The Impact of G Protein-coupled Receptor (GPCR) Structures on Understanding Signal Transduction

DAVID T. LODOWSKI AND KRZYSZTOF PALCZEWSKI

Department of Pharmacology, School of Medicine, Case Western Reserve University, Cleveland, Ohio 44106-4965, USA

1.1 Introduction

Recent advances in the structural study of G protein-coupled receptors (GPCRs) have significantly enriched our understanding of the process of G protein signalling, providing a structural framework to understand the huge volume of biochemical and biophysical studies on GPCRs performed and published over the past 50 years. Determination of X-ray crystal structures provides the most direct methodology for examining the structural aspects of GPCR signalling and careful analysis of these structures reveals mechanistic details shared by all GPCRs (Table 1.1).

1.2 Early Approaches to Analysing GPCR Structure

Prior to the determination of the rhodopsin structure, similarities were noted between the light responsive photopigment, bacteriorhodopsin, and purified bovine rhodopsin, including their Schiff base linkage of chromophore, photoactivation and seven-transmembrane architecture (Figure 1.1A). After structure determination of bacteriorhodopsin via electron crystallography, (reviewed in ref. 18), it was assumed that these two proteins shared a similar architecture. Protein and DNA sequencing of rhodopsin and the β2- adrenergic receptor, coupled with hydropat hy plots, allowed the construction of two-dimensional (2-D) topology models (Figure 1.1B and 1.1C). Further clarification provided by electron crystallographic studies of 2-D crystals of bovine (Figure 1.1D) and invertebrate rhodopsin suggested that the arrangement of helices in bacteriorhodopsin versus rhodopsin and other GPCRs were distinct. These studies also improved predictions of the helical arrangement within the transmembrane bundle (Figure 1.1E). Mutagenesis and biochemical studies allowed location assignments of disulfide bonds, palmitoylation, and phosphorylation sites as well as further clarification of the border between loop and transmembrane regions (reviewed in ref. 42–46) (Figure 1.1F).

1.3 The Crystal Structure of Rhodopsin

While several laboratories were able to obtain crystals of rhodopsin, these studies did not result in the determination of a crystal structure. A postdoctoral fellow and a technician in our laboratory, Dr. Tetsuji Okada and Preston Van Hoosier, were able to improve the purification of rhodopsin from a very high quality preparation of bovine rod outer segment membranes. This provided homogenous bovine rhodopsin at the purity, quantity and concentration necessary to enable growth of diffraction quality crystals. These crystals led to the determination of the first crystal structure of rhodopsin and subsequent rhodopsin structures extended the resolution to 2.2 Å, facilitating subsequent structural studies of additional GPCRs obtained by similar purification methods, as well as those obtained by an alternative purification method. These structures supplied the first true atomistic view of any GPCR, fully defining its transmembrane region and topology, including interhelical contacts. This detailed resolution cemented our knowledge as to the position of the chromophore and its contacts within the transmembrane (TM) region (Figure 1.2A). Using these structures, later biophysical studies probed the mechanisms by which conserved motifs act together to both maintain the GPCR in its inactive state and release these constraints upon activation. The rhodopsin crystal structure has continued to provide a high-resolution template for the development of homology models of other GPCRs, which has even been extended to virtual screening of ligands with these models. The initial rhodopsin X-ray structure also allowed further revision of the 2D model introduced by Hargarve and Argos (Figure 1.2B).

