Emerging structural biology of lipid G protein‐coupled receptors

ABSTRACT

The first crystal structure of a G protein‐coupled receptor (GPCR) was that of the bovine rhodopsin, solved in 2000, and is a light receptor within retina rode cells that enables vision by transducing a conformational signal from the light‐induced isomerization of retinal covalently bound to the receptor. More than 7 years after this initial discovery and following more than 20 years of technological developments in GPCR expression, stabilization, and crystallography, the high‐resolution structure of the adrenaline binding β₂‐adrenergic receptor, a ligand diffusible receptor, was discovered. Since then, high‐resolution structures of more than 53 unique GPCRs have been determined leading to a significant improvement in our understanding of the basic mechanisms of ligand‐binding and ligand‐mediated receptor activation that revolutionized the field of structural molecular pharmacology of GPCRs. Recently, several structures of eight unique lipid‐binding receptors, one of the most difficult GPCR families to study, have been reported. This review presents the outstanding structural and pharmacological features that have emerged from these new lipid receptor structures. The impact of these findings goes beyond mechanistic insights, providing evidence of the fundamental role of GPCRs in the physiological integration of the lipid signaling system, and highlighting the importance of sustained research into the structural biology of GPCRs for the development of new therapeutics targeting lipid receptors.

Abbreviations

agoPAMs

positive allosteric modulator agonists

ApoM

apolipoprotein‐M

BLT1

leukotriene B4 receptor 1

CB1

cannabinoid receptor 1

ECLs

extracellular loops

EP3

prostaglandin E2 receptor 3

GPCRs

G protein‐coupled receptors

GPR40

free fatty acid receptor 1

HDL

high‐density lipoproteins

ICLs

intracellular loops

LPA

lysophosphatidic acid

LPA1

lysophosphatidic acid receptor 1

LPA6

lysophosphatidic acid receptor 6

PAFR

platelet‐activating factor receptor

S1P1

sphingosine‐1‐phosphate receptor 1

Introduction

G protein‐coupled receptors (GPCRs) constitute the largest superfamily of membrane proteins in the human genome. They transduce signals of a large variety of extracellular stimuli such as light, peptides, neurotransmitters, hormones, lipids, and ions, across plasma membranes, and are implicated in the regulation of every physiological system. As a result, they are the target of ~35% of currently prescribed drugs.1, 2

GPCRs can be grouped into five major classes based on their transmembrane domain sequence conservation. Class A GPCRs are by far the most abundant subfamily with 719 out of approximately 800 receptors known to date.3 The first report of a high‐resolution GPCR crystal structure, published in 2000,4 was the light receptor rhodopsin captured in the inactive dark state and stabilized by a covalently bound 11‐cis‐retinal. However, an additional 7 years of extensive technological development was required to solve the first structure of a receptor bound to a diffusible ligand, opening up the GPCR structural biology field.5, 6, 7, 8 Since then, the structures of more than 53 unique receptors have been reported, 46 of which are Class A GPCRs [Fig. 1(A)]. However, the structure determination of GPCRs still remains a significant scientific undertaking, requiring 3–4 years of intense laboratory work on each new receptor. At that rate, complete coverage of the subfamily would take over 100 years.10 Fortunately, even with only ~6% (46 out of 719 receptors) of the Class A receptor structures determined thus far, an unprecedented diversity of ligand‐mediated receptor interaction modes and activation mechanisms is being revealed.8, 10, 11, 12, 13, 14, 15

Figure 1.

G protein‐coupled receptors (GPCR). (A) Coverage of non‐olfactory class A GPCRs grouped respective to ligand type. The crystallized receptors are indicated by a red dot. Adapted from GPCRdb.9 (B) Schematic of the two‐dimensional topology of a GPCR consists of seven transmembrane helical domains, an extracellular N‐terminus, three extracellular loops (ECLs), three intracellular loops (ICLs), an additional intracellular helical domain that is parallel to the membrane plane followed by an intracellular C‐terminus.

