|Home | About | Journals | Submit | Contact Us | Français|
G protein-coupled receptors (GPCRs) are the largest family of proteins in the human genome. They transmit an exogenous signal to the intracellular second messenger cascade by ligand induced conformational changes within a common seven transmembrane helical bundle. Due to their critical role in biology and drug discovery, tremendous effort has been made over the past several decades to understand the mechanism of signal transduction through the cell membrane. Within the last year we have witnessed a relative explosion in the amount of structural information available for the GPCR family with two new structures of opsin in the presence and absence of transducin peptide, four new structures of β-adrenergic receptors and a recent structure of the human adenosine A2A receptor. The new biological insight being gained such as the highly divergent extracellular loops and areas of structural convergence within the transmembrane helices, allows us to chart a course for further investigation into this important class of membrane proteins.
Cellular recognition and transmission of exogenous signals is a central requirement for higher-level functionality of multicellular organisms. Nature has arrived at a few common mechanisms by which cells respond to external stimuli, either internalizing the stimulant directly, as in the case of protons and ions, or by propagating a signal induced by the stimulant through defined response pathways. One of the most common response pathways in the cell is formed by the guanine nucleotide-binding proteins (G proteins) which are involved in second messenger cascades triggered by the G protein-coupled receptors (GPCRs). Until recently, our atomic level understanding of GPCRs has been based on rhodopsin in its inactive state(Palczewski et al., 2000). While groundbreaking, it became apparent over the ensuing eight years that rhodopsin is a highly specialized member of the GPCR family that may not be the ideal representative for drawing generalized conclusions about the other family members (Figure 1). In the past year, the field of GPCR structural biology has enjoyed a renaissance with three new members yielding to crystallization efforts (Cherezov et al., 2007; Jaakola et al., 2008; Rasmussen et al., 2007; Rosenbaum et al., 2007; Warne et al., 2008) and significant inroads made in resolving the activated state of bovine rhodopsin (Park et al., 2008; Scheerer et al., 2008) (Table 1; Figure 2). Together these efforts represent the first examples of what is sure to be a blossoming of information for this important class of membrane proteins and as such we will review what insights may be gained from a structural comparison of the different regions of the GPCRs after introducing the structures themselves.
Since the initial structure of bovine rhodopsin eight years ago, much effort has gone into further characterizing the photocycle of this protein including the elusive activated structure (Li et al., 2004; Okada et al., 2004; Palczewski et al., 2000; Salom et al., 2006). The past year has witnessed great strides in the rhodopsin field and indeed in GPCRs in general with the publication of the opsin structure at 2.9 Å and opsin bound to a peptide from transducin (the main conduit by which rhodopsin transmits signaling information) at 3.2 Å (Park et al., 2008; Scheerer et al., 2008). The opsin structure represents an unliganded form of the receptor and displays significant structural deviation with rhodopsin in its inactive form. Interestingly, no significant structural differences are seen upon binding of opsin to the C-terminal peptide from transducin. This suggests that in the absence of its covalently bound ligand, 11-cis-retinal, opsin establishes an equilibrium close to that of activated receptor that is capable of binding its signaling partner. These two structures push the envelope in the GPCR field one step closer to resolving the active form of the receptor.
