only in β-cells but also in α-cells.
Numerous studies have demonstrated that GLP-1 is a potent and physiological inhibitor of glucagon secretion (5,6). However, the underlying mechanisms are highly controversial. In particular, there is little consensus regarding whether GLP-1R is expressed in glucagon-producing α-cells located in pancreatic islets. Some studies show no GLP-1R protein or mRNA expression in α-cells of normal mice (7), some show GLP-1R is expressed in a small portion of α-cells in adult mice (8,9), and others show clear GLP-1R expression in α-cells, especially during early development and throughout life in mice with α-cell hyperplasia (10,11). To make the topic even more complicated, studies have reported that the commonly used anti–GLP-1R antibodies are neither specific nor sensitive enough to reliably detect GLP-1R expression in tissues (1,12). Due to these controversies, one popular opinion is that GLP-1 does not directly suppress glucagon secretion from α-cells via GLP-1R; instead, GLP-1 works indirectly through insulin—GLP-1 stimulates insulin secretion from β-cells, which subsequently suppress glucagon secretion because insulin is a powerful inhibitor of glucagon secretion (13,14). However, this indirect hypothesis cannot explain the facts that the glucagonostatic effect of GLP-1 is present even when insulin is unstimulated (15,16) or in patients with type 1 diabetes with no residual β-cells (17,18). To reconcile these controversies, more persuasive and definitive strategies are desired.
In this study, we examined GLP-1R expression in α-cells using both antibody- and nonantibody-based methods. Because our data confirmed GLP-1R expression in α-cells, we then generated an α-cell–specific GLP-1R knockout (KO) mouse model (αGLP-1R−/−) and investigated whether α-cell–expressed GLP-1R plays a direct role in suppressing glucagon secretion using the mice and isolated islets.
Research Design and Methods
Animals
All animal experiments followed the guidelines established by the Tulane University Institutional Animal Care and Use Committee, and proper approvals were obtained before the study. The founder Rosa26-yellow fluorescent protein (YFP)-LoxP mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Generation of the humanized (h)GLP-1R–LoxP founder mouse was described previously (19,20), and the mice were maintained at Taconic Biosciences (Cambridge City, IN). The glucagon-cre founder mouse strain (B6.Cg-Tg(Gcg-cre)1Herr/Mmnc, identification #358-UNC) was obtained from the Mutant Mouse Regional Resource Centers, a National Institutes of Health–funded strain repository, and was donated to the Mutant Mouse Regional Resource Centers by Dr. P. Herrera (University of Geneva Medical School) (21). Mouse breeding and genotyping followed standard procedures. Offspring containing the transgenes glucagon-cre and homozygous hGLP-1R–LoxP (fl/fl) are designated as αGLP-1R−/− mice. The littermates containing only glucagon-cre or only hGLP-1R–LoxP were used as controls. Generation of αYFP mice, in which preexisting α-cells are labeled by enhanced YFP, was achieved by breeding glucagon-cre mice with Rosa26-YFP-LoxP mice. Homozygous glucagon-cre and Rosa26-YFP-LoxP mice were used in the study.
Antibodies
The polyclonal goat anti–GLP-1R (mouse) antibody (#TA326758) was obtained from OriGene Technologies (Rockville, MD). Guinea pig anti-insulin antibody (#ab7842) and the mouse anti–GLP-1 monoclonal antibody specific for the active form of GLP-1 (amidated GLP-17–36) (#ab26278) were purchased from Abcam (Cambridge, MA). The rabbit anti-glucagon polyclonal antibody (#2760) was from Cell Signaling Technology (Danvers, MA). Mouse anti–green fluorescent protein monoclonal antibody (#A11120) was purchased from Thermo Fisher Scientific (Waltham, MA). All secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).
Glucose Tolerance Tests
The intraperitoneal glucose tolerance test (ipGTT) was performed essentially as described previously (22). Briefly, after overnight fasting, the mice were injected with glucose (2 g/kg body weight) via i.p. injection. Their blood glucose was measured before glucose injection (time 0) and at 15, 30, 45, 60, 90, and 120 min after the injection using the AlphaTRAK glucose meter (Abbott Laboratories, Chicago, IL). For oral GTT (OGTT), after overnight fasting, the mice were administered with glucose (2 g/kg body weight) orally with a syringe, and their blood glucose was measured the same way as described above for ipGTT.
