Three Distinct Epitopes on the Extracellular Face of the Glucagon Receptor Determine Specificity for the Glucagon Amino Terminus*

The glucagon and glucagon-like peptide-1 (GLP-1) receptors are homologous family B seven-transmembrane (7TM) G protein-coupled receptors, and they selectively recognize the homologous peptide hormones glucagon (29 amino acids) and GLP-1 (30–31 amino acids), respectively. The amino-terminal extracellular domain of the glucagon and GLP-1 receptors (140–150 amino acids) determines specificity for the carboxyl terminus of glucagon and GLP-1, respectively. In addition, the glucagon receptor core domain (7TM helices and connecting loops) strongly determines specificity for the glucagon amino terminus. Only 4 of 15 residues are divergent in the glucagon and GLP-1 amino termini; Ser 2 , Gln 3 , Tyr 10 , and Lys 12 in glucagon and the corresponding Ala 8 , Glu 9 , Val 16 , and Ser 18 in GLP-1. In this study, individual substitution of these four residues of glucagon with the corresponding residues of GLP-1 decreased the affinity and potency at the glucagon receptor relative to glucagon.

Glucagon and GLP-1 1 evolved from a common ancestor by a gene duplication in early vertebrate evolution, and human tissue-specific processing of their common precursor peptide, preproglucagon, generates glucagon in the pancreatic ␣-cells and GLP-1 in the intestinal L-cells (1,2). Activation of hepatic glucagon receptors (GluR) by glucagon stimulates glycogenolysis and gluconeogenesis. Activation of GLP-1 receptors (GLP-1R) on pancreatic ␤-cells by GLP-1 stimulates glucose-induced insulin secretion. Given their biological functions in glucose homeostasis, both receptors are promising targets for the treatment of type II diabetes.
GluR and GLP-1R belong to family B of the 7TM GPCRs, which includes the receptors for peptide hormones of the glucagon/PACAP superfamily: glucagon-like peptide-2 (GLP-2), glucose-dependent insulinotropic polypeptide, pituitary adenylate cyclase-activating polypeptide (PACAP), vasoactive intestinal polypeptide (VIP), growth hormone-releasing hormone, and secretin and other peptide hormones such as calcitonin, corticotrophin-releasing factor, and parathyroid hormone (PTH). The fingerprint of family B 7TM GPCRs is six conserved cysteines that form three disulfide bonds in the N-terminal extracellular domain (Nt-domain) (3). The structural constraints imposed by an identical disulfide bond pattern probably guide the Nt-domain of family B 7TM GPCRs into a common structural fold regardless of limited sequence identity (4,5). The present knowledge about the structure and arrangement of the 7TM helices is based primarily on multiple sequence alignment analyses, although a functional coupling of conserved polar residues of the PTH receptor (PTH1R) suggests the existence of a cooperative helix-helix interface between TM2 and TM7 (6,7).
Structure-activity analyses of peptides of the glucagon/ PACAP superfamily suggest that their entire length is required for optimal biological activity and that the C terminus is primarily involved in receptor binding, whereas the N terminus contains the residues involved in receptor activation (8). The N-terminally modified glucagon analog desHis 1 Glu 9 -glucagon is a potent GluR antagonist, which emphasizes the importance of His 1 and Asp 9 of glucagon in activation of GluR (9). The N termini of GLP-1 and exendin-4 are highly conserved, and the N terminus of exendin-4 is essential for activation of GLP-1R (10). Furthermore, His 7 , Gly 10 , Phe 12 , Thr 13 , and Asp 15 of the GLP-1 N terminus are important for optimal binding and activation of GLP-1R, and they are all conserved in exendin-4 (11). NMR structures of GLP-1, glucagon, and PACAP-  in lipophilic solvents or dodecylphosphocholine micelles agree that the central and C-terminal parts are ␣-helical, often with a central distortion of the helix geometry, whereas the N terminus is a flexible random coil (12)(13)(14). Interestingly, the N terminus of receptor-bound PACAP-(1-21) forms a specific ␤-coil structure, and the PACAP N terminus is important for receptor activation (8,14). The N termini of glucagon and PACAP are highly conserved, and thus the glucagon N termi-nus may form a PACAP-like structure upon binding to GluR.
