Deactivation of light-activated rhodopsin involves phosphorylation of its C-terminus by RK, and subsequent binding of arrestin to block interaction with Gt. Photoactivated rhodopsin eventually decays into (phosphorylated) opsin and all-irarcs-retinal. The role of arrestin in this process was confirmed by knock-out studies . Binding of arrestin is specific for the phosphorylated Mil form with a KD of 50 nM and a bimolecular on-rate of about 0.2 (iM_1 s_1 [196,208]. The major phosphorylation sites of rhodopsin are Ser338, Ser343 and Ser334 . Under extrapolated cellular concentrations, the binding to rhodopsin is fast (of the order of 50 ms) and thus not rate limiting for the overall shut-off reaction sequence , Moreover, self-association of arrestin was proposed as a regulation mechanism . The affinity to opsin is low, as seen by the release of arrestin due to hydroxylamine-induced Mil decay . The rhodopsin-arrestin interaction may be short-lived in vivo; additional experiments in mice indicate that Mll-arrestin complexes dissociate, to allow the reduction of all-ira/M-retinal to all-ira«^-retinol  and dephosphoryla-tion of the receptor by a membrane-associated form of protein phosphatase 2A (PP2A) . It remains to be studied how the necessary release of arrestin is induced. On contact with the phosphorylated C-terminus of rhodopsin, arrestin switches from the inactive to the active state which is capable of binding to rhodopsin . Two crystal structures of arrestin are available [201,202] and show an interesting "double-cap" structure, formed by a N- and a C-terminal domain of arrestin, respectively. However, the two structures disagree in important details such as the location of the N- and C-termini. Both structures probably represent the inactive conformation of the molecule.
By using synthetic peptides and phage display, arrestin residues 109-130 were found to be involved in the interaction with rhodopsin, and indications of additional sites were found , By employing the spectroscopic "extra Mil" assay and overlapping synthetic peptides, three regions of arrestin could be identified which competed with arrestin. These regions display higher affinity than the region previously identified , The respective parts of the sequence are located in both the N- and C-terminal domains of the protein structure, i.e. both caps. Interaction from the side of the receptor appears to involve cytoplasmic loop domains of rhodopsin and a phosphorylation site at the C-terminus . Synthetic peptides corresponding to rhodopsin's cytoplasmic loops competed against R*-arrestin interaction in a co-elution assay; the effect was strongest for a peptide from loop C-III and much weaker for a peptide from loop C-I . Analysis of arrestin binding to rhodopsin mutants suggested an interaction of loops C-I and C-II with arrestin ,
Visual arrestin, and presumably all arrestins, share a conformational switch with G-proteins, operated by contact with the active receptor [170,200, 208-210]. Early indications came from the Arrhenius plots of the arrestin binding reaction monitored via Mil stabilization which yielded a large activation energy, which can be interpreted as a conformational transition in domains of arrestin and/or rhodopsin, linked to the interaction , Limited proteolysis of free and bound arrestin has indeed shown that binding to rhodopsin protects arrestin from the attack of the proteolytic enzyme. This was interpreted as an indication that arrestin bound to phosphorylated rhodopsin and assumes a conformation which is different from that of free arrestin . Heparin  or phytic acid  mimic the effect of photoactivated, phosphorylated rhodopsin to some degree. A highly cationic region beginning with residue 163 was proposed to mediate the interaction with the negatively charged regions of phosphorylated rhodopsin or heparin . Studies on mutated and truncated arrestins [209,212] have confirmed this hypothesis and localized another major binding site for the phosphorylated region of rhodopsin, heparin and phytic acid to the N-terminus of arrestin (residues 1-47). A sequential process was invoked from mutational and synthetic peptide approaches. Based on structural assignments, Sigler and co-workers [198,202,213] have specified a trigger mechanism, in which the phosphorylated C-terminus of the receptor interacts with the "polar core", embedded between the N- and C-terminal domains in the fulcrum of the molecule. Upon this interaction, intramolecular interactions, including a hydrogen-bonded network of buried ion pairs and salt bridges between charged side chains are disrupted, leading to structural changes, possibly involving an en bloc rearrangement of the N- and C-terminal domains . It may seem intriguing that arrestin, as a "blocking" protein, goes through a conformational switch. However, this makes sense because it ensures that no interaction occurs during the amplifying phase of phototransduction. The binding sites identified are distant and do not form a flat surface. A conformational switch may thus be required to allow their simultaneous interaction with the relevant receptor loop structures.
The situation is similar for the G-protein Gt, where two distant sites (the C-termini of the a- and y-subunits, ~45 A apart) are involved in the signal transfer . For arrestin and Gt (see  and , respectively), the simultaneous binding of two receptors at one molecule is unlikely because the titration of the complexes yields 1:1 ratios. A splice variant of arrestin, p44, in which the C-terminal residues 370-404 are replaced by a single alanine, is apparently only present in ROS  and has been reported to inhibit phototransduction in a similar way to the parent protein. However, it interacts not only with phosphorylated, but also with unphosphorylated rhodopsin, although with lower affinity . The lack of the C-terminal sequence also lowers the activation energy of the binding reaction (70 instead of 140 kJ mol-1), and it removes any specificity for the C-terminal structure of the receptor, so that even C-terminally truncated rhodopsin binds to p44. Thus it appears that, by the lack of the C-terminal structure, the conformational switch between active and inactive conformations of arrestin is absent in the splice variant . This fits nicely to a proposal, based on intrinsic fluorescence and circular dichroism data, that the conformational switch involves localized movements of the N- and C-termini of arrestin; these regions may interact in the inactive conformation, and may be separated by interaction with phosphorylated rhodopsin .
