The major determinant of exendin-4/glucagon-like peptide 1 differential affinity at the rat glucagon-like peptide 1 receptor N-terminal domain is a hydrogen bond from SER-32 of exendin-4

Abstract

Background and purpose: Exendin-4 (exenatide, Ex4) is a high-affinity peptide agonist at the glucagon-like peptide-1 receptor (GLP-1R), which has been approved as a treatment for type 2 diabetes. Part of the drug/hormone binding site was described in the crystal structures of both GLP-1 and Ex4 bound to the isolated N-terminal domain (NTD) of GLP-1R. However, these structures do not account for the large difference in affinity between GLP-1 and Ex4 at this isolated domain, or for the published role of the C-terminal extension of Ex4. Our aim was to clarify the pharmacology of GLP-1R in the context of these new structural data.

Experimental approach: The affinities of GLP-1, Ex4 and various analogues were measured at human and rat GLP-1R (hGLP-1R and rGLP-1R, respectively) and various receptor variants. Molecular dynamics coupled with in silico mutagenesis were used to model and interpret the data.

Key results: The membrane-tethered NTD of hGLP-1R displayed similar affinity for GLP-1 and Ex4 in sharp contrast to previous studies using the soluble isolated domain. The selectivity at rGLP-1R for Ex4(9-39) over Ex4(9-30) was due to Ser-32 in the ligand. While this selectivity was not observed at hGLP-1R, it was regained when Glu-68 of hGLP-1R was mutated to Asp.

Conclusions and implications: GLP-1 and Ex4 bind to the NTD of hGLP-1R with similar affinity. A hydrogen bond between Ser32 of Ex4 and Asp-68 of rGLP-1R, which is not formed with Glu-68 of hGLP-1R, is responsible for the improved affinity of Ex4 at the rat receptor.

Figure 1

Competition radioligand binding experiments using [¹²⁵I]-Ex4(9–39) as the tracer and either Ex4(9–39) or Ex4(9–30) as the competing ligands, at (A) hGLP-1R, (B) rGLP-1R, (C) hNT-TM1 and (D) rNT-TM1. The panels each represent a typical example of one of three independent experiments. The receptors derived from rat are highly sensitive to the removal of the C-terminal extension of Ex4, while those from human are not.

Figure 2

Competition radioligand binding experiments at the rGLP-1R using [¹²⁵I]-Ex4(9–39) as the tracer and either Ex4(9–39), Ex4(9–30), Ala32-Ex4(9–39) or Ala33-Ex4(9–39) as the competing ligands. Note that the curves for Ex4(9–39) and Ala33(Ex4(9–39) are overlaid. Curves represent a typical example of one of three independent experiments. Mean pIC₅₀ for Ala33–Ex4(9–39) is 8.01 ± 0.14 (n= 3); see Table 2 for other pIC₅₀ values. Disruption of the putative Trp-cage, via removal of the side chain hydroxyl of Ser-33*, had no effect upon Ex4(9–39) affinity, whereas removal of the Ser-32* hydroxyl reduces the affinity to that of the C-terminally truncated peptide.

Figure 4

Competition radioligand binding experiments using [¹²⁵I]-Ex4(9–39) as the tracer and either Ex4(9–39) (circles) or Ala32–Ex4(9–39) (triangles) as the competing ligands at (A) hGLP-1R and (B) hGLP-1R Glu68–Asp. The panels each represent a typical example of one of three independent experiments. The replacement of the native glutamate in hGLP-1R with Asp confers the ability to select between Ex4(9–39) and Ala32–Ex(9–39).

Figure 3

Competition radioligand binding experiments using [¹²⁵I]-Ex4(9–39) as the tracer and either Ex4(9–39) or Ala32–Ex4(9–39) as the competing ligands at (A) rGLP-1R Asp68–Glu and (B) rGLP-1R Asp68–Ala. Curves represent a typical example of one of three independent experiments. Removal of the hydrogen-bonding capability of Asp68 by replacing it with Ala abolishes the ability of rGLP-1R to select between the two ligands (compare with Figure 1B). The ability to select between the peptides is also absent in the Asp68–Glu mutant, suggesting that the glutamic acid side chain cannot interact with Ser-32*.

Figure 6

Ribbon diagrams illustrating snapshots of the structure of Ex4(9–39) bound to hGLP-1R taken from different points during molecular dynamics simulations. Ex4(9–39) is shown in green, GLP-1R is shown in orange and the atoms of Ser32* of Ex4(9–39) and Glu/Asp68 from GLP-1R are shown as ball and sticks coloured by atom type. Hydrogen bond formation between these groups is indicated (see boxed panels for close-up views). In the wild-type hGLP-1R, Ser32* and Glu68 spend a relatively short amount of time at a hydrogen bonding distance, for example, (A) while, instead, the majority of time is spent over 4 Å apart, for example, (B), a distance too great to allow a hydrogen bond. In Glu68–Asp hGLP-1R, Ser32* and Asp68 have a high probability of forming a distance ideal for hydrogen bond formation, for example (C).

Figure 5

Histogram chart showing calculations from molecular dynamics simulations showing the probability (normalized so that the integral between 0 and infinity is 1) that the side chain oxygen of Ser-32* of Ex4(9–39) and the closest side chain oxygen of Asp/Glu68 of GLP-1R are at a given distance. A peak at 3 Å indicates the atoms are at an ideal distance for hydrogen bond formation to occur. Upper panel: probabilities of distances between atoms of Ser-32* in Ex4(9–39) and Glu68 in GLP-1R. Lower panel: probabilities of distances between atoms of Ser32* in Ex4(9–39) and Asp68 in GLP-1R.

Figure 7

A model for the binding of Ex4 (left panels, A–C) and GLP-1 (right panels, D–F) to either the fully isolated/soluble NTD of GLP-1R expressed in Escherichia coli (A, D), the isolated NTD tethered to the membrane, expressed in HEK-293 cells (NT-TM1, B and E) or the full-length receptor expressed in HEK-293 cells (C, F). The helical structure of the peptide ligands is critical for their high-affinity interaction with the NTD via the ‘H’ interaction. (A) A cartoon representation of the binding of Ex4 (helix) with the isolated NTD of GLP-1R based on the crystal structure. The C-terminal half of the peptide interacts with the NTD, while the high helical propensity of Ex4 enables its helical structure to remain intact, despite the absence of interactions with the N-terminal region of the ligand. However, the binding of GLP-1 with the isolated NTD of GLP-1R occurs with much lower affinity because the lower helical propensity of GLP-1, coupled with the absence of interactions with the N-terminal region of the ligand, results in the more frequent unwinding of the helix, loss of the ‘H’ interaction and consequent dissociation of the ligand (D). Hence, there is a large affinity difference between Ex4 and GLP-1 at the isolated NTD (200- to 400-fold), which reflects the difference in helical propensity rather than the relative strength of the ‘H’ interaction. However, the two ligands bind with much more similar affinity to the isolated NTD tethered to the plasma membrane (B, E). We speculate that this is due to the close proximity of the membrane, which stabilizes the N-terminal region of GLP-1, eliminating the difference in helical propensity, and thus enabling the C-terminal helix to remain intact and interact with the NTD. The addition of the core domain of the receptor (C, F) results in additional interactions with the N-terminal region of the ligands, the ‘N’ interaction, which is stronger for GLP-1 than for Ex4. A third interaction, termed ‘Ex’ (white asterisk), represents a specific interaction between Ser-32* in the C-terminal region of Ex4 and the NTD of rGLP-1R.