The isolated Nt-domain of family B 7TM GPCRs is sufficient for low affinity ligand binding, and it is a critical determinant of ligand selectivity (3,4,15). The Nt-domain of GLP-1R binds exendin-4-(9 -39) with high affinity, and therefore either the C-terminal extension of exendin-4 (the Trp cage) or divergent residues in exendin-4 and GLP-1 increase the affinity of exendin-4 at the Nt-domain of the GLP-1R relative to GLP-1 (16). Additional interactions with the extracellular loops (ECL) and the extracellular end of the 7TM helices may explain the high affinity ligand binding of intact receptors and provide additional determinants of ligand selectivity (17)(18)(19)(20).
The molecular information about receptor-ligand complexes of family B 7TM GPCRs is limited. A two-site binding model has been proposed for the ligand interaction with PTH1R, in which the PTH C terminus interacts with the PTH1R Ntdomain and the PTH N terminus interacts with the PTH1R core domain (21). Peptides of the glucagon/PACAP superfamily may follow a similar binding mechanism (14,22). The conserved Asp 198 in the boundary between TM2 and ECL1 of GLP-1R is important to maintain the binding site for the GLP-1 N terminus (23). Specifically, Asp 3 in the N terminus of VIP and secretin interacts with positively charged residues in the extracellular end of TM2 of VPAC 1 (VIP receptor) and the secretin receptor, respectively (24,25). The corresponding Gln 3 in the N terminus of glucagon most likely interacts with the extracellular end of TM2 of GluR (26). In addition, p-benzoyl-L-phenylalanine in position 22 or 26 of the secretin C terminus cross-links to specific residues in the Nt-domain of the secretin receptor (27).
Family B 7TM GPCRs selectively bind their natural ligands with high affinity, although they may bind homologous ligands with low affinity. GLP-1R binds GLP-1 with high affinity and glucagon with low affinity, and glucagon is a low potency full agonist of GLP-1R (22). The GLP-1R Nt-domain defines almost completely the glucagon/GLP-1 selectivity profile of GLP-1R by selective recognition of the GLP-1 C terminus, and the GLP-1R core domain is not important for glucagon/GLP-1 selectivity (22). In contrast, GluR has a very strong glucagon/GLP-1 selectivity profile and does not cross-react with GLP-1. The GluR Nt-domain selectively recognizes the glucagon C terminus, and the GluR core domain strongly determines specificity for the glucagon N terminus. These results encouraged a search for the selectivity determinants in the GluR core domain. Singlesubstituted glucagon analogs addressing four divergent residues in the glucagon and GLP-1 N termini were analyzed for their ability to bind and activate GluR. Loss-of-function mutations in glucagon were rescued by gain-of-function mutations in GluR. The results show that three distinct epitopes of the GluR core domain determine specificity for the N terminus of glucagon.
The radioligand 125 I-Glucagon (2.2 Ci/mol) was prepared by the chloramine-T method, and 125 I-GLP-1 (2.2 Ci/mol) was prepared by the lactoperoxidase method (29). Both radioligands were purified by reverse-phase high pressure liquid chromatography.
Cell Culture and Transient Receptor Expression-HEK293 cells were maintained in Dulbecco's modified Eagle' medium supplemented with 10% fetal bovine serum (Invitrogen) and penicillin/streptomycin (90 units/ml and 90 g/ml, respectively). Cells were seeded in T75 flasks, transfected with 9 g of DNA using the FuGene TM transfection reagent (Roche Molecular Biochemicals), harvested 24 h after transfection, and used for plasma membrane preparations or applied directly to functional experiments.
