Physical abilities, cognitive skills and many other activities of humans and other organisms are generated by the coordinated actions of millions of cells. Each cell contributes to these activities by interacting with molecules outside itself. Many of these interactions are mediated through G-protein-coupled receptors (GPCRs), which in turn activate the eponymous intracellular G proteins, and so act as the starting points for numerous cellular signalling pathways. Arrestin proteins regulate the activity of these pathways by interacting with GPCRs, thus preventing G-protein-induced signalling and/or inducing additional signalling through G-protein-independent pathways. Two papers1,2 in this issue report the first X-ray crystal structures of arrestins in their active states*.
GPCRs are members of the largest family of membrane proteins. There are more than 800 different GPCRs, most of which are activated by small molecules (agonists). One exception is rhodopsin, a well-characterized GPCR found in the light-responsive cells of the retina. Rhodopsin is activated by photons, which change the isomerization state of retinal, its covalently attached cofactor. But all GPCRs have similar interaction partners downstream: after activation of a GPCR, signalling begins when a G protein binds to the receptor, and is terminated by the binding of an arrestin molecule, which is triggered by the phosphorylation of several amino acids at the carboxy-terminal end of the receptor.
If we understand the specific interactions between biomolecules in space and time, we can picture the molecular events that underpin macroscopic biological processes. Scientists have therefore invented numerous chemical and physical methods to study these interactions, with X-ray crystallography generally having the predominant role. Rhodopsin was the first GPCR to be crystallized and to have its structure solved to high resolution3, and the interactions of this receptor with other molecules in the visual system have been intensively studied. To establish how GPCRs interact with arrestins on the atomic scale, the method of choice would be to crystallize the two proteins together for X-ray analysis. But in most cases this is difficult. The two papers published today report impressive procedures for obtaining crystal structures of activated arrestins for which co-crystallization had failed.
Kim et al.1 (page 142) describe a method for activating p44, a naturally occurring variant of visual arrestin-1 in which 35 amino-acid residues at the C terminus have been replaced by a single alanine residue. In contrast to full-length arrestin-1 (refs 4,5), which binds only to light-activated, phosphorylated rhodopsin, p44 binds to rhodopsin with high affinity regardless of whether the receptor is activated and/or phosphorylated. The same research group had previously shown6 that opsin (retinal-free rhodopsin) behaves much like rhodopsin in the active meta II state — the intermediate that allows the enzyme rhodopsin kinase to phosphorylate the C-terminal domain of rhodopsin so that full-length arrestin-1 can bind and deactivate the receptor. In the current study, Kim and colleagues attempted to crystallize p44 in the presence of opsin. The co-crystallization failed, but the authors did obtain crystals of p44 alone, which they analysed by X-ray crystallography.
The resulting structure revealed major conformational changes in p44 when compared with unactivated full-length arrestin-1. The salient differences are a 21° twist between the amino-terminal and C-terminal domains in p44, and local changes of loop conformations and interacting hydrogen-bonding networks. The researchers attribute these conformational changes to the active form of p44, assuming that the opsin present during crystallization caused p44 to adopt this form. Whether this structure represents fully activated p44 is open to question — the protein is highly flexible, which means that crystal-packing effects might have induced the large conformational changes observed. A previously reported structure7 of p44 crystallized in the absence of opsin revealed much smaller conformational changes than in Kim and co-workers' study.
In the second arrestin paper, Shukla et al.2 (page 137) describe the crystal structure of non-visual β-arrestin-1 in complex with an antibody fragment (Fab30) and a phosphorylated peptide (V2Rpp) that corresponds to 29 amino-acid residues of the C-terminal end of a GPCR (the V2 vasopressin receptor). The authors found that this peptide activates β-arrestin-1, but that they could not crystallize the protein with the peptide alone — Fab30 was also needed to stabilize the activated state of β-arrestin-1 by binding to the arrestin's convex surface. Compared with the unactivated state of arrestin8, the most remarkable change in Shukla and colleagues' structure is the 20° twist between the N-terminal and C-terminal domains, which is similar to that observed in p44 by Kim and co-workers. However, the question arises of whether the C-terminal peptide V2Rpp induces the same conformational changes in β-arrestin-1 as the complete vasopressin receptor would do. Another issue is to what degree crystal packing and Fab30 contribute to the observed conformational changes in the β-arrestin-1.
An answer comes from the superposition of the structure of the unactivated state of visual arrestin-1 with that of β-arrestin-1 (Fig. 1a), and from the superposition of p44 and β-arrestin-1 in their active states (Fig. 1b): the structures are very similar in both cases. The striking structural similarity and the presence of Fab30 in one of the structures rule out crystal-packing effects as a cause of the twist between the N-terminal and C-terminal domains. It is therefore highly probable that Kim et al. and Shukla and colleagues have indeed observed fully active arrestin states. In this respect, the two papers strengthen each other's results, and pave the way for further studies of the structural basis of GPCR–arrestin interactions.
About this article
Nature Reviews Molecular Cell Biology (2013)