Abstract
The V2 vasopressin receptor (V2R) is a class A G protein-coupled receptor (GPCR) and plays a vital role in controlling water homeostasis upon stimulation by the natural peptide arginine vasopressin (AVP). Thus, V2R has attracted intense interest as a drug target for diabetes insipidus, nocturia, and hyponatremia. However, how AVP recognizes and activates V2R remains elusive. Here, we report the 2.6 Å resolution structure of V2R bound to AVP and the stimulatory G protein Gs, determined by cryo-electron microscopy (cryo-EM). In this complex, AVP presents a unique cyclic conformation formed by an intramolecular disulfide bond and engages the orthosteric binding pocket of V2R in a ligand-specific mode. Comparison of the AVP–V2R–Gs complex to previously reported Gs-coupled class A GPCRs reveals distinct structural features, including a smaller outward movement of TM5 and TM6 and the concomitant shift of Gs protein. Our detailed structural analysis provides a framework for understanding AVP recognition and V2R activation, thereby offering a structural template for drug design targeting V2R.
Introduction
Vasopressin type 2 receptor (V2R) belongs to the vasopressin (VP)/oxytocin (OT) receptor subfamily of G protein-coupled receptors (GPCRs), which comprise at least four closely related receptor subtypes: V1aR, V 1bR, V2R, and OTR1. These receptors are activated by arginine vasopressin (AVP, CYFQNCPRG-NH2) and OT (CYIQNCPLG-NH2), two endogenous nine-amino acid neurohypophysial hormones synthesized in the hypothalamus and secreted from the posterior pituitary gland. AVP and OT are thought to mediate a global biologically conserved role in social behavior and sexual reproduction2. Among VP/OT receptors, OT exhibits high selectivity for OTR. Conversely, AVP shows similar affinities to all the four receptors3. Upon stimulation by AVP, V2R plays a central role in controlling water homeostasis4.
V2R is mainly expressed in the renal collecting duct principal cells and mediates the antidiuretic action of AVP by accelerating water reabsorption. In response to hypovolemia or high osmolality, AVP is secreted from the posterior pituitary and binds to V2R in the basolateral membrane of these principal cells, thus stimulates adenylyl cyclase via the stimulatory G protein (Gs). This process promotes cAMP/PKA-mediated trafficking of aquaporin-2 (AQP2) water channels to the apical membrane in a long-term manner, allowing water to freely enter via the medullary osmotic gradient4. Additionally, ectopic expression of AVP and V2R has been reported in breast, pancreatic, colorectal, and gastrointestinal cancers, as well as renal cell and small cell lung carcinomas, indicated of their roles of tumorgenicitys5–7.
Considering its primary role in pathogenesis, V2R is referred to as an attractive drug target for prevention and control of diabetes insipidus, nocturia, hyponatremia secondary to syndrome of inappropriate diuretic hormone secretion (SIADH), and autosomal dominant polycystic kidney disease (ADPKD). Although there has been substantial progress in discovering peptidic and small molecular ligands targeting V2R, desmopressin (dDAVP), a peptidic AVP analogue8, is the only agonist approved for the treatment of diabetes insipidus. Additionally, small molecular V2R antagonists, tolvaptan, conivaptan, and atosiban, are presently applied to clinical management of nephrogenic syndrome of inappropriate diuresis (NSIAD) (https://www.cortellis.com/drugdiscovery).
Extensive efforts have been devoted to understanding the mechanism of peptide recognition and activation of V2R by its natural peptide ligands and to develop therapeutic agents via mutagenesis and structure-activity relationship studies9 and molecular dynamics simulation10, 11. However, except for the structure of OTR bound to a selective small molecular antagonist Retosiban (PDB code: 6TPK)12, no other structures from the VP/OT receptor subfamily, especially an agonist-bound active structure, are available. In this study, we determined a near-atomic resolution cryo-EM structure of the full-length, Gs-coupled human V2R bound to AVP. It reveals a specific recognition mode of AVP by V2R and an unconventional receptor activation mechanism different from that of previously reported Gs-coupled class A GPCRs, thus providing a structural template for mechanistic understanding and V2R-targeted drug design.
Results
Cryo-EM structure of Gs-coupled V2R bound to AVP
The structure of the AVP-V2R-Gs complex was determined to a resolution of 2.6 Å (Fig. 1 and Supplementary information, Table S1). An engineered mini-Gαs protein was used to obtain this complex structure13. The N-terminus of the engineered mini-Gαs was replaced by the corresponding residues of Gαi to facilitate the binding of scFv16. An analogous approach was used to obtain the structures of the 5-HT2A receptor bound to Gαq and the Gα11-coupled muscarinic M1 receptor14, 15.