Hormone Measurements
Insulin, glucagon, and active GLP-1 in blood (serum), culture supernatants, and islet lysates were measured using corresponding ELISA kits (with cat #80-INSMS-E01, 48-GLUHU-E01, and 43-GP1HU-E01, respectively) from ALPCO (Salem, NH) according to the manufacturer’s protocols.
Insulin Tolerance Tests
The mice were fasted for 6 h before insulin tolerance tests (ITTs). After their blood glucose was measured (time 0), each mouse was administered with insulin (Humulin U-100; from Eli Lilly and Company, Indianapolis, IN) via i.p. injection at 0.5 units/kg body weight. Their blood glucose was measured at 15, 30, 60, and 90 min after the insulin injection.
Isolation of α-Cells by Flow Cytometry
Pancreatic islets were isolated from αYFP mice and control mice following standard protocols. All islets were cultured overnight at 37°C in a 5% CO2 humidified incubator to allow recovery. Before FACS, handpicked islets were dispersed by trypsinization (0.25% trypsin for 15 min at 37°C in a 5% CO2 incubator), with gentle pipetting every 5 min. After clumped cells were removed by filtration through a 70-μm cell strainer, cells in suspension were sorted based on the YFP signal using the flow cytometry system BD FACSDiva version 6.1.3.
Arginine-Stimulated Glucagon Secretion
Handpicked islets were preincubated with Krebs-Ringer buffer (KRB) containing 2.5 mmol/L glucose (KRB-2.5G) for 30 min. Solutions in each well were then changed to fresh KRB-2.5G, with or without 20 mmol/L arginine, and incubated for 1 h at 37°C in a 5% CO2 incubator. For experiments involving exendin9-39, 100 nmol/L was included in the solution. The supernatants were collected and processed for glucagon concentration measurements using a glucagon ELISA kit from ALPCO. Islets were lysed and used for insulin content measurements. All data were normalized to insulin content and are expressed as mean ± SEM.
Glucose-Dependent Glucagon Secretion
Freshly isolated and handpicked islets were cultured in 24-well plates in quadruples as described above. For the assay, islets were preincubated in KRB-2.5G for 30 min and then incubated with KRB containing 2.5 mmol/L, 5 mmol/L, or 10 mmol/L glucose for 1 h. The supernatants and islet lysates were collected for glucagon and insulin measurements using the corresponding ELISA kits. All data were normalized to insulin content and are expressed as mean ± SEM.
Immunohistochemistry
The mouse paraffin-embedded tissue slices were processed for immunohistochemistry (IHC) staining as described previously (23). Human paraffin-embedded pancreatic sections were obtained from the Network for Pancreas Organ Donors with Diabetes (nPOD), which is sponsored by JDRF. Briefly, the tissue sections first underwent deparaffinization and antigen retrieval. Then they were incubated with proper primary antibodies, followed by incubation with corresponding secondary antibodies. The FACS-isolated α-cells were first centrifuged to glass slides via Cytospin system. Cells were air dried, fixed with 4% formaldehyde, and processed for anti-YFP and anti-glucagon immunofluorescence staining. For nuclei staining, slides were incubated with DAPI at a concentration of 2 µg/mL for 10 min.
Quantification of GLP-1R+ α-Cells
After immunofluorescence staining, microscopic images of 10–15 islets from each mouse (n = 8 mice per group) were captured. The numbers of GLP-1R+ glucagon+ cells (i.e., GLP-1R+ α-cells) and glucagon+ cells (total α-cells) in each islet were manually counted. The percentage of GLP-1R+ α cells versus the total α-cells was calculated for each islet and then averaged for the mouse. Data from n = 8 mice were used to obtain the mean and SD for each group. The data are expressed as mean ± SD.
