Discovery of a functionally selective serotonin receptor (5-HT1AR) agonist for the treatment of pain

Abstract

The heterotrimeric G protein-coupled serotonin receptor 5-HT_1A receptor (5-HT_1AR) mediates antinociception and may serve as a valuable target for the treatment of pain. Starting from a chemical library, we evolved ST171, a bitopic 5-HT_1AR agonist that revealed highly potent and functionally selective G_i/o signaling without G_s activation and marginal β-arrestin recruitment. ST171 is effective in acute and chronic pain models. Cryo-electron microscopy structures of ST171 bound to 5-HT_1AR in complex with the G_i protein compared to the canonical agonist befiradol bound to complexes of 5-HT_1AR with G_i or G_s revealed that the ligands occupy different exo-sites. The individual binding poses are associated with ligand-specific receptor conformations that were further studied by molecular dynamics simulations, allowing us to better understand ligand bias, a phenomenon that may be crucial to the discovery of more effective and safe G protein-coupled receptor drugs.

Fig. 1.. Discovery of ST171.

(A) Chemical structures of serotonin and befiradol. (B) In-house library investigation of 5-HT_1AR affinity using radioligand binding studies revealing ST162 that was further developed into its analog ST171. (C) Comparison of the binding affinities of serotonin, befiradol, and ST171 at the 5-HT_1AR {displacement of [³H]WAY100635}, 5-HT_2AR {displacement of [³H]ketanserin}, and 5-HT₆R {displacement of [³H]LSD}. Data are shown with ±SEM of 4 to 11 independent experiments.

Fig. 2.. ST171 selectively activates G_i/o proteins via the 5-HT_1AR.

(A) An inhibition of 5-HT_1AR–mediated cAMP accumulation is observed for serotonin, befiradol, and ST171. The stimulation of cAMP accumulation in the presence of PTX suggests an additional G_s response for befiradol and serotonin (serotonin: EC₅₀ = 52 ± 6 nM; befiradol: EC₅₀ = 25 ± 11 nM) but not for ST171. Data were obtained in three to five independent experiments with a CAMYEL assay in CHO-K1 cells. (B) Inhibition of cAMP accumulation by serotonin, befiradol, and ST171 in HEK293A cells deficient of G_s proteins and (C) accumulation of cAMP by serotonin and befiradol but not ST171 in HEK293A cells deficient of G_i/o proteins. Data were obtained with a GloSensor in n = 6 experiments, each performed in triplicates. (D) ST171 inhibits cAMP accumulation caused by 1 μM serotonin in HEK293AΔG_i/o cells and thus demonstrates functional antagonism for the activation of G_s proteins (IC₅₀ = 3.9 ± 0.8 nM). Data were obtained with the GloSensor in n = 3 independent experiments, each performed as triplicates. (E) β-Arrestin 1 and (F) β-arrestin 2 recruitment is monitored by bystander BRET in HEK293T cells with elevated GRK2 levels [n = 4 to 6 (β-arrestin 1) or 11 to 14 (β-arrestin 2) independent experiments]. (G) ST171 is a functional antagonist (IC₅₀ = 3.9 ± 0.5 nM) for 5-HT_1AR–mediated β-arrestin 2 recruitment stimulated with serotonin (1 μM, ~EC₈₀). Data were obtained with the PathHunter assay in the absence of GRK2 in three independent experiments, each performed in triplicates. All data were normalized to the maximum response of serotonin and are indicated as the means ± SEM of n independent experiments.

Fig. 3.. ST171 is analgesic in acute pain tests and antihyperalgesic in mouse models of chronic (neuropathic and inflammatory) pain.

(A) In vivo efficacy of ST171 and befiradol in the von Frey assay. (B to D) Antinociceptive effects of ST171 and befiradol in different thermal assays: 50°C tail flick (B), Hargreaves (C), and 52°C hot plate (D). (E) ST171 and befiradol have no significant antinociceptive effect in 5-HT_1AR KO mice in the hot plate test. (F) Effect of increasing doses of ST171 and befiradol on motor performances in the rotarod assay. (G) Antihyperalgesic effects of ST171 and befiradol in the CFA-induced inflammatory pain model. (H) Antiallodynic effects of ST171 and befiradol in the SNI model of neuropathic pain. All data represent the means ± SD of 5 to 10 animals. The effects of ST171 and befiradol were compared to their respective vehicle in (A) to (D) and compared to SNI in (H) using Student’s t test with *P < 0.05, **P < 0.01, ***P < 0.005, and ****P < 0.001, ns not significant. Differences between groups and doses were measured using the one-way ANOVA in (F) (Tukey’s post hoc test) and two-way ANOVA in (E) (Sidak’s post hoc test) and (G) (Tukey’s post hoc tests).

