GLP-1 metabolite GLP-1(9–36) is a systemic inhibitor of mouse and human pancreatic islet glucagon secretion

Aims/hypothesis Diabetes mellitus is associated with impaired insulin secretion, often aggravated by oversecretion of glucagon. Therapeutic interventions should ideally correct both defects. Glucagon-like peptide 1 (GLP-1) has this capability but exactly how it exerts its glucagonostatic effect remains obscure. Following its release GLP-1 is rapidly degraded from GLP-1(7–36) to GLP-1(9–36). We hypothesised that the metabolite GLP-1(9–36) (previously believed to be biologically inactive) exerts a direct inhibitory effect on glucagon secretion and that this mechanism becomes impaired in diabetes. Methods We used a combination of glucagon secretion measurements in mouse and human islets (including islets from donors with type 2 diabetes), total internal reflection fluorescence microscopy imaging of secretory granule dynamics, recordings of cytoplasmic Ca2+ and measurements of protein kinase A activity, immunocytochemistry, in vivo physiology and GTP-binding protein dissociation studies to explore how GLP-1 exerts its inhibitory effect on glucagon secretion and the role of the metabolite GLP-1(9–36). Results GLP-1(7–36) inhibited glucagon secretion in isolated islets with an IC50 of 2.5 pmol/l. The effect was particularly strong at low glucose concentrations. The degradation product GLP-1(9–36) shared this capacity. GLP-1(9–36) retained its glucagonostatic effects after genetic/pharmacological inactivation of the GLP-1 receptor. GLP-1(9–36) also potently inhibited glucagon secretion evoked by β-adrenergic stimulation, amino acids and membrane depolarisation. In islet alpha cells, GLP-1(9–36) led to inhibition of Ca2+ entry via voltage-gated Ca2+ channels sensitive to ω-agatoxin, with consequential pertussis-toxin-sensitive depletion of the docked pool of secretory granules, effects that were prevented by the glucagon receptor antagonists REMD2.59 and L-168049. The capacity of GLP-1(9–36) to inhibit glucagon secretion and reduce the number of docked granules was lost in alpha cells from human donors with type 2 diabetes. In vivo, high exogenous concentrations of GLP-1(9–36) (>100 pmol/l) resulted in a small (30%) lowering of circulating glucagon during insulin-induced hypoglycaemia. This effect was abolished by REMD2.59, which promptly increased circulating glucagon by >225% (adjusted for the change in plasma glucose) without affecting pancreatic glucagon content. Conclusions/interpretation We conclude that the GLP-1 metabolite GLP-1(9–36) is a systemic inhibitor of glucagon secretion. We propose that the increase in circulating glucagon observed following genetic/pharmacological inactivation of glucagon signalling in mice and in people with type 2 diabetes reflects the removal of GLP-1(9–36)’s glucagonostatic action. Graphical Abstract Supplementary Information The online version of this article (10.1007/s00125-023-06060-w) contains peer-reviewed but unedited supplementary material.


Introduction
Glucagon is one of the body's principal blood glucoseincreasing (hyperglycaemic) hormones [1].In type 2 diabetes, the elevation of plasma glucose results from a combination of insufficient insulin and excessive glucagon [2].Whereas the insulin secretion defect in beta cells has attracted much attention [3], the dysregulation of glucagon secretion in alpha cells represents, by comparison, an understudied area [4].
The incretin hormone glucagon-like peptide 1 (GLP-1) is secreted by the enteroendocrine L cells as GLP-1 .It exerts a strong hypoglycaemic effect by potentiating insulin secretion in beta cells [5] and inhibiting glucagon secretion in alpha cells [6].Although the latter effect may account for as much as 50% of the peptide's hypoglycaemic action [6], the underlying cellular mechanism(s) remain(s) poorly understood.
