Structural basis for the activation and ligand recognition of the human oxytocin receptor

The small cyclic neuropeptide hormone oxytocin (OT) and its cognate receptor play a central role in the regulation of social behaviour and sexual reproduction. Here we report the single-particle cryo-electron microscopy structure of the active oxytocin receptor (OTR) in complex with its cognate ligand oxytocin. Our structure provides high-resolution insights into the OT binding mode, the OTR activation mechanism as well as the subtype specificity within the oxytocin/vasopressin receptor family.

T he neurohypophyseal hormones oxytocin (OT) and arginine vasopressin (AVP) are cyclic peptides that activate an evolutionary ancient subfamily of class A G proteincoupled receptors (GPCRs) 1 , comprising the oxytocin receptor (OTR) and the closely related vasopressin receptors (V 1A R, V 1B R and V 2 R). This OT/AVP hormone system is highly conserved among many species and exerts a central role in the regulation of social cognition, social behaviour, and sexual reproduction 2 .
Currently, several clinical trials are evaluating the efficacy of OTR-mediated signalling through administration of OT to treat malfunctions such as autism-spectrum disorders 3 , anxiety 4 and schizophrenia 5 . While OT itself is an approved peripheral drug in obstetrics 6 , OT-based treatments of socio-behavioural deficiencies, requiring central administration of the hormone, have yet failed, potentially due to its poor drug-like properties 7 and limited penetration through the blood brain barrier 8 . Despite recent advances in finding a non-peptide agonist that is active in an animal model 9 , the high demand for such drugs continues. Up to now, the identification and development of OTR-specific molecules with satisfactory pharmacokinetic properties, favourable biodistribution and specificity has been impeded by the lack of structural information on the OTR:OT signalling complex.
Here, we now report the single-particle cryo-electron microscopy (cryo-EM) structure of the OT-bound human OTR in complex with a heterotrimeric G protein at a resolution of 3.2 Å.

Results and discussion
Receptor and G protein engineering. Initially, structural studies of the OTR:OT signalling complex were hampered by the poor biophysical behaviour of the wild-type OTR (wtOTR). To improve expression and purification yields we included a single stabilising mutation (D153Y), which we previously identified by a next-generation sequencing (NGS)-based in-depth analysis of directed evolution experiments 10 . This mutation enabled a 50fold increase in the yield of purified functional receptor, with very similar agonist binding and signalling behaviour ( Supplementary  Fig. 1a, b). As an additional hurdle, complexes of OTR with an otherwise frequently employed engineered mini-G s 11,12 or a G s/q chimera were not stable and dissociated upon plunge freezing. While G q -based signalling is the main route of OTR activation, the receptor has been shown to also interact with G o and G i 13 , but not with G s . Therefore, we hypothesised that the observed instability of the OTR:mini-G s/q complex may be attributed to unfavourable interactions of OTR with the G s domain, and interactions of the G q α5 helix are not sufficient to overcome this. Thus, to maximally stabilise the OTR active state, we designed a G protein chimera using mini-G o 11 as a basis, substituting the G o α5 helix with the corresponding amino acids of G q . Additionally, we replaced the N terminus with the respective G i1 residues to permit binding of the complex-stabilising single-chain variable fragment 16 (scFv16) 14 . For simplicity, the resulting mini-G protein is denoted G o/q henceforth. These combined engineering efforts finally enabled single-particle cryo-EM analysis of the OTR:OT:G o/q :scFv16 complex at a resolution of 3.2 Å (Fig. 1a,  OT binding mode. In the orthosteric ligand binding pocket, all nine amino acids of OT participate in OTR binding (Fig. 1c). The cyclic part (residues 1-6) is buried deep inside the pocket while the C-terminal tripeptide (residues 7-9) is facing the extracellular loops. Interestingly, the amidated C terminus of Gly 9 , known to be important for activation 15 , is located in proximity of residues E42 1.35 and D100 2.65 ( Fig. 1d; numbers in superscripts correspond to Ballesteros-Weinstein numbering 16 ) of transmembrane helices I and II, which are involved in magnesiumdependent modulation of OT binding 17 . Together with the neighbouring Leu 8 , the extracellular