Main

The kallikrein–kinin system (KKS) is a poorly understood hormonal system involved in the regulation of blood pressure, inflammation, coagulation and pain control1,2,3. The main components of KKS include the metabolic products of kinin peptides, such as bradykinin, kallidin (Lys-bradykinin) and its carboxy-terminal des-Arg metabolites, derived from different kininogen isoforms4. These kinin peptides have highly conserved sequences, with kallidin differing from bradykinin only by an additional N-terminal lysine, while des-Arg10-kallidin lacks the C-terminal Arg relative to kallidin. They are potent vasodilators and proinflammatory peptides that activate bradykinin type 1 (B1R) and type 2 (B2R) receptors, two members of class A G protein-coupled receptors (GPCRs)5.

Hyperactivation of KKS, termed ‘bradykinin storm’, was reported to be closely related to COVID-19 pathogenesis6. Gene expression analyses of the bronchoalveolar lavage fluid from COVID-19 patients revealed a dramatic upregulation of B1R by ~3,000-fold and of B2R by ~200-fold6, respectively. The resulting ‘bradykinin storm’ is thought to be responsible for most of the COVID-19 symptoms, including vascular leakage and pulmonary edema, that are linked with hyperactivation of B1R and B2R7. As such, blockade of B1R and B2R activation has been proposed as a therapeutic option to prevent acute respiratory distress syndrome in patients with COVID-19 (ref. 8).

B1R and B2R bind to kinin-derived peptide hormones and mediate transmembrane (TM) signaling primarily through Gq pathways. B2R is expressed in many normal tissues, whereas B1R expression is only induced in tissues under pathological conditions, such as inflammation1,9. B1R and B2R share 34% identity in their amino acid sequences, which are predicted to form a canonical GPCR fold of seven-transmembrane (7TM) helices, with a conserved peptide-binding pocket10. Nevertheless, kinin peptides show different selectivity for bradykinin receptor subtypes. Specifically, bradykinin is one of the highest affinity kinin-derived peptides for B2R, but exhibits low affinity for B1R, with over 10,000-fold selectivity11,12,13. In contrast, des-Arg10-kallidin displays over 100,000-fold selectivity for B1R over B2R (Fig. 1a)11,12,13. Extensive efforts have been made in defining the pharmacophore of antagonists and the molecular basis of ligand selectivity for kinins and other nonpeptides using biochemical methods and molecular modeling14,15,16,17,18,19. However, the underlying mechanisms for these peptide hormone–receptor subtypes selectivity remain largely unknown due to the lack of structural evidence. Given their important physiological and pathological properties, it is of great value to elucidate molecular mechanisms for peptide recognition and bradykinin receptor activation. Here we report two cryo-EM structures of the B1R–Gq complex bound to des-Arg10-kallidin and the B2R–Gq complex bound to bradykinin. Combined with mutagenesis and functional analyses, our findings provide insight into specific recognition of kinin-derived peptide hormones by B1R and B2R and the molecular basis for receptor activation and Gq protein coupling.

Fig. 1: Cryo-EM structures of the des-Arg10-kallidin–B1R–Gq and bradykinin–B2R–Gq complexes.
figure 1

a, Schematic illustration of subtype selectivity for kinin and Gq protein coupling of bradykinin receptors. Sequences of des-Arg10-kallidin and bradykinin are shown. b,c, Orthogonal views of the density map (b) and model (c) for the des-Arg10-kallidin–B1R–Gq complex. d,e, Orthogonal views of the density map (d) and model (e) for the bradykinin–B2R–Gq–scFv16 complex. Des-Arg10-kallidin is shown in cyan, des-Arg10-kallidin-bound B1R in salmon; bradykinin is displayed in orange, bradykinin-bound B2R in purple. The Gq heterotrimer is colored by subunits: Gαq, yellow; Gβ, pale green; Gγ, tomato; scFv16, gray.

Full size image

Results

Structure determination of kinin-bound B1R and B2R

To stabilize B1R–Gq and B2R–Gq complexes, we applied the NanoBiT tethering method, a general strategy that has been used to obtain the structures of several GPCR–G protein complexes (Supplementary Fig. 1)20,21,22. An engineered Gαq chimera was generated on the basis of the mini-Gαs/q71 scaffold with its N terminus replaced by corresponding sequences of Gαi1 to facilitate the binding of scFv16 (Supplementary Fig. 2)23,24. This analogous approach had been used in the structure determination of the 5-HT2AR–Gq complex25. Unless otherwise specified, Gq refers to Gq chimera used in the structure determination. Meanwhile, both B1R and B2R bear a tryptophan mutation at position 3.41 (F1263.41W for B1R and C1463.41W for B2R, superscripts refer to Ballesteros–Weinstein numbering26), a known mutation that enhanced GPCR thermal stabilization27,28. Both complexes were efficiently assembled on the membrane by coexpressing receptors with Gαq, Gβ1 and Gγ2 subunits (Extended Data Figs. 1a and 2a).