Topologically, GPCRs are composed of an extracellular N‐terminal domain, a seven‐transmembrane (7TM) helical domain (helices I–VII), interleaved with three intracellular (ICLs) and three extracellular loops (ECLs), and an intracellular C‐terminal domain that often includes an amphiphilic helix running parallel to the plasma membrane [Fig 1(B)]. GPCRs transduce chemical signals across the cell membrane by binding to extracellular ligands that induce or stabilize specific receptor conformations. These conformations in turn allow intracellular effectors, such as the heterotrimeric G proteins and arrestins, to be engaged at the intracellular side of the receptor where they initiate intracellular signaling cascades, and regulate receptor activity and cellular trafficking8 (Fig. 2). The extracellular domains often play a regulatory role in ligand access into the binding pocket, whereas the intracellular domains participate in coupling to various effectors. For most Class A receptors, the orthosteric binding pocket is located on the extracellular side, inside the 7TM bundle. This location allows the ligand to act on residues that could stabilize slightly different transmembrane configurations, especially that of helices VI and VII, yielding to the opening of an intracellular cleft for the engagement of intracellular effectors. The mechanism by which a ligand acts on a receptor and the resulting consequences on the intracellular effector activity is defined by pharmacologists as the efficacy of the ligand. For example, a ligand that stabilizes the receptor in an active state is defined as an agonist, which increases the engagement of effectors and induces a net positive signal in the cell. In contrast, a ligand that stabilizes the receptor in an inactive state is defined as an inverse agonist, which reduces the basal level of signaling in the cell. Finally, a ligand that binds and stabilizes the receptor without impacting the basal level of signaling in the cell is defined as a neutral antagonist.

Figure 2.

Signaling of GPCRs. Ligand interactions with the receptor's binding site triggers the intracellular engagement of signaling effectors, such as the heterotrimetic G proteins or the arrestins, leading to a downstream intracellular signaling cascades and physiological responses. Cannabinoid receptor 1 (CB1) (green ribbons) and agonist ligand AM11542 (green spheres) are from PDB access code 5xra. Heterotrimeric G protein αi1β1γ2 (blue ribbons and spheres) is from PDB access code 1gp2. Arrestin‐3 (orange ribbons) is from PDB access code 5tv1.

Bioactive lipids have important physiological roles and lipid‐binding receptors from the Class A GPCR subfamily are of great therapeutic value16, 17, 18, 19, 20 [Fig. 3(A), and Tables 1, 2, 3]. Yet, lipid‐binding GPCRs represent some of the most difficult receptors to study experimentally, because the physico‐chemical properties of the lipids can (i) in some cases, lead to an elevated non‐specific ligand binding; (ii) require the use of additive scaffolding adjuvants to deliver the ligand on‐target in vitro; and (iii) sometimes require long incubation periods to reach radioligand binding equilibrium because of their inherently long dissociation half‐life. Given the more difficult pharmacology associated with this GPCR subfamily, achieving high‐resolution structures of lipid of receptors using X‐ray crystallography has become an essential complementary approach to better understand their mechanisms of action, while providing medicinal chemists with an important tool to inform design of more diverse ligands. Therefore, here we present the most recent advances in X‐ray structural biology of lipid binding Class A GPCRs in order to shed light on bioactive lipids’ action on their cognate receptors.

Figure 3.

Two‐dimensional structures of important lipid receptor ligands. (A) Bioactive lipids and (B) synthetic derivatives. The ligands are classified according to their reported efficacy. The target receptor is indicated in parenthesis below compound's name.

Table 1.

