The antidepressant drug vilazodone is an allosteric inhibitor of the serotonin transporter

Abstract

Depression is a common mental disorder. The standard medical treatment is the selective serotonin reuptake inhibitors (SSRIs). All characterized SSRIs are competitive inhibitors of the serotonin transporter (SERT). A non-competitive inhibitor may produce a more favorable therapeutic profile. Vilazodone is an antidepressant with limited information on its molecular interactions with SERT. Here we use molecular pharmacology and cryo-EM structural elucidation to characterize vilazodone binding to SERT. We find that it exhibits non-competitive inhibition of serotonin uptake and impedes dissociation of [³H]imipramine at low nanomolar concentrations. Our SERT structure with bound imipramine and vilazodone reveals a unique binding pocket for vilazodone, expanding the boundaries of the extracellular vestibule. Characterization of the binding site is substantiated with molecular dynamics simulations and systematic mutagenesis of interacting residues resulting in decreased vilazodone binding to the allosteric site. Our findings underline the versatility of SERT allosteric ligands and describe the unique binding characteristics of vilazodone.

Fig. 1. Characterization of VLZ binding to SERT WT and S1 mutants.

a Inhibition potency of VLZ (black) and S-CIT (brown) compared to 5-HT (gray). The affinity for VLZ is 5-fold higher than for S-CIT in inhibiting [³H]5-HT transport with K_i values of 1.06 [0.90; 1.25] and 5.21 [4.19; 6.48] nM, respectively (n = 6). b Equilibrium binding inhibition for S-CIT and VLZ displacement of [³H]S-CIT. The K_i for VLZ is 7-fold higher than for S-CIT (0.68 [0.56; 0.83] and 5.13 [2.85; 3.76] nM, respectively). c, d The affinity for S-CIT but not VLZ is affected by mutations in the SERT S1 site. S-CIT binding is decreased by 7- and 192- and 625-fold by the mutations Y95F (purple), I172M (blue) and S438T (green), respectively relative to WT (dotted line of data in a). The mutations do not significantly affect the affinity for VLZ. V_MAX, K_M, K_i, and n-values for 5-HT, S-CIT, and VLZ binding to SERT WT and S1 mutants are shown in the Supplementary Table 1. Experiments are performed in triplicates on intact COS7 cells transiently expressing SERT WT or mutants. Data are shown as means ± S.E. (error bars), n = 4–17. Source data are provided as a Source Data file.

Fig. 2. Indications of VLZ as an allosteric inhibitor of SERT WT.

a Saturation uptake experiments for [³H]5-HT transport as a function of increasing VLZ concentrations (0–15 nM) are compatible with a non-competitive inhibition for VLZ. Colors of lines and symbols represent various concentrations of VLZ defined in legend. The V_MAX for [³H]5-HT decreases with increasing VLZ concentration. Data are shown in percent of 5-HT V_MAX in the absence of VLZ (V_MAX in the absence of VLZ is 4899 ± 312 cpm/min/10⁵ cells). The K_M-value for 5-HT is largely unaffected by increasing VLZ concentrations (Supplementary Table 2). Non-specific [³H]5-HT binding was assessed in the presence of 1 µM paroxetine. Experiments are performed in triplicates on intact COS-7 cells transiently expressing SERT, n = 3–8. b Representative experiment for the assessment of VLZ allosteric potency determined by its influence on [³H]IMI dissociation. The slope for [³H]IMI dissociation ranges from 0.528 ± 0.012 min⁻¹ in the absence of VLZ (ctrl, black line) to 0.045 ± 0.019 min⁻¹ in the presence of 1120 nM VLZ (blue line). Data are converted to the dose-response curve in (b) to determine the allosteric potency (see Methods). c Allosteric potency of VLZ determined as concentration-dependent inhibition of [³H]IMI and [³H]S-CIT dissociation. The radioligands were prebound to SERT WT prior to the addition of VLZ in the indicated concentrations. Data are plotted as the effect of VLZ on the [³H]IMI and [³H]S-CIT dissociation rate (k_[VLZ]) relative to the rate in the absence of VLZ (k_[ctrl]). The allosteric potency of VLZ is ~17-fold higher in the presence of prebound [³H]IMI (circles, tan) relative to [³H]S-CIT (triangles, brown), with IC₅₀ = 14.1 [11.8; 16.8] and 250 [203; 307] nM, respectively (mean [S.E. interval]), n = 10 ([³H]IMI) and 7 ([³H]S-CIT). Data are shown as means ± S.E. (error bars). [³H]cmpd: [³H]compound (being either [³H]IMI or [³H]S-CIT). Source data are provided as a Source Data file.

