The crystal structures described so far revealed detailed molecular structure of a GPCR in the inactive ground state but did not provide any information about the mechanism of activation or subsequent rearrangements of the helical domains upon activation. The first data on activated states of rhodopsin came from 2D crystals of the photointermediate MI, which were obtained by cryo - EM [ 40] . Although these structures were all of rather low resolution (2.7-5.5 Â) , they still enabled comparison with the dark (ground) state of rhodopsin. In these structures, no large (rigid body) movements or rotations of helices had occurred. The 2D map revealed movements of the side chains close to the retinal binding pocket. For example, a clear density bulge was present in the middle of TM6 facing the P --onone ring in the MI structure, which was not present in the ground state. This is close to the location of Trp265, suggesting movement of this residue in the MI state. 3D crystal structures of bathorhodopsin and lumirhodopsin were obtained by trapping these photolyzed states at low temperature [41, 42] . In the higher resolution 3D structure of lumirhodopsin, the main difference was in the middle of TM3, suggestive of an outward movement of the polypeptide backbone. However, this change was not propagated to the intracellular side associated with G protein activation. In the lumirhodopsin structure, retinal is almost completely in the all--rans conformation. The P-ionone ring is displaced in the direction toward TM3 and TM4. Displacement of the P--onone ring causes a number of perturbations of surrounding residues, which results in the breakage or weakening of electrostatic restraints between helices. These include hydrogen bonding sites, which are highly conserved across other GPCRs including Asn55-Asp83, and hydrogen bonding to Asn302 in the NPxxY motif.
Rhodopsin is inherently unstable in its photoactivated deprotonated state so Salom et al.  attempted to get round this problem by selecting crystals in the ground state that remained stable upon exposure to light at room temperature. Most crystals lose diffraction when illuminated; however, the group was able to identify crystallization conditions that generated two crystal forms that could withstand photoactivation. One was a rhombohedral structure that diffracted to 3.7 A but lost resolution upon light, whereas the other was a trigonal form that diffracted to 4.1-4.2 A before and after photoactivation. Interestingly, these crystals could still activate G protein when dissolved in detergent, although only to a limited extent. These were the first crystals to contain all -trans-rhodopsin and represented an MII-like intermediate stage in the activation of the receptor. Unfortunately, the low resolution of the structures meant that locations of the main-chain carbon atoms could not be precisely determined and locations of the side chains could not be resolved. Nevertheless, some conclusions could be drawn; both cytoplasmic loops became more disordered upon photoactivation, indicating movement in this area; in addition, as seen in the previous structure, there were no large displacements of the individual helices.
The lack of movement of individual helices in these structures did not fit with those predicted from other indirect methods, such as electron pair spin resonance (EPR) studies on spin-labeled cysteine mutants [44, 45]. It is probable that the crystal lattice limits the magnitude of changes observed in these structures or that the additional energetic barriers have not been overcome to enable the formation of the functionally active conformation despite the presence of all-trans-retinal and the deprotonated Schiff base.
11-c's-retinal is a strong inverse agonist, which holds rhodopsin in its inactive state. Subsequent to receptor activation, all-trans- retinal is released from its binding site, and a new light-sensitive rhodopsin is generated through the binding of a new 11-a's-retinal molecule. The transiently formed apoprotein, in the absence of any ligand, is opsin. In the absence of 11-c's-retinal, opsin can bind and activate G protein, and is therefore considered a constitutively active receptor [46, 47]. Like other ligand-free GPCRs, opsin is very unstable and difficult to purify. In 2008, Park et al.  were able to purify native opsin from bovine rod disk membranes. These formed colorless crystals in P-D-octylglucopyranoside, which changed to red upon addition of 11-c's-retinal, a process that is also observed with ROS membranes. A 2.9 A structure was determined, which represented the first activated structure of a GPCR. In this structure, the protein is arranged as a dimer with a TM1/8 interface. There was very little difference in the arrangement of TM1-TM4 in opsin compared to the ground state of rhodopsin. Larger changes were found for TM5-TM7, in particular, at the ends of the helices, resulting in a rearrangement of both the second and third cytoplasmic loops, C2 and C3, respectively. There was also a significant movement of TM5 at the cytoplasmic end toward TM6. TM6 itself was moved outward from the center of the helix bundle, and the cytoplasmic end of TM6 was shifted by as much as 6-7 A outward, pivoting at Trp265.