List of Lipid‐Binding GPCR Structures

Receptors	Ligand	Efficacy	Resolution (Å)	PDB access code	Reference
S1P1	ML056a	Antagonist	3.35	3v2w	21
S1P1	ML056	Antagonist	2.8	3v2y
PAFR	SR27417a	Antagonist	2.8	5zkp	22
PAFR	ABT491a	Inverse agonist	2.9	5zkq
BLT1b	BIIL260a	Antagonist	3.7	5x33	23
LPA6b	APO	N/A	3.2	5xsz	24
LPA1	ONO9780307a	Antagonist	3.0	4z34	25
LPA1	ONO9910539	Antagonist	2.9	4z35
LPA1	ONO3080573	Antagonist	2.9	4z36
GPR40	Compound 1	agoPAM	2.76	5kw2	26
GPR40	MK8666a	Partial agonist	2.2	5tzr	27
GPR40	AP8a + MK8666	agoPAM + partial agonist	3.22	5tzy
GPR40	TAK‐875a	Partial agonist	2.33	4phu	28
CB1	AM841a	Agonist	2.95	5xr8	29
CB1	AM11542	Agonist	2.8	5xra
CB1	Taranabant	Antagonist	2.6	5u09	30
CB1	AM6538	Antagonist	2.8	5tgz	31
EP3	Misoprostola	Agonist	2.5	6m9t	32

^a Also see chemical structure in Figure 3.

^b Structure solved is for non‐human variant of receptor.

Table 2.

Some Physiological Roles of Lipid Receptors with Known Structures

Receptors	Physiological and pathophysiological roles	Reference
S1P1	Immune cells trafficking; vascular development; vascular and endothelial integrity	19
PAFR	Acute inflammation; platelet aggregation	33
BLT1a	Bone and immune cells chemotaxis; peritonitis; asthma; rheumatoid arthritis; atherosclerosis; osteoporosis; bacterial infection clearance	34
LPA6a	Hair growth; vascular stability	35, 36
LPA1	Neural development; bone homeostasis; pain; hydrocephalus; autoimmune disorders	16
GPR40	Insulin secretion; incretin secretion	20
CB1	Psychiatric disorders; addiction; eating disorders/appetite	18
EP3	Pain; fever; cough; labor; pro‐thrombotic; airway inflammation; vasoconstriction	37, 38, 39

^a Non‐human receptors crystallized. The physiological effects reported are for human.

Table 3.

Recent Active and Inactive Clinical Trials on Drugs Targeting Lipid Receptors

Receptors	Drugs	Synonyms	Medical condition	Reference
S1P1	Fingolimod	Gilenya® FTY720	Multiple sclerosis; stroke, uveitis, schizophrenia, amyotrophic lateral sclerosis; Rett's syndrome	19
S1P1	Mocravimod	KRP‐203	Ulcerative colitis; systematic lupus erythematosus
S1P1	Ponesimod	ACT‐128800	Multiple sclerosis; psoriasis; graft rejection
S1P1	Cenerimod	ACT‐334441	Systemic lupus erythematosus
S1P1	Siponimod	BAF312	Multiple sclerosis; dermatomyositis; polymyositis
S1P1	Ozanimod	RPC‐1063	Multiple sclerosis; ulcerative colitis; Crohn's disease
S1P1	Amiselimod	MT‐1303	Multiple sclerosis; systemic lupus erythematosus; Crohn's disease; psoriasis
S1P1	Etrasimod	APD334	Ulcerative colitis
S1P1	GSK2018682		Multiple sclerosis	16
PAFR	None
BLT1	Etalocib	VML‐295 LY‐293111	Solid tumors; non‐small cell lung cancer; pancreatic cancer	40
BLT1	Amelubant	BIIL 284	Chronic obstructive pulmonary disease; asthma; rheumatoid arthritis; cystic fibrosis
BLT1	Moxilubant	CGS‐25019C; LTB‐019	Chronic obstructive pulmonary disease
LPA6	None
LPA1	SAR100842		Systemic sclerosis; fibrotic diseases	16
LPA1	BMS‐986020		Idiopathic pulmonary fibrosis
GPR40	Falsiglifam	TAK‐875	Type II diabetes	20
GPR40	AMG837		Type II diabetes
GPR40	LY2881835		Type II diabetes
GPR40	LY2922470		Type II diabetes
CB1	Dronabinol	(−)‐trans‐Δ9‐tetrahydrocannabinol; Marinol	Anorexia; refractory nausea and vomiting	18
	Nabilone	Cesamet™	Pain; nausea/vomiting
	Rimonabant	SR141716; Acomplia®; Zimulti®	Obesity; chemotherapy‐induced nausea	41
	Taranabant	MK‐0364	Obesity
	Otenabant	CP‐945598	Obesity	42
	Ibipinabant	SLV319, BMS‐646256	Obesity
	Surinabant	SR147778	Nicotine addiction/smoking cessation	43
EP3	None
GPR55	Epidiolex	Cannabidiol	Epilepsy	44