The final structure of the AVP–V2R–Gs complex contains all residues of AVP (residues 1-9), the Gαs Ras-like domain, Gβγ subunits, Nb35, scFv16, and the V2R residues from T31 to L339 8.57 (superscripts refer to Ballesteros–Weinstein numbering16) (Fig.1, Supplementary information, Fig. S1). We also used the NanoBiT tethering strategy, which was initially introduced in the structural studies of the class B GPCR-Gs complexes17, 18, to stabilize the AVP–V2R–Gs complex. These modifications have little effect on the pharmacological properties of V2R (Supplementary information, Table S2). The majority of amino acid side chains, including AVP, transmembrane domain (TMD), all flexible intracellular loops (ICLs) and extracellular loops (ECLs) except for ICL3 and G185-G188 in ECL2, were well resolved in the model, refined against the EM density map (Supplementary information, Figs. S2 and S3). Thus, the complex structure can provide detailed information on the binding interface between AVP and helix bundle of the receptor, as well as the interface between Gs heterotrimer and the receptor. The TMD is surrounded by an annular detergent micelle mimicking the natural phospholipid bilayer (Fig.1a). Three cholesterol molecules surrounding the GPCR transmembrane domain (TMD) are shown in the final structure (Fig. 1b).
Molecular recognition of AVP by V2R
AVP occupies an orthosteric binding pocket in the TMD bundle composed of all TM helices and all ECLs (Fig. 2). The most notable conformational feature of AVP is the tocin ring formed by a disulfide bridge between the first and sixth cysteine residues, presenting a “spoon-like” conformation (Fig. 2b). This distinctive architecture is also shared with OT and the synthesized peptide dDAVP19, 20, and is the minimal common requirement for the biological activity of these peptides21. The cyclic spoon head inserts deeply into the TMD core, while the C- erminus spoon tail stretches toward the ECLs of the receptor (Figs. 2a and b).
The distinguished cyclic conformation of AVP and the binding pocket’s physicochemical environment render a peptide-specific AVP binding mode (Fig. 2 and Supplementary information, Table S3 and S4). Q962.61 forms a stabilizing H-bond network with the N-terminal amine of Cys1P and the sidechain amine of Q1193.32 which in turn stacks against the Tyr2P sidechain. Q1193.32 H-bonds with K1163.29, which forms an additional H-bond with the backbone CO group of Tyr2P. These amino acids (Q962.61, K1163.29, Q1193.32, and Cys1P) form a stabilizing H-bond network (Fig. 2c). The alanine mutations of Q961.61 and Q1193.32 significantly diminish the AVP-induced cAMP accumulation, supporting our hypothesis that this H-bond network is critical to AVP-induced V2R activation (Fig. 2f and Supplementary information, Table S3). Tyr2P makes hydrophobic interactions with Q922.57, Q962.61, Q1193.32, F2876.51, M3117.39, L3127.40, and A3147.42 (Figs. 2c and e). Besides hydrophobic interactions, the hydroxyl group of Tyr2P H-bonds to the main-chain oxygen of L3127.40, whereas the main-chain CO group of Tyr2P forms a H-bond with Q1744.60 (Figs. 2c and e). Q1744.60, together with Q922.57, Q962.61, and Q1193.32, make substantial contributions to AVP-induced activation of V2R supported by our alanine mutagenesis analysis (Figs. 2f, g, and Supplementary information, Table S3). Phe3P buries in a hydrophobic cleft constituted by M1203.33, M123 3.36, Y2055.38, V2065.39, I2095.42, F2876.51, F2 8 86.52, and Q29 1 6.55, of which F2886.52 is closely involved in AVP-induced V2R activation (Figs. 2c, e, g, and Supplementary information, Table S3). Other polar contacts are observed between Gln4P, R2025.35, and the backbone and sidechain oxygens of Q2916.55, as well as Asn5P and the main-chain of A194ECL2 (Figs. 2d and e). Arg8P forms a part of a complex salt bridge with D331.28, E401.32, and K1001.32 (Figs. 2d and e). The K1001.32A mutation significantly decreased the AVP-induced cAMP accumulation, although its sidechain carbon atoms make week hydrophobic contacts with the Cys1P-Cys6P disulfide bond (Figs. 2f, and Supplementary information, Table S3). Additionally, the C-terminal amide of AVP points to ECL2, stacking against the R104ECL1 guanidinium group and H-bonding to the backbone CO group of D103ECL1 (Figs. 2d and e).