surface of the orthosteric pocket is lined by Pro 7 and Asn 5 , which pack against extracellular loop 3 (ECL3) and ECL2, respectively. Leu 8 is oriented towards the extracellular space, explaining why position 8 is best suited for the attachment of fluorophores in OT 18 . Gln 4 is the only residue pointing out perpendicularly from the ring plane and stabilizes the cyclic ring position through a hydrogen bond to Q295 6.55 . Ile 3 is buried in a hydrophobic pocket formed by side-chain residues of transmembrane helices IV, V, and VI. The critical contribution of this hydrophobic pocket is underlined by the observed reduction in OT potency when mutating the main contact residues I201 5.39 , I204 5.42 and F291 6.51 of the receptor to alanine (Fig. 1d, e and Supplementary Table 2). Tyr 2 penetrates deep into a crevice at the bottom of the orthosteric pocket formed by residues from helices II, III, VI and VII. While the carboxy group of Tyr 2 interacts with Q171 4.60 , the phenol ring engages in hydrophobic interactions with Q92 2.57 and F291 6.51 , and the hydroxyl group forms a hydrogen bond to the backbone amide oxygen of L316 7.40 . The importance of these interactions explains the loss of potency when either Q171 4.60 or F291 6.51 are mutated to alanine (Fig. 1d, e). Finally, Cys 1 , which stabilizes the OT ring conformation through a disulphide with Cys 6 , contacts with its backbone oxygen a polar cluster consisting of residues Q96 2.61 , K116 3.29 , and Q119 3.32 observed in the OT/AVP family, consistent with earlier mutagenesis studies 19 (Fig. 1d, e).
OT mediated receptor activation. A comparison of the OTR active-state structure with the previously reported inactive-state structure of the OTR in complex with the small-molecule antagonist retosiban 17 enabled us to identify the molecular changes involved in receptor activation (Figs. 2 and 3). In contrast to OT, retosiban only partially occupies the region of the orthosteric pocket, where the cyclic part of OT binds (Fig. 2b, c). Nonetheless, the OT-induced helical rearrangements in the orthosteric pocket are relatively small, reflected by the subtle change of the pocket volume between the active and inactive state (Fig. 2d, e). OT interacts with residues F291 6.51 and F292 6.52 at the bottom of the binding pocket. This interaction induces a rearrangement of F291 6.51 , initiating the large outward movement of helix VI through a series of side-chain reorientations in conserved microswitches, including W288 6.48 (CWxP motif), F284 6.44 (PIF motif) and Y329 7.53 (NPxxY motif) (Fig. 3c, d). These rearrangements ultimately lead to the breakage of the interaction between T273 6.33 and R137 3.50 of the DRY motif (DRC in OTR), and the reorientation of D136 3.49 into a position enabling direct contact with the α5 helix of G o/q (Fig. 3e).
Early functional studies on OT derivatives identified Cys 1 , Tyr 2 and Gln 4 as centrally involved ligand residues in receptor activation. For example, alkylation of the Tyr 2 hydroxy group led to decreased agonistic activity 20 , suggesting an important role of this residue, consistent with our structural data. Remarkably, we observe a local unfolding of helix VII at the extracellular receptor side in the region of L316 7.40 , creating a pronounced kink which is stabilised by a hydrogen bond formed between Tyr 2 of OT with the backbone oxygen of L316 7.40 (Fig. 3b). Importantly, a similar helix VII conformation was also observed in active-state structures of V 2 R 21 ( Supplementary Fig. 5), suggesting that partial helix VII unfolding is a feature of the OT/AVP family receptor activation. Sequence alignments of the four receptors of the OT/AVP family reveal that all receptors share a conserved kink region, with the exception of position 7.42, where both OTR and V 2 R share an alanine, whereas V 1A R and V 1B R carry a glycine. To test if a kink region carrying a glycine is compatible with the observed helix VII reorientation, we determined ligand binding affinity and receptor activation of an OTR mutant where we mutated A318 7.42 to glycine. We observe only little differences in activity and binding affinity, supporting an unaltered activation mechanism. It appears that glycine might potentially destabilise the local helical conformation and even facilitate ligand binding, as indicated by the slightly improved affinity and potency of OT to A318 7.42 G. Accordingly, we find that removal of the glycine in V 1A R by mutation G337 7.42 A has the inverse effect ( Supplementary Fig. 5, Supplementary Tables 2  and 3).