The structure of the des-Arg10-kallidin–B1R–Gq complex was defined with 633,636 final particles from 3,681,755 initial particles to a global nominal resolution of 3.0 Å (Fig. 1b,c, Table 1 and Extended Data Fig. 1). The structure of the bradykinin–B2R–Gq complex was determined with 664,416 final particles from 3,460,328 initial particles to a global nominal resolution of 2.9 Å (Fig. 1d,e, Table 1 and Extended Data Fig. 2). The overall conformation comparison shows highly similarity between two receptors, with a root mean squared deviation (r.m.s.d.) of 1.0 Å. Kinin peptides, TM bundles, extracellular loops (ECLs) and intracellular loops (ICLs), except ICL3, of both receptors show clear densities, enabling near-atomic modeling for the two complexes. The majority of amino acid side chains were well resolved in the refined final model (Fig. 1b–e and Extended Data Fig. 3). Thus, these two structures can provide detailed structural information of the peptide-binding pockets and receptor-Gq interaction interfaces.

Table 1 Cryo-EM data collection, model refinement and validation statistics
Full size table

Molecular basis of bradykinin recognition by B2R

Bradykinin (RPPGFSPFR) occupies the orthosteric binding pocket comprising TM helices and ECLs, except for TM1 and ECL3 (Extended Data Fig. 4 and Supplementary Fig. 3). It presents an S-shaped overall conformation, with its C terminus inserting deeply into the transmembrane domain (TMD) core (Fig. 2a). This S-shaped fold is stabilized by two intramolecular hydrogen bonds between the main chain of G5b and R10b, as well as the backbone of S7b and R10b (Fig. 2b).

Fig. 2: The bradykinin-binding pocket in B2R.
figure 2

a, Cross-section of the bradykinin-binding pocket in B2R. The cryo-EM density of bradykinin is highlighted. Side chains of the residues are displayed as sticks. Bradykinin is displayed in orange, B2R is colored in purple. b–d, Detailed interactions of bradykinin with residues in B2R. The binding sites of R2b and R10b (b), P3b-P8b (c) and F9b (d) are shown. Hydrogen bonds and salt bridge are depicted as red dashed lines.

Full size image

The sequence of bradykinin features two positively charged arginines residing at both the N and C terminus and the majority of hydrophobic amino acids at the middle segment of the peptide. The N-terminal R2 in bradykinin (refers to R2b) constitutes a stabilizing polar interaction network with ECL2, TM6 and TM7. The side chain of R2b forms polar interactions with Y201ECL2 and E204ECL2. Its main chain NH group makes polar interactions with E3077.28 and D3117.32, while its backbone CO group builds a hydrogen bond with R2976.62 (Fig. 2b). The C-terminal R10b is also involved in polar interactions between bradykinin and B2R. Although the density of the guanidino group of R10b is weak (Fig. 2a), it is indicative that the side chain of R10b forms a hydrogen bond with T3147.35, which is supported by diminished activity of bradykinin for B2R with the T3147.35A mutation (Fig. 2b, Extended Data Fig. 5 and Supplementary Table 1). Besides R2b and R10b, the main chain CO group of G5b makes a hydrogen bond with R1964.64, which forms a salt bridge with the free carboxylic acid group of R10b (Fig. 2c). The polar interaction network is essential for bradykinin-induced B2R activation, since substituting R1964.64 with alanine entirely abolishes the activity of bradykinin (Extended Data Fig. 5 and Supplementary Table 1).

P3b, P4b, F6b, P8b and F9b face hydrophobic environments within the B2R TMD pocket. P3b and P4b interact with the aromatic ring of Y201ECL2 and Y215ECL2, respectively (Fig. 2c). Y201ECL2, together with F121ECL1, makes hydrophobic contact with F6b, which is also supported by the mutagenesis analysis (Fig. 2c, Extended Data Fig. 5 and Supplementary Table 1). P8b is surrounded by hydrophobic residues of TM2 (W1132.60), TM3 (I1373.32) and TM7 (F3197.40) (Fig. 2c). F9b inserts deeply into a potent hydrophobic core comprised of residues in TM2 (L1062.53), TM3 (I1373.32), TM5 (L2285.42), TM6 (W2836.48 and F2866.51) and TM7 (Y3227.43) (Fig. 2d). Alanine mutations of these hydrophobic residues, except for I1373.32, show a notable impact on bradykinin-induced B2R activation, indicating a potentially critical role of these hydrophobic residues near F9b for bradykinin binding or B2R activation (Extended Data Fig. 5 and Supplementary Table 1). Together, these detailed structural analyses provide important information to better understand the recognition mechanism of bradykinin by B2R.