Fig. 3. Cryo-EM structure of SERT:IMI:VLZ complex.

a The overall reconstruction of the SERT:15B8-Fab:IMI:VLZ complex at 3.65 Å (PDB: 7LWD). Right panel shows the zoomed allosteric and central sites. SERT is colored in light blue and the 15B8 Fab is shown in green. IMI and VLZ densities are shown in yellow and red, respectively. b Surface representation of the outward-open structure of SERT, viewed parallel to the membrane. The open extracellular pathway and the closed intracellular pathway are displayed. IMI and VLZ are shown as ball sticks. c Zoomed view of the IMI (yellow) binding site. IMI is shown fit into a density feature at the S1 site with subsite A, B, and C and the involved SERT backbone structures (light blue) annotated. Parts of SERT have been removed for clarity. d VLZ (red) is shown fit into a density feature at the allosteric site with the central SERT backbone structures annotated. Parts of SERT have been removed for clarity. e Comparison of the IMI (yellow) binding pose with the binding poses for S-CIT (light magenta, PDB code:5I73) and paroxetine (gray, PDB code:5I6X) within the S1 site. The main chain position of SERT from the IMI:VLZ complex is shown in light blue. f Comparison of VLZ (red) with S-CIT (light magenta, PDB code:5I73) and Lu AF60097 (gray). The main chain position of SERT from the IMI:VLZ complex is shown in light blue.

Fig. 4. Detailed representation of the binding sites for IMI and VLZ in SERT.

a Overall view of SERT:IMI:VLZ complex in cartoon representation; IMI (yellow) and VLZ (red) are depicted as spheres. The “eye” represents the angle view depicted in (b) and (c). b Detailed representation of the IMI binding site in S1. IMI binding is mainly formed by side chain residues in TM1, 3, 6, 8, and 10. Residues that interact with IMI are annotated and shown in violet sticks. IMI is shown in yellow sticks and VLZ as gray sphere. c Detailed representation of the VLZ binding site in S2. Residues that interact with VLZ are annotated and shown in violet sticks, Arg104 and Glu493 are shown in pink. VLZ is shown in red sticks and IMI as gray sphere. The salt bridge formed between Arg104 and Glu493 is shown as purple dashed lines. TM: Transmembrane, EL: Extracellular loop, IL: Intracellullar loop.

Fig. 5. MD simulations of possible VLZ binding poses.

Zoomed view of VLZ modeled into the allosteric binding site of SERT. Parts of SERT TMs and bound IMI are removed for clarity. a The refined cryo-EM SERT structure (blue ribbons) with VLZ (magenta) superimposed onto the final MD simulated SERT structure (orange ribbons) with VLZ (orange) in the hypothezised pose. b Structure of the refined SERT:VLZ complex modeled with the 180° flipped VLZ in its binding site and superimposed onto the final MD simulated SERT structure (gray ribbons) with VLZ (gray) in the flipped pose. The MD simulations-derived structures of (c) SERT with VLZ bound at the hypothesized pose and (d) the flipped pose, both superimposed on the VLZ electron density from the cryo-EM structure. e, f As in c and d, viewed from the plane of the membrane, showing that the indole ring in the hypothesized pose could explain the density feature in the cryo-EM structure. TM: Transmembrane.

Fig. 6. Effect of SERT S2 mutants on VLZ allosteric potency.

a Zoomed interactions between VLZ and SERT. Schematics are generated by LIGPLOT + 1.4. Each eyelash motif indicates a hydrophobic contact. b Effect of mutating residues in the S2 site on allosteric potency for VLZ. Point mutation of residues, shown in the cryo-EM structure to interact with VLZ, all significantly (P < 0.001, one-way ANOVA with Tukey’s multiple comparisons test) decreases its allosteric potency to various degrees relative to SERT WT (dotted line, data shown in Fig. 2b). See allosteric potencies for SERT WT and mutants in Table 1. Experiments are performed on membrane preparations from COS-7 cells transiently expressing the indicated SERT mutants. Data are shown as means ± S.E. (error bars), n = 3–10. Wild-type is shown with a black dotted line. Allosteric site mutants are shown using colored symbols and solid lines defined in the legend. See Table 1 for quantitative data including allosteric potency and n-value for the individual experiments. Source data are provided as a Source Data file.