Inactive structures of the vasopressin V2 receptor reveal distinct binding modes for Tolvaptan and Mambaquaretin toxin

Abstract

Inhibitors of the arginine-vasopressin (AVP) V2 receptor (V2R) are key therapeutic compounds for treating hyponatremia or polycystic kidney diseases. Rational drug design based on experimental G protein-coupled receptor structures is a powerful avenue to develop better drugs. So far, the lack of inhibitor-bound V2R structures has impaired this strategy. Here we describe the cryo-electron microscopy structures of the V2R in complex with two selective inverse agonists, the non-peptide Tolvaptan (TVP) and the green mamba snake Mambaquaretin toxin (MQ1). Both ligands bind into the orthosteric binding site but with substantial differences. TVP binds deeper than MQ1, and directly contacts the toggle switch residue W284^6.48 in the transmembrane domain 6. The Kunitz-fold toxin displays extensive contacts with extracellular and transmembrane residues. As anticipated from TVP and MQ1 pharmacological properties, both structures represent inactive V2R conformations. Their comparison with those of the active AVP-bound V2R reveals the molecular mechanisms modulating receptor activity. The mini-protein MQ1-bound V2R structure suggests a new pharmacology approach for treating water homeostasis and renal diseases.

Fig. 1. Structures of the inactive TVP-V2R and MQ1^K39A-V2R complexes.

a Cryo-EM density map of TVP-V2R-BRIL-Fab-Nb complex and b corresponding 3D structure as ribbon representation. TVP is colored in gold, V2R in light blue, BRIL in gray, anti-BRIL Fab in clear green, and Anti-Fab Nb in salmon. A close-up view of the inverse agonist-receptor interaction is shown. For a better view of TVP, TM7 has been hidden. c Cryo-EM density map of MQ1^K39A-V2R-BRIL-Fab-Nb complex and d corresponding 3D structure as ribbon representation. MQ1^K39A is shown in raspberry, V2R in green, BRIL, Fab, and Nb are colored as in (a, b). A close-up view of the toxin-receptor interaction is depicted.

Fig. 2. Binding poses of TVP and MQ1^K39A in the V2R orthosteric ligand pocket.

a, b Side and top view of the TVP-V2R binding interface. V2R is in light blue, TVP in gold. Receptor residues within a distance of 4 Å of the ligand are highlighted in sticks, and numbered following the Ballesteros-Weinstein nomenclature. The toggle switch W^6.48 is shown in orange. H-bonds and halogen bond are illustrated as dashed lines. c, d Side and top views of the MQ1^K39A-V2R binding interface. V2R is in green, and the toxin in raspberry. Residues within a 4 Å distance from the receptor and from the toxin are highlighted in sticks, and numbered as in (a, b). H-bonds are illustrated as dashed lines. e TVP-V2R and MQ1^K39A-V2R were aligned onto the receptor structures, color scheme is equivalent to that of (a–d). The partial overlap of TVP and MQ1^K39A is shown. f TVP-V2R structure was aligned onto that of the Retosiban-OTR structure (pdb-6tpk). OTR is in gray, V2R in light blue. The non-peptide antagonist Retosiban is in clear green. Overlap of TVP and Retosiban at the bottom of the binding pocket is shown. The toggle switch W^6.48 is illustrated in orange.

Fig. 3. Comparison of active AVP-bound and inactive inverse agonist-bound structures of the V2R.

a TVP-bound and AVP-bound (pdb-7dw9) structures were aligned (central panel) and major differences in conformations are shown (close-up views). Inactive V2R with TVP is in light blue, TVP is in gold, active V2R with AVP is in violet, AVP is in clear green. On the left, zoom images of AVP/TVP overlap (top), extracellular (middle) and intracellular (bottom) TM6 extremities are shown. On the right, zoom images of extracellular part of TM1 (top) and disruption of TM7 (bottom) are illustrated. Distances (in Å) between the position of reference residues in the active and inactive structures are shown as red dashed lines: A294^6.58, V266^6.30, R38^1.33, and A314^7.42 were chosen. b MQ1^K39A-bound and AVP-bound (pdb-7dw9) structures were aligned (central panel), and major differences in conformations are shown (close-up views). Inactive V2R with MQ1^K39A is in green, the toxin is in raspberry. The zoom images are equivalent to those shown in (a) the reference TM residues chosen for measuring distances (red dashed lines) are equivalent. c Overlap of AVP and TVP represented as spheres. d Overlay of AVP and MQ1^K39A displayed as spheres.