Molecular basis of des-Arg10-kallidin recognition by B1R

Des-Arg10-kallidin (KRPPGFSPF) shows high selectivity for B1R over B2R. Compared with bradykinin, des-Arg10-kallidin shares a conserved middle segment (RPPGFSPF) and sits in an almost identical orthosteric binding pocket with a similar S-shaped conformation (Fig. 3a, Extended Data Fig. 4 and Supplementary Fig. 3). Nevertheless, distinct interactions are observed between two peptides and corresponding receptor subtypes, proving the basis for their receptor selectivity as described below.

Fig. 3: The des-Arg10-kallidin-binding pocket in B1R.
figure 3

a, Cross-section of the des-Arg10-kallidin-binding pocket in B1R. The cryo-EM density of des-Arg10-kallidin is highlighted. Side chains of the residues are displayed as sticks. b, A sequence and conformation comparison of des-Arg10-kallidin and bradykinin. Two intramolecular hydrogen bonds are depicted as red dashed lines. Des-Arg10-kallidin is displayed in cyan, and bradykinin is colored in orange. c, Detailed interaction between K1k and residues in B1R. d, Comparison of the binding mode of R2b and R2k. The movement of ECL2 in B1R relative to that in B2R are highlighted in a red arrow. The salt bridges are shown as red dashed lines. Side chains of des-Arg10-kallidin and residues in two receptors are shown as sticks.

Full size image

In contrast to bradykinin, des-Arg10-kallidin has an additional lysine (K1k) at its N terminus but lacks arginine that is located at the C terminus of bradykinin (R10b) (Fig. 3b). Two extra intramolecular hydrogen bonds exist between K1k and the backbone CO group of R2k, as well as the backbone CO group of P3k and NH group of F6k, causing a minor conformational change of des-Arg10-kallidin (Fig. 3b). The additional N-terminal K1k forms a polar interaction with E2736.58, E2877.28 and D2917.32 (Fig. 3c). Intriguingly, these residues are conserved in B2R (D2936.58, E3077.28 and D3117.32, Supplementary Fig. 3), which may explain the comparable B2R activation potency of Lys-bradykinin relative to bradykinin12. Compared with R2b in bradykinin, the equivalent R2k in des-Arg10-kallidin presents distinct interactions with TM1 and ECL2. R2k forms a cation-pi interaction with W331.28 of B1R, while R2b pushes W531.28 away from the binding pocket owing to the steric hindrance (Fig. 3d). Additionally, ECL2 of B1R displays a smaller shift towards the peptide-binding pocket relative to B2R, which may be attributed to the lack of a corresponding salt bridge observed between R2b and E204ECL2 in B2R (Fig. 3d).

Molecular basis of kinin selectivity for B1R and B2R

Comparison of the binding modes between the two kinin peptides provides a framework for understanding kinin peptide selectivity by B1R and B2R. The free carboxylic acid backbone of F9k engages a positively charged binding pocket and forms electrostatic interactions with K1183.33 and R2025.38 in B1R, which are not conserved in B2R (Fig. 4a). The cognate residues S1383.33 and T2245.38 in B2R fail to create a similar electrostatic environment, raising a hypothesis that the electrostatic pocket consisting of K1183.33 and R2025.38 is the determinant for selective binding of des-Arg10-kallidin to B1R over B2R. This hypothesis is supported by our mutagenesis studies showing that single or combined substitutions of K1183.33 and R2025.38 in B1R with serine and threonine, the equivalent residues in B2R, abolished the activity of des-Arg10-kallidin (Fig. 4b,f and Supplementary Table 2). Our results are consistent with previous reports showing that K1183.33 attracts the negative charge of the C terminus of B1R-selective peptides and serves as a key residue in the selectivity of C-terminal des-Arg kinin peptides for B1R18,29.