A “DCWA” sequence in ECL2 is highly conserved throughout VP/OT receptor sub-family but is not a feature of other class A GPCRs22. In this conserved sequence, W193ECL2 and A194ECL2 directly contact with AVP, of which W193ECL2 anchors Asn5P as a lid (Fig. 2h). Interestingly, when comparing structures of the AVP– V2R-Gs complex with that of OTR in inactive state12, a steric clash can be observed between Asn5P and W188ECL2 of OTR, the cognate residue of W193ECL2 in V2R, which may probably confer a notable conformational transition of ECL2 from β-hairpin to a flexible loop upon AVP binding (Fig. 2h). Although D191ECL2 does not directly interact with AVP, it may form a H-bond to the F105ECL1 amide and a salt-bridge to R104ECL1, which in turn, stabilize ECL1 and ECL2 conformation, thereby facilitating the interactions between ECL1 and the C-terminal amide of AVP (Fig. 2i). The significance of D191ECL2 to AVP-induced V2R activation is supported by our alanine mutagenesis analysis (Supplementary Table S3). The disulfide bond conserved across class A GPCRs is formed between C194ECL2 located in this sequence and C1123.25 and may contribute to the stabilization of ECL2 conformation (Fig. 2i).
This structure also provides a template to study the selectivity of AVP and OT (Supplementary information, Fig. S4). OT has two amino acid substitutions at positions 3 and 8, Phe3P to Ile3P, and Arg8P to Leu8P, respectively (Supplementary information, Fig. S4a). It exhibits a much lower selectivity against V2R, with an over 500-fold decrease in binding affinity as opposed to AVP3. Interestingly, residues surrounding Phe3P and Arg8P are closely associated with AVP-induced V2R activation (Supplementary information, Table. S3), indicating the environment faced by amino acids at positions 3 and 8 are probably critical to OTR activation. By docking OT into the V2R binding pocket, a weaker hydrophobic interaction network is created for Ile3P compared to its equivalent residue Phe3P in AVP (Supplementary information, Fig. S4b). Additionally, the hydrophobic leucine at position 8 breaks the polar interactions between Arg8P of AVP and receptor (Supplementary information, Fig. S4c). Besides, Leu8P faces a relatively weak hydrophobic environment composes of R32, D331.28, L361.31, A371.32, and E401.35. Thus, the peptide-binding pocket of V2R defines a relatively unfavorable binding environment for OT in contrast to AVP, which may hamper OT binding.
Collectively, AVP interacts with V2R in a ligand-specific manner. These observations provide a rationale for understanding V2R recognition by AVP and OT, as well as a structural template for V2R-targeted drug discovery.
Activation of V2R by AVP
A comparison of the AVP–V2R–Gs complex structure to that of OTR in the inactive state sheds light on the conformational changes involved in activation of V2R: The cytoplasmic end of TM6 in Gs-coupled V2R undergoes a notable outward displacement, the hallmark of GPCR activation (Fig. 3a). V2R conforms to a common activation pathway that directly links the ligand-binding pocket to G-protein coupling regions in class A GPCRs23, including the AVP-induced turning of the rotamer “toggle switch” W2846.48, which translates into the rotation and outward movement of TM6 (Fig. 3b). Several distinct features on active V2R conformation can be observed compared to other previously reported class A GPCR structures, suggesting that V2R may undergo a receptor-specific activation.
To the best of our knowledge, residues P5.50, I3.40, and F6.44, the conserved “P-I-F” micro-switch, undergoes a conformation rearrangement during receptor activation24. Indeed, this “P-I-F” micro-switch of V2R exhibits a substantial rearrangement compared to inactive OTR, reflecting the activation state of V2R (Fig. 3c). The sequence alignment analysis shows that the residue F6.44 is evolutionally conserved among 74% of class A GPCRs, while the hydrophobic residues Ile/Leu/Val account for residues at position 3.40 in 74% of class A GPCRs25, thus providing a hydrophobic environment for the packing of TM3-5-6. By contrast, F’3.44 and I3.40 are substituted by polar residues Y2806.44 and S1273.40 in V2R, respectively (Figs. 3d and e). Compared to the Gs-coupled β2 adrenergic receptor (β2AR, PDB code: 3SN6)26, the distinct physicochemical environment in V2R facilitates the formation of H-bonds between Y2806.44 and S1273.40, as well as Y2806.44 and the backbone CO group of V2135.46, which probably stabilize the active conformation of the receptor (Fig. 3d). The disease-associated mutations S1273.40F and Y2806.44C deactivate V2R, offering experimental evidence to support these residues’ putative role in V2R activation27, 28. The alanine mutation of F2846.44 abolished the binding and activation of OTR, while the replacement of F2846.44 in OTR by the V2R-equivalent tyrosine slightly decreased the V2R activation by AVP but converted AVP from a partial to a full agonist29. These data are consistent with the hypothesis that polar interactions among these unconventional “P-I-F” residues may differentiate the activation mode between V2R and OTR.