A flurry of activity has occurred within the last year surrounding β-adrenergic receptor (βAR) structural biology efforts. Two initial structures of β2AR where released in quick succession, the higher resolution complete model was crystallized as a fusion protein with T4-lysozyme using a nanovolume cholesterol doped lipidic cubic phase crystallization method (Cherezov et al., 2007). The majority of the protein was observed at 2.4 Å resolution including the extracellular loops and the ligand binding site allowing modeling of the receptor’s interactions with the partial inverse agonist carazolol. Genetic fusing of T4-lysozyme between TM V and TM VI, effectively replaced the disordered third intracellular loop increasing the available surface area potential for crystal contacts. The receptor displayed near native binding characteristics with essentially identical affinity for antagonists as compared to the non-fusion protein and a slight increase in affinity for agonists (Rosenbaum et al., 2007). Structural comparison to the 3.5 Å β2AR in complex with a Fab fragment bound to a structural epitope at the cytoplasmic base of TM V and TM VI that was crystallized using the bicelle crystallization method (Rasmussen et al., 2007) showed minimal structural deviations from β2AR-T4-lysozyme strengthening the assertion that the T4-lysozyme fusion had little effect on the structural properties of the antagonist bound receptor. First proposed as a strategy in the crystallization of lactose permease (Kaback et al., 1994; Prive and Kaback, 1996; Prive et al., 1994), the β2AR-T4 lysozyme structure represents the first conclusive demonstration of the general usefulness of the strategy in structural studies. Furthermore, the method appears to be quite general with any membrane protein possessing a disordered loop potentially being amenable to this technique. However, due to the importance of this loop in G-protein binding interactions, the fusion protein strategy will likely not be amenable to structural studies of the signaling mechanism of GPCRs. Furthermore the success of this technique may be dependent on the specific G-protein interaction partners that a particular receptor partakes and the resulting topology of the intracellular region as the squid rhodopsin structures would suggest (Murakami and Kouyama, 2008; Shimamura et al., 2008). Despite these difficulties, this structure could very well serve as a harbinger for the field of GPCR structural biology in combining the strategy of insertion of fusion proteins in loops of membrane proteins to promote crystallizability with the emerging technology of lipidic cubic phase crystallization (Caffrey, 2000, 2003; Cherezov et al., 2006; Landau et al., 2003; Landau and Rosenbusch, 1996; Nollert et al., 2002).
Shortly after the initial breakthroughs, an additional structure of β2AR-T4 lysozyme with a stabilizing point mutation, E122W, was solved in complex with the inverse agonist timolol (Hanson et al., 2008; Roth et al., 2008). Based on different crystal packing interactions, this structure established the very novel discovery of a specific binding site for cholesterol between transmembrane helices II, III and IV. This cholesterol binding interaction (cholesterol consensus motif, CCM) appears to be conserved across multiple members of the class A GPCR family based on sequence similarity among the four residues participating in the tightest binding interactions (Hanson et al., 2008). Cholesterol has long been thought to be involved in modulating membrane protein function by two fundamentally different mechanisms; in modifying bulk membrane properties such as fluidity and curvature or by direct interactions with the membrane protein in question (Pucadyil and Chattopadhyay, 2006). The new CCM site is a potential allosteric binding site, in close proximity to the orthosteric binding site of many small molecule therapeutics targeting a variety of class A GPCRs. Thus, this finding may open new avenues for drug discovery by directly targeting this area with sterol like molecules that retain binding to the four residues comprising the CCM motif and introducing specificity by designing binding interactions with less well conserved residues surrounding the CCM.
The turkey β1AR structure was solved in the presence of cyanopindolol, a strong antagonist (Warne et al., 2008). Stabilization of the protein by systematic mutagenesis coupled to a thermal stability assay was necessary to achieve high resolution diffraction (Serrano-Vega et al., 2008). The overall structure is very similar to that of human β2AR with the exception of the second intracellular loop (ICL2) which forms a short helical segment in β1AR whereas it is random coil in β2AR.
The recent publication of the adenosine A2A structure at almost exactly the one-year anniversary of the initial βAR structures underlines the general applicability of the approaches outlined above. This structure at 2.6 Å provides a well resolved picture of the adenosine ligand-binding site, which has shifted position relative to rhodopsin and the βARs and shows even greater helical shifts relative to rhodopsin than β2AR. Aside from the biological insights gained from the adenosine A2A structure, it provides a fundamental validation of the process by which it and two of the four β-adrenergic structures were generated.