1.4 Conformational Intermediates of Rhodopsin

Because rhodopsin is relatively easily purified in its native state from bovine retina, a considerable amount of biochemical data exists pertaining to all aspects of the rhodopsin activation pathway. Through the use of specific temperatures, wavelengths of light, chemicals and pH, it is possible to isolate various photointermediate states which display characteristic absorbance maxima (Figure 1.3). Research by indirect methods predating the structural determination of rhodopsin relied on this photointermediate cascade to define functional intermediates which could be characterized (Figure 1.3). However, some of these photointermediates were identified under low temperature conditions which might trap a molecular species that does not occur physiologically, rendering the functional relevance of these photointermediate states open to question. Use of electron paramagnetic resonance (EPR) measurements upon attainment of the meta II state provided evidence for a longstanding proposed mechanism for the activation of GPCRs, which portrayed large-scale rigid body movements of helices to occur upon activation. However, later refinements of these initial studies reduced the scale of these proposed movements to more thermodynamically feasible scales.

Structures of several photointermediate states of rhodopsin determined by Xray crystallography as well as by electron crystallography have revealed changes in both the conformation of the chromophore upon photoactivation as well as the scale of changes accompanying photoactivation. Structures of early photointermediates of rhodopsin, batho- and lumirhodopsin, do not exhibit any significant changes within their protein backbones; only changes in the conformation of the chromophore are observed (Figure 1.4). Determination of a low resolution structure of a photoactivated rhodopsin containing the characteristic deprotonated Schiff base linkage of the meta II activated state by our laboratory revealed that only small-scale changes were needed to achieve this state. More recent work has revealed the conformation of the inactive apo-protein, opsin, in complex with a peptide which is proposed to induce a conformation resembling the activated state. Structural superposition reveals that the photoactivated meta II state falls midway between ground state rhodopsin and opsin.

In analysing structural information, it is important to consider that opsin is very inefficient at activating G protein compared with meta II Rho and that GPCR peptides are not equivalent to the proteins from which they are derived. While it is tempting to state that the peptide induces a conformation akin to the meta II active state, there is a fundamental disconnect; contacts that rhodopsin makes with transducin are much more extensive than this isolated helix, and docking of Gt using the position of this peptide would result in a collision with the lipid bilayer. The wasp venom derived mastoparan peptide also very efficiently activates G protein subunits, but it is unlikely that a co-crystal structure of mastoparan and G protein would reveal novel aspects of G protein activation which are germane to the GPCR–G protein interaction and nucleotide exchange.

1.5 Crystal Structures of Other GPCRs

More recent groundbreaking work has employed heterologously expressed GPCRs engineered to stabilize their conformations to maximize crystallization success. Squid rhodopsin also required modification for successful crystallization, even though it was obtained from a native source (Figure 1.5A). Proteolysis to remove a large cytoplasmic domain allowed crystallization and consequent structural determination. Due to limited success with Fab fragment complexes in stabilizing the β2-adrenergic receptor for crystallization, it was proposed that mutations which confer stability or reduce conformational variability might assist in obtaining diffraction quality crystals.

In the case of the β2-adrenergic and A2A-adenosine receptors, heterologous expression was carried out to obtain the respective proteins with particular mutations, including the critical insertion of a T4-lysozyme into the corresponding C-III loops (Figure 1.5B and 1.5C, respectively); further stabilization was achieved with the addition of an antagonist to potentiate the inactive states of these receptors during crystallization. Crystallization of the β1-adrenergic receptor also employed an antagonist for stabilization, but also relied on mutations which were presumed to stabilize the structures based on positive increases in thermal stability (Figure 1.5D). It is readily apparent that all these structures share the same global fold; they differ significantly only in their cytoplasmic and extracellular loops and ligand binding sites (Figure 1.5A–D).

