The excessive light illumination of mammalian retina is known to induce oxidative stress and photoreceptor cell death linked to progression of age-related macular degeneration. The photochemical damage of photoreceptors is suggested to occur via two apoptotic pathways that involve either excessive rhodopsin activation or constitutive phototransduction, depending on the light intensity. Both pathways are dramatically activated in the absence of rhodopsin desensitization by GRK1. Previously, we have shown that moderate illumination (halogen lamp, 1,500 lx, 1-5 h) of mammalian eyes provokes disulfide dimerization of recoverin, a calcium-dependent regulator of GRK1. Here, we demonstrate under in vivo conditions that both moderate long-term (metal halide lamp, 2,500 lx, 14 h, rat model) and intense short-term (halogen lamp, 30,000 lx for 3 h, rabbit model) illumination of the mammalian retina are accompanied by accumulation of disulfide dimer of recoverin. Furthermore, in the second case we reveal alternatively oxidized derivatives of the protein, apparently including its monomer with sulfinic group. Histological data indicate that thiol oxidation of recoverin precedes apoptosis of photoreceptors. Both disulfide dimer and oxidized monomer (or oxidation mimicking C39D mutant) of recoverin exhibit lowered α-helical content and thermal stability of their apo-forms, as well as increased Ca2+ affinity. Meanwhile, the oxidized monomer and C39D mutant of recoverin demonstrate impaired ability to bind photoreceptor membranes and regulate GRK1, whereas disulfide dimer exhibits notably improved membrane binding and GRK1 inhibition in absence of Ca2+. The latter effect is expected to slow down rhodopsin desensitization in the light, thereby favoring support of the light-induced oxidative stress, ultimately leading to photoreceptor apoptosis. Overall, the intensity and duration of illumination of the retina affect thiol oxidation of recoverin likely contributing to propagation of the oxidative stress and photoreceptor damage.

Caveolin-1 is a cholesterol-binding scaffold protein, which is localized in detergent-resistant membrane (DRM) rafts and interacts with components of signal transduction systems, including visual cascade. Among these components are neuronal calcium sensors (NCSs), some of which are redox-sensitive proteins that respond to calcium signals by modulating the activity of multiple intracellular targets. Here, we report that the formation of the caveolin-1 complex with recoverin, a photoreceptor NCS serving as the membrane-binding regulator of rhodopsin kinase (GRK1), is a redox-dependent process. Biochemical and biophysical in vitro experiments revealed a two-fold decreased affinity of recoverin to caveolin-1 mutant Y14E mimicking its oxidative stress-induced phosphorylation of the scaffold protein. At the same time, wild-type caveolin-1 demonstrated a 5-10-fold increased affinity to disulfide dimer of recoverin (dRec) or its thiol oxidation mimicking the C39D mutant. The formation of dRec in vitro was not affected by caveolin-1 but was significantly potentiated by zinc, the well-known mediator of redox homeostasis. In the MDCK cell model, oxidative stress indeed triggered Y14 phosphorylation of caveolin-1 and disulfide dimerization of recoverin. Notably, oxidative conditions promoted the accumulation of phosphorylated caveolin-1 in the plasma membrane and the recruitment of recoverin to the same sites. Co-localization of these proteins was preserved upon depletion of intracellular calcium, i.e., under conditions reducing membrane affinity of recoverin but favoring its interaction with caveolin-1. Taken together, these data suggest redox regulation of the signaling complex between recoverin and caveolin-1. During oxidative stress, the high-affinity interaction of thiol-oxidized recoverin with caveolin-1/DRMs may disturb the light-induced translocation of the former within photoreceptors and affect rhodopsin desensitization.