Intriguingly, a distortion of TM7 helix was observed between L3127.40 and S3157.43, demonstrating a unique structural feature that is different from other class A GPCRs solved so far (Fig. 3f). This conformation appears to be stabilized by the H-bond between Tyr2P of AVP and the main-chain CO group of L3127.40 in the receptor. This polar contact draws TM7 closer to the core of the peptide-binding pocket (Fig. 3f). [Phe2]AVP, a synthesized AVP analog with a substitution of Tyr by Phe abolishing this H-bond, decreased V2R activity by 28-fold30, implying that the polar interaction between Tyr2P and the receptor is critical to AVP-induced V2R activation.
Furthermore, the cytoplasmic end of TM4 is one helical turn shorter than other Gs-coupled GPCRs, thereby releasing a longer ICL2 (Fig. 3g). The latter protrudes towards TM1 and adopts a C-shaped conformation (Fig. 3g). This distinct conformation is stabilized by a H-bond formed between H802.45 and the backbone CO group of H155ICL2 (Fig. 3h). The C-shaped loop is further stabilized by a polar interaction network constituted by the side chains of N157ICL2 and S152ICL2, as well as the backbones of G153ICL2 and W156ICL2 (Fig. 3h). These interactions may stabilize ICL2 in a specific conformation and probably affect the ICL2-Gαs interaction pattern.
G protein coupling by V2R
The interface between V2R and Gs heterotrimer consists of four transmembrane helices (TM2, TM3, TM5, and TM6), ICL2, and helix 8. The outward movement of the cytoplasmic ends of TM5 and TM6 open a cytoplasmic cavity together with TM2, TM3, and helix 8 to accommodate the α5 helix of Gαs. Compared to other solved Gs-coupled class A GPCRs, the AVP-V2R-Gs complex shows a distinctive GPCR-Gs coupling feature (Fig. 4a and b). The cytoplasmic ends of TM5 and TM6 in V2R display a smaller amplitude of outward displacements than the Gs-coupled β2AR (PDB code: 3SN6; 5.8 Å for TM5 and 8.7 Å for TM6)26 and adenosine A2A receptor (A2AR, PDB code: 5G53; 2.5 Å for TM5 and 7.2 Å for TM6 measured at Cα carbon of residues 5.67 and 6.30, respectively) (Fig. 4a)31. In addition, the α5 helix of Gαs in the AVP–V2R–Gs complex shifts 0.5 helical turns away from the 7TM core (Fig. 4a). These structural differences altogether create a smaller cytoplasmic cavity to accommodate Gs protein. Indeed, the solvent-accessible surface area (SASA) of the V2R-Gαs interface (943.32 Å2) is smaller than that of β2AR-Gαs (1,030 Å2) and A2AR-G%s interfaces (1,276 Å2) (Fig. 4c). A noticeable shift of Gs protein in the AVP–V2R–Gs complex occurs (Fig. 4a and b) that may partly be caused by relatively inward positions of TM5 and TM6, thus pushing the entire Gs heterotrimer shifts in the same direction.
The major binding interface between V2R and Gs protein is between the α5 helix of the Gαs subunit and the cytosolic core formed by TMD (Fig. 4d). In this interface, E392 of Gαs polar interacts with three consecutive serines in receptor helix 8 (S3 2 98.47, S3308.48, and S3318.49). Another polar interaction can be observed between R380 and E2315.64. These polar interactions may contribute to stabilizing the V2R-Gαs interface, which is crucial to GPCR-G protein coupling. One additional interface relates to ICL2 that interacts with α5 helix, αN-β1 junction, and the top of the β3-strand of Gαs, presumably stabilized by hydrophobic contacts (Fig. 4e).
The GPCR-Gs coupling profiles of V2R, β2AR, and A2AR point to several distinct features (Fig. 4f). ICL2 of V2R makes more extensive hydrophobic interactions with Gαs compared to β2AR and A2AR (Fig. 4f). The superposition of the three also exhibits a 1-2 helical turn shorter TM5 cytoplasmic end in V2R (Fig. 4b). Since proline and glycine are well-known α-helix breakers that disrupt α-helical backbone conformation’s regularity, a consecutive P-G-P sequence (P238, G239, and P240) may terminate the extension of the TM5 (Supplementary information, Fig. S5). Thus, compared to Gs-coupled β2AR and A2AR, a shorter TM5 in V2R produces fewer interactions with Gαs (Fig. 4b and f).
Implication of disease-causing mutations
Numerous mutations of V2R were identified and closely associated with human diseases32. Gain-of-function mutations in the gene encoding the V2R (AVPR2), including R1373.50C/L, F2295.62V, and I1303.43N, cause NSIAD, leading to hyponatremia and related clinical symptoms in infants33–35. Conversely, loss-of-function mutations of AVPR2 account for almost 90% incidence rate of X-linked congenital nephrogenic diabetes insipidus (NDI)36, 37.