1.6 Sequence Similarities and Conserved Motifs within GPCRs

The structural similarity shared by all GPCR structures determined to date is quite high and is expected to extend to other GPCRs whose structures have not yet been determined. However, the sequence homology shared by over 900 GPCRs is limited to a number of conserved motifs that are likely to play similar functional roles in most, if not all, GPCRs (Figure 1.6). When all the reported structures of GPCRs are superposed, and the locations of crystallographically observed solvent (water) along with the side chains to which they are hydrogen bonded are examined, it is readily apparent that these waters occupy similar positions within the structures and that the sequence homology of the side chains with which they interact is high (Figure 1.7). Because of this high degree of 'water homology', these waters and the residues to which they are hydrogen bonded most likely comprise a network along which the activation signal is passed upon agonist binding or photoactivation. Radiolytic footprinting studies performed by our laboratory, in conjunction with the Chance laboratory, indicate that such waters do not freely exchange with bulk solvent, further supporting their roles as non-covalently bound prosthetic groups. Meta-analysis of all GPCR structures links these clusters with conserved sequences such as the D(E)RY and NPxxYx(5,6)F motifs that are thought to play crucial roles in receptor activation.

1.7 The D(E)RY Motif and GPCR Activation

The ionic lock or D(E)RY region represents an energetic barrier which must be broken in order to reach the activated state. Most but not all GPCRs contain this motif, suggesting that it plays an important but not wholly indispensible role in the activation process. A conserved glutamic or aspartic acid residue located in helix-VI (E247 in Rho) makes a hydrogen bonding interaction with a conserved Arg (R135 in Rho) residue within helix-III (Figure 1.8). Disruption of this bond is largely considered a hallmark of progression from the meta I to the meta II state. Protonation of the acidic residue within this motif is thought to accompany activation. This hydrogen bond was disrupted in all recently determined structures of non-rhodopsin GPCRs (although bound to inverse agonists/antagonists), leading to much speculation that this lock was unique to rhodopsin and that, while the sequence motif was present in most GPCRs, it might not actually act as a functional 'lock'. This supposition prompted a series of molecular dynamics simulations that revealed that, for each of these receptors, the ionic lock — while disrupted in the crystal structure reforms when the restraints imposed by the T4-lysozyme fusion or crystal lattice are removed (Figure 1.8). However, partial occupancy of this broken ionic lock state might be an explanation for the agonist independent activation observed in these receptors.

1.8 The NPxxYx(5,6)F Motif within GPCRs

The NPxxYx(5,6)F motif is also highly conserved among members of class A GPCRs (Figure 1.9). Mutational analysis indicates that this region is important for receptor endocytosis and transport, interaction with small GPCR interacting proteins such as ADP ribosylation factors (ARFs) and Rho A, as well as for G protein binding. The hydrogen bonding interaction between this motif's Tyr residue with a conserved Asn is postulated to directly affect G protein coupling. Furthermore, hydrophobic interactions between the Tyr and Phe residues link helix-8 to the end of helix-VII, allowing changes in position of helix-VII to induce movements of helix-8 upon activation. In higher resolution structures of rhodopsin, a cluster of three waters appear to link helices H-I, H-II and H-VI. Further in-depth examination of bound water interactions in this vicinity reveals that similar networks of waters interact with the Asn residue (Figure 1.9). However, this interaction between N73 and Y306 observed in rhodopsin crystal structures is not seen in any other GPCR structure.

1.9 Ligand Binding Domains of GPCRs

In the human body, GPCRs respond to many diverse agents. Activation results from absorption of a single photon in the case of rhodopsin, binding of a small hormone molecule (e.g. β1/β2-adrenergic receptors), proteolysis of a tethered ligand (e.g. thrombin receptor), binding of a peptide (e.g. vasopressin receptor) or even binding of an entire protein (e.g. follicle-stimulating hormone receptor). It has been estimated that upwards of 60% of all current therapeutics act on or modulate GPCR-mediated signalling events, emphasizing the medical importance of this group of proteins. Given the high diversity of GPCR activating compounds, it is hardly surprising that there are multiple sites for agonist binding. In many cases (including all structurally determined GPCRs), the ligand binding site is located within the transmembrane region (Figure 1.10A–D). However, for peptide and protein hormones the ligand binding site often is either on the extracellular face or extends to entire domains attached to the transmembrane region (Figure 1.10E and 1.10F). Limited structural data are available for a few of these extracellular domains, examples of which are shown in Figure 1.10.