RT-PCR
Total RNA was extracted from the purified mouse α-cells or intact islets using the RNeasy Mini Kit from Qiagen (Valencia, CA). RT was performed with ∼170 ng RNA as template and oligo (dT) as primer using the iScript cDNA Synthesis Kit from Bio-Rad (Hercules, CA) according to the manufacturer’s protocol. The GLP-1R cDNA fragment was then amplified by PCR using 2 μL of the RT reactions and mouse (m)GLP-1R–specific primers, which include forward primer 5′-CATCCACCTGAACCTGTTTG-3′ (starting at position 544 nucleotides of mGLP-1R cDNA) and reverse primer 5′-TGGAGACCACTATGCAGATG-3′ (starting at position 1,010 nucleotides of mGLP-1R cDNA). The PCR product was analyzed by 1.5% agarose/ethidium bromide gel electrophoresis.
Statistical Analysis
The statistical analyses were performed with the SAS 9.4 software. The Student t test was used to determine the significance of differences between two groups, and one-way ANOVA was used for multiple groups. The data are expressed as mean ± SEM unless indicated otherwise. P < 0.05 was considered statistically significant.
Results
Detection of GLP-1R Expression in Pancreatic α-Cells Using a Specific Anti–GLP-1R Antibody
Recent studies revealed that the anti–GLP-1R antibodies from major commercial sources were not specific or not sensitive enough (1,12,24). To determine whether GLP-1R is expressed in pancreatic α-cells, it is critical to identify a reliable antibody. We found that the goat anti–GLP-1R antibody from OriGene (cat #TA326758) was an excellent candidate. The antibody was raised against the sequence C-SKRGERNFPEEQ, which is located at the extracellular domain of mGLP-1R immediately adjacent to the cell membrane (Fig. 1A). The sequence is highly conserved between mGLP-1R and hGLP-1R, with only two amino acid residues different (Fig. 1A). A Basic Local Alignment Search Tool (BLAST) search shows no other proteins containing similar epitopes. We tested the specificity of this anti–GLP-1R antibody by performing Western blotting analysis (Fig. 1B). The anti–GLP-1R antibody recognized a specific band at ∼53 kDa, the expected size for the GLP-1R protein, in the positive control lane but not in the negative control lane. In addition, the antibody detected highly specific bands in lysates from human islets and mouse organs such as lung and pancreas. The band in human islets and mouse pancreas corresponded to the GLP-1R molecular weight (∼53 kDa), whereas the band in the lung tissue is ∼110 kDa, which corresponded to a dimerized or a heavily glycosylated version of GLP-1R. The key conclusion is that the GLP-1R antibody is highly specific in islets and the pancreas and can thus be reliably used for this study.
We next examined GLP-1R expression in mouse and human pancreatic tissues using IHC staining (Fig. 1C). The GLP-1R antibody showed clean staining in the islets, and GLP-1R was detected not only in β-cells but also in α-cells. No GLP-1R signal was detected in the exocrine pancreas (outside the islets) or in somatostatin-expressing δ-cells (data not shown). The GLP-1R staining was completely blocked when its immunizing peptide was present (Fig. 1C), confirming the specificity of the staining.
Verification of GLP-1R Expression in Pancreatic α-Cells Using an Antibody-Independent Strategy
To further demonstrate GLP-1R expression in pancreatic α-cells, we purified α-cells and examined GLP-1R expression at the mRNA level by RT-PCR (Fig. 1D–G). To purify primary α-cells, we first established a mouse model in which existing α-cells were labeled by enhanced YFP, namely αYFP mice. Immunofluorescence staining confirmed that all α-cells expressed YFP and that YFP was only expressed in α-cells (Fig. 1D). We then isolated islets from the αYFP mice, dissociated them into single cells, and purified YFP+ α-cells using FACS (Fig. 1E). Immunostaining of the FACS-isolated cells showed that all isolated cells were positive for both YFP and glucagon, demonstrating they were pure α-cells (Fig. 1F). We also costained the purified α-cells with GLP-1R, glucagon, and YFP antibodies, and the results showed that all purified α-cells expressed GLP-1R (Supplementary Fig. 1).
We examined GLP-1R expression on the mRNA level by extracting total RNA from the purified α-cells and performing RT-PCR using GLP-1R–specific primers. Total islet RNA was used as a positive control because GLP-1R is expressed in β-cells. As shown in Fig. 1G, we obtained a PCR product corresponding to the designed size of the GLP-1R fragment in the extracts of both purified α-cells and intact islets. The PCR product was also purified and processed for sequence analysis, and the results confirmed that it was part of mGLP-1R cDNA (data not shown). Together, these data demonstrate that GLP-1R is indeed expressed in pancreatic α-cells.
