The quest for high-resolution G protein-coupled receptor–G protein structures

G protein-coupled receptors (GPCRs) are integral membrane proteins that mediate many responses of cells to external stimuli (1). The fundamental role of GPCRs in regulating human cellular processes makes them high-value targets for drugs (2). GPCRs are highly dynamic entities, and thus, determination of high-resolution structures of GPCRs in complex with their binding partners by X-ray crystallography or by cryoelectron microscopy (cryo-EM) techniques is a delicate experimental undertaking despite tremendous progress in recent years (3). In this issue of PNAS, Mafi et al. (4) explore an alternative approach, molecular dynamics simulations, as a means to elucidate the atomistic structure of the human κ-opioid receptor (κOR) in complex with a potent agonist and the heterotrimeric Gi protein. To obtain structural information of GPCR–agonist–G protein complexes by computational methods is a welcome addition to existing experimental techniques.

On activation by a ligand from the extracellular side, GPCRs interact with heterotrimeric G protein(s) and/or with arrestin(s) on the cytoplasmic side, initiating downstream signaling (1). A given agonist can stabilize a particular subset of receptor conformations favoring the binding of a particular G protein. Other ligands stimulate both G protein and arrestin pathways, activate multiple G proteins, or drive signaling through arrestin. The greater efficacy toward one or the other route is called ligand bias (5). The ability of some ligands to stimulate both pathways may cause undesired effects of drugs directed at GPCRs. For example, morphine acts on opioid receptors and is highly effective for pain relief via G protein signaling but causes severe side effects via the arrestin pathway (6). Understanding of the structural changes during GPCR activation, signaling, and regulation is necessary to conceptualize the complex pharmacology of GPCRs and exploit this knowledge for the discovery of biased ligands.

Structures of GPCR–G protein complexes have been determined by X-ray crystallography and single-particle cryo-EM (Tables 1 and 2). The first approach has proven to be exceedingly difficult, and there is only one example of a crystal structure of a receptor coupled to a heterotrimeric G protein (7). The design of so-called minimal G proteins (8) facilitated the crystallization of two more GPCR complexes in their active conformation. Despite the availability of an array of tools, X-ray structure determination of GPCR–G protein complexes remains intractable as the growth of well-ordered three-dimensional (3D) crystals continues to present a major obstacle for many membrane protein targets. However, active-state crystal structures of several GPCRs have been obtained using nanobodies, the variable portion of heavy-chain camelid antibodies (9) (Table 2). Nanobodies can stabilize particular conformational states of a receptor, mimicking a G protein by binding to the G protein binding site. Receptor–nanobody structures have been highly informative in explaining the conformational changes in GPCRs that lead to activation (10).

Table 1.

GPCR–G protein complexes

Receptor	G protein	Resolution (Å)	Stabilizing antibody	PDB ID code
Crystallography
Class A GPCRs
Adenosine A2a receptor	Mini-Gs	3.4		5G53
β2AR	Gαs Gβγ	3.2	Nb35	3SN6
β2AR	Receptor fusion with GαsCT	3.7		6E67
Opsin	GαtCT peptide	3.2		3CAP
Rhodopsin	Mini-Go	3.1		6FUF
Cryo-EM
Class A GPCRs
Adenosine A1 receptor	Gαi Gβγ	3.6		6D9H
Adenosine A2a receptor	Mini-Gs Gβγ	3.8	Nb35	6GDG
Cannabinoid CB1 receptor	Gαi Gβγ	3.0, 3.0	scFv16	6N4B, 6KPG
Cannabinoid CB2 receptor	Gαi Gβγ	3.2, 2.9	scFv16	6PT0, 6KPF
Formyl peptide receptor 2	Gαi Gβγ	3.2	scFv16	6OMM
Muscarinic M1 receptor	Gα11 Gβγ	3.3	scFv16	6OIJ
Muscarinic M2 receptor	GαoA Gβγ	3.6	scFv16	6OIK
μOR	Gαi Gβγ	3.5	scFv16	6DDE, 6DDF
Neurotensin receptor 1	Gαi Gβγ	3.0	scFv16	6OS9, 6OSA
Orphan GPR52	Mini-Gs Gβγ	3.3	Nb35	6LI3
Rhodopsin	Gαi Gβγ	4.5	Fab_50	6CMO
Rhodopsin	Gαi Gβγ	4.4	Fab16	6QNO
Rhodopsin	Gαt Gβγ	3.3, 3.9	±Nb35*	6OY9, 6OYA
Serotonin 5-HT1B receptor	Mini-Go Gβγ	3.8		6G79
Class B GPCRs
Calcitonin gene-related peptide receptor	Gαs Gβγ	3.3	Nb35	6E3Y
Calcitonin receptor	Gαs Gβγ	4.1	Nb35	5UZ7
Corticotropin-releasing factor receptor 1	Gαs Gβγ	2.9, 3.0	Nb35	6P9X, 6PB0
Corticotropin-releasing factor receptor 2	Gαs Gβγ	2.8	Nb35	6PB1
Glucagon receptor	Gαs Gβγ	3.1	Nb35	Ref. 21
Glucagon-like peptide 1 receptor	Gαs Gβγ	3.3, 4.1, 3.0	Nb35	6B3J, 5VAI, 6ORV
Pituitary adenylate cyclase-activatingpeptide 1 receptor	Gαs Gβγ	3.0, 3.5, 3.6	Nb35	6P9Y (22)
Parathyroid hormone receptor 1	Gαs Gβγ	3.0	Nb35	6NBF, 6NBH, 6NBI
Class F GPCRs
Smoothened	Gαi Gβγ	3.9	Fab50	6OT0

The reader is referred to the respective publications for engineered modifications of the GPCRs and G proteins. A stabilizing antibody prevents dissociation of the GPCR–G protein complex but is not a G protein mimetic. For example, nanobody (Nb) Nb35 packs at the interface of the Gβ and Gαs subunits, preventing dissociation of the complex (7). β₂AR, β₂-adrenergic receptor; mini-G, minimal G protein; PDB, Protein Data Bank.