Analysis of the extracellular regions of all three representative sub-classes (rhodopsin, adrenergic, adenosine) of GPCRs reveals a great deal of topological divergence and highlights the under-appreciated importance of this domain (Figure 3). For instance, rhodopsin has extensive secondary and tertiary structure in this region, which serves to completely occlude the binding site from solvent access. The N-terminus of rhodopsin along with ECL2 forms a four-stranded β-sheet with additional interactions between ECL3 and ECL1 (Figure 3A). There is one disulfide bridge that has been shown to be essential for the normal stability function of rhodopsin (Richards et al., 1995) and there is evidence that incorporating further disulfide bridges through genetic engineering can enhance the thermal stability (Standfuss et al., 2007). The extensive tertiary interactions between all four components of the extracellular region represents a very well defined structural unit that severely restricts access of solvent to the retinal binding pocket.
The extracellular region of the βARs is very open in comparison to rhodopsin. The most prominent feature is a short helical segment within ECL2 that is supported by limited interactions with ECL1 and two disulfide bridges, one with a random coil segment of ECL2 C-terminal to the helical stretch and the other fixing the entire loop to the top of TM III. The random coil section of ECL2 forms the top of the ligand-binding pocket, which is only partially occluded by the extracellular region (Figure 3B). In contrast to rhodopsin where the entire extracellular region forms a compact folded unit, ECL3 in the β-adrenergic family forms no interactions with ECL1 or ECL2 and the entire 28 residue N-terminus is completely disordered in the four structures solved to date. It would appear that the extracellular region of the β-adrenergic family has evolved to allow access to the ligand-binding site. Based on the interactions observed between ECL2 and the bound ligand, it is possible that this loop is even more disordered in the absence of the ligand allowing even greater access to the binding pocket.
The extracellular region of the adenosine receptors is highly constrained by four disulfide bridges and multiple polar and van der Waals interactions between the three loops. Three of the four disulfide bridges constrain the position of ECL2 by anchoring the loop to ECL1 and the top of TM III. Nevertheless, the tip of ECL2 is highly flexible and not observed in the electron density maps. It appears that the three disulfide bridges serve to stabilize a short helical segment of ECL2 just N-terminal of TM V (Figure 3C) which presents two important residues, Phe168 and Glu169 for ligand binding interactions (Figure 4C). ECL3 contains an intraloop disulfide bridge between Cys259 and Cys262 whose function is unknown, however based on the structure, we speculate that it constrains the position of His264 which forms a polar interaction with Glu169. Thus, the extracellular region of the adenosine family appears to have evolved to form multiple ligand binding interactions.
Sequence conservation among class A GPCRs is highest within the transmembrane regions. Thus it is not surprising that the helical bundle orientation and packing is similar among the structures solved to date. However, it is instructive to analyze this region in some detail for two main reasons: 1. Through structural comparison one can map the regions in the receptors that are most similar. 2. The areas of structural divergence within the helical segments can inform differences in receptor pharmacology. We have carried out an alignment of the transmembrane regions of all five representative GPCRs and identified a common structural core of 97 residues with an average Cα RMSD of 1.3 Å (Table 2). Calculation of the RMSD over all residues in the transmembrane helices after aligning only the common structural core indicates that the non-core residues tend to be highly structurally divergent, increasing the total transmembrane RMSD considerably (Table 2). As such these divergent positions are likely to be responsible for differences in signaling and ligand binding properties observed across the class A receptors.
The position of the antagonist-binding pocket varies significantly as a function of the receptor. It is useful to compare the two alternate binding pockets to that of rhodopsin, both in terms of relative position in reference to specific helices and in more absolute terms using the membrane plane as a reference point. As was previously described, the β-adrenergic binding pocket is quite similar to that of rhodopsin (Cherezov et al., 2007). The position does not vary considerably with alternate ligands or between different sub-types of different species (Hanson et al., 2008; Warne et al., 2008). As a representative ligand the aliphatic tail of carazolol follows a very similar path as that of rhodopsin (Figure 4A and 4B). The ligand in both cases extends from TM VII, where it participates in polar interactions for carazolol and is covalently bound in the case of rhodopsin, to the TM V/ TM VI interface. Here the position of retinal and carazolol deviate as retinal extends deeper into the receptor to interact with Trp265 (a tryptophan is in general conserved at this position in class A receptors and is thought to be involved in receptor signaling) whereas carazolol forms more extensive interactions with TM V.
