Structure and dynamics determine G protein coupling specificity at a class A GPCR

Abstract

G protein-coupled receptors (GPCRs) exhibit varying degrees of selectivity for different G protein isoforms. Despite the abundant structures of GPCR-G protein complexes, little is known about the mechanism of G protein coupling specificity. The β₂-adrenergic receptor is an example of GPCR with high selectivity for Gαs, the stimulatory G protein for adenylyl cyclase, and much weaker for the Gαi family of G proteins inhibiting adenylyl cyclase. By developing a Gαi-biased agonist (LM189), we provide structural and biophysical evidence supporting that distinct conformations at ICL2 and TM6 are required for coupling of the different G protein subtypes Gαs and Gαi. These results deepen our understanding of G protein specificity and bias and can accelerate the design of ligands that select for preferred signaling pathways.

Fig. 1.. Ligand efficacy at the β₂AR.

(A and B) Luminescence GTPase-Glo assay. Ligand efficacy reflects the ability of the ligand-bound receptor to promote turnover of the G protein cycle. The reaction starts when β₂AR, bound to different ligands, is incubated with Gs (A) or Gi (B). At the end of the reaction, lower bars correspond to higher GTPase activity. (A) Salmeterol is a subtype-selective β₂AR partial agonist for Gs activation relative to other agonist such as epinephrine and formoterol (****P < 0.0001). LM189 is as efficacious as epinephrine at Gs turnover (ns, P = 0.96). In (B), salmeterol is more efficacious than epinephrine for β₂AR coupling to Gi (**P = 0.003). LM189 is more efficacious than epinephrine and salmeterol at coupling to Gi (****P = 0.0001). Experiments were performed as biological triplicates and results were plotted using GraphPad Prism. P values were calculated using the unpaired t test analysis on GraphPad Prism, assuming Gaussian distributions. ns = P > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Data are represented as the mean ± SD. (C) Structures of β₂AR ligands. From the left, epinephrine (adrenaline) is the endogenous catecholamine neurotransmitter. Salbutamol is a partial agonist belonging to the short-acting β₂AR agonists (SABAs). Salmeterol is a long-acting β₂AR partial agonist (LABA), which exhibits a long duration of action and is used in the chronic management of asthma. Salmeterol has the same saligenin head group as salbutamol. LM189 was developed by replacing the saligenin moiety of salmeterol with a catechol group (red square).

Fig. 2.. Structure of the LM189-bound β₂AR-Gi complex.

(A) Cryo-EM density map of the LM189-bound β₂AR-Gi-scfv16 complex. (B) Cryo-EM structure of LM189-bound β₂AR-Gi complex and crystal structure of BI-167107–bound β₂AR-Gs (4) (3SN6). β₂AR bound to Gi is colored orange, Gαi is in olive green, Gβ is in cyan, Gγ is in magenta, and scfv-16 is in gray. β₂AR bound to Gs is colored green, and Gαs is in slate. (C) Relative orientation of receptor TM6 and α5 helix of the β₂AR-Gi (orange), β₂AR-Gs (green), and μOR-Gi (blue) (3) complex structures. (D) Top: α5 engagement of Gi and the interactions formed with the ICL2, TM3, and TM5 of β₂AR. β₂AR is colored orange, and Gαi is in olive green. Bottom: α5 engagement of Gs and the interactions formed with the ICL2, TM3, and TM5 of β₂AR. β₂AR is colored green, and Gαs is in slate. Polar contacts within 4 Å are highlighted. (E) Structural differences between the LM189-bound β₂AR-Gi complex and BI-167107–bound β₂AR-Gs (4). Top left blue panel: The PAM cmpd-6FA (42, 43) (yellow), binding at the top of ICL2. Lower left magenta panel: When bound to Gi, ICL2 of β₂AR adopts a slightly different conformation compared to Gs. Top right red panel: By monitoring TM6 rotation at Glu²⁶⁸, the TM6 helix is slightly less rotated in the β₂AR-Gi structure compared to β₂AR-Gs (~3 Å). Carazolol-bound inactive-state β₂AR is colored gray (2RH1) (44), salmeterol-bound β₂AR in complex with Nb71 is colored wheat (6MXT) (38), LM189-bound β₂AR-Gi complex is colored orange, and BI-167107–bound β₂AR-Gs (3SN6) (4) is colored green. Lower right brown panel: TM5 of the β₂AR-Gi complex is one helix turn shorter than TM5 of the β₂AR-Gs structure, avoiding a steric clash between the tip of TM5 and the connecting loop between the a4 helix and b6 strand of Gi.

Fig. 3.. Ligand-binding pocket of the LM189-bound β₂AR-Gi complex.