Fig. 4.. Cryo-EM structures reveal the overall architecture of ST171 bound to 5-HT_1AR-G_i1 and befiradol bound to 5-HT_1AR-G_i1 and 5-HT_1AR-G_s protein complexes.

Cryo-EM maps (A to C) and models (D to F) of [(A) and (D)] the ST171-bound 5-HT_1AR-G_i1 protein complex with two cholesterol molecules and PI4P (light pink), [(B) and (E)] the befiradol-bound 5-HT_1AR-G_i1 protein complex, and [(C) and (F)] the befiradol-bound 5-HT_1AR-G_s protein complex, each with one cholesterol molecule and PI4P (light pink). (G to I) Zoom-in of the densities corresponding to the ligands viewed from the side [left, as in (D) to (F)] and top of the receptor (right, 90° rotation) view for (G) ST171, (H) befiradol in the 5-HT_1AR complex with G_i, and (I) befiradol in the complex of 5-HT_1AR with G_s.

Fig. 5.. ST171 and befiradol engage the 5-HT_1AR-G_i1 complex via distinct extended binding pockets located close to TM2 and TM7.

Top view of (A) ST171 and (B) befiradol binding poses and close-up comparison to the serotonin binding mode (PDB: 7E2Y; orange) in the orthosteric pocket. (C and D) Surface representation of ST171-bound (light green) and befiradol-bound (gray) 5-HT_1AR illustrates the different extended binding pockets addressed by the ligands.

Fig. 6.. The benzoxazinone of ST171 contributes to the ligand-specific stabilization of the 5-HT_1AR.

(A) Conformation of Y96^2.64 in the experimental cryo-EM structures of the befiradol-bound (PDB: 8PKM), serotonin-bound (PDB: 7E2Y), and ST171-bound (PDB: 8PJK) 5-HT_1AR-G_i complexes. (B) Progression of the torsion angle -C_α-C_β-C_γ-C_δ- of the Tyr96^2.64 side chain over the course of 2 μs for one of five independent simulations of the respective ligand-bound ternary 5-HT_1AR complexes (simulation restraint on the G_i protein interface). (C) The cryo-EM structure of 5-HT_1AR in complex with ST171 suggests the presence of a hydrogen bond between the amide group of Q97^2.65 and the NH group of W387^7.40. (D) Inward shift of TM2 and outward shift of TM1 and TM7 of the ST171-bound 5-HT_1AR-G_i structure (light green) in comparison to the befiradol-bound structure (light gray) (extracellular view) with highlighted residues Q97^2.65 and W387^7.40. (E) Progression of the distance between Q97^2.65 and W387^7.40 is shown over the course of 2 μs for one of five independent simulation runs of the respective ligand-bound 5-HT_1AR as a ternary complex (restraint on the G_i protein interface).

Fig. 7.. Befiradol distinctly engages the extended binding pocket depending on the coupling partner.

(A) Top view of the befiradol binding pose within the G_s protein–bound 5-HT_1AR complex and (B) comparison to the binding pose in the 5-HT_1AR-G_i1 protein complex. (C) The extended binding pocket residues F112^3.28 and M92^2.60 exhibit different conformations in the cryo-EM structure with ST171 compared to the structures with befiradol bound to 5-HT_1AR in complex with the G_i and G_s proteins, respectively. (D to G) Mutation F112^3.28W alone and in combination with M92^2.60F substantially decreases befiradol’s potency for the activation of G_i- and G_s-mediated signaling and β-arrestin 2 recruitment at the 5-HT_1AR, while the signaling profile of ST171 is not affected. For both ligands, the relative efficacies remain unchanged compared to serotonin. Data were obtained with the cAMP GloSensor measuring the inhibition of forskolin-promoted cAMP accumulation in (D) parental HEK293A and (E) HEK cells deficient of G_s proteins. (F) In HEK cells deficient in G_i/o proteins, an accumulation of cAMP was observed with befiradol. (G) β-Arrestin 2 recruitment was monitored by BRET in HEK293T cells with elevated GRK2 levels. Data are normalized to the maximal response of serotonin (100%) and shown with ±SEM of n = 3 to 14 independent experiments.