The peptide agonist-binding site of the glucagon-like peptide-1 (GLP-1) receptor based on site-directed mutagenesis and knowledge-based modelling

Abstract

Glucagon-like peptide-1 (7-36)amide (GLP-1) plays a central role in regulating blood sugar levels and its receptor, GLP-1R, is a target for anti-diabetic agents such as the peptide agonist drugs exenatide and liraglutide. In order to understand the molecular nature of the peptide-receptor interaction, we used site-directed mutagenesis and pharmacological profiling to highlight nine sites as being important for peptide agonist binding and/or activation. Using a knowledge-based approach, we constructed a 3D model of agonist-bound GLP-1R, basing the conformation of the N-terminal region on that of the receptor-bound NMR structure of the related peptide pituitary adenylate cyclase-activating protein (PACAP21). The relative position of the extracellular to the transmembrane (TM) domain, as well as the molecular details of the agonist-binding site itself, were found to be different from the model that was published alongside the crystal structure of the TM domain of the glucagon receptor, but were nevertheless more compatible with published mutagenesis data. Furthermore, the NMR-determined structure of a high-potency cyclic conformationally-constrained 11-residue analogue of GLP-1 was also docked into the receptor-binding site. Despite having a different main chain conformation to that seen in the PACAP21 structure, four conserved residues (equivalent to His-7, Glu-9, Ser-14 and Asp-15 in GLP-1) could be structurally aligned and made similar interactions with the receptor as their equivalents in the GLP-1-docked model, suggesting the basis of a pharmacophore for GLP-1R peptide agonists. In this way, the model not only explains current mutagenesis and molecular pharmacological data but also provides a basis for further experimental design.

Keywords: GLP-1; GPCR; diabetes; hormone; molecular model; receptor.

Figure 1. Sequence and numbering of ligands and receptor residues discussed in this paper

(A) The aligned amino acid sequences of the peptides discussed in the present paper. Note that GLP-1 starts with residue His-7*, whereas the remaining peptides begin at residue 1. (B) A schematic topological representation (generated using GPCRDB Tools, http://tools.gpcr.org/) of GLP-1R, annotated to show the regions mutated in the present study (blue: single mutations, data in main body of paper; red: remaining sites from the initial screen of double mutations, data in Supplementary Material). Residue numbers of the most interesting sites are highlighted.

Figure 2. Pharmacological analysis of two TM7 single mutant receptors compared with wild-type GLP-1R: Arg-380^7.34–Ala, Lys-383^7.37–Ala

(A) Concentration–response curves for GLP-1-induced cAMP accumulation in Flp-In HEK293 cells expressing either hGLP-1R or the mutated receptors. US, unstimulated. (B) Radioligand competition binding curves, using 75 pM ¹²⁵I-EX4(9–39) and unlabelled GLP-1 as competitor, for membranes derived from Flp-In HEK293 cells expressing either wild-type GLP-1R or mutated receptors. Replicates are in triplicate with errors representing S.E.M.

Figure 3. Pharmacological analysis of four mutant receptors compared with wild-type GLP-1R: Lys-288–Ala^4.64, Asn-300–Ala^ECL2, Trp-306–Ala^5.36 or Arg-310–Ala^5.40

(A) Concentration–response curves for GLP-1-induced cAMP accumulation in Flp-In HEK293 cells. US, unstimulated (no GLP-1 added). (B) Radioligand competition binding curves, using 75 pM ¹²⁵I-EX4(9–39) and unlabelled GLP-1 as competitor, for membranes derived from Flp-In HEK293 cells expressing either wild-type GLP-1R or one of four single site mutant receptors. Replicates are in triplicate with errors representing S.E.M.

Figure 4. A side view of the GLP-1-docked GLP-1R model from between TM5 and TM6

GLP-1R is shown in cartoon form in cyan, with the side chains of the eight residues highlighted by the mutagenesis (Table 1) shown as red sticks. The ligand is shown with its surface in yellow, with six residues highlighted by colour and single residue codes.

Figure 5. A side view of the GLP-1-docked GLP-1R model from between TM1 and TM2

GLP-1R is shown in cartoon form in cyan, with the side chains of four residues highlighted by mutagenesis in the literature (Table S1) shown as green sticks. The ligand is shown with its surface in yellow, with three residues highlighted by colour and single residue codes.

Figure 6. Views of the GLP-1R model docked with “model 11” from pdb code 2N0I

(A) View of the GLP-1R model from between TM5 and TM6 for comparison with Figure 4. (B) View of the GLP-1R model from between TM1 and TM2 for comparison with Figure 5. The ligand is the cyclic constrained synthetic 11-residue analogue of GLP-1 based on model 11 of pdb code 2N0I, and is shown as space-fill in yellow, but with four conserved residues highlighted by colour and single residue codes.