Despite vast knowledge of the molecular mechanisms underlying photochemical damage of photoreceptors, linked to progression of age-related macular degeneration, information on specific protein targets of the light-induced oxidative stress is scarce. Here, we demonstrate that prolonged intense illumination (halogen bulb, 1500 lx, 1-5 h) of mammalian eyes under ex vivo (cow) or in vivo (rabbit) conditions induces disulfide dimerization of recoverin, a Ca(2+)-dependent inhibitor of rhodopsin kinase. Western blotting and mass spectrometry analysis of retinal extracts reveals illumination time-dependent accumulation of disulfide homodimers of recoverin and its higher order disulfide cross-linked species, including a minor fraction of mixed disulfides with intracellular proteins (tubulins, etc.). Meanwhile, monomeric bovine recoverin remains mostly reduced. These effects are accompanied by accumulation of disulfide homodimers of visual arrestin. Histological studies demonstrate that the light-induced oxidation of recoverin and arrestin occurs in intact retina (illumination for 2 h), while illumination for 5 h is associated with damage of the photoreceptor layer. A comparison of ex vivo levels of disulfide homodimers of bovine recoverin with redox dependence of its in vitro thiol-disulfide equilibrium (glutathione redox pair) gives the lowest estimate of redox potential in rod outer segments under illumination from -160 to -155 mV. Chemical crosslinking and dynamic light scattering data demonstrate an increased propensity of disulfide dimer of bovine recoverin to multimerization/aggregation. Overall, the oxidative stress caused by the prolonged intense illumination of retina might affect rhodopsin desensitization via concerted disulfide dimerization of recoverin and arrestin. The developed herein models of eye illumination are useful for studies of the light-induced thiol oxidation of visual proteins.

Recently, we have found that calcium binding proteins of the EF-hand superfamily (i.e., a large family of proteins containing helix-loop-helix calcium binding motif or EF-hand) contain two types of conserved clusters called cluster I ('black' cluster) and cluster II ('grey' cluster), which provide a supporting scaffold for the Ca2+ binding loops and contribute to the hydrophobic core of the EF-hand domains. Cluster I is more conservative and mostly incorporates aromatic amino acids, whereas cluster II includes a mix of aromatic, hydrophobic, and polar amino acids of different sizes. Recoverin is EF-hand Ca2+-binding protein containing two 'black' clusters comprised of F35, F83, Y86 (N-terminal domain) and F106, E169, F172 (C-terminal domain) as well as two 'gray' clusters comprised of F70, Q46, F49 (N-terminal domain) and W156, K119, V122 (C-terminal domain). To understand a role of these residues in structure and function of human recoverin, we sequentially substituted them for alanine and studied the resulting mutants by a set of biophysical methods. Under metal-free conditions, the 'black' clusters mutants (except for F35A and E169A) were characterized by an increase in the α-helical content, whereas the 'gray' cluster mutants (except for K119A) exhibited the opposite behavior. By contrast, in Ca2+-loaded mutants the α-helical content was always elevated. In the absence of calcium, the substitutions only slightly affected multimerization of recoverin regardless of their localization (except for K119A). Meanwhile, in the presence of calcium mutations in N-terminal domain of the protein significantly suppressed this process, indicating that surface properties of Ca2+-bound recoverin are highly affected by N-terminal cluster residues. The substitutions in C-terminal clusters generally reduced thermal stability of recoverin with F172A ('black' cluster) as well as W156A and K119A ('gray' cluster) being the most efficacious in this respect. In contrast, the mutations in the N-terminal clusters caused less pronounced differently directed changes in thermal stability of the protein. The substitutions of F172, W156, and K119 in C-terminal domain of recoverin together with substitution of Q46 in its N-terminal domain provoked significant but diverse changes in free energy associated with Ca2+ binding to the protein: the mutant K119A demonstrated significantly improved calcium binding, whereas F172A and W156A showed decrease in the calcium affinity and Q46A exhibited no ion coordination in one of the Ca2+-binding sites. The most of the N-terminal clusters mutations suppressed membrane binding of recoverin and its inhibitory activity towards rhodopsin kinase (GRK1). Surprisingly, the mutant W156A aberrantly activated rhodopsin phosphorylation regardless of the presence of calcium. Taken together, these data confirm the scaffolding function of several cluster-forming residues and point to their critical role in supporting physiological activity of recoverin.