These reported disease-associated mutations are located in three major regions of V2R: the ligand-binding pocket, the G protein-coupling site, and the central region connecting these two regions (Supplementary information, Fig. S6)23, 32. The activating L3127.40S mutation and inactivating mutations R2025.35C, F2876.51L, A2946.58P, and M3117.39V all sit in the AVP binding pocket. It is notable that L3127.40 sits at the site of the unique distortion of TM7 mentioned above. Substitutions of residues in ECLs to cysteines (R104ECL1C, R106ECL1C, and R181ECL2C) led to loss-of-function mutations38, which are putatively attributed to incorrect disulfide bonds formation by these cysteines39. A majority of disease-associated mutations occur in the central region linking the bottom of the ligand-binding pocket with the G protein-coupling site. Several mutations arise at the highly conserved micro-switches (S1273.40F and Y2806.44C in P-I-F, N3217.49K/Y and Y3257.53A in NPxxY, R1373.50H/C/L in D/ERY, as well as D852.50N and N3177.45S in Na+ pocket). Not surprisingly, these mutations can affect V2R activation since they are related to key residues responsible for subsequent G protein-coupling23. In the G protein-coupling region, R1373.50 and S3298.47 directly interact with Gs protein in our structure. Interestingly, mutations in R1373.50 can result in opposite phenotypes (R1373.50C/L for gain-of-function and R1373.50H for loss-of-function), probably due to the mutation-elicited changes in the physicochemical environmentunsuitable to TM association and Gs coupling.
Collectively, our findings provide a structural basis for V2R mutation-associated diseases. It is worth noting that the mechanisms involved are complex. Besides functional disorder, mislocalization by impaired intracellular trafficking and abnormal receptor expression disorder are also implicated in diseases. Finally, most V2R mutations are likely associated with combined disorders32. Systematic investigations are required to enrich our knowledge in this regard.
Discussion
The near-atomic resolution single-particle cryo-EM structure of Gs-coupled V2R bound to its endogenous ligand AVP reported in this study shows that the peptide adopts a “spoon-like” conformation with its featured tocin ring inserting into the TMD bundle and the C-terminus facing the binding pocket. This tocin ring is connected by a disulfide bridge formed between the first and sixth cysteine residues. Obviously, the cyclic backbone is shared with OT. It is conceivable that the distinct physicochemical environment in the peptide-binding pocket may lead to the low selectivity of OT to V2R. Our findings thus provide a framework for a better understanding of ligand recognition by V2R. Structural comparison with other Gs-coupled class A GPCRs sheds light on the basis of AVP-induced V2R activation. Although V2R conforms to a common activation mechanism, AVP-induced distortion of TM7 and a unique polar network formed by equivalent residues in the “P-I-F” motif may differentiate activation of V2R and OTR from other class A GPCRs. A smaller amplitude of the outward displacement of TM5 and TM6, as well as the concomitant shift of Gαs, not only distinguish V2R from its counterparts but also represent a diversified Gs coupling mechanism. Of note is that ICL3 of V2R/OTR was shown to be a determinant in G protein coupling40. Considering ICL3 is invisible in our structure, this conclusion deserves verification by additional structural information.
Our findings also offer a structural template for designing better ligands targeting V2R. dDAVP is an AVP analog with moderate selectivity to V1b and OT but is devoid of V1a binding. It has a sound antidiuretic effect. However, it also increases the risk of hyponatremia caused by a long plasma half-life and the resulting clearance delay from the kidney41. Efforts have been made to design potent V2R agonists with improved selectivity and pharmacokinetic profiles42. Such an endeavor will certainly be benefited from the structure information present in this paper.
Materials and Methods
Cell culture
Spodoptera frugiperda 9 (Sf9) insect cells (Expression Systems) were grown in ESF 921 serum-free medium (Expression Systems) at 27°C and 120 rpm. HEK 293T cells were purchased from American Type Culture Collection (ATCC), cultured in Dulbecco’s modified Eagle’s medium (DMEM, Life Technologies) supplemented with 10% fetal bovine serum (FBS, Gibco), and maintained in a humidified chamber with 5% CO2 at 37°C.
Constructs of V2R and Gs heterotrimer
The human wild-type (WT) V2R gene was codon-optimized for Sf9 expression and synthesized by Sangon Biotech. To facilitate the expression and purification, the full-length V2R DNA was cloned into a modified pFastBac vector (Invitrogen), which contains an N-terminal hemagglutinin (HA) signal peptide followed by a 10 × His-tag and a b562RIL (BRIL) epitope before the receptor. To improve the homogeneity and stability, the NanoBiT tethering strategy was applied by fusing a LgBiT subunit (Promega) at the receptor C-terminus using homologous recombination (CloneExpress One Step Cloning Kit, Vazyme) (Supplementary information, Fig. S1)17.