Functional Assay-HEK293 cells transiently expressing the desired wild type or chimeric receptor were harvested and resuspended in assay buffer (Flashplate®; PerkinElmer Life Sciences) to a cell density of 1.8 ϫ 10 6 cells/ml. Peptides were diluted in phosphate-buffered saline with 0.02% Tween 20. Cells (50 l) and peptides (50 l) were mixed in 96-well Flashplates® (PerkinElmer Life Sciences), gently agitated for 5 min, and incubated for 25 min at room temperature. The resulting intracellular level of cAMP was measured according to the supplier's manual, and data were analyzed by nonlinear regression analysis using Prism®, (GraphPad Software, Inc.).
Binding Assay-Binding experiments were performed as previously described (22). Briefly, a freshly thawed plasma membrane preparation of HEK293 cells transiently expressing the desired wild type or chimeric receptor was incubated for 2 h at 30°C in the presence of peptide and tracer. Bound and unbound peptide/radioligand were separated, and filters were washed twice in cold incubation buffer and left to dry. Finally, the amount of bound radioligand was determined using a Packard ␥-counter, and data were analyzed by nonlinear regression analysis using Prism®, (GraphPad Software, Inc.).  (Table I). Whole cells transiently expressing the human GLP-1R gave a functional response with half-maximal stimulation (EC 50 ) of the adenylate cyclase at 12 pM GLP-1, and the GLP-1 analogs were full agonists and equipotent with GLP-1 (Table I). In glucagon, Ser 2 , Gln 3 , Tyr 10 , and Lys 12 were substituted with the corresponding residues of GLP-1 (Fig. 1). The glucagon analogs were analyzed for their ability to bind and activate GluR. In competition binding, glucagon displaced 125 I-glucagon from GluR with an IC 50 value of 2.1 nM, and the glucagon analogs displaced 125 I-glucagon with significantly higher IC 50 values ( Fig. 1A and Table I). Whole cells transiently expressing the human GluR, responded functionally with half-maximal stimulation at 11 pM of glucagon, and the potencies of the glucagon analogs were significantly lower ( Fig. 1B and Table I). The results showed that the individual substitutions of Ala 8 , Glu 9 , Val 16 , and Ser 18 in the GLP-1 N terminus with the corresponding glucagon residues had only subtle effects on affinity or potency at GLP-1R relative to GLP-1. Conversely, the individual substitutions of Ser 2 , Gln 3 , Tyr 10 , and Ser 12 in the glucagon N terminus with the corresponding GLP-1 residues decreased the affinity and potency at GluR relative to glucagon. Chimera A-The chimeric receptor, chimera A, consists of the GluR Nt-domain and the GLP-1R core domain (Fig. 1) (22). The single-substituted glucagon analogs were analyzed for their ability to bind and activate chimera A. In competition binding, glucagon displaced 125 I-glucagon from chimera A with an IC 50 value of 1.2 nM (Fig. 1C and Table I). The IC 50 values of Ala 2 -glucagon and Glu 3 -glucagon were lower than that of glucagon, whereas the IC 50 values of Val 10 -glucagon and Ser 12glucagon were similar to that of glucagon ( Fig. 1C and Table I).

The Glucagon and GLP-1 Receptors-In
In functional experiments with chimera A, all of the glucagon analogs were equipotent with glucagon ( Fig. 1D and Table I).
The results obtained with GluR and chimera A showed that substitution of the GluR core domain with the GLP-1R core domain increased the affinity and potency of Ala 2 -glucagon, Glu 3 -glucagon, Val 10 -glucagon, and Ser 12 -glucagon relative to glucagon.
