Glucose-dependent insulinotropic polypeptide (also known as gastric inhibitory peptide, GIP), glucagon-like peptide-1 (GLP-1) and glucagon (GCG) are peptide hormones responsible for glucose homeostasis. Their cognate receptors, GIPR, GLP-1R and GCGR, belong to class B1 G protein-coupled receptor (GPCR) family. Successful application of various GLP-1 mimetics to treat type 2 diabetes mellitus (T2DM) and obesity highlights the clinical value of this group of drug targets. However, development of GIPR- and GCGR-based therapeutics has encountered drawbacks due to the complexity of physiology associated with GIP and GCG. For example, GIP stimulates insulin secretion but also increases GCG levels, while the latter has a parallel role in elevating energy expenditure and blood glucose.
It was reported that the weight loss property of most GLP-1 analogs, except for semaglutide administered subcutaneously once weekly of smoother pharmacokinetics, is hampered by the therapeutic window because of the dose-dependent side-effects. Chimeric peptides consisting of amino acids from GIP and GLP-1 were then designed to maximize their metabolic benefits. Additional consideration was given to GCG for its role in energy expenditure. Therefore, multi-targeting or unimolecular peptides possessing combinatorial agonism at GIPR, GLP-1R and GCGR have been extensively explored and more than a dozen peptides including two GIPR/GLP-1R dual agonists, ten GLP-1R/GCGR dual agonists and five GIPR/GLP-1R/GCGR triagonists have entered into clinical development. Of them, two pioneered multi-targeting agonists, tirzepatide (LY3298176) and peptide 20 (MAR423) have attracted significant attention from both academic and industrial communities. Tirzepatide is an investigational once-weekly GIPR/GLP-1R dual agonist with a profound therapeutic superiority in reducing blood glucose and body weight beyond several approved drugs such as semaglutide and dulaglutide in multiple head-to-head clinical trials. Peptide 20, a GIPR/GLP-1R/GCGR triagonist (phase 1 clinical trials completed) with balanced potency at the three receptors, is evolved from a GLP-1R/GCGR dual agonist through iterative sequence refinement and modification. It reversed glucose dysregulation without detrimental effects on metabolically healthy animals and reduced body weight, lowered fasting blood glucose, decreased glycosylated hemoglobin (HbA1C), improved glucose tolerance, and protected pancreatic islet architecture in diabetic fatty Zucker rats.
The aim of this work is to understand molecular mechanisms of the dual and triple agonism conferred by tirzepatide and peptide 20. Thus, we determine five cryo-electron microscopy (cryo-EM) structures, including GIPR and GLP-1R bound with tirzepatide and GIPR, GLP-1R and GCGR bound with peptide 20, all in complex with G s proteins at global resolutions of 3.4 Å, 3.4 Å, 3.1 Å, 3.0 Å and 3.5 Å, respectively. Integrated with pharmacological and clinical data, this work reveals the structural basis of peptide recognition by each receptor and provides important information for the design of better drugs through combinatorial agonism.
The tirzepatide–GIPR–G s, tirzepatide–GLP-1R–G s, peptide 20–GIPR–G s, peptide 20–GLP-1R–G s and peptide 20–GCGR–G s structures were determined by the single-particle cryo-EM approach with overall resolutions of 3.4 Å, 3.4 Å, 3.1 Å, 3.0 Å, and 3.5 Å, respectively. Apart from the α-helical domain of Gα s, the presence of bound tirzepatide and peptide 20, individual receptor and heterotrimeric G s in respective complex was clearly visible in all five EM maps, thereby allowing unambiguous modeling of the secondary structure and side chain orientation of all major components of the complexes.
Tirzepatide has two non-coded amino acid residues at positions 2 and 13 (Aib, α-aminoisobutyric acid), and is acylated on K20 P (P indicates that the residue belongs to the peptide) with a γGlu-2×OEG linker and C18 fatty diacid moiety. The first 30 and 29 amino acids of tirzepatide were modeled for the tirzepatide–GIPR–G s and tirzepatide–GLP-1R–G s complexes, respectively.
Peptide 20 contains two modifications: A2 P with Aib and K10 P that is covalently attached by a 16-carbon acyl chain (palmitoyl; 16:0) via a gamma carboxylate (γE spacer). The γE spacer and palmitic acid (C16:0) were well resolved in the final models of peptide 20–GLP-1R–G s and peptide 20–GCGR–G s, while only the γE spacer was modeled for peptide 20–GIPR–G s with high-resolution features. The first 30, 29, and 28 amino acids of peptide 20 were modeled for the peptide 20–GIPR–G s, peptide 20–GLP-1R–G s and peptide 20–GCGR–G s complexes, respectively.
As shown in Fig. 2a, the tirzepatide–GIPR–G s and peptide 20–GIPR–G s complex structures closely resembled that of the GIP–GIPR–G s complex with Cα root mean square deviation (RMSD) values of 0.5 and 0.4 Å, respectively. Notable conformational differences were observed in the positions of peptide C-terminal half and the surrounding extracellular loop 1 (ECL1) and extracellular domain (ECD), indicative of GIPR-associated ligand specificity. Through two mutations (M14 P L and H18 P A), the dense contacts between ECL1 (residues 194 to 211) and GIP were disrupted by peptide 20, as seen from the buried surface area that decreased from 406 Å 2 for GIP to 278 Å 2 for peptide 20. Consequently, ECL1 adopted a more relaxed conformation, making peptide 20 straighter by shifting its tip toward the transmembrane domain (TMD) core by 4.2 Å (measured by the Cα of L27 P). Similar movement was also seen for the C-terminal half of tirzepatide (2.1 Å measured by the Cα of I27 P). As far as the N terminus is concerned, GIP and tirzepatide were stabilized by multiple strong contacts with the TMD core through a common N terminus (Y1 P-A/Aib2 P-E3 P), while that of peptide 20 (H1 P-Aib2 P-Q3 P) formed weaker interactions with the TMD core by abolishing the hydrogen bond