Following its release, GLP-1  is quickly degraded by dipeptidyl peptidase 4 (DPP-4) to form the metabolite GLP-1 , which lacks insulin-releasing capacity [7], and only 10-15% of the GLP-1 released in the gut reaches the pancreatic islets [5].We hypothesised that the degradation product GLP-1  mediates GLP-1's glucagonostatic effect.Previous in vivo work performed under normoglycaemic conditions has failed to document any effects (stimulatory or inhibitory) of GLP-1(9-36) under normoglycaemic conditions [7][8][9].Here, we have compared the glucagonostatic effects of GLP-1  and  in vitro using physiological (pmol/l) concentrations of the peptides and over a wider range of glucose concentrations in isolated mouse and human islets (some from donors with type 2 diabetes) and extend these data to the in vivo situation with a particular focus on the counterregulatory increase in plasma glucagon during insulin-induced hypoglycaemia.

Methods
Animals and islet isolation Most studies were conducted in 8-to 16-week-old both male and female C57BL6/J (Envigo, IN, USA) or C57Bl6/N mice (Charles River, MA, USA), which were fed chow (Global Diet no.2016, Harlan-Teklad).No differences in the secretory function of the two strains were observed.In addition, Glp1r −/− mice [10] were used (sex-and age-matched wild-type littermates were used as controls).A transgenic reporter mouse model that expresses a genetically encoded Ca 2+ indicator GCaMP6-fast variant [11] (GCaMP6f) specifically in alpha cells (Gcg-GCaMP6f) was generated by crossing Gcg CreERT2 mice [12] (42277-JAX; Jackson Laboratory) (which express a tamoxifen-inducible form of Cre from the endogenous pre-proglucagon gene) to the Rosa26 GCaMP6f mice [13] (028865; Jackson Laboratory).To induce nuclear accumulation of Cre recombinase and GCaMP6f expression in alpha cells, Gcg-GCaMP6f mice were fed daily with tamoxifen (T5648; Sigma; 20 mg/ml in corn oil) through oral gavage for 5 days.Islets from Gcg-GCaMP6f mice were typically isolated 10 days after tamoxifen induction.The mice were killed by cervical dislocation, the pancreas quickly resected and pancreatic islets isolated by liberase (Sigma) digestion.All experiments in mice were conducted in accordance with the UK Animals (Scientific Procedures) Act (1986), the Ethical Committee at the University of Göteborg or the Veterinary Office of Canton de Vaud.
Human islets Human pancreatic islets were isolated (with ethical approval and clinical consent) as previously described [14].Islets from the pancreases of 18 healthy donors and five donors with type 2 diabetes were used (see electronic supplementary material [ESM] Table 1; ESM Human islets checklist).

Measurements of islet hormone secretion
Mouse islets were used acutely, except for studies with PTX, wherein islets were treated overnight with the toxin.Human islets were maintained in culture for up to 48 h in RPMI medium containing 10% (vol./vol.)FCS, 1% (vol./vol.)penicillin/ streptomycin and 5 mmol/l glucose.Experiments were conducted as previously described [15].Islets were incubated in 0.3 ml EC1 or EC2 media (ESM Table 2) supplemented with glucose and other reagents as indicated.Glucagon was determined by ELISA (Mercodia).The fractional glucagon release in isolated islets was 0.39±0.06%/h (mean value ± SEM of 15 preparations using 150 mice).For comparison, glucagon secretion in the intact mouse pancreas perfused at the physiological rate (~0.3 ml/min) measured at 1 mmol/l glucose using the same assay was 86±17 pg/min (n=10) (i.e.5.16±1.02ng/h).Total pancreatic glucagon content was 1.16±0.06μg (n=6).Thus, the rate of glucagon secretion normalised to the content in the perfused pancreas was estimated as 0.44±0.08%/h, in fair agreement with that obtained in the isolated islets.
Insulin and somatostatin were determined by RIA (Millipore and Diasource ImmunoAssays, respectively) as described previously [16,17].In these experiments, rates of release are expressed as % of contents unless otherwise indicated.