Dissection of the Glucagon Receptor Core Domain-Small segments of the GluR core domain were substituted with the corresponding segments of GLP-1R. Ala 2 -glucagon, Glu 3 -glucagon, and Ser 12 -glucagon were analyzed for their ability to bind and activate three chimeric receptors: ChTM2, ChECL2, and ChECL3 (Fig. 1). In ChTM2, the extracellular end of TM2 of GluR was substituted with the corresponding segment of GLP-1R. In competition binding, glucagon displaced 125 I-glucagon from ChTM2 with an IC 50 value of 1.4 nM (Fig. 1E and Table II). The IC 50 value of Glu 3 -glucagon was lower than that of glucagon, whereas the IC 50 values of Ala 2 -glucagon and Ser 12 -glucagon were higher than that of glucagon. On whole cells transiently expressing ChTM2, Glu 3 -glucagon was equipotent with glucagon, whereas Ala 2 -glucagon and Ser 12 -glucagon were less potent than glucagon ( Fig. 1F and Table II). The results obtained with GluR and ChTM2 showed that substitution of the extracellular end of TM2 of GluR with the corresponding segment of GLP-1R increased the affinity and potency of Glu 3glucagon relative to glucagon, Ala 2 -glucagon, and Ser 12 -glucagon. In addition, the affinity but not the potency of Ala 2 -glucagon had increased relative to glucagon and Ser 12 -glucagon.
In ChECL2, the second extracellular loop and the extracellular ends of TM4 and TM5 were substituted with the corresponding segments of GLP-1R. In competition binding, glucagon displaced 125 I-glucagon from ChECL2 with an IC 50 value of 3.1 nM (Fig. 1G and Table II). The IC 50 value of Ser 12 -glucagon was similar to that of glucagon, whereas the IC 50 values of Ala 2 -glucagon and Glu 3 -glucagon were higher than that of glucagon. On whole cells transiently expressing ChECL2, Ser 12glucagon was equipotent with glucagon, whereas Ala 2 -glucagon and Glu 3 -glucagon were less potent than glucagon ( Fig. 1H and Table II). The results showed that substitution of the second extracellular loop and the extracellular ends of TM4 and TM5 with the corresponding segments of GLP-1R specifically increased the affinity and potency of Ser 12 -glucagon relative to glucagon, Ala 2 -glucagon, and Glu 3 -glucagon.
In ChECL3, the third extracellular loop and the extracellular ends of TM6 and TM7 were substituted with the corresponding segments of GLP-1R. In competition binding, glucagon displaced 125 I-glucagon from ChECL3 with an IC 50 value of 1.9 nM (Fig. 1I and Table II). The IC 50 value of Ala 2 -glucagon was lower than that of glucagon, whereas the IC 50 values of Glu 3glucagon and Ser 12 -glucagon were higher than that of glucagon. On whole cells transiently expressing ChECL3, Ala 2 -glucagon was equipotent with glucagon, whereas Glu 3 -glucagon and Ser 12 -glucagon were less potent than glucagon (Fig. 1J and Table II). The results showed that substitution of the third extracellular loop and the extracellular ends of TM6 and TM7 with the corresponding segments of GLP-1R specifically increased the affinity of Ala 2 -glucagon relative to glucagon, Glu 3glucagon, and Ser 12 -glucagon. In functional experiments, the potency of Ala 2 -glucagon and to a lesser extent Ser 12 -glucagon had increased relative to glucagon and Glu 3 -glucagon.
We were unable to generate a chimeric receptor that specifically increased the affinity and potency of Val 10 -glucagon; therefore, Val 10   Ala 2 -glucagon was slightly higher than that of glucagon ( Fig.  2C and Table II). In functional experiments with whole cells transiently expressing the Asp 385 -Glu mutant, the potency of Ala 2 -glucagon was slightly higher than that of glucagon ( Fig.  2D and Table II). With the other GluR point mutants, the affinity and potency of Ala 2 -glucagon were lower than the affinity and potency of glucagon, in a manner similar to GluR (Table II). This is illustrated by the Ser 379 3 Phe mutant (Fig.  2, A and B). Subsequently, the Asp 385 3 Glu and Ser 379 3 Phe mutants were analyzed in binding and functional experiments with Glu 3 -glucagon and Ser 12 -glucagon. At both mutants, the affinity and potency of Glu 3 -glucagon and Ser 12 -glucagon were significantly lower than the affinity and potency of glucagon.