Fig. 4. Structural comparison of conserved motifs in inactive and active conformations of V2R.

a Alignment of TVP-V2R (top) or of MQ1^K39A-V2R (bottom) with AVP-V2R (pdb-7dw9) is shown. The color scheme is equivalent to that of Fig. 3. Microswitches along the helix bundle are highlighted as sticks. b Close-up views of activation motifs of the V2R are presented: CWxP (toggle switch), PSY (transmission motif), NPxxY (tyrosine toggle switch), and DRH (ionic lock) from top to down. Residues participating in these motifs are highlighted as sticks and numbered as superscript following the Ballesteros-Weinstein nomenclature.

Fig. 5. Functional role of key residues in V2R ligand binding.

a Affinities (Ki) of AVP, b TVP, and c MQ1^K39A for the wild-type (WT) V2R and the different receptor mutants were calculated from competition binding experiments using the benzazepine-red antagonist as a tracer (see “Methods”). Dashed lines indicate the mean Ki of each ligand for the wild-type V2R. Data are means ± SEM from 3 to 6 individual experiments, each performed in triplicates. For the WT V2R, n = 6 for all AVP, TVP and MQ1^K39A competition assays. For mutant W99A, n = 4 in all competition assays. For all other mutants, n = 3 in all competition assays. Statistical significance was assessed using one-way ANOVA, comparing all mutants to the wild-type receptor: ns, not significant p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001. Here are the precise values of p. In a E184A, p > 0.9999; R202A, p = 0.0327; I209A, p = 0.0918; all other mutants, p < 0.001. In b Q92A, p = 0.1272; Q96A, p = 0.0201; D103A, p = 0.0221; M120A, p = 0.0101; E184A, p = 0.4160; W193A, p = 0.9941; R202A, p = 0.2045; M311A, p = 0.0136, all other mutants, p < 0.001. In c Q96A, p = 0.9999; M120A, p = 0.0025; E184A, p = 0.0595; I209A, p = 0.9724; W284A, p = 0.1251; Q291A, 0.0985; M311A, p = 0.6222; S315A, p = 0.0030, all other mutants, p < 0.001. Source data are provided as a Source data file. d Graph representation of the effects of V2R mutations on ligand binding. The receptor mutants are classified according to their effect toward the 3 ligands AVP, TVP, and MQ1^K39A. To highlight the most significant effects, only a loss or a gain >5-fold as compared to the value for the WT V2R was considered. The effects of D103A, R104A, E184A, I209A, Q291A and S315 were less than 5-fold. Results from the competition assays and from saturation assays (benzazepine-red tracer) are illustrated on the left and on the right, respectively.

Fig. 6. Kunitz-fold versus three-finger toxin interactions to GPCRs.

Alignment of the MT7-muscarinic M1 receptor (M1R) structure (pdb-6wjc) with the inactive MQ1^K39A-V2R structure. The V2R is shown in green, the MQ1^K39A in raspberry, the M1R in salmon, and the MT7 in light blue. Orthogonal views are displayed from the side of the TM bundle (top panels) or from the extracellular space (bottom panels). The three-finger MT7 toxin and the MQ1^K39A Kunitz-fold toxin are represented as ribbons (top panels) and as transparent surfaces (bottom panels).

Fig. 7. GPCR binding mode with a Kunitz-fold ligand and mechanism of V2R inhibition.

a Aromatic interactions between TVP and V2R are shown. V2R is illustrated as ribbons and in light blue, TVP in gold. Only TM6 is not transparent. Direct aromatic/aromatic contacts between the different moieties of TVP and F287^6.51 and W284^6.48 are highlighted in red dashed lines. b Structure of MQ1^K39A-V2R complex is shown from the side of the TM bundle (top panel) or from the extracellular space (bottom panel). MQ1^K39A is depicted in raspberry as ribbons and transparent surface. The ring of receptor toxin-interacting residues from all TM excepted TM1 is shown in green. c Aromatic network between the MQ1^K39A and V2R is proposed. The scheme color is equivalent to that of (b). The pi-stacking (T-type) contact between the toxin F18 and F307^7.35 of V2R, as well as aromatic ring contacts between F307^7.35 in TM7, F287^6.51, and W284^6.48 in TM6 are illustrated as red dashed lines.