Overall structures of lipid receptors

Since 2012, 18 X‐ray crystal structures from eight unique lipid binding GPCRs have been solved in complex with different antagonists, agonists and allosteric modulators (Table 1 and Figs. 3 and 4). All share the common 7TM domains as previously reported for other GPCRs. Interestingly, in contrast to the structures of receptors from other subfamilies, seven out of the eight receptors exhibit a tight folding of the extracellular domain that restricts ligand access to the orthosteric binding site from the extracellular space (Fig. 4). Previously, an enclosed orthosteric binding pocket was observed only in the structure of rhodopsin, which has a long hairpin ECL2 covering the entire retinal binding site and forming a tight lid along with an N‐terminal hairpin structure that lies on top.4 Interestingly, similar folding of an extended ECL2 hairpin was also seen in the recent X‐ray crystal structure of the prostaglandin E2 receptor 3 (EP3) in complex with misoprostol32 [Fig. 4(A)]. Further, alignment of the two structures shows that their ECL2s overlay well. However, in contrast to rhodopsin, the N‐terminus of EP3 is unfolded. For the other six available lipid receptor structures, the actual folding of their extracellular domains is distinct from rhodopsin and EP3. Three of them, sphingosine‐1‐phosphate receptor 1 (S1P1), lysophosphatidic acid receptor 1 (LPA1), and platelet‐activating factor receptor (PAFR), display an α‐helical N‐terminal domain that folds on top of the binding site [Fig. 4(B)]. While lysophosphatidic acid receptor 6 (LPA6) has free binding site access from extracellular space, and the binding pocket of GPR40 is covered with a lid formed by tight interactions between ECL1 and the ECL2 hairpin structure [Fig. 4(A)]. Similarly, the ligand‐binding site in Leukotriene B4 receptor 1 (BLT1) is occluded [Fig. 4(A)], albeit to a lesser extent, by the ECL2 hairpin fold and several long chain polar residues from ECL2 and the tips of helices V‐VI‐VII. Interestingly, the cannabinoid receptor 1 (CB1) also exhibits an N‐terminal domain that folds over the orthosteric binding site and includes a partial α‐helical structure in three out of four available structures [Fig. 4(C)], a fold that apparently depends on N‐terminal interactions with the ligand.

Figure 4.

Overall crystal structures of lipid receptors in complex with ligands. (A and B) Receptors displaying a (A) β‐strand lid or (B) N‐terminal α‐helical lid on the binding site. (C) CB1 shows both features with an α‐helical lid in the agonist‐bound state (see Figure 2) and β‐strand lid in the antagonist‐bound state. Receptors are shown as ribbons and ligands as spheres: prostaglandin E2 receptor 3 (EP3) bound to misoprostol (PDB access code 6m9t; white); GPR40 bound to TAK875 (PDB access code 4phu; blue), apo lysophosphatidic acid receptor 6 (LPA6) (PDB access code 5xsz; yellow), leukotriene B4 receptor 1 (BLT1) bound to BIIL260 (PDB access code 5x33; magenta), sphingosine‐1‐phosphate receptor 1 (S1P1) bound to ML056 (PDB access code 3v2y; green), lysophosphatidic acid receptor 1 (LPA1) bound to ONO9780307 (PDB access code 4z34; purple), platelet‐activating factor receptor (PAFR) bound to ABT491 (PDB access code 5zkq; orange), and CB1 bound to AM6538 (PDB access code 5tgz; brown).