An engineered Gs construct (G112) was designed based on mini-Gs that was used in the crystal structure determination of A2AR–mini-Gs complex31. By replacing the N-terminal histidine tag (His6) and TEV protease cleavage site with the N-terminal eighteen amino acids (M1-M18) of human Gi1, this chimeric Gs was capable of binding to scFv16, which was used to stabilize the GPCR–Gi or –G11 complexes14–45. Additionally, replacing the GGSGGSGG linker at the position of original Gαs α-helical domain (AHD, V65-L203) with that of human Gi1 (G60-K180) provided the binding site for Fab_G50, another antibody fragment which was used to stabilize the rhodopsin-Gi complex46. Furthermore, three mutations (G226A, L272D, and A366S) were incorporated through site-directed mutagenesis as previously described to further increase the dominant-negative effect by stabilizing the Gαβγ heterotrimer47, 48. These modifications enabled the application of different nanobodies or antibody fragments to stabilize the receptor-Gs complex, although Nb35 and scFv16 were used together during the AVP–V2R–Gs complex formation and stabilization in this study. Rat Gβ1 was modified to fuse a SmBiT subunit17, 49 (peptide 86 and also named as HiBiT, Promega) with a 15-amino acid (15AA) polypeptide linker (GSSGGGGSGGGGSSG) at its C-terminus. The engineered Gs (G112), Gβ1-15AA-HiBiT, and bovine Gγ2 were cloned into the pFastBac vector, respectively. scFv16 was cloned into a modified pFastBac vector which contains a GP67 secretion signal peptide at its N-terminus. For constructs used in functional assays, they were all cloned into the pcDNA3.1 vector (Invitrogen) with an N-terminal Flag (DYKDDDD) tag proceeded by a HA signal sequence.
Expression and purification of nanobody35
Nanobody35 (Nb35) with a C-terminal 6 × His-tag was expressed in E.coli BL21 (DE3) bacteria and cultured in TB medium supplemented with 2 mM MgCl2, 0.1% (w/v) glucose and 50 μg/mL ampicillin to an OD600 value of 1.0 at 37°C, 180 rpm. Nb35 expression were then induced by adding 1 mM IPTG into the cultures and grown for 5 h at 37°C. Cells were then harvested by centrifugation (4,000 rpm, 20 min) and Nb35 protein was extracted and purified by nickel affinity chromatography as previously described50. Eluted protein was concentrated and subjected to a HiLoad 16/600 Superdex 75 column (GE Healthcare) pre-equilibrated with buffer containing 20 mM HEPES, pH 7.5 and 100 mM NaCl. The monomeric fractions supplemented with 30% (v/v) glycerol were flash frozen in liquid nitrogen and stored in −80°C until use.
Expression and purification of the AVP–V2R–Gs complex
Baculoviruses were prepared using the Bac-to-Bac Baculovirus Expression System (Invitrogen). Sf9 insect cells were cultured to a density of 3 × 106 cells per mL and co-infected with His10-BRIL-V2R-LgBiT, engineered Gs (G112), Gβ1-15AA-HiBiT, Gγ2, and scFv16 baculoviruses at a 1:1:1:1:1 ratio. The cells were then harvested by centrifugation 48 h post-infection and stored in −80°C for future use.
The frozen cells were thawed on ice and resuspended in lysis buffer containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 10% (v/v) glycerol, 10 mM MgCl2, 5 mM CaCl2, 100 μM TCEP (Sigma-Aldrich) and supplemented with EDTA-free protease inhibitor cocktail (Bimake). Cells were lysed by dounce homogenization and complex formation was initiated with the addition of 10 μg/mL Nb35, 25 mU/mL Apyrase (Sigma-Aldrich) and 10 μM vasopressin (GenScript) for 1.5 h at room temperature (RT). The membrane was then solubilized by adding 0.5% (w/v) lauryl maltose neopentyl glycol (LMNG, Anatrace) and 0.1% (w/v) cholesterol hemisuccinate (CHS, Anatrace) for 2 h at 4°C. The sample was clarified by centrifugation at 30,000 rpm for 30 min and the supernatant was then incubated with TALON resin (Clontech) supplemented with 10 mM imidazole for 3 h at 4°C. After incubation, the resin was collected by centrifugation (600 g, 10 min) and loaded into a gravity flow column, followed by first wash with five-column volumes of 20 mM HEPES, pH 7.5, 100 mM NaCl, 10% (v/v) glycerol, 5 mM MgCl2, 2 μM vasopressin, 25 μM TCEP, 10 mM imidazole, 0.1% (w/v) LMNG and 0.02% (w/v) CHS and then, with fifteen-column volumes of 20 mM HEPES, pH 7.5, 100 mM NaCl, 10% (v/v) glycerol, 5 mM MgCl2, 2 μM vasopressin, 25 μM TCEP, 25 mM imidazole, 0.03% (w/v) LMNG, 0.01% (w/v) glyco-diosgenin (GDN, Anatrace), and 0.008% (w/v) CHS. The protein was finally eluted with ten-column volumes of 20 mM HEPES, pH 7.5, 100 mM NaCl, 10% (v/v) glycerol, 5 mM MgCl2, 2 μM vasopressin, 25 μM TCEP, 200 mM imidazole, 0.03% (w/v) LMNG, and 0.01% (w/v) GDN. The purified AVP–V2R–Gs complex was concentrated using a Amicon Ultra centrifugal filter (molecular weight cut-off of 100 kDa, Millipore) and then subjected to a Superdex 200 Increase 10/300 GL column (GE Healthcare) that was pre-equilibrated with buffer containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 2 mM MgCl2, 100 μM TCEP, 5 μM vasopressin, 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN, 0.0002% (w/v) digitonin (Anatrace), and 0.0002% (w/v) CHS. The monomeric fractions of the complex were collected and concentrated to 4-5 mg/mL for cryo-EM examination.