In this partially activated form of opsin, the ionic lock between Asp135 and Glu247 was broken, and two new interactions stabilized the ends of TM5 and TM6. There was also a deviation of TM7, which includes the NPxxY motif, causing Tyr306 to rotate into the helix. A surprising finding in this structure was the appearance of two openings close to the retinal binding pocket: one between the extracellular ends of TM5 and TM6, and one between TM1 and TM7. Intriguingly, these two openings may be the entrance and exit route for retinal channeling into and out of the binding site. Prior to this structure, it was hypothesized that the extracellular cap might be displaced to allow entry to the ligand binding site. This entry route to the binding site between TM5 and TM6 may be a general route for other Family A GPCRs that bind hydrophobic ligands. Although this is a partially active form of the receptor, the relationship between this structure and the fully activated R* state is not clear. This awaits the crystal structure of MII in complex with a G protein.
To date, the most convincing data on the transitions, which occur upon activation, come not from X-ray crystallography, but from a new version of EPR known as double electron-electron resonance (DEER). Using this method, 16 pairs of nitroxide spin labels were introduced, one pair at a time, by site-directed mutagenesis and chemical modification, into the cytoplasmic ends of the helices of rhodopsin [49, 50] . Interspin distances were then measured in the inactive and light-activated state. This study was facilitated by the development of an optimized protocol, which included detergent (dodecyl maltopyranoside, DDM), low pH, and high concentrations of glycerol together with rapid cooling in liquid nitrogen to trap the MII conformation. The key findings of these experiments were the outward displacement of TM6 by 5 A. Smaller movements were also observed in TM1, TM7, and the C -t erminal domain. The negative data in this study were also important, namely, that no changes were observed in TM2, TM3, TM4, and TM5, which form part of a core helical bundle that does not change upon photoactivation.
Subsequent to the publication of the opsin structure, the same group  was able to obtain a structure of activated opsin in complex with an 11-amino acid peptide derived from the extreme C-terminus of the transducin Gat subunit (Ga-CT) solved to 3.2 A. In the structure, the Ga-CT is seen binding in an a-helical conformation to a crevice formed by the movement of TM5 and TM6. An important feature of the structure is the role of Arg135 from the E(D)RY motif. As described above, in the activated opsin structure, the ionic lock is broken, and Arg135 no longer interacts with Gly134 but instead interacts with Tyr233 in TM5. This causes the arginine side chain to swing into the center of the binding crevice for Ga-CT, where it interacts with the carbonyl of Cys347 on the G protein C- t erminus. As in the opsin structure, Tyr306 (TM7) extends into the helical bundle below Arg135, where it may act to stabilize the position of TM6.
A particularly interesting feature of the Scheerer et al. structure is the longrange changes in the receptor through to the ligand binding site as a result of binding of the G protein peptide. A network of stabilizing interactions is formed among Lys296 on TM7 in the retinal binding site, Ser186 and Glu181 in the E2 loop, and Tyr268 in TM6. Thus, it appears that both ligand and/or G protein can stabilize the activated state of the receptor. This structure has provided the first evidence for a model of signal transduction whereby the C-terminal a5 helix of Gat acts as a "transmission rod," such that binding of the C-terminal tail to the receptor alters the juxtaposition of the a5 helix and the remainder of the G protein, resulting in receptor-catalyzed GDP release. It remains to be determined how the active receptor first recognizes and binds the GDP-bound G protein.
Was this article helpful?