Generation of the α-Cell–Specific GLP-1R KO (αGLP-1R−/−) Mouse Model
To fully understand the role of GLP-1/GLP-1R in α-cells, we next generated αGLP-1R−/− mice using the cre-loxP system. The glucagon-cre line was used because our data demonstrated that glucagon-cre–mediated recombination is highly efficient and specific for α-cells in the pancreas (Fig. 1D). For Glp1r-LoxP mice, we used the novel hGLP-1R mouse model (19,20). The hGLP-1R mice were created by knocking the hGLP-1R cDNA into the mouse Glp-1r locus so that it is under the control of native mGLP-1R regulatory elements and also disrupts mGLP-1R expression (Fig. 2A). The hGLP-1R cDNA is flanked by LoxP sites, allowing Cre-LoxP–mediated deletion of hGLP-1R, and by design, the deletion generates a frame-shift mutation in the mGLP-1r gene, resulting in the loss of both hGLP-1R and mGLP-1R in targeted cells. We bred the glucagon-cre mice with hGLP-1R–LoxP mice and screened for genotypes containing glucagon-cre and homozygous hGLP-1R–LoxP (fl/fl), which are identified as αGLP-1R−/− mice. Immunofluorescence staining shows GLP-1R expression was eliminated from nearly all α-cells of the αGLP-1R−/− mice, whereas GLP-1R expression in β-cells was not affected compared with the wild-type (WT) mice (Fig. 2B). Because previous studies have observed low recombination efficiency mediated by glucagon-cre in some mouse models (25,26), we examined the efficiency of glucagon-cre–mediated GLP-1R ablation in our system. Specifically, we quantified the percentage of GLP-1R+ α cells. Our data show that GLP-1R was detected in 9.4% ± 2.1% of α-cells in the KO mice, whereas GLP-1R expression in the WT mice was readily detectable in 93.6% ± 2.7% of α-cells, suggesting GLP-1R was ablated from ∼90% of α-cells in the αGLP-1R−/− mice. Of note, the glucagon promoter is also active in intestine L cells, where it drives the production of GLP-1, but GLP-1R is not expressed in L cells (8). So the KO should not affect L cells.
GLP-1R Deletion From α-Cells Causes an Increase in Nonfasting Glucagon Level in Circulation
We next monitored the αGLP-1R−/− (KO) mice and their WT littermates for their growth, blood glucose, and general wellness. Initial observations suggested there was no significant difference between the KO mice and the control mice. Food intake and urinary glucose concentrations were also similar between the KO and WT mice (Supplementary Fig. 2). However, when we measured circulating hormones, we found that glucagon levels (random, nonfasting) in both females and males were significantly higher in the KO mice than in the WT littermates (Fig. 2F), whereas blood glucose, insulin, and GLP-1 levels were similar between the KO and WT mice (Fig. 2C–E). These data demonstrate that GLP-1R deletion from α-cells increases nonfasting glucagon secretion and that this effect is not mediated indirectly by insulin.
Interestingly, under the fasting condition, we found no significant difference in glucagon concentrations between the KO and WT mice (Fig. 2I). Blood glucose and insulin secretion in the fasting condition were also similar between these mice (Fig. 2G and H). Fasting GLP-1 concentrations were too low to be detected for all mice. These data indicate that αGLP-1R plays a direct role in suppressing glucagon secretion at the nonfasting/fed state.