Oxytocin/Vasorpessin receptor family subtype specificity. OT differs only in two positions from the closely related AVP (Ile 3 and Leu 8 in OT vs. Phe 3 and Arg 8 in AVP) (Fig. 4a). While these differences suffice to render OT specific for the OTR over the vasopressin receptors, the OTR itself is not selective between OT and AVP 22 . To investigate possible contributions to subtype selectivity, we compared the OT-bound OTR structure to the previously published AVP-bound V 2 R structures 21,23,24 (Fig. 4). Both OT and AVP adopt a similar orientation in the orthosteric pocket of their respective receptor with highest similarity observed for the agonist's cyclic portion, where Phe 3 of AVP penetrates only marginally deeper than Ile 3 of OT (Fig. 4). The largest structural differences are found for positions eight and nine of the ligand in the tripeptide C terminus. Leu 8 and Gly 9 in OT are located along the ring plane, with Gly 9 facing helix I (Figs. 1d and 4). Conversely, Arg 8 and Gly 9 in AVP, which in each reported V 2 R:AVP structure have been modelled differently, are facing away from the ring plane, enabling contacts to residues of ECL1, ECL3 and the N terminus of V 2 R. In the OTR, helix I adopts a position further away from the central axis of the receptor compared to V 2 R. This wider opening of the orthosteric pocket in OTR enables binding of the OT C-terminal residues Leu 8 and Gly 9 in the observed conformation. In V 2 R, however, OT binding would be sterically compromised due to a clash of Gly 9 with helix I. In contrast, the bound conformation of AVP in V 2 R is compatible with binding to OTR. Therefore, the positioning of helix I might explain why the OTR is not selective between OT and AVP, but OT binding is specific to OTR and it does not bind to V 2 R. In V 2 R, helix I is packed in a more compact manner in the helix bundle compared to the OTR, so it would clash with Gly 9 if OT were to be expected to adopt the identical binding conformation as observed in complex with OTR.
In variance to long-standing models 22,25 , we do not observe an interaction of OT with R34 N terminus or F103 ECL1 of the OTR. This might indicate an allosteric effect, or in the case of R34, which does not show clear density, a more dynamic interaction. We speculate that this could additionally contribute to subtype selectivity in the OT/AVP receptor family.
n.a. On the intracellular receptor side, the main conformational differences between OTR and V 2 R are found in the positioning of the helix VII-VIII transition region and the elongation of helix V of V 2 R compared to OTR ( Supplementary Fig. 6). Both regions contribute to G protein binding, and the respective differences are likely a feature of the diverse binding modes between the receptors and the interacting Gα subunits (V 2 R:G s and OTR:G o/q ).
G protein interaction. In the signalling complex of the activated OTR, the α5 helix of G o/q is bound in a crevice constituted by helices II, III, V, VI and VII at the receptor intracellular side, where the G protein C-terminal residues E350 and Y351 engage in hydrogen bonding with R73 2.38 and D136 3.49 of OTR (Fig. 5a, d). Compared to reported receptor:G s/q complexes 12,26,27 , the G q α5 helix in OTR:G o/q is rotated away from helix VI (Fig. 5b). This rotation cannot be attributed to differences in chimera design, as there is a perfect structural alignment of the individual G α subunits (Fig. 5e). Most importantly, the same rotation is also distinct from GPCR:G o strcutures [28][29][30] (Fig. 5c). Instead, the orientation of the G q α5 helix in OTR:G o/q very much resembles that of G s coupled to V 2 R, with the latter penetrating less deeply into the V 2 R intracellular TMD crevice, suggesting the receptor is governing the α5 orientation (Fig. 5f).
