Probing Arrestin Conformation
By: Carol A. Rouzer, VICB Communications
Published: November 5, 2012
Site-specific spin labeling, combined with molecular modeling, reveals the intricacies of conformational changes of a key signaling protein.
The seven transmembrane domain receptors are the largest superfamily in the human genome, comprising ~3% of all protein-coding genes. Named for their common structural feature - seven transmembrane α- helical domains (Figure 1) - these proteins serve as receptors for a vast number of external stimuli, including many hormones, neurotransmitters, and even light. They also share a common mechanism for signal transduction. Upon binding of its specific ligand (hormone, neurotransmitter, etc.), the receptor undergoes a conformational change to its activated state. The activated receptor then interacts with one or more members of the G protein family. It is the G protein that carries the signal into the cell through its interaction with multiple effector proteins. Thus, the seven transmembrane domain receptors are also commonly referred to as G protein-coupled receptors.
Figure 1. Generalized structure of a seven trans-membrane domain receptor (also known as a G protein-coupled receptor) embedded in a lipid bilayer. Image kindly provided by the Gurevich lab.
A common structure and mechanism for signal transduction suggests a common mechanism for termination of signaling, which in the case of G protein-coupled receptors is mediated by the arrestin family of proteins. Once a G protein-coupled receptor has been activated by ligand binding, it is subjected to phosphorylation at key intracellular serine and threonine residues. The phosphates attached to active receptor attract an arrestin protein, which binds and blocks the further interaction of the receptor with G proteins, initiates a second round of signaling, and targets the receptor to coated pits for internalization (Figure 2). Although crystal structures for arrestin proteins are available, the mechanism by which they recognize activated and phosphorylated receptors is not well understood. This led Vanderbilt Institute of Chemical Biology members Vsevolod Gurevich and Jens Meiler to team up with the Wayne Hubbell lab at UCLA to investigate arrestin-receptor interactions [M. Kim et al., (2012) Proc. Natl. Acad. Sci. U.S.A., published online October 22, DOI: 10.1073/pnas.1216304109].
Figure 2. Ligand binding to a receptor (R) leads to phosphorylation and arrestin binding. The arrestin-receptor complex is then internalized. The receptor may be targeted for degradation in lysosomes or returned to the plasma membrane. Image reproduced from L. Pan et al. (2003) J. Biol. Chem., 278, 11623. Copyright 2003 ASBMB.
Much of what we know about G protein-coupled receptors comes from studies on rhodopsin (Rh), the light receptor found in rod cells in the retina. This is also true for arrestins, which were first discovered in the visual system. Mammals express four subtypes of arrestin, and two of these, arrestin-1 and arrestin-4, are specific to the eye. The other two, arrestin-2 and arrestin-3, are ubiquitously expressed, indicating that this mechanism of G protein-coupled receptor regulation has widespread physiological relevance. For their studies, the Gurevich, Hubbell, and Meiler team focused on Rh, and its interaction with arrestin-1.
Previous X-ray crystal structures of unbound arrestin-1 had revealed a tetramer in which each subunit exhibits a distinct conformation. All four conformations comprise two cup-like domains linked by a hinge. The differences in the conformations reside primarily in the positions of three loops located between residues 68 and 79, residues 155 and 165, and residues 337 and 347. Since prior data had suggested that arrestins undergo a conformational change upon binding to an activated receptor, the research team hypothesized that the three flexible loops might be key drivers of this proposed change. To test their hypothesis, the researchers constructed twenty-five arrestin-1 molecules, each bearing nitroxide side chains at two different positions within the molecule (Figure 3). In each case, one side chain was placed in a rigid portion of the protein to serve as a reference point, while the other was placed in a region predicted to be more flexible. The presence of the nitroxide labels allowed the team to use double electron-electron resonance (DEER) to measure the distance between the labeled sites in the unbound basal state and after binding to activated and phosphorylated Rh (P-Rh*). The resulting data provided the median distance between the two labels in each state of the protein as well as a distribution of distances around the median. A broad distance distribution suggests flexibility in the protein in the region involving the labels, while a narrow distribution suggests greater rigidity. A change in median distance between the free and unbound state indicates a conformational change involving the labeled residues.