Ligand access to lipid receptors

Restricted ligand access to the receptors’ binding site is a new feature for GPCRs, which may reflect the physico‐chemical properties of the lipids and the physiological context of their cellular delivery. Owing to their amphiphilic nature, lipids can enter the receptor either from the plasma membrane or directly access it from the extracellular space. Moreover, lipids often act through protein chaperones that transport them to the site of action on a cell that contains the GPCR to be activated [Fig. 5(A)]. For example, S1P is transported in the blood mainly by binding to apolipoprotein‐M (ApoM) in the complex of high‐density lipoproteins (HDL), an important class of lipid transporters. An ApoM‐S1P complex is engaged by the common lymphoid progenitors cells in the bone marrow and activates S1P1 signaling to restrain lymphopoiesis and adaptive immune response.45 Autotaxin, a secreted lysophospholipase D that synthesizes lysophosphatidic acid (LPA) in exosomes, forms an autotaxin/LPA/exosome delivery system that then merges with membranes of LPA receptor‐containing cells to induce LPA receptor signaling.46, 47 This carrier system has an important role in the physiological integration of LPA receptor signaling regulation, as the inhibition of autotaxin activity by synthetic and natural steroids (like bile salts) is known to negatively affect LPA receptor signaling.48 Interestingly, autotaxin has also been proposed to deliver LPAs from the extracellular milieu directly to the orthosteric binding site in LPA receptors through a hydrophobic channel on the enzyme.49 In the context of lipid receptors, the X‐ray crystal structures help reveal these key lipid entry mechanisms. The surface representations of the enclosed lipid receptor structures show that there are different access tunnels to the ligand‐binding sites, suggesting that ligand entry could occur from the extracellular space for LPA1, CB1, and BLT1 [Fig. 5(B)], or directly through the plasma membrane into the TM region [Fig. 5 (C)] for S1P1, GPR40, and PAFR. Interestingly, although the LPA6 structure did not contain a ligand, analysis shows that a ligand could have open access from either the membrane or extracellular space [Fig. 5(D)]. Additionally, no openings to suggest a pathway for ligand entry were revealed in the EP3 structure [Fig. 5(E)]. Clearly the significance of the ligand access channel on these receptors and the precise mechanisms of lipid delivery remains a key area for future exploration, and the available X‐ray crystal structures of these receptors help provide an important first step toward a broader physiological understanding of the lipid signaling system.

Figure 5.

Hypothetical mechanisms of ligand entry into the orthosteric binding sites of lipid receptors. (A) Free diffusing bioactive lipids can access the binding site from the extracellular space and by lateral diffusion in the plane of the membrane. Transporters and enzymes, such as high‐density lipoprotein (HDL) and autotaxin (PDB access code 3nkp), are recruited to the cellular surface and distribute the bioactive lipids in the receptor's local membrane area or directly channel the bioactive lipids into the receptor.45, 49 The crystal structures of lipid receptors suggest a putative ligand entry mechanism in the orthosteric receptor binding site from the classical (B) extracellular side and from the (C) lateral membrane plane of the receptors or (D) both. (E) EP3 receptor in complex with misoprostol show no tunnel that suggests a ligand entry site. The red arrows indicate the putative ligand entry site. The receptors are represented as surfaces and the ligand as spheres. The structures are displayed according to the same color and PDB access code as in Figure 4.