Cryo-EM data collection and images processing
The freshly purified AVP–V2R–Gs complex (3.0 μL) at a final concentration of 4.2 mg/mL was applied to glow-discharged holey carbon grids (Quantifoil R1.2/1.3, 300 mesh), and subsequently vitrified using a Vitrobot Mark IV (ThermoFisher Scientific). Cryo-EM images were collected on a Titan Krios microscope (ThermoFisher Scientific) equipped with a K2 Summit direct electron detector (Gatan) and a GIF Quantum energy filter (slit width 20 eV) in the Cryo-Electron Microscopy Research Center at Southern University of Science and Technology. A total of 4,796 movies were automatically acquired using SerialEM51 in super-resolution counting mode at a pixel size of 0.42 Å and with a defocus values ranging from −0.5 to −2.0 μm. Movies with 32 frames each were collected at a dose of 8 electrons per pixel per second over an exposure time of 5.28 s, resulting in an accumulated of dose of 60 electrons per Å2 on sample.
All cryo-EM data were processed using RELION-3.152 as shown in the flowchart of Supplementary Fig. S2. Dose-fractionated image stacks were subjected to beam-induced motion correction and dose-weighting using MotionCor2.153. The Contrast transfer function (CTF) parameters for each micrograph were determined by CTFFIND-4.154 and micrographs whose maximum estimated resolution was worse than 4 Å were excluded, leaving 4,611 micrographs for further processing. Particle selection, two-dimensional (2D) and three-dimensional (3D) classifications were performed on a binned dataset with a pixel size of 0.84 Å. Auto-picking yielded 2,583,920 particle projections which were extracted downscaled and subjected to two-rounds of reference-free 2D classification to discard particles with false positives or poorly defined classes, resulting in 1,762,197 particles for 3D processing. This subset of particles was selected and re-extracted without downscaling to produce a 3D initial model for further consecutive rounds of 3D classification. With a 40 Å low-pass filter of initial model, two rounds of 3D classification yielded one well-defined subset of 577,084 particles to be subjected to CTF refinement and three-rounds Bayesian Polishing before the final 3D auto-refinement and sharpening was applied. The final refinement generated a map with an indicated global resolution of 2.6 Å, with 577,084 projections at a Fourier shell correlation of 0.143. Local resolution was determined using the Resmap package with half maps as input maps55.
Model building and refinement
The density map was automatic post-processed using DeepEMhancer56 to improve the EM map quality before model building. The initial V2R model was generated by an online homology modeling server, SWISS- MODEL57, using the oxytocin receptor (OTR) structure (PDB code: 6TPK) as a template. The mini-Gs heterotrimer (mini-Gs, Gβ1 and Gγ2) and Nb35 taken from GPR52–mini-Gs–Nb35 complex (PDB code: 6LI3), scFv16 taken from CB2–Gi–scFv16 complex (PDB code: 6PT0) and vasopressin taken from typsin-vasopressin complex (PDB code: 1YF4) were used as initial models. All models were fitted into the EM density map using UCSF Chimera58, followed by iterative rounds of manual adjustment and automated rebuilding in COOT59 and PHENIX60, respectively. The model was finalized by rebuilding in ISOLDE61 followed by refinement in PHENIX with torsion-angle restraints to the input model. The final model statistics were validated using Comprehensive validation (cryo-EM) in PHENIX60 and provided in the supplementary information, Table S1. All structural figures were prepared using Chimera58, Chimera X44 and PyMOL (https://pymol.org/2/).