With the recent elucidation of the A2A adenosine structure, we see that the ligand binding pocket can assume a very different location to that of rhodopsin and the βARs (Jaakola et al., 2008). In addition to shifting to the interface of TM VI and TM VII, ZM241385 forms extensive interactions with ECL2 and is rotated relative to carazolol and retinal, so that the long axis of the ligand is perpendicular to the plane of the plasma membrane (Figure 4C). The binding pocket is also much higher relative to the modeled position of the plasma membrane with almost half of the ligand completely exposed to bulk solvent (Lomize et al., 2006).
Based on sequence conservation and the initial bovine rhodopsin structure, it has been assumed that a conserved triplet of residues termed the DRY motif (usually consisting of an aspartate or glutamate followed by an arginine and tyrosine) at the intracellular base of TM III found in most class A GPCRs, takes part in a conserved interaction with a glutamate residue at the base of TM VI (Okada et al., 2004; Palczewski et al., 2000; Vogel et al., 2008). This proposed conserved interaction was termed the ionic lock and thought to be an important component in stabilizing the inactive state of most members of the class A GPCR family. The determination of the β2AR cast doubts on this assumption due to the finding that despite the presence of the DRY motif and glutamate residue, no interaction was observed. Now with the solution of both the β1AR and adenosine A2A, neither of which have the ionic lock interaction, one can draw the conclusion that the ionic lock observed in rhodopsin is not a universal interaction among the class A receptors. The question remains, however, what is the purpose of the conserved triplet of residues, if not to stabilize the inactive state of GPCRs through ionic interactions? It is interesting to note that the unliganded form of rhodopsin (opsin) undergoes conformational changes especially in the intracellular region that can mimic some of the features observed in the other GPCR structures (Figure 5). For instance, in rhodopsin helix V extends one turn below the plane of the membrane, as determined by the database OPM (Lomize et al., 2006). Upon loss of retinal, however helix V becomes extended by two turns effectively shortening ICL3, perhaps constraining the movement of helix VI pulling it outward and away from the base of TM III and the DRY motif. While it is difficult to interpret this region due to the presence of the T4L fusion protein, which replaces ICL3, it is apparent that in both β2AR and A2A helix V is 1.5 turns longer than that of rhodopsin, perhaps having a similar effect on TMVI and the ionic lock (Figure 5). This analysis is substantiated by β1AR, which did not possess a fusion protein and also does not have an ionic lock interaction. Instead in these receptors the DRY region is interacting with ICL2 by means of a polar hydrogen bond between the aspartate residue and either a serine or tryptophan on the loop. One interesting correlation points to the type of interaction in this area having implications for levels of basal activity (Jaakola et al., 2008). For instance, in both A2A adenosine and β1AR, ICL2 consists of a short helical segment and the DRY aspartate interacts with a conserved tyrosine at a similar position on this helix and both have a low basal activity (Birnbaumer et al., 1994; Zezula and Freissmuth, 2008). However, despite the presence of a tyrosine at the same position in β2AR, the DRY aspartate interacts with a serine two positions C-terminal to the conserved tyrosine and has a relatively high level of basal activity (Birnbaumer et al., 1994). If this correlation holds, our understanding of the conformational changes that take place upon agonist binding together with the structural information already accumulated on G protein subunits will further our understanding of the interactions that take place among different class A GPCRs and their downstream effectors (Table 1).
It is said that form follows function in structural biology. This is a particularly prescient statement in the context of GPCR structural biology as this class of receptors has evolved naturally to induce signaling along many pathways dependent on ligand binding, cellular localization and allosteric modulation. This functional diversity however, is at the core of the difficulties in structurally resolving aspects of their biology, making development of novel approaches a necessity. Within the last year we have witnessed a veritable explosion in the amount of structural information available on this class of pharmaceutically important integral membrane proteins. Due to this exponential increase in information over the last year the field has progressed from doubting the viability of GPCR structural biology to anticipation of the next receptor to yield to this type of analysis and new discovery.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.