Fig. 4: Molecular basis of kinin peptides selectivity for bradykinin receptors.
figure 4

a, Detailed interaction between F9k and residues in B1R. F9k and residues in B1R and the corresponding residues in B2R are shown as sticks. b, Effects of mutations in the F9k-binding pocket on calcium responses. Data are presented as mean ± s.e.m. of three independent experiments. c, The binding site of R10b. Residues in the R10b-binding site in B2R and cognate residues in B1R are shown. The steric clash between side chain of R10b and residues in B1R are highlighted as black dashed ovals. Polar interactions in a and c are shown as red dashed lines. d, A larger R10b-binding pocket in B2R relative to B1R. The pockets are highlighted as black dashed ovals. e, Effects of mutations in the R10b-binding pocket on calcium responses. e, Schematic model of the molecular basis of kinin peptide selectivity for bradykinin receptors. The chemical structures of F9k in des-Arg10-kallidin as well as F9b and R10b in bradykinin are displayed. Residues in the positively charged pocket of B1R and the large binding pocket of B2R are highlighted. Data for b and e are displayed in Supplementary Table 2. Each data point presents mean ± s.e.m. of three independent experiments. WT, wild type. Source data for b and e are available online.

Source data

Full size image

Compared with T2245.38, F2866.51 and D2936.58 in B2R, the cognate residues in B1R (R2025.38, Y2666.51 and E2736.58) are bulkier, resulting in insufficient space for interaction with the side chain of R10b (Fig. 4c,d). The role of these residues in B2R selectivity for bradykinin is identified by swapping functional analysis. The triplicate swapping of T2245.38/F2866.51/D2936.58 with the cognate residues with larger side chains in B1R remarkably impaired bradykinin activity (Fig. 4e and Supplementary Table 2). This finding suggests that a larger pocket consisting of T2245.38, F2866.51 and D2936.58 is crucial to bradykinin selectivity for B2R over B1R (Fig. 4f). Together, these data reveal the determinants of bradykinin receptor selectivity between bradykinin and des-Arg10-kallidin.

However, when mutating these kinin selectivity-related residues to cognate ones, only the T2245.38 R mutation in B2R showed slightly increased activity of des-Arg10-kallidin. Other residue substitutions did not cause substantially increased activities of bradykinin and des-Arg10-kallidin for B1R and B2R, respectively (Extended Data Fig. 5i,j). It seems that only swapping the residues in the electrostatic pocket in B1R or a larger pocket in B2R failed to make the two receptors possess high affinity for kinins. Thus, we believe that the residues in these two pockets are not entirely responsible for kinin selectivity.

Activation mechanism of B1R and B2R

A structural comparison of B1R and B2R complexes to their closely related angiotensin II receptor type 1 (AT1R) in the inactive (PDB 4YAY)30 and active states (PDB 6OS0)31 sheds light on the basis of bradykinin receptor activation. The structural comparison demonstrates that both B1R and B2R adopt fully active conformations similar to the active AT1R (Fig. 5a). Compared with the inactive AT1R, they show a remarkable outward displacement of the cytoplasmic end of TM6, a hallmark of class A GPCR activation, along with an inward movement of the TM7 cytoplasmic end (Fig. 5a,b)32.

Fig. 5: Activation mechanism of bradykinin receptors.
figure 5

a,b, Structural superposition of two active bradykinin receptors, inactive AT1R (PDB 4YAY), and active AT1R (PDB 6OS0) from the side (a) and cytoplasmic (b) views. The movement directions of TM6 and TM7 in bradykinin receptors relative to inactive AT1R are highlighted as red arrows. AT1R, angiotensin II receptor type 1. Inactive AT1R, active AT1R, B1R and B2R are colored in gray, yellow, salmon and purple, respectively. c–f, Conformational changes of the conserved ‘micro-switches’ upon receptor activation, including toggle switch (c), PIF (d), DRY (e) and NPxxY (f) motifs. F9b/F9k-triggered conformational changes of W6.48 and Y7.43 are highlighted. The conformational changes of residue side chains are shown as red arrows upon receptor activation. The complex structures were aligned by the receptors.

Full size image

Although bradykinin and des-Arg10-kallidin present different binding selectivity, they may activate bradykinin receptors through a common mechanism. The side chains of F9b and F9k insert into a conserved hydrophobic crevice at the bottom of the peptide-binding pocket and trigger rotameric switch of W6.48, the toggle switch residue, which further facilitates the swing of F6.44 and initiates the rotation of TM6 (Fig. 5c). Meanwhile, the steric clash between F9k/F9b and F/Y7.43 would drive the latter swinging away from the receptor helical core and the inward shifting of the cytoplasmic end of TM7 (Fig. 5c). [Leu9, des-Arg10]kallidin, in which F9k of des-Arg10-kallidin is substituted with a smaller bulky amino acid (leucine), loses its agonistic activity with conversion to a B1R antagonist, supporting the critical role of F9k in B1R activation33,34. The switches of W6.48 and F/Y7.43 further trigger the active-like conformational changes of ‘micro-switch’ residues (toggle switch W6.48 and PIF, DRY and NPxxY motifs), leading to an agonism signal transduction to the cytoplasmic end of the receptor (Fig. 5c–f).