cAMP accumulation assay
HEK 293T cells were transiently transfected with different V2R constructs: HA-Flag-V2R(1-371), HA-Flag-His10-BRIL-V2R(1-371)-LgBiT, or HA-Flag-V2R mutants. The cells were digested by 0.02% (w/v) EDTA, resuspended by stimulation buffer (1 × HBSS, 5 mM HEPES, 0.5 mM IBMX, and 0.1% (w/v) BSA, pH 7.4) 24h after transfection, and seeded at a density of 3,000 cells per well into 384-well plates (PerkinElmer). Cells were stimulated by different concentrations of AVP for 40 min at RT. The reaction was terminated by addition of 5 μL Eu-cAMP tracer and 5 μL ULight-anti-cAMP (diluted by cAMP detection buffer). After incubation at RT for 1h, cAMP signals were detected by measuring the fluorescence intensity at 620 nm and 650 nm using an EnVision multilabel plate reader (PerkinElmer).
Receptor surface expression
Cell-surface expression levels of WT or mutants V2R were quantified by flow cytometry. HEK 293T cells were seeded at a density of 6 ×105 per well into 6-well culture plates. They were digested 24 h post-transfection by 0.02% (w/v) EDTA and blocked with 5% (w/v) BSA in PBS at RT for 15 min, followed by the incubation with 1:300 diluted anti-Flag primary antibody (Sigma) at RT for 1 h. Cells were then washed by 1% (w/v) BSA in PBS three times followed by incubation 1:1000 donkey-anti-mouse Alexa Fluor 488 conjugated secondary antibody (Invitrogen) at 4°C in the dark. After another three washes, cells were resuspended by 200 μL 1% (w/v) BSA in PBS for detection by a NovoCyte flow cytometry (Agilent).
Molecular docking
The entire process was done in Schrodinger Suite 2017-4. The AVP–V2R–Gs complex structure was prepared by Protein Preparation Wizard62, hydrogen atoms were added using Maestro and protonation states were assigned using PROPKA. The directions of polar hydrogens were optimized and the whole structure was minimized. Because the peptides to be analyzed have over 100 atoms thus are not suitable for traditional molecular docking, we performed MMGBSA63 calculation to estimate relative binding affinity for a list of congeneric ligands using the Prime MM-GBSA module. The analog of AVP, OT was gradually mutated on the basis of AVP after several rounds of energy minimization. Besides, the residues within 4 Å of peptides were further subjected to relax with the sampling method “Minimize” in the Prime MM-GBSA module.
Statistical analysis
All functional data were analyzed using Prism 7 (Graphpad) and presented as means ± S.E.M. from at least three independent experiments. Concentration–response curves were evaluated with a three-parameter logistic equation to determine the pEC50 and Emax values. The significance was determined by one-way ANOVA in Prism 7 and P value < 0.01 was considered statistically significant.
Author contribution
F.Z. designed the expression constructs, purified the AVP–V2R–Gs complex, prepared the final samples for negative stain, cryo-EM grid preparation, and data collection toward the structure, and participated in figure and manuscript preparation; F.Z. and X.M. performed specimen screening by negative-stain EM, cryo-EM data collection, and map calculations; F.Z., X.M. X.Z. built the structure model; F.Z. and T.C. refined the structure model; C.Y. conducted functional studies under the supervision of D.Y. who also participated in data analysis and manuscript editing; Q.Z. and X.H. carried out docking analysis and participated in manuscript preparation; W.Y. engineered the mini-Gs protein; F.Z. and Y.J. prepared the bulk of figures, performed the structural analysis, and drafted the manuscript; Y.J., H.E.X., and M.-W.W. initiated the project, supervised the studies, analyzed the data, and wrote the manuscript with inputs from all co-authors; P.W. supervised the EM studies.
Competing Interests
The authors declare no competing interests.
Acknowledgements
We thank all staff members of the Cryo-EM Centre, Southern University of Science and Technology for their assistance in data collection. This work was partially supported by the National Natural Science Foundation of China (31770796 to Y.J., 81872915 to M,-W.W., 31600606 to X.Z., and 81773792 to D.Y.); the National Science & Technology Major Project “Key New Drug Creation and Manufacturing Program” (2018ZX09711002-002-002 to Y.J., 2018ZX09735-001 to M.-W.W., and 2018ZX09711002-002-005 to D.Y.); Ministry of Science and Technology (China) grant (2018YFA0507002 to H.E.X.); Shanghai Municipal Science and Technology Major Project (2019SHZDZX02 to H.E.X.); CAS Strategic Priority Research Program (XDB37030103 to H.E.X.); Start-up funding by Fudan University (Q.Z.); Wellcome Trust (209407/Z/17/Z); the National Key R&D Program of China (2016YFA0501100 to X.Z.); Guangdong Provincial Key Laboratory of Brain Connectome and Behavior (2017B030301017 to X.Z.); CAS Key Laboratory of Brain Connectome and Manipulation (2019DP173024 to X.Z.)