Perfused mouse pancreas Briefly, the aorta was ligated above the coeliac artery and below the superior mesenteric artery and then cannulated.The pancreas was perfused at 1.34 µl min −1 mg −1 pancreas weight using an Ismatec Reglo Digital MS2/12 peristaltic pump.Pancreatic weight was estimated from whole body weight as previously described [18,19].The perfusate was maintained at 37°C using a Warner Instruments temperature control unit TC-32 4B in conjunction with an in-line heater (Warner Instruments P/N 64-0102) and a Harvard Apparatus heated rodent operating table.The effluent was collected in intervals of 1 min into 96-well plates kept on ice and containing aprotinin.Samples were subsequently stored at −80°C pending analysis of glucagon content (using the Mercodia assay).
PKA activity measurements PKA activity was measured using AKAR3 as previously described [22].The cells were continuously superfused at a rate of 60 μl/min with EC4 (ESM Table 2).Alpha cells were identified as those exhibiting a response to 10 μmol/l adrenaline (epinephrine) [23].
Cytoplasmic free Ca 2+ concentration imaging For livecell cytoplasmic free Ca 2+ concentration ([Ca 2+ ] i ) imaging experiments, Gcg-GCaMP6f islets were immobilised to a poly-l-lysine-coated coverslip fixed in a custom-built imaging chamber filled with EC3 (ESM Table 2).[Ca 2+ ] i was measured in islets from mice expressing GCaMP6f in alpha cells.[Ca 2+ ] i imaging experiments were then performed using an inverted LSM 510 confocal microscope (Zeiss) controlled with ZEN Black (Zeiss), using a ×40/1.3oil immersion objective.Time-lapse images were collected every 0.98 s with a frame size of 256×256 pixels, with the bath solution (EC3) perfused at a rate of 0.4 ml/min and heated to 37°C.GCaMP6f was excited by an argon laser (488 nm) and emission was collected at 510 nm.
[Ca 2+ ] i imaging videos were analysed using the Fiji imaging processing package.The mean fluorescence (F) of each region of interest was normalised to baseline signal (F 0 ) and expressed as F/F 0 before exporting into ClampFit (version 10.7; Molecular Devices, CA, USA), where baseline was corrected and AUC was calculated.
Total internal reflection fluorescence microscopy imaging of granule mobility Cells were imaged using total internal reflection fluorescence (TIRF) microscopy (AxioObserver Z1 with a ×100/1.45objective [Carl Zeiss] and a diode-pumped solid state laser at 491 nm).For these measurements, EC1 was used (ESM Table 2).Images from cells infected with preproglucagon promoter (PPPG)-neuropeptide Y (NPY)enhanced green fluorescent protein (EGFP) virus were recorded using an electron-multiplying charge-coupled device camera (Photometrics Evolve) using ZEN blue [24].Single images of cells were acquired to measure the number of docked granules.Incoming and outgoing granules during visiting, docking and undocking were determined from multiple frame movies as described previously [25].Candidate docking or undocking events were found manually as granules that approached the TIRF field with an axial component and became laterally confined for at least two frames.We defined docking as granules that remained confined for at least 1 min.Visitors were those granules that remained for <1 min after appearing at the plasma membrane.Undocking was defined as slow movement of a previously docked granule away from its docking site [25].Granule density was calculated using a script that used the built-in 'find maxima' function in ImageJ (version 1.53c; http:// rsbweb.nih.gov/ ij) for spot detection [26].
Plasma glucose and glucagon measurements C57Bl6N mice were fasted for 5 h prior to the ITTs.GLP-1(9-36) (100 µg/ kg body weight) was injected intraperitoneally before or with insulin as indicated.The relative high dose of GLP-1  was used because of high basal levels of the peptide.In the control experiments, mice were injected with the saline solvent (154 mmol/l NaCl).Tail-vein blood glucose and glucagon levels were monitored before and during injection of insulin (0.75 U/kg body weight in PBS; Actrapid; Novo Nordisk).Plasma glucagon was measured using a glucagon ELISA (10-1281-01; Mercodia, Uppsala, Sweden; CV<7%).Blood glucose levels were measured using a glucometer (Accu-Chek; Roche).Total GLP-1 was measured by ELISA (81508; Crystal Chem, Zaandam, the Netherlands; CV<10%).