Structural comparison of B1R and B2R with their closely related class A GPCR member AT1R in the active state (PDB 6OS0)31 suggests a common mechanism of receptor activation. The bound endogenous peptide hormones des-Arg10-kallidin, bradykinin and angiotensin II share conserved C-terminal phenylalanine, which inserts into the peptide-binding pockets of corresponding receptors at a comparable depth (Extended Data Fig. 6a,b). Moreover, although differing in side chain orientations, these phenylalanines are buried within a similar hydrophobic environment, indicating a universal activation mechanism of these closely related GPCRs (Extended Data Fig. 6c).

Structural superposition of Gq-coupled B1R and B2R complexes with the Gq-coupled 5-HT2AR (PDB 6WHA)25 and G11-coupled M1R (PDB 6OIJ)35 by receptors shows nearly identical conformations of TM6 and TM7 (Extended Data Fig. 7a), suggest that Gq-coupled receptors have a similar conformation to that of G11-coupled receptors. In addition, compared with these two Gq/11-coupled receptors, the helix 8 of B1R and B2R is closer to the Gβ subunit, which may be attributed to the intramolecular salt bridge formed between K8.53 and E2.40 (Extended Data Fig. 7b). On the G protein side, the α5 helix of Gαq shifts 4 Å for both B1R and B2R compared with that of G11-coupled M1R (measured at Cα of YH5.23) and moves a half-helical turn upward towards the cytoplasmic cavity of the receptor (2 Å for B1R and 3 Å for B2R, respectively) relative to the Gq-coupled 5-HT2AR. Meanwhile, the Gαq N termini of B1R and B2R undergo notable shifts as seen across the Gq/11-coupled class A GPCRs (Extended Data Fig. 7c).

Discussion

Bradykinin receptors are involved in various clinical symptoms and their use as therapeutic targets remains the focus of extensive investigations. Recently, decoding the bradykinin inflammatory pathway in COVID-19, known as the ‘bradykinin storm’, highlights the implications of bradykinin receptor modulators as a potential treatment for COVID-19. In this study, we determined two Gq-coupled structures of the human B1R and B2R bound to selective kinin peptides: des-Arg10-kallidin and bradykinin, respectively. In combination with functional analyses, these structures enhance our understanding of the molecular basis of kinin peptide recognition and activation of B1R and B2R.

Intriguingly, it has recently been predicted that des-Arg10-kallidin and bradykinin show distinct V- and S-shaped conformations, respectively. These distinct conformations result in different presentations of the N and C termini of kinin peptides towards their receptors18. In our structural model, bradykinin adopts an overall similar S-shaped conformation, but notably differs in the orientation of the C-terminal charged arginine (R10b), with an overall r.m.s.d. of 1.8 Å (Extended Data Fig. 8a). The side chain of R10b sits in the gap between TM6 and TM7 and forms a salt bridge with T3147.35. In contrast, R10b points to TM5 and forms a salt bridge with E2215.35 in the predicted model. It is worth noting that des-Arg10-kallidin in the structural model shows an entirely different conformation, presenting an S-shaped but not a V-shaped fold, with an r.m.s.d. of 3.4 Å (Extended Data Fig. 8b). Even so, the middle segment of des-Arg10-kallidin (P3k-F6k) in both our structure and NMR model displays a similar β-turn-like conformation, which might be stabilized by the intramolecular hydrogen bond made by the backbone NH of F6k and CO of P3k, as observed in the B1R complex structure. A des-Arg10-kallidin analog with the methylated amide of F6k, which may disturb this hydrogen bond and central β-turn, showed a 1,000-fold lower binding affinity for B1R than the native peptide, indicating the importance of this intramolecular interaction in maintaining the conformation stability of des-Arg10-kallidin18. It was predicted that the C-terminal segment of bradykinin (S7b-R10b) adopted a β-turn conformation, which might be one of the requirements for high affinity to B2R17. The β-turn constitutes a molecular basis for designing B2R ligands, including the only approved B2R antagonist, icatibant16,17. Consistently, we also observed a similar β-turn conformation, which is stabilized by two intramolecular hydrogen bonds in the B2R structure. This β-turn forces F9b to insert deeply into the TMD core and engage with hydrophobic residues at the bottom of the B2R pocket. Our alanine mutagenesis analysis on these hydrophobic residues further supports a potential role of the β-turn conformation of bradykinin in B2R activation.

