Structural determinants of dual incretin receptor agonism by tirzepatide

Abstract

SignificanceTirzepatide is a dual agonist of the glucose-dependent insulinotropic polypeptide receptor (GIPR) and the glucagon-like peptide-1 receptor (GLP-1R), which are incretin receptors that regulate carbohydrate metabolism. This investigational agent has proven superior to selective GLP-1R agonists in clinical trials in subjects with type 2 diabetes mellitus. Intriguingly, although tirzepatide closely resembles native GIP in how it activates the GIPR, it differs markedly from GLP-1 in its activation of the GLP-1R, resulting in less agonist-induced receptor desensitization. We report how cryogenic electron microscopy and molecular dynamics simulations inform the structural basis for the unique pharmacology of tirzepatide. These studies reveal the extent to which fatty acid modification, combined with amino acid sequence, determines the mode of action of a multireceptor agonist.

Keywords: G protein coupled receptor (GPCR); GIP receptor; GLP-1 receptor; structure; tirzepatide.

Fig. 1.

Cryo-EM structure determination of the GIPR/GIP, GIPR/tirzepatide, and GLP-1R/tirzepatide. (A) Overall structures of GIPR (orange)/GIP (yellow) (3.2-Å resolution), GIPR (magenta)/tirzepatide (blue) (3.1-Å resolution), and GLP-1R (slate blue)/tirzepatide (green) (2.9-Å resolution). Additional subunits of the complexes are colored as the following: GsαiN18, salmon; Gβ, dark green; Gγ, gray; Nb35, brown; ScFv16, violet-brown. (B, C) The interaction of GIP (B) or tirzepatide (C) and the TM domain of GIPR. Residues that are involved in interactions are shown as sticks, and the residues that contribute most significant interactions are labeled. Hydrogen bonds were labeled as dashes. (D) Difference of the residue on position 7 of GIP and tirzepatide. (E) The interaction of GIP and tirzepatide with the ECD and ECL1 of the GIPR.

Fig. 2.

Differential binding of tirzepatide (TZP) versus GLP-1 at the GLP-1R. (A) The overall orientation of TZP bound to the GLP-1R (GLP-1R slate blue; Tzp, green) compared with GLP-1R (cyan)/GLP-1 (gray) (PDB: 6 × 18), GIPR (orange)/GIP (yellow), and GIPR (magenta)/Tzp (blue). (B) TZP is positioned further away from the ECL2 of the GLP-1R, and Arg299^ECL2 does not interact with TZP. Residues that are involved in significant interactions are shown as sticks. Hydrogen bonds are labeled as dashes. (C) The bulkiness of Tyr1^Tzp results in a different rotamer of Trp306^5.36, disrupting the interactions between TM5 and ECL3 of the GLP-1R.

Fig. 3.

MD simulations of the GIPR and the GLP-1R in complex with TZP. (A) Chemical structure of the lipid modification and its attachment to Lys20^Tzp of TZP. (B) Frequency of hydrogen bond interactions observed during the MD simulations by the lipid chain with the peptide sequence of TZP (Left panel) or the receptors (Right panel). Results for the GIPR are represented by the blue bars and the GLP-1R data as red. “X” denotes the absence of hydrogen bond interactions of the corresponding residues in the receptor complex. (C) The radius of gyration distribution of the lipid chain over two 500-nsec MD runs in the GIPR (blue) and GLP-1R (red) complexes, respectively. (D) Overlays of MD snapshots of TZP in GIPR (Left) and GLP-1R (Right). Ten snapshots at 100-nsec intervals from each complex system are displayed. The peptide portion of TZP is shown as orange ribbons, and the lipid chain is shown as yellow sticks.

Fig. 4.