The αGLP-1R Deletion Impairs i.p. Glucose Tolerance That Is Attributable to Abnormal Glucagon Secretion
To further examine the effect of αGLP-1R−/−, we performed GTTs. When glucose was administered by i.p. injection, the KO female mice exhibited glucose intolerance starting at 3–4 months of age (Fig. 3A and B). In addition, OGTT results showed no significant differences between the KO and WT mice for males or females (Supplementary Fig. 3). To examine whether the impaired i.p. glucose tolerance in female αGLP-1R−/− mice was due to impaired glucose-stimulated insulin secretion (GSIS), we measured circulating insulin during ipGTT and found no difference in GSIS between the KO and WT mice (Fig. 3C). Insulin secretion by the KO mice was even higher than the control mice at 30 min, probably because of the higher glucose level that stimulated more insulin secretion. Because glucose intolerance occurred at early time points of the ipGTT, and insulin secretion is biphasic, we next examined the first phase of insulin secretion that peaks at ∼2 min after glucose administration (27) and observed no difference among these mice (Fig. 3D). Interestingly, when circulating glucagon levels were examined, we found that glucagon was stimulated 15 min after glucose administration in the KO mice but not in the control mice (Fig. 3F). This suggests that αGLP-1R deletion has altered glucagon secretion in response to glucose increase, which could explain their impaired glucose tolerance during ipGTT. Furthermore, ITTs showed no significant differences between the KO mice and the WT mice, suggesting insulin sensitivity is not impaired by αGLP-1R KO (Fig. 3E).
Interestingly, male αGLP-1R−/− mice did not show impaired i.p. glucose tolerance even at 1 year of age (Supplementary Fig. 4A, and data not shown). However, similar to the female KO mice, the males had normal GSIS but abnormal glucagon secretion after i.p. glucose administration (Supplementary Fig. 4B and C). The male KO mice also had normal insulin sensitivity as assessed by ITT (Supplementary Fig. 4D). The sex dimorphism in glucose intolerance appears to lie on different responses to glucagon dysregulation between female and male mice.
Taken together, these data demonstrate that α-cell–specific GLP-1R KO does not affect β-cell function or insulin sensitivity; instead, it altered glucagon secretion in response to glucose changes in both male and female mice.
The Effect of αGLP-1R Deletion on Glucagon Secretion in Isolated Islets
We next used isolate islets to investigate whether αGLP-1R deletion affects glucagon secretion directly. Shown in Fig. 4 are results from female islets, and similar phenomena were observed in male islets (data not shown). GSIS assay with the isolated islets demonstrated that αGLP-1R KO did not affect β-cell function (Fig. 4A and B), consistent with the in vivo observations. To test α-cell function, we performed arginine-stimulated glucagon secretion (ASGS) assays. The data show that arginine significantly stimulated glucagon secretion in both the WT islets and the KO islets (Fig. 4C). Interestingly, basal glucagon concentrations were significantly lower in the KO islets than that of the WT islets, whereas glucagon concentrations after arginine stimulation were similar between the groups. This led to a substantially higher arginine stimulation index for the KO islets than for the WT islets (Fig. 4D).
Emerging evidence continues to support that GLP-1 is produced and secreted locally within islets (28,29). Our data show that GLP-1R deletion from α-cells did not interfere with intraislet GLP-1 production or secretion (Supplementary Fig. 5). To further assess whether αGLP-1R deletion affected α-cell function, we performed ASGS assays in the presence of GLP-1R antagonist exendin9-39. Exendin9-39 was expected to suppress insulin secretion by blocking GLP-1R that is expressed on β-cells, which in turn increases glucagon secretion due to the removal of insulin-mediated glucagon suppression. Indeed, exendin9-39 alone was able to increase glucagon secretion in WT and KO islets, and the effect of exendin9-39 and arginine appeared to be additive in promoting glucagon secretion (Fig. 4C and D). Taken together, the data demonstrate that GLP-1 regulation of glucagon secretion in isolated islets occurs both directly (via GLP-1R on α-cells) and indirectly (via its potentiation of insulin secretion from β-cells).
It should be noted that ASGS is normally performed in the presence of 2.5 mmol/L glucose (hypoglycemic) conditions in which glucagon secretion is highly stimulated (and arginine can further stimulate glucagon secretion). Without arginine stimulation, glucagon secretion at 2.5 mmol/L glucose in the KO islets was significantly lower than that of WT islets, suggesting GLP-1R deletion from α-cells impaired glucagon stimulation at the hypoglycemic state. This is in contrast with the in vivo data in which the KO mice failed to suppress glucagon secretion at the nonfasting state (∼180 mg/dL = 10 mmol/L glucose) (Fig. 2F). To see whether glucose accounts for the difference, we used isolated islets to examine glucagon secretion under glucose conditions equivalent to hypoglycemic (2.5 mmol/L), fasting (5 mmol/L), and nonfasting (10 mmol/L) states. As shown in Fig. 4E and F, insulin secretion was stimulated to similar levels in the KO and WT islets with increasing glucose concentrations, whereas glucagon secretion was in sharp contrast between the KO and WT islets. As expected, glucagon secretion in the WT islets reduced with increasing glucose concentrations; however, glucagon secretion in the KO islets was significantly lower at 2.5 mmol/L glucose and significantly higher at 10 mmol/L glucose than in the WT islets (Fig. 4E). These data indicate that GLP-1R activation in α-cells could play a bidirectional role—either inhibitory or stimulatory—for glucagon secretion depending on glucose conditions.