In the experiments involving the use of REMD2.59,mice were injected intraperitoneally with the antagonist (5 mg/ kg body weight in PBS) or vehicle 24 h prior to insulininduced hypoglycaemia.Following the completion of the experiment, the mice were euthanised and the pancreases were resected and weighed.The pancreases were then homogenised, sonicated in acid ethanol and stored at 4°C for glucagon extraction.

Statistical analysis
All data are reported, unless otherwise stated, as mean values ± SEM for the indicated number of experiments using islets from multiple mice.For hormone release studies, islets were isolated and pooled from 4-16 mice.These islets were then subdivided into groups of 12-20 size-matched islets.Each unique set of islets counted as an experiment with the experiments repeated on 2-4 days.Similarly, each unique group of human islets was treated as an experiment.Because the experiments were conducted over more than a decade by different investigators and using mice of different ages, glucagon secretion rates have been (for display) normalised to the mean rate of secretion at 1 mmol/l glucose (expressed as pg/islet × h).In the Ca 2+ imaging and granule docking experiment, each cell represents an experiment making sure that cells from multiple donors/animals were used.For the in vivo experiments and the perfused pancreas, each mouse count represents an experiment.For two groupings, a t test was conducted.For multiple comparisons, one-way ANOVA was conducted.All statistical tests were performed using Graphpad Prism (version 9.1.0;graphpad.com).It was ascertained that the data were normally distributed.
In separate experiments, GLP-1(7-36) produced a dosedependent stimulation of somatostatin secretion that was detectable already at 10 pmol/l (Fig. 1c).Insulin secretion was also stimulated by GLP-1  but this effect only attained statistical significance at concentrations ≥100 pmol/l.No stimulatory effects of GLP-1(9-36) (10-10,000 pmol/l) on somatostatin or insulin secretion were observed (Fig. 1d).These data make it likely that the glucagonostatic effect of GLP-1(9-36) reflects a direct action rather than an indirect (paracrine) effect on the alpha cells.However, part of the glucagonostatic effect of GLP-1(7-36) might be exerted via somatostatin and/or insulin.
Collectively, these data suggest that the Glp1r-independent effects of GLP-1(7-36) are exerted following its degradation to GLP-1(9-36) and is mediated by activation of G i/o .In addition, GLP-1(7-36) exerts Glp1r-dependent effects, which are prevented following genetic ablation of the receptor, possibly via paracrine signals originating from the beta and delta cells (Fig. 3d).

GLP-1(9-36) inhibits PKA activity by a G i/o -dependent mechanism
The GCGR can activate both stimulatory (G s ) and inhibitory (G i/o ) GTP-binding proteins [33][34][35] and may therefore mediate the PTX-sensitive effects of GLP-1 .Published RNA-seq data of mouse and human alpha and beta cells indicate that expression levels of Gcgr/GCGR are much lower in alpha cells than in beta cells [36,37].Nevertheless, GCGR immunoreactivity was detected in 34±4% of the glucagonpositive alpha cells (as calculated from the images in Fig. 4a).
Our data are in agreement with a recent report using the same antibody, the specificity of which is suggested by the loss of immunoreactivity in islets from Gcgr −/− mice [20].
We determined the GTP-binding protein coupling of human GCGR expressed in HEK293T cells.GCGR potently coupled to both G s and G i/o proteins in response to glucagon (ESM Fig. 4a-f).Notably, GLP-1(9-36) increased dissociation of the inhibitory GTP-binding protein G oA (Fig. 4b) with similar potency but lower efficacy (~10%) than glucagon (ESM Table 4).