Lipid receptor activating transitions

CB1 is the only lipid receptor that has been determined in the presence of both agonists and antagonists to date, allowing for the analysis of allosteric transitions associated with receptor activation (Table 1).29, 30, 31 First, the binding pocket of CB1 displays a large induced‐fit that accommodates the transition from antagonist to agonist binding [Fig. 6(A)]. This transition involves inward movements of helices I and II with respect to the 7TM bundle along with a clockwise rotation of helix II. This movement effectively collapses the binding site in the region of helices I–II–VII and repositions the bulky aromatic residues Phe170^2.57 and Phe174^2.61 to form a new binding site stabilized by the agonist. The reduction in ligand binding pocket volume from 822 to 384 ų is the largest conformational change seen for an agonist and antagonist pair.5, 50, 51, 52 However, despite the modeling of anandamine in the binding pocket of the CB1 agonist‐bound structure, the synthetic agonists used to obtain the structure of CB1 are more similar to the phytocannabinoid Δ9‐tetrahydrocannabinol than to the human endocannabinoid. A structure of CB1 with the human endocannabinoid is required to confirm that the changes associated with CB1 activation are conserved with endogenous ligands (Fig. 3).

Figure 6.

Example of a binding site and activating transition in a lipid receptor. (A) Extracellular view of CB1 receptor in complex with the agonist AM841 (PDB access code 5xr8, active) and the antagonist AM6538 (PDB access code 5tgz, inactive). The inward movement of helices I–II and rotation of helix II reduce the binding site volume in the region formed by helices I, II, and VII. Helix II movement and rotation reposition residues Phe170^2.57 and Phe174^2.61 to form the agonist binding site floor. Together with the N‐terminal domain that adopts a partial helical conformation, the new binding site configuration is refolded to accommodate only the agonist. (B) Close‐up view of the activating twin microswitch transition implicating residues Phe200^3.36 and Trp356^6.48. (C) Intracellular view of the CB1 receptor activating the transition. Unlocking the toggle switch residue Trp356^6.48 frees the outward movement of helix VI opening the effector coupling intracellular cavity. CB1 receptor bound to the agonist is shown in green cartoon (receptor) and sticks (AM841 ligand). CB1 receptor bound to the antagonist is shown in brown cartoon (receptor) and sticks (AM6538 ligand). (D) Misoprostol tight fit in EP3 receptor orthosteric binding pocket. EP3 receptor is shown as gray surface and misoprostol as light gray spheres. EP3 receptor in complex with misoprostol is from pbdid 6m9t

Class A GPCRs contain conserved structural microswitches that are important for the coordination of TM movements upon receptor activation.12 Interestingly, in the antagonist‐bound CB1 structure the position of the conserved toggle switch residue Trp356^6.48 is stabilized by the interaction with Phe200^3.36 [Fig. 6(B)]. In the active‐state CB1 structure, the agonist and Phe170^2.57 stabilize a 90° shift in the Phe200^3.36 rotamer conformation, allowing Trp356^6.48 and the bottom of helix VI to move away in an outward motion from the 7TM bundle. This leads to the opening of the intracellular binding cavity for engagement of effectors [Fig. 6(C)]. The microswitch acts as a dual toggle switch for CB1 activation, which has not been observed previously. Overall, the change in positions of helices I, II, and VI between the inactive and active states is important, significantly affecting the binding site shape to the point where steric constraints preclude the fitting of agonists in the inactive state and antagonists in the active state. Moreover, the arrangement of the N‐terminal domain folding seems to depend on ligand efficacy, as the agonist‐bound CB1 structures show an α‐helical folding of the N‐terminal domain, while the antagonist‐bound CB1 structures show little or no α‐helical content, but a greater protrusion of the N‐terminal domain in the binding pocket. Given the role of the N‐terminal domain on ligand access for lipid receptors, this binding site plasticity may be a common feature for lipid receptors to accommodate the large variety of natural lipid structures and their metabolites.53, 54, 55