Bradykinin receptors exhibit exquisite selectivity for bradykinin and des-Arg10-kallidin, and our findings provide a framework to depict the subtype selectivity of kinin peptides for B1R and B2R. It was found that the residue environments surrounding the C termini amino acids of bradykinin and des-Arg10-kallidin are critical to such selectivity. K1183.33 and R2025.38, which constitute a positively charged pocket interacting with the free carboxylic acid group of F9k, are determinants of the preference of des-Arg10-kallidin for B1R. Additionally, the hydrogen bond between R2025.38 and Y2666.51 in B1R creates an inaccessible space for bradykinin. Meanwhile, the smaller side chains of T2245.38, F2866.51 and D2936.58 in B2R, relative to equivalent residues in B1R, create a larger pocket space to accommodate R10b, thereby revealing the molecular basis of the higher B2R selectivity by bradykinin. Coincidentally, the NMR model predicted the same hydrogen bond between R2025.38 and Y2666.51 (ref. 18). A molecular modeling study on B1R and B2R also speculated that the smaller size of T5.38 in B2R, relative to cognate residue R5.38 in B1R, allows nonpeptide antagonists to access an aromatic pocket composed of W6.48, F6.51 and Y7.43, which is inaccessible for B1R15,16,19. However, R/T5.38 in B1R and B2R cannot hamper the engagement of kinins with this aromatic pocket, which accommodates F9k/F9b in our structures. Conversely, R/T5.38 is involved in the binding of R10b to a larger pocket in B2R, which partially determines bradykinin selectivity for B2R over B1R. Additionally, we further propose a common activation mechanism for B1R and B2R, through structural comparison with AT1R. With an in-depth knowledge of ligand selectivity and receptor activation, new opportunities will arise to design potent and efficacious modulators of B1R and B2R for the treatment of inflammation, cardiovascular disorders and COVID-19.

Methods

Construct cloning

Homo sapiens B1R, residues 2–350 of the 353 residues (UniProt accession: P46663) with an N-terminal thermostabilized apocytochrome b562RIL (BRIL)36 and a C-terminal LgBiT were cloned into pFastBac. Before BRIL, there are HA and FLAG tags followed by a His10 (H10) tag, as well as a TEV cleavage site. For Homo sapiens B2R (UniProt accession: P30411), residues 40–370 of the 391 residues with an N-terminal HA tag followed by BRIL and a C-terminal LgBiT were cloned into pFastBac using homologous recombination (CloneExpress One Step Cloning Kit, Vazyme). Another TEV cleavage site and the tandem maltose-binding protein tag after LgBiT were added to both B1R and B2R constructs to facilitate expression and purification. The engineered Gαq construct was generated on the basis of mini-Gs/q71 (ref. 23) with two dominant-negative mutations (corresponding to G203A and A326S)37 to decrease the affinity of nucleotide binding. The N-terminal 1–18 amino acids and the α-helical domain of the mini-Gs/q71 were replaced by the corresponding sequences of the human Gαi1, providing possible binding sites for two antibody fragments scFv16 and Fab-G50, respectively35,38. Rat Gβ1 with an N-terminal His6 tag was followed by HiBiT at its C terminus. The engineered Gαq, Gβ1 and bovine Gγ2 were cloned into the pFastBac vector (Invitrogen).

Expression and purification of Nb35

Nb35 (ref. 39) with a C-terminal His6 tag was expressed in Escherichia coli BL21 (DE3) bacteria, and cultured in LB medium with 50 μg ml−1 ampicillin to an optical density (OD600) value of 0.6–1.0 at 37 °C, 180 r.p.m. IPTG (1 mM) was added to induce expression at 27 °C, 180 r.p.m for 8 h. E. coli bacteria were then collected by centrifugation (4,000 r.p.m., 20 min) and disrupted in 20 mM HEPES, pH 7.4, 100 mM NaCl, 10% glycerol and 1 mM PMSF. Cell pellets were removed by centrifugation (8,000 r.p.m., 30 min) and the supernatant was purified by nickel affinity chromatography (Ni Smart Beads 6FF, Smart Life Sciences). The resin was washed with 30 column volumes of buffer containing 20 mM HEPES pH 7.4, 100 mM NaCl, 25 mM imidazole, 10% glycerol, and the eluted protein was collected with buffer containing 20 mM HEPES pH 7.4, 100 mM NaCl, 200 mM imidazole and 10% glycerol. The eluted Nb35 was concentrated and subjected to a HiLoad 16/600 Superdex 75 column (GE Healthcare) pre-equilibrated with buffer containing 20 mM HEPES pH 7.4 and 100 mM NaCl. The monomeric fractions were collected and stored with 30% (v/v) glycerol at −80˚C for future use.