In conclusion, we report here the structure of the human OTR:OT signalling complex, providing insights into the subfamily-specific OTR activation mechanism and an unexpected G protein binding mode as well as the detailed OT binding mode, thereby enhancing our understanding of the subtype specificity within the closely related oxytocin and vasopressin receptor family. After the present manuscript had been submitted, a related structure with a different G protein has appeared, reaching similar conclusions 31 . Together, these findings are expected to greatly facilitate the development of novel therapeutics for the treatment of a variety of OTR-implicated diseases.

Methods
Design of complex constructs. The sequences of scFv16 14 and of wild-type human OTR (wtOTR), codon-optimised for expression in Spodoptera frugiperda (Sf9) (Cterminally truncated after residue 359), were cloned into a modified pFL vector (MultiBac system, Geneva Biotech) for Sf9 expression. The resulting expression constructs contained a melittin signal sequence, followed by a FLAG-tag, a His 10tag, and a human rhinovirus 3C protease cleavage site N-terminal to the gene of interest. To increase OTR purification yield, the mutation D153Y was introduced into the truncated wtOTR sequence, as identified previously 10 . It shows very similar KD and EC50 for oxytocin (OT) (Supplementary Fig. 1). The mutant was generated by sequence-and ligation-independent cloning as previously described in detail 32 . To generate an Gα o/q subunit that would allow interaction with scFv16 30 , the N-terminal 18 amino acids of G i1 were introduced to the engineered mini-G o1 12 11 . To generate G q -like interactions, residues H5.16, H5.17, H5.18, H5.22, H5.23, H5.24, and H5.26 (according to the common Gα numbering) 33 in the C-terminal helix were mutated to corresponding amino acids of G q . Finally, the engineered Gα o/q chimera sequence was cloned into one pFL vector, together with Gβ 1 (including an N-terminal non-cleavable His 10 -tag) and with Gγ 2 , with each gene under the control of its own polyhedrin promoter.
The receptor-bound resin was washed with 20 column volumes (CVs) of wash buffer I containing 50 mM Hepes (pH 7.5), 500 mM NaCl, 10 mM MgCl 2 , 5 mM Protein-containing fractions were combined and concentrated to 0.5 ml using a 50-kDa molecular weight cutoff (MWCO) Vivaspin 2 concentrator (Sartorius), and desalted by buffer exchange on a PD MiniTrap G-25 column (Cytiva) equilibrated with G25 buffer containing 50 mM Hepes (pH 7.5), 150 mM KCl, 10% (v/v) glycerol, 0.01% (w/v) LMNG, 0.001% (w/v) CHS, and 50 μM OT. To remove the N-terminal affinity tags and to deglycosylate the receptor, OT-bound receptor was treated with His-tagged 3 C protease and His-tagged peptide N-glycosidase (PNGase) F (both prepared in-house) overnight. To collect cleaved receptor, the reaction was incubated with Ni-nitrilotriacetic acid (Ni-NTA) resin (Cytiva) for 1 h, cleaved receptor was collected as the flow-through, then concentrated to~3 to 5 mg/ml with a 50-kDa MWCO Vivaspin 2 concentrator, and directly used for complex formation. Protein purity and monodispersity were assessed by LDS-polyacrylamide gel electrophoresis and analytical size exclusion chromatography (SEC) using a Nanofilm SEC-250 column (Sepax).
Purification of G o/q . Purification of the engineered heterotrimeric G protein was carried out similarly to receptor purification, with small adaptations. All buffers used were supplemented with 10 μM guanosine diphosphate (GTP, MilliporeSigma) and 100 μM tris(2-carboxyethyl)phosphine) (TCEP, Thermo Fisher Scientific). In contrast to the receptor purification, monovalent cation concentration never exceeded 150 mM, and all buffers were devoid of any receptor ligands and iodoacetamide. Enriched G protein-containing membranes were washed by Dounce homogenisation without high salt concentrations in physiological buffer, containing 10 mM Hepes (pH 7.5), 150 mM NaCl, 20 mM KCl, 10 mM MgCl 2 , and protease inhibitors. Solubilisation and immobilisation on TALON IMAC resin was performed as described above.
