Structural basis of GABAB receptor-Gi protein coupling

Abstract

G-protein-coupled receptors (GPCRs) have central roles in intercellular communication^1,2. Structural studies have revealed how GPCRs can activate G proteins. However, whether this mechanism is conserved among all classes of GPCR remains unknown. Here we report the structure of the class-C heterodimeric GABA_B receptor, which is activated by the inhibitory transmitter GABA, in its active form complexed with G_i1 protein. We found that a single G protein interacts with the GB2 subunit of the GABA_B receptor at a site that mainly involves intracellular loop 2 on the side of the transmembrane domain. This is in contrast to the G protein binding in a central cavity, as has been observed with other classes of GPCR. This binding mode results from the active form of the transmembrane domain of this GABA_B receptor being different from that of other GPCRs, as it shows no outside movement of transmembrane helix 6. Our work also provides details of the inter- and intra-subunit changes that link agonist binding to G-protein activation in this heterodimeric complex.

Fig. 1. Cryo-EM structure of GABA_B–G_i complex.

a, b, Cryo-EM map (a) and model (b) of the baclofen- and BHFF-bound GABA_B–G_i1 complex.

Fig. 2. Asymmetric activation of GABA_B.

a, Side, extracellular, and intracellular views of the superposed structures of the agonist-bound (PDB 6UO9) and the agonist- and PAM-G_i-bound (agonist/PAM-bound) GABA_B, aligned by the TMD of GB1. b, Conformational changes of the TMD of GB2 between antagonist-bound (PDB 7C7S) and agonist- and PAM–G_i-bound structure. c, Magnified views of the critical residue F568, the bulky side chain of which undergoes a substantial rotation upon activation and causes the TM3 shifting. d, Baclofen-induced IP1 accumulation of wild-type (WT) and F568A-mutant GABA_B using the chimeric Gα protein Gα_qi9. Data are mean ± s.e.m. from six independent experiments, performed in technical triplicate. e, Magnified views of the ‘ionic lock’ located in the cytoplasmic TMD of GB2. f, Baclofen-induced IP1 accumulation of wild-type GABA_B and several forms of GABA_B with substitutions in the ionic lock region using Gα_qi9. Data are mean ± s.e.m. from at least three independent experiments, performed in technical triplicate. Source data

Fig. 3. GABA_B–G_i coupling and G-protein selectivity.

a, The G_i1 binding pocket in GABA_B, which is mainly formed by three intracellular loops of GB2. GB2, green; Gα_i1, yellow. b, c, Detailed interactions of the ICL2 and TM3 of GB2 with Gα_i (b), and of ICL1 and ICL3 with Gα_i (c). d, Baclofen-induced IP1 accumulation using Gα_qi9. Bars represent differences in calculated E_max and basal activity or potency (pEC₅₀) for each mutant as a percentage of the maximum in wild type. Data are mean ± s.e.m. from at least three independent experiments, performed in technical triplicate and analysed using one-way analysis of variance with Dunnett’s multiple comparison test to determine significance (compared with wild type). ND, not determined; NS, not significant. e, The C^G.H5.23 and G^G.H5.24 residues in the C-terminal α5 helix of Gα_i are involved in the selective coupling between GABA_B and G_i protein. The α5-helix structures of G_s (PDB 5VAI), G_q (PDB 6WHA) and the GABA_B-bound G_i were aligned. f, g, Effect of C^G.H5.23 (f) and G^G.H5.24 (g) mutations in Gα_i on GABA_B–G_i coupling using NanoBiT G-protein dissociation assay. Data are mean ± s.e.m. from at least three independent experiments. Source data

Fig. 4. Distinct G_i binding model of GABA_B.

a, b, Orientations of the α5 helix in G_i protein when coupling to GABA_B, cannabinoid receptor 1 (CB1) (class A), glucagon receptor (GCGR) (class B) and  smoothened (SMO) (class F). Structures were aligned by the TMDs; only the TMD of GB2 is shown, for clarity. GABA_B-bound, yellow; CB1-bound α5, PDB 6N4B; GCGR-bound α5, PDB 6LML; SMO-bound α5, PDB 6OT0. c, Schematics of the two types of pocket that are involved in G-protein recognition. GABA_B, green; monomeric GPCR, blue; G_i, yellow. d, Superposition of GABA_B-bound G_il with the GDP-bound G_il. GDP-bound G_il, PDB 1GP2. e, Structural comparison of the GABA_B-bound G_il with CB1-, GCGR- and SMO-bound G_il. G proteins are coloured as in a, b.

Extended Data Fig. 1. Purification of the GABA_B–G_i1 complex.

a Pharmacology of wild-type GABA_B and the purification construct (EM) in a baclofen-mediated NanoBiT-G-protein dissociation assay. Data are mean ± s.e.m. from four independent experiments, performed in technical triplicate. b, Flow chart of the purification steps for the GABA_B–G_i1 complex. GABA_B was expressed in HEK293F cells. Heterotrimeric G_i1 and scFv16 were expressed in Hi5 cells. c–e, Size-exclusion chromatography profile (c), SDS–PAGE gel (d) and the negative-staining electron microscopy analysis (e) of the purified GABA_B–G_i1 complex. Source data

Extended Data Fig. 2. Cryo-EM data processing of the GABA_B–G_i complex.

a, Representative cryo-EM micrograph (from 13,483 movies) and 2D class averages (from 16 classes) of the GABA_B–G_i1 complex. b, Flow chart of cryo-EM data processing. c, Gold-standard Fourier shell correlation curves of the globally refined GABA_B–G_i1 complex and the locally refined GABA_B and G_i1.