In Vivo Glucagon Versus Blood Glucose Data Confirms αGLP-1R Regulates Glucagon Secretion in a Glucose-Dependent Bidirectional Manner
In Fig. 2, we observed that the αGLP-1R−/− mice had significantly higher glucagon secretion at the nonfasting condition and similar glucagon secretion after overnight fasting compared with the WT controls. The discovery of the bidirectional role of αGLP-1R in isolated islets promoted us to examine the αGLP-1R−/− mice further in hypoglycemic and postprandial (fed) conditions. Hypoglycemia (defined as blood glucose <70 mg/dL) was induced by extended fasting (∼40 h). As shown in Fig. 5A and B and consistent with islet studies, glucagon levels were significantly lower in the female KO mice than in the WT mice, suggesting the KO mice failed to stimulate glucagon secretion in the hypoglycemic condition. For the postprandial (fed) condition, blood was drawn 1 h after lights off, when most mice should have woken up and eaten. Glucagon levels were significantly higher in the KO mice than in the WT mice, suggesting the KO mice failed to inhibit glucagon secretion in the postprandial (fed) condition (Fig. 5C and D). Furthermore, to have a full picture on how blood glucose affected glucagon secretion, we plotted each mouse’s glucagon concentration against its blood glucose (Fig. 5E). Included in the plot are data from female mice after extended fasting, overnight fasting, nonfasting, and feeding. The trend lines showed a similar pattern to what was observed in the isolated islets. Male KO and WT mice had similar trends in the analysis (Supplementary Fig. 6). Taken together, these data demonstrate that GLP-1R in α-cells is essential for glucagon stimulation in the hypoglycemic condition and glucagon suppression in the high glucose condition and confirm that GLP-1R regulates glucagon secretion in a glucose-dependent bidirectional manner.
Discussion
The mechanism whereby GLP-1 regulates glucagon secretion has been a longstanding question mainly because 1) it is uncertain whether α-cells express GLP-1R, and 2) it is technically difficult to investigate direct GLP-1 effects on glucagon suppression versus its indirect effect through insulin stimulation. In this study, we attempted to clarify this matter by using novel systems. First, we confirmed the expression of GLP-1R in α-cells using antibody-dependent and antibody-independent methods. We then generated α-cell–specific GLP-1R KO mice and examined the effects in vivo and in isolated islets. Our data show that αGLP-1R deletion disturbs glucose-dependent glucagon secretion, whereas insulin secretion is not affected, demonstrating GLP-1R plays a direct role in regulating glucagon secretion. More interestingly, we discovered that αGLP-1R plays a bidirectional role, either stimulatory or inhibitory, in glucagon secretion depending on glucose levels.
The controversy on GLP-1R expression in α-cells was largely attributable to the use of different antibodies or different methods (7,8,10,30). Most of the commonly used GLP-1R antibodies have been reported to lack specificity (1,12), and some GLP-1R antibodies may also lack sufficient sensitivity. Indeed, we found that a widely used hGLP-1R antibody generated a much weaker signal than the OriGene (#TA326758) antibody in immunofluorescence staining of normal human pancreatic tissue (data not shown). This may explain why some studies did not detect GLP-1R expression in α-cells or only in a portion of α-cells. Our study confirms the specificity and sensitivity of the anti–GLP-1R antibody (#TA326758) from OriGene, as assessed by Western blotting, immunohistochemistry staining, with or without antigen blocking peptide, and immunostaining of the αGLP-1R−/− pancreatic tissue. More importantly, we were able to demonstrate GLP-1R expression at the mRNA level using purified primary α-cells, circumventing the use of an GLP-1R antibody. Interestingly, a previous study using transgenic mice in which GLP-1R+ cells are labeled with a fluorescence reporter red fluorescent protein showed only a small fraction of α-cells are GLP-1R+ (8), whereas our data show nearly all α-cells express GLP-1R (Supplementary Fig. 1). The discrepancy is likely due to differences in labeling efficiency and detection sensitivity. In our study, YFP labeled nearly all α-cells, and the subsequent FACS isolation and antibody detection were both highly specific and efficient (Fig. 1 and Supplementary Fig. 1).