Purification of scFv16. ScFv16 was expressed and purified as described before 12       centrifugation, then pH-adjusted by addition of Hepes (pH 7.5). Metal-chelating agents from the cells and medium were quenched by incubation with 1 mM CoCl 2 and 5 mM CaCl 2 for 1 h at 22°C. Resulting precipitates were removed by centrifugation, and the filtrated supernatant was loaded onto a Co 2+ -loaded HiTrap IMAC HP column (Cytiva). The column was washed with 20 CVs of buffer A containing 20 mM Hepes (pH 7.5), 150 mM NaCl, 1 mM MgCl 2 , 4 mM ATP, and 5 mM imidazole, followed by 20 CVs of buffer B containing 20 mM Hepes (pH 7.5), 150 mM NaCl, and 30 mM imidazole. The protein was eluted with buffer C [20 mM Hepes (pH 7.5), 150 mM NaCl, and 300 mM imidazole]. Monomeric fractions were pooled, concentrated using a 10-kDa MWCO Amicon Ultra concentrator (Merck), and imidazole was removed by applying the concentrate to a PD-10 desalting column (Cytiva) equilibrated with G25-buffer containing 50 mM Hepes (pH 7.5), 150 mM NaCl, 10% (v/v) glycerol, 0.01% (w/v) LMNG, and 0.001% (w/v) CHS. ScFv16 was incubated for 3 h with His-tagged 3 C protease and His-tagged PNGase F for removal of affinity tags and deglycosylation. After incubation with Ni-NTA resin for 1 h, cleaved scFv16 was collected as the flowthrough, concentrated with a 10-kDa MWCO Amicon Ultra concentrator, and further purified using a Superdex 200 10/300 column equilibrated with SEC buffer.
Monomeric fractions were pooled, concentrated to~6 to 7 mg/ml, flash-frozen in liquid nitrogen, and stored at −80°C. Right before complex formation, scFv16 was thawed and the buffer was exchanged on a PD MiniTrap TM G-25 column (Cytiva) equilibrated with G25-buffer.
Single-particle cryo-EM data processing. All image stacks were binned to generate a pixel size of 0.65 Å followed by motion-correction and dose-weighting using MotionCor2 34 (version 1.4). All images were contrast transfer function corrected using Gctf 35 (version 1.06), as implemented in cryoSPARC 36 (version 3.0.1). Subsequent image processing steps were performed in cryoSPARC. Initial particle selection was done using the automated blob picker on 100 micrographs, using a 100 Å minimum and 150 Å maximum particle diameter, to extract a total of 38,166 particles. Next, the extracted particles were subjected to one round of 2D classification, into 200 classes, from which 7 classes were selected and used as a template for the automatic particle picking process. A total of 3,062,337 particles were extracted from the first 6,450 micrographs, followed by a round of 2D classification that resulted in 200 classes. Finally, a round of 3D reconstructions and classification produced 6 classes. A total of 3,504,800 particles were extracted from the second data-set of imagestacks, followed by a round of 2D classification, split into 200 classes, and a round of 3D reconstructions and classification into 3 classes. The particles from the best classes from both datasets were then joined together and subjected to one abinitio round of 3D reconstructions split into 6 classes. A final data-set of 392,370 particles from the best 3D classes were subjected to local and global CTF refinements, followed by a 3D non-uniform refinement. The final density map was resolved to 3.25 Å, after map sharpening, as determined by gold-standard Fourier shell correlation using the 0.143 criterion. Local resolution estimation was performed using cryoSPARC.