Expression and purification of scFv16

ScFv16 with a C-terminal His8 tag was expressed in High Five insect cells and purified as previously described24. In detail, the cells infected with scFv16 virus for 48 h were removed by centrifugation (2,000 r.p.m., 20 min). The supernatant was balanced with Tris pH 8.0 and then quenched with chelating agents (1 mM NiCl2 and 5 mM CaCl2) at room temperature for 1 h. Precipitates were removed by centrifugation (8,000 r.p.m., 30 min) and the supernatant was purified by nickel affinity chromatography as described above.

Expression and purification of B1R/B2R–Gq complex

High Five cells were infected with viruses of the receptor (B1R or B2R), Gαq, Gβ1 and Gγ2 in the ratio of 1:1:1:1 for 48 h at 27 ˚C. The cell pellets were lysed by dounce homogenization in 20 mM HEPES pH 7.4, 100 mM NaCl, 10 mM MgCl2, 5 mM CaCl2, 10% glycerol and EDTA-free protease inhibitor cocktail (TargetMol). The supernatant was then centrifuged at 30,000 r.p.m. for 30 min to collect the membranes. The washed membranes were re-suspended in 20 mM HEPES pH 7.4, 100 mM NaCl, 10 mM MgCl2, 5 mM CaCl2, 10% glycerol, 40 μM peptide (des-Arg10-kallidin for B1R and bradykinin for B2R, respectively, GenScript), 25 mU ml−1 apyrase (Sigma-Aldrich), 100 μM TCEP (Sigma-Aldrich), EDTA-free protease inhibitor cocktail and 20 μg ml−1 Nb35, and incubated at 4˚C overnight. For the B2R–Gq complex, an additional 20 μg ml−1 scFv16 was added. After incubation, 0.5% (w/v) n-dodecyl-β-d-maltopyranoside (DDM, Anatrace) and 0.1% (w/v) cholesteryl hemisuccinate (CHS, Anatrace) were used for solubilization at 4˚C for 3 h. The supernatant was collected by centrifugation at 30,000 r.p.m. for 30 min and then incubated with dextrin resin (Dextrin Beads 6FF, Smart Life Sciences) at 4˚C for 3 h. The resin was collected by centrifugation at 500g for 8 min, loaded onto a gravity flow column and washed with 10 column volumes of buffer containing 20 mM HEPES pH 7.4, 100 mM NaCl, 10 mM MgCl2, 5 mM CaCl2, 10% glycerol, 100 μM TCEP, 40 μM peptide, 0.05% (w/v) DDM and 0.01% (w/v) CHS. The detergent of washing buffer was then displaced by 0.1% (w/v) lauryl maltose neopentylglycol (LMNG, Anatrace) and 0.02% (w/v) CHS for 10 column volumes washing, followed by 0.03% (w/v) LMNG, 0.01% (w/v) glyco-diosgenin (GDN, Anatrace) and 0.008% (w/v) CHS for 20 column volumes washing. His-tagged TEV protease was then added and incubated with resin at 4 ˚C overnight. The flow-through was collected and concentrated with an Amicon Ultra Centrifugal Filter (MWCO 100 kDa) and loaded onto a Superdex 200 10/300 GL column (GE Healthcare) with running buffer containing 20 mM HEPES pH 7.4, 100 mM NaCl, 2 mM MgCl2, 100 μM TCEP, 40 μM peptide, 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN and 0.0002% (w/v) CHS. The fractions of monomeric protein complex were collected and concentrated with an Amicon Ultra Centrifugal Filter (MWCO 100 kDa) by 30–50-fold for sample preparation and detection by cryo-EM.

Cryo-EM grid preparation and data collection

For cryo-EM grid preparation of the bradykinin–B2R–Gq complex, 3 μl of purified protein (21 mg ml−1) was loaded onto a glow-discharged holey carbon grid (Quantifoil, Au300 R1.2/1.3) using a Vitrobot chamber (FEI Vitrobot Mark IV). Cryo-EM images were collected by a FEI Titan Krios at 300 kV accelerating voltage equipped with a Gatan K3 Summit direct electron detector at the Center of Cryo-Electron Microscopy Research Center, Shanghai Institute of Materia Medica, Chinese Academy of Sciences. Micrographs were recorded with a pixel size of 1.045 Å. In total 3,437 movies were obtained at a dose of 80 electrons per Å2 for 36 frames.