The results showed that the point mutation Asp 385 3 Glu in the extracellular end of TM7 specifically increased the affinity and potency of Ala 2 -glucagon relative to glucagon, Glu 3 -glucagon, and Ser 12 -glucagon.

DISCUSSION
The glucagon analogs bound and activated GluR with lower affinity and potency than glucagon. Substitution of the entire GluR core domain with the GLP-1R core domain rescued the affinity and potency of all of the glucagon analogs relative to glucagon. In addition, the corresponding GLP-1 analogs bound and activated GLP-1R with the same affinity and potency as GLP-1. Apparently, the GluR core domain selectively recognized Ser 2 , Gln 3 , Tyr 10 , and Ser 12 of the glucagon N terminus and discriminated the corresponding residues of the GLP-1 N terminus. In contrast, the GLP-1R core domain (GLP-1R and chimera A) potently accommodated both the GLP-1 and glucagon N termini, although the substituted residues in the GLP-1 and glucagon analogs may interact differently with the GLP-1R core domain than the corresponding residues of native GLP-1 and glucagon.
Dissection of the GluR core domain identified three distinct epitopes of the GLP-1R core domain that rescued the affinity and potency of specific glucagon analogs. The extracellular end of TM2 (ChTM2) rescued Glu 3 -glucagon, ECL2 and the proximal segments of TM4 and TM5 (ChECL2) rescued Ser 12 -glucagon, and ECL3 and the proximal segments of TM6 and TM7 (ChECL3) rescued Ala 2 -glucagon. ChTM2, ChECL2, and ChECL3 did not compromise glucagon binding or potency, which confirmed the structural integrity of the chimeric receptors. Two nonexclusive explanations may account for these results: 1) the GLP-1R segments of ChTM2, ChECL2, and ChECL3 interact directly with Glu 3 , Ser 12 , and Ala 2 , respectively, and/or 2) the GLP-1R segments of ChTM2, ChECL2, and ChECL3 are required to maintain a local binding site conformation that accommodates the interaction with Glu 3 , Ser 12 , and Ala 2 , respectively. Nevertheless, it is difficult to explain the results without considering the proximity of the GLP-1R segments in ChTM2, ChECL2, and ChECL3 with Glu 3 , Ser 12 , and Ala 2 of the glucagon analogs, respectively. In addition, the corresponding segments of GluR define the strong glucagon/ GLP-1 selectivity profile of the GluR core domain by selective recognition of the glucagon N terminus.
Most family B 7TM GPCRs have two positively charged residues in TM2, whereas GluR has only one and a neutral hydrophobic residue in place of the other (Lys 187 and Ile 194 in human GluR). Furthermore, the glucagon/PACAP superfamily peptides have either Asp or Glu in position 3 except glucagon, which has a Gln. The K187R/I194K GluR mutant rescued both affinity and potency of Asp 3 -glucagon relative to glucagon (26). The additional positive charge of the K187R/I194K GluR mutant and the chimeric receptor ChTM2 probably accommodates the extra negative charge of Asp 3 -glucagon and Glu 3 -glucagon, respectively. In fact, Glu 3 -glucagon bound ChTM2 with higher affinity than glucagon but was equipotent with glucagon in functional experiments with ChTM2. The small discrepancy in affinity versus potency suggests that Glu 3 of Glu 3 -glucagon provides a binding determinant for interaction with ChTM2 that is not equally favorable for activation of ChTM2. However, the discrepancy is small compared with the total rescue of Glu 3 -glucagon by ChTM2. The corresponding Asp 3 of secretin and VIP interact with the conserved Arg and Lys in TM2 of the secretin and VPAC 1 receptor, respectively, and the interaction with Arg in the center of TM2 is important for receptor activation (24,25). Site-directed mutagenesis of a highly conserved His in the cytoplasmic end of TM2 leads to constitutive activity of several family B 7TM GPCRs (30 -32). Collectively, it appears that TM2 is important for agonist binding and activation of family B 7TM GPCRs. Ile 194 in TM2 of GluR may serve as a selectivity determinant that prevents access of homologous peptides to the activation site of GluR.