Model building. An initial model was created by docking the individual complex components into the cryo-EM map using the "fit in map" routine in UCSF Chimera 37 (version 1.15). The following structures from the Protein Data Bank (PDB) were used: OTR (PDB ID: 6TPK), Gα o (PDB ID: 6WWZ), Gβ 1 γ 2, scFv16 (PDB ID: 6OIJ), and AVP (PDB ID: 7DW9). All initial model components were manually rebuilt in Coot 38 (version 0.9.7), followed by several rounds of manual real-space refinement in Coot and real-space refinement with the software package Phenix.real_space_refine from Phenix 39 (version 1.20.1-4487). The quality of the models was assessed using MolProbity 40 before PDB deposition. PyMOL (version 2.5) was used for visual inspection, model comparison and figure preparation. IP1 accumulation assays. Ligand-induced IP1 accumulation and ligand-binding experiments were measured using transiently transfected Human Embryonic Kidney (HEK) 293 T/17 cells (American Type Culture Collection). The cells were cultivated in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific) supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml, Milli-poreSigma), and 10% (v/v) foetal calf serum (BioConcept) and maintained at 37°C in a humidified atmosphere of 5% CO 2 and 95% air. Transient transfections were performed with TransIT-293 (Mirus Bio) according to the manufacturer's instructions. Full-length OTR, truncated V 1A R (residues 1-378 with C-terminal mutation T378K) and mutants thereof were directly cloned into a mammalian expression vector containing an N-terminal SNAP-tag (pMC08; Cisbio). Cells were transfected and directly seeded at 7500 cells per well in poly-D-lysine-coated white 384-well plates (Greiner).
To compare IP1 accumulation a homogeneous time-resolved fluorescence (HTRF) signalling assay was performed, adapting a previously described protocol 41 . The cells were washed 48 h after transfection with phosphate-buffered saline (PBS) and stimulation buffer (Cisbio) and subsequently incubated for 1 h at 37°C with a concentration range of oxytocin (Psyclo Peptide Inc.) diluted in stimulation buffer. The IP1 accumulation was determined using the HTRF IP-One Kit (Cisbio) according to the manufacturer's protocol. Fluorescence intensities were measured on a Spark fluorescence plate reader (Tecan). To generate concentrationresponse curves, data were analysed by a three-parameter logistic equation in GraphPad Prism software (version 9.2.0).
Whole-cell ligand binding assays. Ligand-binding experiments were performed on whole cells for comparison of affinities for wild-type and receptor mutants using an HTRF binding assay as previously described 17 . Forty-eight hours posttransfection the cells were labelled with 50 nM SNAP-Lumi4-Tb (Cisbio) in labelling buffer (20 mM Hepes (pH 7.5), 100 mM NaCl, 3 mM MgCl 2 , and 0.05% (w/v) bovine serum albumin (BSA)) for 1.5 h at 37°C. The cells were washed two times with labelling buffer and two times with assay buffer (20 mM Hepes (pH 7.5), 100 mM KCl, 3 mM MgCl 2, and 0.05% (w/v) BSA)) and subsequently incubated for 1 h at room temperature with a concentration range of fluorescently labelled peptide HiLyte Fluor 488-Orn 8 oxytocin (Eurogentec) in assay buffer. Fluorescence intensities were measured on a Spark fluorescence plate reader with an excitation wavelength of 340 nm and emission wavelengths of 620 nm and 510 nm for Tb 3+ and the fluorophore HiLyte Fluor 488, respectively. The ratio of fluorescence resonance energy transfer (FRET) donor and acceptor fluorescence intensities was calculated (F510 nm/F620 nm). Nonspecific binding was determined in the presence of a 1000-fold excess of unlabelled oxytocin. Data were analysed by global fitting to a one-site saturation binding equation with the GraphPad Prism software.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Atomic coordinates of the OTR:G o/q :OT:scFv16 complex have been deposited in the PDB under the accession code 7QVM. Cryo-EM maps used have been deposited in the EMDB found under code EMD-14180. Source data for ligand binding and receptor activation are provided with this paper.