Extended Data Fig. 3. Flexibility analysis of GABA_B–G_i1 coupling.

a, Multibody refinement and principal component analysis of the relative orientations of GABA_B and G_i1. The GABA_B–G_i1 consensus map and the body masks of GABA_B and G_i1 are shown. b, Contribution of individual eigenvectors to the total variance in rotation and translation between GABA_B and G_i1. The first and second eigenvectors explain more than 50% of the variance observed and are highlighted in red. c, Histograms of the amplitudes along the first and second eigenvectors. d, Motion represented by the first and second eigenvectors.

Extended Data Fig. 4. Analysis of the quality of the cryo-EM map.

a, Global fitting of the GABA_B–G_il structure into the composite cryo-EM density map. b, Fourier shell correlation curves of the model versus the map. c, Cryo-EM densities and the fitted atomic models are shown. GB1 in red; GB2 in green; Gα_i1 in yellow; baclofen in magenta; BHFF in blue.

Extended Data Fig. 5. Structural comparisons of the determined GABA_B–G_i complex with the previously reported low-resolution B2a state GABA_B–G_i complex, the agonist-bound, and the agonist- and PAM-bound GABA_B.

a, Structural comparison between the low-resolution GABA_B–G_i complex in B2a state (grey) and this study determined GABA_B-G_i structure (green). b, Structural comparisons of the G_i-bound GABA_B (ago/PAM–G_i) (green) with the agonist-bound (ago) (PDB 6UO9) (grey) and agonist- and PAM-bound GABA_B (ago/PAM) (PDB 6UO8) (blue).

Extended Data Fig. 6. Intra-subunit conformational changes of the TMD of GB2 upon activation.

a, Overlay of the structures of the TMD of GB2 in antagonist-bound (antago) (PDB 7C7S) (yellow), agonist-bound (ago) (PDB 6UO9) (sky blue), and agonist- and PAM–G_i-bound (ago/PAM-G_i) (green) states. b, c, Overlay of the different states of the TM6 of GB1 (antago, grey; ago/PAM–G_i, red) (b) and the TM6 of GB2 (antago, yellow; ago, blue; ago/PAM–G_i, green) (c). d, e, NanoBiT G-protein dissociation assay of GABA_B with alterations of the residues that are involved in activation. Data are mean ± s.e.m. from at least three independent experiments, performed in technical triplicate. Source data

Extended Data Fig. 7. G_i activation and signalling assays.

a, Interface of GABA_B and G_i1 protein. GB2, green; G_i1, yellow. The interaction interface between GB2 and G_i1 is in red. b–e, Agonist-induced IP1 accumulation assay of the wild-type and the G_i1-binding-pocket mutant GABA_B. f, g, E_max and basal activity for each mutant relative to wild type, detected by IP1 accumulation assay and presented as dot plots. Data are mean ± s.e.m. from at least three independent experiments, performed in technical triplicate and analysed using one-way ANOVA with Dunnett’s multiple comparison test to determine significance (compared with wild type). h, Sequence alignment of the final five residues in the α5 helix among different Gα proteins. i, j, NanoBiT G-protein dissociation assay of the D350^G.H5.22 (i) and F354^G.H5.26 (j) mutant G_il. Data points in b–g, i, j are mean ± s.e.m. from at least three independent experiments, performed in technical triplicate. Source data

Extended Data Fig. 8. Comparison of the G_i binding pocket among class-A, -B, -C and -F GPCRs.

a–c, Parallel comparisons of the G_i binding pocket between the GABA_B and class-A CB1 (a), class-B GCGR (b) and class-F SMO (c) receptors. Four structures were aligned by the class-A TMD as a reference, as in Fig. 2. A comparison of the indicated two receptors is shown. Colours for α5 are: GABA_B-bound, yellow; CB1-bound (PDB 6N4B), orange–red; GCGR-bound (PDB 6LML), sky blue; and SMO-bound (PDB 6OT0), magenta.

Extended Data Fig. 9. Proposed model of GABA_B activation.

a–d, Schematic of the essential steps for GABA_B activation. e, Comparison of the relative bending of GB2 subunit in the agonist- and PAM–G_i-bound (ago/PAM-G_i) (green), the agonist-bound (ago) (PDB 6UO9) (blue), and antagonist-bound (antago) (PDB 7C7S) (yellow) when aligned on the GB2 VFT. f, The transmembrane domain rearrangement of GABA_B during activation. The antagonist-bound (antago) (PDB 7C7S) (yellow), agonist-bound (ago) (PDB 6UO9) (blue), and the agonist- and PAM–G_i-bound (ago/PAM–G_i) (green) structures of the TMD of GABA_B were aligned by the TMB of GB1. g, PAM binding in agonist- and PAM–G_i-bound GABA_B (ago/PAM–G_i).