Arrestin and Gt appear to use different mechanisms of microscopic (i.e. site to site) recognition. In contrast to the G-protein, where the Gta and farnesylated Gty C-terminal sequences are identified as binding sites, all arrestin domains are intramolecular stretches. Remarkably, the Gta and Gty C-termini, when prepared as synthetic peptides, have the capacity to recognize the Mil species and to distinguish it from the other intermediates. None of the numerous arrestin peptides examined displayed such specificity , Only the parent structure appears to provide the recognition specificity, presumably by nonlinear binding domains on the side of the receptor, which form only in concert with the active arrestin conformation.
Like Gt, monomeric RK (G-protein receptor kinase, GRK1) binds to cytoplasmic loops of R* forming a stable complex with a dissociation constant of 0.5 jiM [196,216] (for review on properties of RK see ). By binding of ATP its affinity for the activated receptor is increased by a factor of ca. 10  allowing phosphorylation of the rhodopsin C-terminus. Depending upon the conditions, up to 7-9 phosphates are transferred in vitro. In vivo, a single phosphate per receptor is sufficient to quench signal transduction [217,218]. Autophosphorylation of RK and phosphorylation of the receptor lowers its affinity to R* and allows dissociation of the R*#RK-complex. Like transgenic mice expressing truncated rhodopsin molecules lacking the C-terminal phosphorylation sites , transgenic mice lacking RK show single-photon responses which are larger and longer lasting than normal . The data suggest that phosphorylation of R* by RK is solely responsible for normal rhodopsin deactivation in the dark-adapted rod.
RK and Gt have several important determinants of their signaling state in common: (i) rhodopsin regenerated with retinal analog ll-cw-9-desmethyl retinal (9-dm rhodopsin) is a poorer substrate for the kinase , as it is a poorer catalyst of Gt activation ; (ii) pH/rate profiles of Gt activation or phosphorylation reflect the protonation of a surface proton acceptor (presumably Glu134) ; (iii) reversal of the residues Glu134/Arg135 (part of the highly conserved ERY motif) in mutant rhodopsin is deleterious to the interaction of Gt or RK with rhodopsin [176,223]; and (iv) in loop mutants of rhodopsin, the lack of loops C-II or C-III abolishs the binding of the enzyme, while mutations in loop C-III affect both binding and catalysis . These results would suggest that the signaling state for both proteins coincides with the Mil photo-product, in agreement with the two-step mechanism of R* formation. Intrigu-ingly, however, the 'extra Mil' assay does not reflect any preference of RK for any of the M states . When Mil formation is measured in the presence of both RK and Gt, the Gt-dependent enhancement effect is reduced, consistent with the notion that the binding of RK destabilizes the MII'Gt-complex by competition. This demonstrates that RK can bind to MI and probably to all metarhodopsin forms. A possible explanation for these observations would be that kinase can bind to all M states of rhodopsin but needs the protonated form, MIIb, to perform the actual phosphorylation step. The idea that formation of MI, the precursor of Mil, provides the trigger for the generation of the kinase substrate  was confirmed by studies on intact retina, in which MI but not Mil was allowed to form by warming up from low temperatures, followed by subsequent photolysis. The resulting photoproduct was a substrate for the kinase. These experiments have also shown that photolysis restores the spectral identity of rhodopsin but not the conformational changes that trigger phosphorylation (in remarkable analogy to the reverted meta product discussed above; ). It is an open question whether constitutively active mutants of rhodopsin, which activate Gt in the absence of the chromophore, can activate RK [225,226].
The mapping of binding sites of RK to active rhodopsin is not yet complete but yielded the involvement of loops C-I, C-II and C-III, as concluded from partial digestion  and studies on mutant rhodopsin [223,228]. When studying the role of loops C-II and C-III, it turned out that loop C-II is involved in binding, whereas C-III has a role in both binding and catalysis .
In vitro data of the kinetics of R**RK-complex formation yielded a K^ of 0.5 raM and a bimolecular rate constant of 1 [iM-1 s_1 . However, these data are of limited value because the necessary detergent solubilization is likely to affect the kinetic properties of RK. Kinetic constants of RK-rhodopsin interaction were derived from the behavior of the photoresponse depending on the number of R* formed per disk membrane . In model calculations, the R*-RK interaction in vivo was described with two subsequent steps, binding and phosphorylation/dissociation. Based on this model, R* binding obeys different kinetics dependent on whether the kinase is substrate saturated or not, yielding reaction times for the binding of RK and the phosphorylation step under in vivo conditions (0.25 and 1 s, respectively ). Such a mechanism would transcend the classical notion of a constant characteristic lifetime of activated rhodopsin , Not only arrestin, but also RK competes with Gt for binding to active rhodopsin . This "pre-arrestin function" has implications in understanding the shut-off mechanism of rhodopsin because, together with the above-mentioned different R* binding kinetics for saturated/ subsaturated RK, it affects the lifetime of R* and thus the photoresponse.
Was this article helpful?