*Engineered nanobody 35.

Table 2.

GPCR complexes with G protein mimicking nanobodies

Receptor	G protein mimicking nanobody	Resolution (Å)	PDB ID code
Crystallography
Class A GPCRs
Angiotensin II type 1 receptor	Nb.AT110i1	2.9, 2.7, 2.8	6DO1, 6OS0, 6OS1, 6OS2
β1AR	Nb80, Nb6B9	3.0–3.2	6H7J, 6H7L, 6H7M, 6H7O
β2AR	Nb80	3.5	3P0G
β2AR	Nb6B9	2.8	4LDE, 4LDL, 4LDO
Muscarinic M2 receptor	Nb9-8	3.5	4MQS, 4MQT
κOR	Nb39	3.1	6B73
μOR	Nb39	2.1	5C1M
Cytomegalovirus GPCR US28	Nb7	2.9	4XT1
Class F GPCRs
Smoothened	NbSmo8	2.8	6O3C

The reader is referred to the respective publications for engineered modifications of the GPCRs. β₁AR, β₁-adrenergic receptor; β₂AR, β₂-adrenergic receptor; PDB, Protein Data Bank.

Cryo-EM does not require crystals, and revolutionary innovations made it now possible to determine structures of the relatively small GPCR–G protein complexes by single-particle imaging (11, 12). In the past few years, a number of new structures of GPCRs coupled to heterotrimeric G proteins have been determined (Table 1), advancing our understanding of how GPCRs couple to G proteins. Remarkably, cryo-EM enables the visualization of distinct conformational states in a single specimen sample as observed in some GPCR–G protein complex preparations (13–15). Successful structure determination by cryo-EM requires specimen preparations that can withstand potential denaturation at the air–water interface during grid preparation, and inherent flexibility between the complex components must be minimized to obtain stable 3D classes (3). Thus, only robust complexes yield good quality electron density maps. As in X-ray crystallography, sample quality and especially, stability determine success or failure of structure determination (perhaps more so) in cryo-EM.

The analgesic properties of opiates are primarily mediated by the μ-opioid receptor (μOR) (6, 16). Together with the κOR and δ-opioid receptor as well as the nociceptin receptor, they constitute an endogenous opioidergic system of GPCRs with intracellular signaling pathways of remarkable complexity. Activation of μOR by opioids triggers signaling through the Gi/o protein pathway ensuing analgesia and sedation. However, many opioids also trigger signaling through arrestin molecules; this alternative pathway leads to the adverse effects of opioid analgesics, such as tolerance, respiratory suppression, and constipation. Thus, enormous medicinal chemistry efforts have been dedicated to developing biased opioid compounds that afford analgesia without the lethal side effects (6).

In the search for safer analgesics, the κOR has emerged as an alternative target. The structure of an active-state κOR in complex with a G protein-mimicking nanobody has been determined (17), but a structure of the G protein-bound κOR has not yet been solved. Conformation-specific nanobodies are excellent substitutes for G proteins (1), but small structural differences have been noted comparing the G protein and nanobody complexes of the β₂-adrenergic receptor (7, 10) and the μOR (16, 18), respectively. To derive a κOR–Gi protein model, Mafi et al. (4) use molecular dynamics simulations using as starting templates the active-state κOR from the κOR–nanobody crystal structure (17) and the Gi protein from the μOR–Gi cryo-EM structure (16). Two aspects guided the treatment of the templates. Because the κOR construct used for crystallization was engineered to facilitate crystal formation, Mafi et al. (4) revert the κOR sequence to the wild type. Also, because the Gi structure of μOR–Gi did not resolve all residues of the Gαi-α helical domain, Mafi et al. (4) build this domain from the cryo-EM structure of rhodopsin complexed with Gi (19). Extensive molecular dynamics simulations resulted in a κOR–Gi model embedded in a palmitoyl-oleoyl-phosphatidylcholine lipid bilayer. The comparison of the simulated κOR–Gi model with the X-ray κOR-nanobody structure revealed a further contraction of the ligand binding pocket in the simulated κOR–Gi model and a slight alteration of the agonist MP1104 binding pose. The intracellular receptor domain is more open compared with the X-ray κOR–nanobody structure to accommodate the Gi protein, consistent with observations in other GPCR complexes (7, 10, 16, 18). The Gi protein and κOR interact through three strong anchors, and the C-terminal α5 helix of Gαi engages in extensive contacts with the receptor, stabilizing its active conformation (4). The validity of the molecular dynamics simulations approach taken by Mafi et al. (4) is assessed by simulating the κOR–nanobody crystal structure (17), building a simulated active-state μOR–Gi model from the μOR–nanobody crystal structure (18) and optimizing the cryo-EM μOR–Gi structure (16). Relevant comparisons yielded aspects consistent with class A receptor activation characteristics.

The molecular dynamics simulations work presented by Mafi et al. (4) required the input from experimentally determined structures. Crystal structures typically trap the lowest-energy states within an ensemble of conformations, and cryo-EM structures result usually from specimens of robust distinct states. However, NMR studies have hinted at transiently populated receptor conformations not amenable to the current methods of structural studies (20). The future challenge will be to capture those intermittent receptor states structurally by computational means, essential for the understanding of the plethora of ligand-specific signaling responses of GPCRs.