Detection of GLP-1R on α-cells argues for a direct role of GLP-1R in regulating glucagon secretion, which is confirmed by our study using αGLP-1R−/− mice and isolated islets. Our study is in agreement with previous studies showing the glucagonostatic effect of GLP-1 is present even when insulin is undetectable or unstimulated (15,17,18). For instance, in patients with type 1 diabetes with undetectable insulin, GLP-1 infusion has been found to reduce glucagon secretion during hyperglycemia and also decrease ASGS (17). In addition, Guenifi et al. (15) showed that GLP-1 substantially inhibits arginine-induced glucagon secretion in perfused rat pancreas without affecting insulin or somatostatin secretion in the same setting. We therefore conclude that GLP-1 suppresses glucagon secretion directly in a glucose-dependent manner, although the indirect effect from insulin is also present.
The most intriguing discovery in this study is that GLP-1 not only suppresses glucagon secretion but also stimulates glucagon secretion, depending on the glucose levels. Under normal conditions, glucagon secretion is highly responsive to glucose variations: glucagon secretion is minimal when the glucose level is 6–10 mmol/L, increases when glucose is 3–6 mmol/L, and is maximally stimulated when glucose is 2–3 mmol/L (31). The glucagonostatic effect of GLP-1 is well documented in both fasting and postprandial states (6), but little is known about its effect in hypoglycemic conditions. Our study reveals that GLP-1 can be glucagonotropic when glucagon secretion is needed at very low glucose levels. This novel discovery has important clinical implications because it helps explain the clinical observations that patients with diabetes on GLP-1R agonists have fewer hypoglycemic events (32–34). GLP-1 therefore appears to facilitate glucagon secretion in response to glucose changes under physiological conditions: it stimulates or suppresses glucagon secretion based on physiological needs.
Interestingly, although αGLP-1R KO caused similar glucagon dysregulation effects in both female and male mice, glucose intolerance was only observed in female mice. This sex dimorphism appears to be attributable to the different responses to increased glucagon concentrations—the males’ glucose homeostasis is more resistant to glucagon dysregulation than females’. Glucagon is part of glucose counterregulatory responses that involve several other hormones (such as epinephrine and growth hormones) and the autonomic nervous system (35). Previous studies have shown significant sex dimorphism between men and women, with men having more potent metabolic and neuroendocrine counterregulatory responses than women (36–38). In addition, the male hormone testosterone has been shown to activate an extranuclear androgen receptor in β-cells and increase cellular cAMP that subsequently amplifies the incretin effect of GLP-1 (39). Therefore, it is possible that glucagon dysregulation in the male KO mice is compensated (or blunted) by other components of the counterregulatory responses or by the action of male hormones. Furthermore, we would like to note that in this study, consistent with the sex dimorphism in counterregulatory responses, we found that inducing hypoglycemia in male mice was very difficult: only 2 of 10 WT and 6 of 10 KO male mice became hypoglycemic with extended fasting, whereas 9 of 11 WT and 11 of 13 KO female mice became hypoglycemic under the same extended fasting condition (Fig. 5 and Supplementary Fig. 6). Nonetheless, the KO males became more susceptible to hypoglycemia induction than their WT controls, suggesting glucagon dysregulation had an effect on the males’ responses, although it was not sufficient to cause obvious glucose intolerance.