Another example of the binding site plasticity among lipid receptors is represented in the structure of agonist‐bound EP332 [Figs. 4(A) and 5(E)]. In this active‐state structure, the binding cavity is tightly enclosed around the small misoprostol agonist with very little additional empty space [Fig. 6(D)]. Interestingly, long synthetic ligands do not fit in the structure suggesting that an allosteric transition occurs to accommodate their binding. Docking simulation on the EP3 structure using longer agonists provided evidence that Arg333^7.40, a residue conserved in most prostanoid receptors and co‐ordinating the acid moiety of the prostanoid α‐chain, moves away to open a ligand tunnel between helices I and VII [Fig. 6(E)]. Overall, it is not surprising that substitution of the acid group of small prostanoids by bulkier moieties would induce a transition in the polar region of the binding site, as the acid group coordinates important interactions among ECL2, helices I–II and VII. Although the polar subpocket of EP3’s binding site is predicted to have a high degree of plasticity and may accommodate a large variety of chemical entities, the relationship between these transitions and the receptor's activation it is not clear.

Allosteric modulation of lipid receptors

Many allosteric modulators targeting GPCRs have been developed.56 In the context of lipid receptors, this mode of efficacy may provide a good alternative to the restricted ligand access in the orthosteric binding site. The four X‐ray crystal structures of GPR40 provide insights on possible allosteric ligand drug design strategies and on the mechanism of allosteric efficacy [Figs. 4(B) and 7].26, 27, 28 Importantly, GRP40 structures have revealed two different binding pockets in the receptor. The first site is at the interface of helices III–IV and ECL2 with the compounds lying partially inside the 7TM bundle and partially protruding outside in the membrane [Figs. 4(B) and 7(A,C–D)]. The second site is completely outside the 7TM bundle facing the lipid bilayer and consisting of the interface of helices III–IV–V and ICL2 [Fig. 7(B–C,F)]. MK8666 and TAK‐875 compounds bind to the first site and are both reported as partial agonists. Interestingly, TAK‐875 was identified as a partial agonist, but also characterized as an allosteric ligand when tested against γ‐linolenic acid, a natural GPR40 ligand57 [Fig. 3(A)]. Despite the fact that orthosteric sites are classically described as inside the 7TM bundle, the inner volume of the first binding site is very small, and it would not fully accommodate a saturated C12‐16 or unsaturated C16‐18 lipid. In contrast, it has been suggested using molecular modeling of GPR40 that the γ‐linolenic acid binds in the second site.26 However, since the compounds that bind to the second site (Compound 1 and AP8) are positive allosteric modulator agonists (agoPAMs), the location of the orthosteric binding site for native lipid ligands in GPR40 is still unclear. Additionally, the two sites show different efficacies on activation of heterotrimeric G proteins, with Compound 1 in the second site activating Gq and Gs pathways in contrast to TAK‐875 in the first site that was reported to activate only the Gq pathway. This difference in heterotrimeric G protein efficacy seems to account for the dual induction of insulin and GLP‐1 secretion by Compound 1.26

Figure 7.

Allosteric modulation of lipid receptors. (A–C) GPR40 structures in complex with (A) MK8666 (PDB access code 5tzr), (B) compound 1 (PDB access code 5kw2) and (C) both MK8666 and AP8 (PDB access code 5tzy) showing two distinct allosteric sites. (D and E) Overlay of GPR40 in complex with MK8666 (green ribbons and sticks), and Compound 1 (blue ribbons and sticks). The conformational changes between the two structures are indicated by red arrows and dotted lines. Residue Leu190^5.46’s induced‐fit allows allosteric ligand AP8 to bind the receptor between the helices V and VI on the outer surface of the seven transmembrane bundle facing the lipid bilayer. Binding of AP8 induces the folding of ICL2, probably favoring the engagement of intracellular effectors.