For the des-Arg10-kallidin–B1R–Gq complex, 7.9 mg ml−1 of purified protein was used for cryo-EM grid preparation, as described above. Cryo-EM images were collected by Titan Krios G3i at 300 kV accelerating voltage at Shuimu BioSciences. The microscope was operated in super-resolution counting mode at a pixel size of 0.54 Å, and a total of 2,779 movies were obtained at a dose of 61.8 electrons per Å2 for 32 frames.

Cryo-EM data processing three-dimensional reconstruction

Cryo-EM data of the des-Arg10-kallidin–B1R–Gq and bradykinin–B2R–Gq complexes were processed using RELION v.3.1.0 (ref. 40). The detailed flowchart of data processing is shown in Extended Data Figs. 1b–d and 2b–d. Dose-fractionated image stacks for both complexes were subjected to dose-weighting and beam-induced motion correction using MotionCor v. 2.1 (ref. 41). Contrast transfer function (CTF) was carried using CTFFIND v.4.1 (ref. 42). The subsequent particle selection, two-dimensional (2D) and three-dimensional (3D) classifications for the des-Arg10-kallidin–B1R–Gq and bradykinin–B2R–Gq complexes were performed on a binned dataset with a pixel size of 1.08 Å and 1.045 Å, respectively.

For the des-Arg10-kallidin–B1R–Gq complex, autopicking yielded 3,681,755 particle projections that were subjected to two rounds of reference-free 2D classifications to extract particles in well-defined classes. The extracted 2,486,616 particles were used to generate a 3D initial model for further processing. With the initial model, two rounds of maximum-likelihood-based 3D classifications were carried out, resulting in one well-defined subset with 796,080 particle projections. A further two rounds of 3D classifications were conducted with the mask on the receptor and G protein, respectively, in which 633,636 particles were subjected to 3D autorefinement, three rounds of CTF refinement and Bayesian polishing. A map with an indicated global resolution of 3.0 Å at a Fourier shell correlation (FSC) of 0.143 was generated from the final 3D refinement, and subsequently post-processed by DeepEMhancer43.

For the bradykinin–B2R–Gq complex, autopicking yielded 3,460,328 particle projections that were subjected to two rounds of reference-free 2D classifications, producing 2,623,863 particle projections for further processing. With the initial model, three rounds of maximum-likelihood-based 3D classifications were carried out, resulting in one well-defined subset with 664,416 particle projections. A map with an indicated global resolution of 2.9 Å at a FSC of 0.143 was generated from the final 3D refinement, and subsequently post-processed by DeepEMhancer. Local resolutions for density maps of the des-Arg10-kallidin–B1R–Gq and bradykinin–B2R–Gq complexes were determined using the ResMap package with half maps as input maps44.

Model building and refinement

Homology models of active-state B1R and B2R were built by SWISS-MODEL45 using AT1R (PDB 6OS0)31 and AT2R (PDB 6JOD)46 as template models for the receptors, respectively. The Gq heterotrimer was built on the basis of the corresponding G protein of the 5-HT2AR-Gq (PDB 6WHA)25 complex as a template. All models were fitted into the cryo-EM density map using Chimera47 followed by a manual adjustment in Coot48. The generated final model was refined in Phenix49. The final refinement statistics are provided in Table 1. Structural figures were prepared with Chimera, ChimeraX50 or PyMOL (https://pymol.org/2/).

Calcium mobilization assay

HEK 293T cells were transiently transfected with wild-type or mutant bradykinin receptors after being seeded into 96-well plates at a density of 50,000 cells per well and incubated for 24 h at 37 °C in 5% CO2. The cells, from which medium had been removed, were reacted with 2 mM Fluo-4-AM in HBSS and 20 mM HEPES supplemented with 2.5 mM probenecid for 45 min at 37 °C in 5% CO2. Different concentrations of bradykinin or des-Arg10-kallidin were added, followed by an analysis of intracellular calcium mobilization on a FLIPR instrument (Molecular Devices) with excitation at 470–495 nm and emission at 515–575 nm. Data were normalized to the baseline response of the ligand.

Statistics

All functional study data were analyzed using GraphPad Prism v.8.0 (Graphpad Software) and shown as mean ± s.e.m. from at least three independent experiments. Concentration–response curves were evaluated with a three-parameter logistic equation. The significance was determined with two-tailed Student’s t-test, and P < 0.05 was considered statistically significant.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.