(A to C) The orthosteric pocket (left) and exosite (right) of β₂AR bound to LM189 (A), epinephrine (B), and salmeterol (C). (A) LM189 is colored yellow, and β₂AR in the β₂AR-Gi structure is colored orange. (B) Epinephrine is colored pink, and epinephrine-bound β₂AR in complex with Nb80 (4LDO) (67) is colored aquamarine. (C) Salmeterol is colored blue, and salmeterol-bound β₂AR in complex with Nb71 (6MXT) (38) is colored slate. H-bonds are showed as dashed lines. (D and E) Molecular dynamics simulations of active-state β₂AR-Gi bound to LM189 or epinephrine. (D) Rotamer analysis of Ser^5.46 and Asn^6.55 of LM189-bound (orange, top) and epinephrine-bound (aquamarine, bottom) β₂AR-Gi. (E) Histograms represent hydrogen-bond formation frequencies as a fraction of time in three 2-μs MD simulations. β₂AR-Gi is represented in orange, and epinephrine-bound β₂AR-Gi is in aquamarine. (F and G) MD simulations at the intracellular cavity of β₂AR. (F) Left: Comparison of the cryo-EM structure of the LM189-bound β₂AR (orange) and a representative MD snapshot of the LM189-bound β₂AR (yellow). Right: Plot shows the progression of the distance between the Cγ of Asp130^3.49 and OH of Tyr141^ICL2 over the course of 4 μs. (G) Left: Comparison of the epinephrine-bound β₂AR model (pink, based on the BI-167107–bound β₂AR crystal structure, PDB: 3SN6) and a representative MD snapshot of the epi-bound β₂AR model (wheat). Right: Plot shows the progression of the distance between the Cγ of Asp130^3.49 and OH of Tyr141^ICL2 over the course of 4.9 μs.

Fig. 4.. CW-EPR studies of ICL2 of β₂AR.

(A) A minimal-cysteine version of the β₂AR with an acetamido-PROXYL spin-label side chain shown at the mutated Q142C residue on ICL2 for EPR studies. Inactive-state β₂AR (2RH1) (44) is colored gray, and active-state β₂AR (3SN6) (4) is colored green. (B) Superimposed CW-EPR spectra of β₂AR in the apo, LM189-bound, and Gi- and Gs-protein bound conditions. The regions of the low-field line dominated by mobile and immobile components are indicated. (C) Superimposed CW-EPR spectra in the apo, LM189-bound, and BI-167107–bound receptor in complex with Gi. (D) Superimposed CW-EPR spectra in the apo, LM189-bound, and BI-167107–bound receptor in complex with Gs. All spectra are area normalized and color coded as indicated.

Fig. 5.. Investigations of TM6 conformational dynamics.

(A and B) Fluorescence spectroscopy measurements of β₂AR TM6 conformations. (A) Side and intracellular views of β₂AR labeled with mBBr on Cys265 of TM6. Inactive-state β₂AR (2RH1) (44) is colored gray, and active-state β₂AR (3SN6) (4) is colored green. (B) Steady-state fluorescence emission spectra of mBBr-labeled β₂AR purified in LMNG/CHS in the presence and absence of ligands. The spectra are normalized relative to apo (unliganded) receptor (gray). (C to F) TM4/6 DEER measurements of β₂AR purified in LMNG/CHS in the presence of ligands and G proteins. (C) Side and intracellular views of the receptor labeling sites on TM4 and TM6. Inactive-state β₂AR (2RH1) (44) is colored gray, and active-state β₂AR (3SN6) (4) is colored green. (D) Ligand dependence of TM4/6 distance distributions. (E) G protein dependence of distance distributions of BI-167107–bound receptor. (F) G protein dependence of distance distributions of the LM189-bound receptor. Distance distributions are color coded as indicated.

Fig. 6.. SmFRET distributions of TM6 conformational dynamics.

(A to C) SmFRET measurements of β₂AR labeled on TM4/6 with donor and acceptor fluorophores. (A) Unliganded (apo, N = 269) receptor, in black, mostly populates the inactive states (~0.9 and 0.7), corresponding to high FRET. BI-167107 (blue, N = 265), a β₂AR full agonist, mostly populates the intermediate state of the receptor (~0.6) and marginally stabilizes a state at ~0.2 FRET efficiency. The LM189-bound receptor (red, N = 181) populates the intermediate state at ~0.6 to a smaller degree, in favor of the low-FRET population (~0.3). (B) Representative FRET traces and transitions in the apo, BI-167107, LM189, and the LM189-bound receptor coupled to Gi and scfv-16. (C) Upon Gi coupling, the active state (~0.3), corresponding to low FRET, becomes predominant (light blue, N = 105).