GLP-1(9-36) depletes docked granules pool
The fact that GLP-1(9-36) remains glucagonostatic regardless of whether glucagon secretion is triggered by low glucose alone, membrane depolarisation or receptor-induced elevation of cAMP suggests it may act late in the secretory process, possibly at the level of exocytosis itself.We used total TIRF microscopy to study the mean near-membrane granule trafficking [24].Under control conditions (1 mmol/l glucose alone), the near-membrane granule density in alpha cells was ~0.5 μm −2 (Fig. 7a).GLP-1(9-36) produced a time-dependent 40% reduction of the number of docked granules, an effect that was maximal after 10 min.Given that GLP-1(9-36) inhibits glucagon secretion, it is unlikely that the reduction in the number of docked granules reflects stimulation of exocytosis.Indeed, under control conditions (1 mmol/l glucose alone), the decrease in granule density was <4%.GLP-1(9-36) instead reduced granule density by inhibiting granule docking and stimulating granule undocking, while and glucagon (red) in a mouse pancreatic islet.GCGR staining was merged with glucagon staining to test the co-localisation.GCGR and glucagon double-positive cells are indicated with white arrows.Scale bar, 10 μm.(b) Effects of increasing concentrations of glucagon or GLP-1(9-36) (logarithmic scale) on dissociation of the G oA GTPbinding protein α-subunit from GCGRs expressed in HEK293T cells using the TRUPATH biosensor platform.Effects are expressed as the ligand-induced change in BRET (relative to that in the absence of any peptides; ΔBRET, y-axis) against concentration of glucagon or GLP-1(9-36) (x-axis).Data representative of 4 and 3 replicates for glucagon and GLP-1(9-36), respectively.See also ESM Table 4.
We reasoned that high endogenous circulating levels of GLP-1(9-36) limit the glucagonostatic action of exogenous GLP-1 .We tested insulin-induced hypoglycaemia in mice pretreated with a GRA.REMD2.59 reduced basal plasma glucose by ~2 mmol/l (ESM Fig. 6d) and increased circulating glucagon by 700% (Fig. 9c,d).Notably, GLP-1(9-36) had no glucagonostatic effect in the presence of REMD2.59.We determined the relationship between plasma glucose and glucagon in the absence and presence of REMD2.59 (Fig. 9e) and obtained slopes of −24±3 ng/mmol and −80±21 ng/mmol in the absence and presence of REMD2.59,respectively (p=0.0115).This corresponds to an increase of ~225% of plasma glucagon when corrected for the change in glucose concentration.Both relationships intercepted the x-axis at ~9 mmol/l glucose, close to the normal plasma glucose concentrations in fasted mice (ESM Fig. 6b, d).Thus, REMD2.59 principally stimulates glucagon secretion under hypoglycaemic conditions.The REMD2.59induced increase in plasma glucagon correlated with a 27% reduction in pancreatic glucagon content (Fig. 9f).There was a negative correlation between pancreatic glucagon content and basal plasma glucagon (ESM Fig. 6e).
GRAs have been reported to increase circulating glucagon in vivo by hepatic hyperaminoacidaemia [52,53].We tested the effects of a cocktail of 3 and 6 mmol/l AAs.When applied at 1 mmol/l glucose, 3 mmol/l AAs resulted in a transient (5 min) stimulation of glucagon secretion.After the initial stimulation, glucagon secretion returned to the pre-stimulatory level and subsequently raising the AAs to 6 mmol/l was without stimulatory effect (ESM Fig. 6f).Plasma glucose (mmol/l)

Discussion
We show that the GLP-1 metabolite GLP-1(9-36), previously assumed to lack biological activity, exerts a strong glucagonostatic effect both in vivo and in vitro.GLP-1(9-36)'s glucagonostatic effect operates in parallel with that of GLP-1 , as illustrated schematically in Fig. 10a.According to this model, GLP-1(7-36) (at least at physiological concentrations; c.f. [42]) principally regulates glucagon secretion by paracrine mechanisms [54] resulting from activation of the GLP-1Rs in beta and delta cells.GLP-1(9-36), formed by the removal of the two N-terminal residues by DPP-4, leads to the activation of an inhibitory G protein (G i/o ) and suppression of glucagon secretion.GLP-1(9-36) has a much longer t½ in circulation than GLP-1(7-36) [55] and the two peptides  may regulate systemic metabolism with different kinetics following their release from the gut.Genetic ablation of GLP-1Rs in alpha cells results in glucose intolerance in vivo and suppression of glucagon secretion in isolated islets at low glucose concentrations [56].These observations raise the possibility the ligand GLP-1(7-36) itself, by increasing intracellular cAMP, exerts a stimulatory rather than an inhibitory effect on glucagon secretion [57].Following its degradation, the glucagonotropic effect of GLP-1(7-36) will be superseded by the inhibitory effect of the metabolite GLP-1 .The identity of the membrane receptor that mediates