There are distinct changes in GPR40 conformation upon binding ligands to either of the two sites. Ligand occupancy in the first site produces a movement of helix III and the folding of part of ECL2 to accommodate binding [Fig. 7(D)]. While ligand binding in the second site shifts the upper parts of helices IV and V. The change in helices IV and V is important as it leads to the displacement of the Leu190^5.46 side chain, effectively removing the space constraint for ligand binding [Fig. 7(E)]. Moreover, the presence of the ligand induces the folding of ICL2 through the ligand's interaction with Tyr114, a structural change that accounts for the dual efficacy of the second ligand site.26 Given the importance of ICL2 for effector coupling, the efficacy from the second site occupancy probably occurs through the direct promotion of effector engagement.50, 58, 59 Other GPCR crystal structures, such as the glucagon receptor60 and purinergic P2Y1 receptor,61 also display allosteric binding pockets facing the lipid bilayer. However, GPR40 is the first GPCR that shows a structurally well‐defined allosteric binding site facing the lipid bilayer. Since lipids freely diffuse in the membrane, it would not be surprising if some lipid receptors evolved to transduce the signal from a native lipid binding location outside the 7TM bundle. Additional lipid receptor structures are needed to fully understand the action of natural lipids on their receptors.

Future directions and conclusion

Lipid‐binding GPCRs play important physiological roles and are important active pharmaceutical targets for the treatment of many pathophysiological conditions (Tables 2, 3). The recent emergence of X‐ray crystal structures of lipid‐binding GPCRs has already provided a new understanding of these lipid's interactions with GPCRs from allosteric modulation to signal transduction (Tables 1 and 2). A tool, such as X‐ray crystallography, that allows the interpretation of the atomic details of lipids’ action on their receptor is important, as lipid receptors generally display a more complex pharmacology than receptors of other free‐circulating hormones and have an ability to mediate the response from a broader range of ligands.62 Also, lipids can be metabolized, a process that changes their selectivity for their cognate receptors and their physiological roles.55, 63 X‐ray structural biology will play a key role in decoding this process at the receptor level. For example, the fact that the CB1‐binding endocannabinoid anandamide can be phosphorylated to switch selectivity toward the LPA1 receptor had been suggested by the analysis of the LPA1 X‐ray crystal structure.25, 55, 63 The data gathered from the few available structures of lipid‐binding receptors show that they possess a higher binding site plasticity that allows for the promiscuity of lipid action. A better understanding of such plasticity can be achieved using X‐ray crystallography of lipid‐binding GPCRs, and would be an essential guide for the design of next generation drugs targeting this GPCR family.

Comment added in proof

The authors acknowledge the recent publication of additional lipid receptor structures, which were not available to us at the time of writing this manuscript, but have appeared online concurrently with the EP3 structure that was included here. These are the prostanoid receptor EP3 (pdb access code 6ak3)^a, prostaglandin E receptor EP4 (pdb access codes 5yhl, 5ywy)^b and thromboxane A2 receptor (pdb access codes 6IIU, 6IIV).^c

References

a. Morimoto K, Suno R, Hotta Y, Yamashita K, Hirata K, Yamamoto M, Narumiya S, Iwata S, Kobayashi T (2019) Crystal structure of the endogenous agonist‐bound prostanoid receptor EP3. Nat Chem Biol 15:8–10. https://doi.org/10.1038/s41589-018-0171-8.

b. Toyoda Y, Morimoto K, Suno R, Horita S, Yamashita K, Hirata K, Sekiguchi Y, Yasuda S, Shiroishi M, Shimizu T, Urushibata Y, Kajiwara Y, Inazumi T, Hotta Y, Asada H, Nakane T, Shiimura Y, Nakagita T, Tsuge K, Yoshida S, Kuribara T, Hosoya T, Yukihiko S, Nomura N, Sato M, Hirokawa T, Kinoshita M, Murata T, Takayama K, Yamamoto M, Narumiya S (2019) Ligand binding to human prostaglandin E receptor EP4 at the lipid‐bilayer interface. Nat Chem Biol 15:18–16. https://doi.org/10.1038/s41589-018-0131-3.

c. Fan H, Chen S, Yuan X, Han S, Zhang H, Xia W, Xu Y, Zhao Q, Wu B (2019) Structural basis for ligand recognition of the human thromboxane A2 receptor. Nat Chem Biol 15:27–33. https://doi.org/10.1038/s41589-018-0170-9.

Conflict of interests

The authors declare no competing interests related to the work described in this manuscript.