Figure 3. Ribbon diagram of one of the subunits in the crystal structure of arrestin-1. The cup-like domains are shown in green (C domain) and gray (N domain), and the interdomain hinge that connects them is also labeled. The positions of all nitroxide labels are marked. Labels at sites 60, 85, 173, 227, 240, 244, 267, 272, and 348 serve as stable reference sites. Note labels at 72, 74, 157, and 344, which are present in flexible loops visualized in the crystal structure. Reproduced from M. Kim, et al. (2012) Proc. Natl. Acad. Sci. U.S.A., published online Oct. 22, DOI: 10.1073/pnas.1216304. Copyright 2012, M. Kim et al.
The research team used the DEER data combined with the crystal structure data to create Rosetta models of the arrestin-1 conformation as a tetramer in solution and bound to P-Rh*. The model of the solution tetramer predicted that, unlike the crystal structure, all subunits adopt a single conformation. Prior studies had suggested that the two cup-like domains of the protein might fold around P-Rh* in a clam shell-like fashion. However, this hypothesis was not supported by the DEER data and modeling results, which indicated that the relative positions of the cup-like domains do not change much upon receptor binding. In fact, most of the observed movement occurred in the flexible loops. Of particular interest was the “finger loop” (residues 68 to 79), which was known from previous work to be involved in receptor binding. Labels placed on residues 72 and 74 in this loop showed distance changes upon P-Rh* binding of up to 8.5 Å, and a decrease in flexibility suggested that the finger loop is stabilized by a direct interaction with the receptor. In addition, labels placed at positions 157 and 344 moved away from and toward, respectively, the central crest of the protein in the P-Rh*-bound conformation, confirming that flexible loops at residues 155 to 165 and residues 337 to 347 were also involved in a receptor binding-dependent conformational change. These loops, however, retained their flexibility in the bound conformation (Figure 4).
Figure 4. Ribbon drawings of the superimposed Rosetta models of the solution tetramer and P-Rh*-bound conformations of arrestin-1. The positions of nitroxide labels in the flexible loops are indicated with spheres, and the motion of these loops on receptor binding can easily be seen. Reproduced from M. Kim, et al. (2012) Proc. Natl. Acad. Sci. U.S.A., published online Oct. 22, DOI: 10.1073/pnas.1216304. Copyright 2012, M. Kim et al.
A striking result provided by the DEER data concerned a fourth loop located near the finger loop. A label placed in position 139 of this loop revealed movement of up to 17 Å on binding to P-Rh*; however, the retention of flexibility of the loop in the bound state suggested that it is not directly involved in a receptor interaction. Deletion of the loop increased binding to P-Rh*, but it also reduced selectivity for P-Rh* over Rh* and P-Rh and decreased thermal stability of the protein. In the unbound state, the C-terminus of arrestin-1 interacts strongly with β-strand I of the N-domain (Figure 5). This interaction is broken in the P-Rh*-bound state. Deletion of the 139 loop also disrupted the C-terminus-β-strand interaction, suggesting that this loop helps to stabilize the conformation of the unbound state of the protein.
Figure 5. Ribbon drawings of the unbound conformation of arrestin-1 showing the position of the 139 loop, the finger loop, and the C-terminal tail (red) interacting with β-strand I. Reproduced from M. Kim, et al. (2012) Proc. Natl. Acad. Sci. U.S.A., published online Oct. 22, DOI: 10.1073/pnas.1216304. Copyright 2012, M. Kim et al.
Together the data support the hypothesis that flexible loops, as observed in the crystal structure of arrestin-1, are primarily responsible for the conformational change that occurs on receptor binding. Three out of four of these loops retain their flexibility in the receptor-bound state, indicating that they are not directly involved in P-Rh* interactions. The surface of arrestins available for protein-protein interactions is much larger than the arrestin-interacting surface of P-Rh*, suggesting that, even after P-Rh* binding, much of the arrestin protein remains available for interaction with other signaling molecules.