the latter effect remains to be conclusively established but our data suggest it is distinct from the GLP-1R.There have been multiple reports that GLP-1(9-36) exerts effects on the central nervous system, cardiovascular system, gastrointestinal tract, liver and muscle that persist in the absence of GLP-1R [55] and the existence of an alternative GLP-1R has therefore been inferred.However, attempts to identify it have failed and there is no obvious candidate in the genome [58,59].Our findings suggest that the GLP-1R-independent effects of GLP-1 may involve the activation of GCGRs.Although Gcgr is expressed at very low levels, the immunocytochemistry suggests that the protein is present in alpha cells (although below the detection level in 66% of the alpha cells).It is possible that the transcription of the gene is periodic [60] and that the short-lived mRNA (minutes/hours [61]) may thus be absent most of the time, whereas the long-lived protein (days [62]) remains present and functional.Indeed, the immunocytochemical data are supported by the functional data using a highly specific antagonist (REMD2.59)[43,44].The response to GLP-1  reported by the TRUPATH assay may seem minute.However, this assay measures the coupling between G protein-coupled receptors (GPCRs) and individual G proteins in an experimental system; under physiological conditions in alpha cells, downstream signalling cascades might amplify small signals.Accordingly, not many receptors need to be occupied to elicit the maximum inhibitory response.This concept also explains how activation of G i/o by GLP-1(9-36) supersedes the stimulatory effect of G s activation by β-adrenoceptor and GLP-1R activation.It is notable that neither the GLP-1R antagonist exendin(9-39) nor the GRA REMD2.59 had any (major) effect on glucagon secretion at 1 mmol/l glucose, making it unlikely that glucagon secretion is under significant autocrine control by glucagon or GLP-1   [63,64] from alpha cells.It remains to be elucidated exactly how the GCGRs can respond to GLP-1  in the presence of the high intra-islet glucagon levels.
Previous studies have failed to document any glucagonostatic action of GLP-1  in vivo [7][8][9], seemingly at variance with the findings reported here.However, three factors may explain this discrepancy.First, the effect is glucose-dependent and GLP-1(9-36) exerts its predominant effect under hypoglycaemic conditions.Second, GLP-1(9-36) acts by granule undocking, a process that develops over 5-10 min and its glucagonostatic effect is therefore delayed.Third, because of high circulating GLP-1 levels, glucagon secretion will be under tonic suppression in vivo.From the dose-inhibition curves established in vitro, circulating GLP-1(9-36) (~30 pmol/l) can be expected to inhibit glucagon secretion by up to 80%, making it difficult to observe any additional suppression by exogenous administration (especially under normoglycaemic conditions when glucagon secretion is already strongly reduced) (schematic Fig. 10b).Conversely, reversal of GLP-1(9-36)'s glucagonostatic effect might explain the dramatic elevation of circulating glucagon observed after pharmacological/genetic inhibition of the GCGRs [43,53,65,66], an effect that has been attributed to AA-induced stimulation of alpha cell proliferation.However, the acute effects of GRAs on circulating AAs are small (+25% [66]); AAs only transiently stimulate glucagon secretion and yet GRA treatment results in a dramatic elevation of circulating glucagon without increasing pancreatic glucagon content.Collectively, these observations militate against the idea that AA-induced alpha cell proliferation accounts for the high circulating levels of glucagon observed acutely upon administration of GRAs, an effect which we instead attribute to the removal of GLP-1(9-36)'s suppressor effect.Comparing the slopes of the relationships between plasma glucose and glucagon in the absence and presence of REMD2.59 suggests that glucagon secretion in vivo is reduced by 70%.From the dose-inhibition curves obtained in vitro we can estimate that this equates to a GLP-1(9-36) concentration of 30 pmol/l, in remarkably good agreement with the plasma GLP-1 concentration observed in vivo.We therefore propose that GLP-1(9-36) plays an important and previously unrecognised role as a systemic inhibitor of glucagon secretion (but not the only one).This concept by no means is incompatible with the finding that inhibition of GCGR signalling also results in alpha cell proliferation but we emphasise that the latter effect operates on a much longer time scale than the acute effect we now describe (days/weeks rather than minutes/hours).If anything, the marked elevation of circulating glucagon we observed following overnight pretreatment with REMD2.59 was associated with a reduction of pancreatic glucagon content.