The combined analyses of GluR mutants and single-substituted glucagon analogs demonstrated the functional significance of the correlated substitution of residues in glucagon (Gln 3 ) and GluR (Ile 194 ). Glucagon is highly conserved during evolution, and specifically position 3 is occupied by Gln in vertebrates except bony fish, where position 3 is occupied by either Gln, Asp, or Glu. In bony fish, glucagon and GLP-1 have overlapping biological activities, and GLP-1 acquired the incretin function after the divergence of bony fish and mammals (33). Specificity of receptor-ligand pairs most likely evolved to ensure distinct physiological functions, and therefore the change of selection pressure on position 3 of glucagon may reflect the divergence of the biological activities of glucagon and GLP-1.
Eight divergent residues in the region defined by the GLP-1R segment of ChECL3 was investigated by site-directed mutagenesis of GluR. Only the substitution of Asp 385 with the corresponding Glu of GLP-1R rescued the affinity and potency of Ala 2 -glucagon without disturbing the pharmacological profile of the other glucagon analogs relative to glucagon. It is difficult to explain the rescue of affinity of Ala 2 -glucagon by a direct interaction with Glu 385 in the Asp 385 3 Glu GluR mutants. It seems more likely that the Asp 385 3 Glu mutation stabilized a local binding site conformation that preferably interacted with Ala 2 -glucagon. Nevertheless the results strongly suggest proximity between Ala 2 of Ala 2 -glucagon and the extracellular end of TM7 of the GluR mutant Asp 385 3 Glu.
The dissection of the GluR core domain provides three constraints that orient the glucagon N terminus with respect to the structural elements of the GluR core domain. Given the structural integrity of the GluR mutants in this study, we suggest a binding model in which Ser 2 of glucagon is close to Asp 385 of TM7, Gln 3 of glucagon is close to Ile 194 on TM2, and FIG. 3. Binding model of the glucagon N terminus and the GluR core domain. Ser 2 of glucagon is close to the extracellular end of TM7 of GluR, Gln 3 is close to the extracellular end of TM2, and Lys 12 is close to ECL2 and/or the proximal helices TM4 and/or TM5. Lys 12 is close to ECL2 and/or proximal segments of TM4 and/or TM5 (Fig. 3). The proximity of Ser 2 with TM7 and Gln 3 with TM2 is consistent with the potential helix-helix interface between TM2 and TM7 of PTH1R. Accordingly, the high level of amino acid identity of the predicted 7TM helices of family B 7TM GPCRs probably reflects an arrangement into a conserved helical bundle structure in which TM2 is close to TM7. Proximity of TM2 and TM7 is consistent with a rhodopsin-like arrangement of the 7TM helices, although the orientation may be either clockwise or counterclockwise (34).
The position of ECL2 relative to the extracellular end of TM3 is probably constrained by a conserved disulfide bond of many family A and B 7TM GPCRs, and ECL2 is involved in peptideagonist binding of family A 7TM GPCRs (34 -38). ECL2 of the secretin and VPAC 2 receptors has been implicated in ligand binding, and specifically four residues in the N-terminal half of ECL2 contribute significantly to the secretin/VIP selectivity profile of the secretin receptor (17,20). On the basis of the results presented here, we will attempt to identity the specific residue(s) in ECL2, TM4, and/or TM5 of GluR that determines specificity for Lys 12 of glucagon.