The C20 diacid fatty acid moiety of TZP impacts incretin receptor binding affinity. Mechanistic pharmacology studies investigating receptor binding and signal transduction were performed using TZP, an analog thereof lacking the lipid moiety (TZP^ΔC20), and a derivative that also contains histidine in place of tyrosine at position 1 (TZP^ΔC20,Y1H). (A) For each ligand, competitive inhibition of [¹²⁵I]-GIP (–42) binding was determined using membranes isolated from HEK293 cells expressing the human GIPR. Binding of TZP is shown to be equivalent to that of native GIP. Removal of the C20 diacid fatty acid chain increased the affinity of the ligand, while changing tyrosine to histidine weakened binding to the receptor. (B) Ligand-induced GTPγS binding of Gα_s was performed using the GIPR-expressing membranes. Removal of the lipid moiety from TZP resulted in a modest increase in potency for inducing activation of the Gα_s versus TZP and GIP, but the TZP^ΔC20,Y1H analog showed both reduced efficacy and potency in comparison. (C) Agonist-stimulated cAMP production was measured in human GIPR expressing HEK293 cells. In line with the increase in binding affinity and potency to activate Gα_s, absence of the lipid moiety led to increased potency for stimulating GIPR-mediated cAMP accumulation, while weaker activity was observed upon replacement of the tyrosine at position 1. (D) For the GLP-1R, competitive inhibition of [¹²⁵I]-GLP-1(7-36)NH₂ binding was determined using membranes isolated from HEK293 cells expressing the human GLP-1R. Binding of both TZP and TZP^ΔC20 is shown to be ∼fivefold weaker than that of GLP-1, but changing tyrosine to histidine at position 1 restored the binding affinity to that of the native peptide. (E) Ligand-induced GTPγS binding of Gα_s was performed using the GLP-1R-expressing membranes. Compared with GLP-1, TZP is shown to be a partial agonist at stimulating Gα_s. Removal of the lipid moiety from TZP resulted in an increase in the efficacious response, with a slightly further elevation observed for the TZP^ΔC20,Y1H analog. (F) Agonist-stimulated cAMP production was measured in human GLP-1R-expressing HEK293 cells. The potency of TZP at stimulating cAMP accumulation is ∼20-fold weaker than that of GLP-1. The absence of the lipid moiety improved potency and the nonlipidated parent peptide containing the histidine displayed activity that is indistinguishable from that of GLP-1. A derivative of glucagon-like peptide-1 containing tyrosine in place of histidine at position 7 (GLP-1^H7Y) was used as a control. Data presented are representative of n ≥ 3 independent experiments. Summarized data are shown in SI Appendix, Table 2. log M; log Molar.

Fig. 5.

Biased pharmacology of TZP at the GLP-1R occurs through a composite effect on signaling efficacy by the N-terminal tyrosine and the C20 diacid fatty acid moiety. Mechanistic pharmacology studies investigating receptor binding and signal transduction were performed using TZP, an analog thereof lacking the lipid moiety (TZP^ΔC20), and a derivative that also contains histidine in place of tyrosine at position 1 (TZP^ΔC20,Y1H). Non-G protein signaling by TZP at the GLP-1R was investigated using assays for GRK2 (A) and β-arrestin (B) recruitment. In both systems, TZP is shown to be a weak, partial agonist in comparison to GLP-1. Removal of the C20 diacid fatty acid moiety improved the responses, and the absence of the lipid moiety in combination with replacing tyrosine with histidine resulted in activity that is nearly fully efficacious. (C) Ligand-induced internalization of the GLP-1R was assessed using changes in the cell surface presentation of SNAP-tagged receptor in HEK293 cells. Relative to GLP-1, TZP is shown to be a weak, partial agonist at inducing internalization of the GLP-1R. Consistent with restoring β-arrestin recruitment, the TZP^ΔC20 and TZP^ΔC20,Y1H derivatives are shown to proportionally improve ligand-induced receptor internalization. A derivative of glucagon-like peptide-1 containing tyrosine in place of histidine at position 7 (GLP-1^H7Y) was used as a control. Data presented are representative of n ≥ 3 independent experiments. Summarized data are shown in SI Appendix, Table 2. (D–L) Representative confocal images of pancreatic islets labeled with fluorescently tagged GLP-1, TZP, TZP^ΔC20,Y1H, or GIP. Red Fluorescence (or green fluorescence for GIP) was detected following incubation of islets from wild-type (D, G, J) or Glp-1r null (F, I, L) mice with 30 nM of GLP-1^AF647, TZP^AF647, TZP^{ΔC20,Y1H-AF647}, or (D-Ala2)GIP^AF488 for 30 min. (E, H, and K) An additional set of islets from wild-type mice were preincubated with 2 µM GLP-1R antagonist exendin-4_(9-39) prior to treatment with the fluorescently labeled ligands. Nuclei are stained in blue with Hoechst 33342. log M; log Molar. (Scale bars, 10 µm.)