The finding that GLP-1(9-36) acts by reducing the number of docked granules provides a simple and unifying explanation for its capacity to inhibit glucagon secretion regardless of whether it is evoked by low glucose, membrane depolarisation or a β-adrenergic agonist.The observations that both ω-agatoxin and diazoxide mimicked the effect of GLP-1(9-36) on granule docking suggest that Ca 2+ influx via P/Q-type Ca 2+ channels during alpha cell electrical activity [60] promotes granule docking in alpha cells.
In type 2 diabetes, the capacity of GLP-1(9-36) to deplete the docked pool of granules in alpha cells and suppress glucagon secretion was abolished.Type 2 diabetes is associated with elevated circulating levels of glucagon, especially when related to plasma glucose levels [67].Our data suggest that reduced capacity of GLP-1(9-36) to exert its glucagonostatic effect might contribute to this defect and exacerbate the hyperglycaemia caused by impaired insulin secretion.

Fig. 4
Fig.4 Glucagon receptors in alpha cells and their activation by GLP-1.(a) Glucagon receptor (GCGR) immunoreactivity in alpha cells.Double immunofluorescence staining of GCGR (green) and glucagon (red) in a mouse pancreatic islet.GCGR staining was merged with glucagon staining to test the co-localisation.GCGR and glucagon double-positive cells are indicated with white arrows.Scale bar, 10 μm.(b) Effects of increasing concentrations of glucagon or GLP-1(9-36) (logarithmic scale) on dissociation of the G oA GTPbinding protein α-subunit from GCGRs expressed in HEK293T cells using the TRUPATH biosensor platform.Effects are expressed as the ligand-induced change in BRET (relative to that in the absence of any peptides; ΔBRET, y-axis) against concentration of glucagon or GLP-1(9-36) (x-axis).Data representative of 4 and 3 replicates for glucagon and GLP-1, respectively.See also ESM Table4.

Fig. 5 Fig. 6 Fig. 7
Fig. 5 GLP-1(9-36) inhibits PKA activity and glucagon secretion by GCGR-dependent mechanism.(a) Effects of increasing concentrations of GLP-1(9-36) (staircase) on PKA activity in individual alpha cells in intact islets under control conditions (n=420 cells from 3 mice) and after pretreatment with PTX (n=785 cells from 3 mice).Data are mean values ± SEM.(b) PKA activity in alpha cells in response to 10 and 100 pmol/l GLP-1(9-36) in the presence of 100 nmol/l of L-168049 (n=420 cells from 3 mice).In (a, b) responses have been normalised to basal conditions prior to the addition of the agonists.(c) Box plots of changes in PKA activity in response to 10 pmol/l GLP-1(9-36) under control conditions and after pretreatment with PTX (n=785 cells from 3 mice) or in the presence of 100 nmol/l of L-168049.**p<0.01vs basal level (evaluated by Friedman ANOVA, Nemenyi post hoc test); † p<0.05 vs GLP-1(9-36) in the absence of L-168049 (Kruskal-Wallis ANOVA, Nemenyi's post hoc test).Black lines represent medians and the boxes indicate first and third quartiles.(d) As for (b) but testing the effects of 10 pmol/l and