Furthermore, increased glucagon levels in the αGLP-1R−/− mice appeared to have only mild effects on glucose metabolism even in female mice because glucose intolerance was only observed during ipGTT, not when assessed by fasting or postprandial glucose levels. Food intake and urinary glucose output are also similar between the KO and WT mice (Supplementary Fig. 2). On one hand, the mild glucose intolerance caused by glucagon upregulation may be explained by normal insulin secretion and insulin sensitivity in the αGLP-1R−/− mice, which readily counters glucagon-induced glucose change. On the other hand, glucagon elevation may have other physiological effects in addition to glucose metabolism, such as amino acid metabolism (40). Indeed, glucagonoma patients have excess glucagon in circulation, and their symptoms are not necessarily associated with disturbed glucose metabolism; instead, they often have accelerated ureagenesis and amino acid turnover (40). Animal models or human patients with defective glucagon receptor have high levels of glucagon, exhibit pancreatic swelling due to α-cell hyperplasia, and also show elevated amino acid levels (41). In light of these observations, we examined amino acid levels in the αGLP-1R−/− mice and found that they had significantly increased amino acid levels compared with the control mice (Supplementary Fig. 7), consistent with glucagon playing a role in amino acid metabolism in addition to glucose metabolism.
Another interesting observation in this study is that oral glucose challenge does not cause any glucose intolerance in the αGLP-1R−/− mice. Oral glucose stimulates GLP-1 secretion from the gut, so one would expect the αGLP-1R KO mice to show impaired tolerance during OGTT. However, this is not the case. Similar observations are also made in the β-cell–specific GLP-1R KO mice, which exhibited impaired glucose tolerance in ipGTT but not in OGTT (4). This can be explained by an emerging concept involving intraislet GLP-1 (4,28,29,39). Specifically, it appears that the GLP-1 molecule that acts on α- (and β-) cells is not from the gut (via circulation) but is locally produced within the islets (i.e., intraislet GLP-1). GLP-1 has very short half-life, so it is difficult for the gut-produced GLP-1 to reach pancreatic islets in time and in sufficient amounts to make a significant effect on α- (and β-) cell function and thus cause any difference in glucose tolerance. Indeed, our current study has shown intraislet GLP-1 expression is not affected by αGLP-1R KO (Supplementary Fig. 5). In addition, the GLP-1R antagonist exendin9-39 was able to increase glucagon secretion in the culture of isolated islets (Fig. 4C and D), supporting the presence of active islet-derived GLP-1 in the culture.
In summary, our study has demonstrated that GLP-1/GLP-1R in α-cells regulates glucagon secretion in a glucose-dependent bidirectional manner: it stimulates or inhibits glucagon secretion based on physiological needs. The novel discovery provides a fresh point of view on how GLP-1/GLP-1R-based therapy benefits diabetes treatment and ameliorates hypoglycemia.
Article Information
Acknowledgments. The authors thank the core facility in Tulane Hypertension and Renal Center of Excellence and Tulane Pathology Core Laboratory for their excellent services.
Funding. G.E.F. was partially supported by the Susan Harling Robinson Fellowship in Diabetes Research. The study was supported by National Institute of Diabetes and Digestive and Kidney Diseases grants DK-081463 and DK-107412 (H.Wu). F.M.-J. was supported by R01-DK-074970, DK-107444, and a U.S. Department of Veterans Affairs VA Merit Review Award (BX003725). This research was performed with the support of the Network for Pancreatic Organ donors with Diabetes (nPOD; RRID:SCR_014641), a collaborative type 1 diabetes research project sponsored by JDRF (nPOD: 5-SRA-2018-557-Q-R) and The Leona M. & Harry B. Helmsley Charitable Trust (grant no. 2018PG-T1D053). The content and views expressed are the responsibility of the authors and do not necessarily reflect the official view of nPOD. Organ Procurement Organizations (OPO) partnering with nPOD to provide research resources are listed at http://www.jdrfnpod.org//for-partners/npod-partners/.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. Y.Z. designed and performed experiments and analyzed data. K.R.P., G.E.F., and R.G. performed experiments and analyzed data. W.X., L.U.N., and A.F.Z. helped with the experiments and data collection. V.A.F., H.Wa., and F.M.-J. contributed to the discussion and reviewed and edited the manuscript. K.W.S. assisted with the experimental design, contributed to the discussion, and edited the manuscript. H.Wu designed the experiments, analyzed data, and wrote the manuscript. H.Wu 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 78th Scientific Sessions of American Diabetes Association, Orlando, FL, 22–26 June 2018.