|Home | About | Journals | Submit | Contact Us | Français|
The vertebrate photoreceptor is the epitome of polarized neurons, containing two specialized compartments—the outer segment and the inner segment, connected by a narrow non-motile cilium. The outer segment of rod and cone photoreceptors is principally dedicated to capturing light and converting the energy of a photon into a change in membrane potential. The primary function of the inner segment is to provide the metabolic and synthetic demands of the photoreceptors. In order to maintain this high degree of specialization, molecules are routinely targeted to their appropriate compartment during protein synthesis. However, in addition to this relatively slow transport process, photoreceptors have a much more rapid process whereby some molecules are rapidly moved between the inner segment and outer segment through the connecting cilium in response to the light adaptational state of the eye. This translocation process has been conclusively demonstrated for two molecules involved in the phototransduction cascade—transducin and arrestin (Broekhuyse et al. 1985; Mangini and Pepperberg 1988; Whelan and McGinnis 1988; Sokolov et al. 2002; Peterson et al. 2003).
Recent studies have shown that the light-driven movement of arrestin is independent of activation of the phototransduction cascade since translocation is normal in transgenic mice that are deficient for transducin (Mendez et al. 2003; Zhang et al. 2003). Various ideas have been proposed regarding the potential function of protein translocation. These hypotheses include a role in the regulation of photoreceptor sensitivity (McGinnis et al. 1991; Sokolov et al. 2002), and a protective function in preventing light damage (Elias et al. 2004).
In this study, we address some of the mechanistic questions regarding arrestin translocation. Specifically we ask whether the light-driven redistribution of arrestin is a consequence of arrestin movement, or whether de novo protein synthesis coupled with arrestin degradation plays a significant role in the apparent redistribution of arrestin. In addition, we utilize a variant of arrestin that is unable to bind to rhodopsin to investigate the role of arrestin binding to light-activated phosphorhodopsin in the translocation of arrestin. Finally, we address whether vertebrate arrestins rely on an affinity for phosphoinositol lipids for translocation.
Transgenic Xenopus expressing a fusion of GFP at the C-terminus of Xenopus arrestin (xAr-GFP) were obtained from breeding pairs of adult transgenic Xenopus as previously described (Peterson et al. 2003) and dark-adapted overnight. Tadpoles were placed in 0.1x tadpole Ringers with or without 100 μM cycloheximide (CHI) for 1 h. Tadpoles then either remained in the dark, or were exposed to laboratory lighting for 45 min or for 240 min. At this point, tadpoles were euthanized in 0.025% benzocaine, one eye removed, the cornea punctured with a scalpel, and placed in 100 μL 1x tadpole Ringers with 20 mM glucose and 70 μCi 35S-labeled methionine and cysteine (Amersham) for 2 h at room temperature. After rinsing, the eye was disrupted by vigorous pipetting and vortexing in 50 μL 1x Laemmli sample buffer (Laemmli 1970), and viscosity reduced by the addition of 25 units of benzonase (Novagen). An aliquot of the extract was separated on 12% SDS-PAGE, the gel stained with Coomassie Brilliant blue, and the dried gel exposed to x-ray film to detect incorporated radiolabeled proteins.
The contralateral eye was fixed in methanolic formaldehyde as described (Peterson et al. 2003), and processed for confocal microscopy to detect the localization of the arrestin/GFP fusion using the endogenous fluorescence of GFP.
The ten amino acid myc tag (EQKLISEEDL) was incorporated into Xenopus arrestin using overlapping PCR products that introduced the cDNA for the myc-tag between the codons for Leu-76 and Thr-77. The overlapping products were combined and the complete cDNA amplified using primers against the 5′ and 3′ ends of the arrestin cDNA, incorporating XhoI and NotI sites, respectively. This product was cloned into the XhoI and NotI sites of the XOPS1.3 vector under the control of the 1.3 kb Xenopus rod opsin promoter (Tam et al. 2000). The cDNA for GFP was inserted immediately prior to the arrestin stop codon at an introduced NheI site. Transgenic animals expressing this myc-tagged xAr-GFP (xAr-myc-GFP) were prepared by nuclear transplantation (Kroll and Amaya 1996).
For in vitro studies, a His(6) tag was incorporated by PCR at the 5′ end of the cDNA immediately after the initiating ATG for xAr-GFP and xAr-myc-GFP, and then cloned into the shuttle vector pPIC-ZA for heterologous expression in Pichia pastoris (Dinculescu et al. 2002). Expressed proteins were purified over nickel-agarose (Ni-NTA, Qiagen), followed by heparin agarose. In vitro binding assays with rhodopsin were performed using unphosphorylated and phosphorylated rod disc membranes prepared from Rana catesbeiana retinas as described for bovine disc membranes (McDowell 1993).
Multiple alignment of Drosophila, bovine, and Xenopus arrestins was performed using ClustalX (Thompson et al. 1997). Two of the three lysines (Lys-228 and Lys-231) identified as important for promoting an association with inositol phospholipids in Drosophila (Lee et al. 2003) are conserved in bovine arrestin and were changed to alanines by site-directed mutagenesis. Because sequence conservation is not perfect, two additional lysines were substituted with alanines to insure disruption of the potential inositol phospholipid binding site, creating the four lysine mutant K232A/K235A/K236A/K238A (4K→A) in bovine visual arrestin. The 4K→A bovine arrestin mutant was expressed in Pichia, purified to homogeneity, and tested for binding to inositol phospholipids immobilized on nitrocellulose strips (PIP Strips, Echelon, Inc.). PIP-Strips were blocked for 60 min with 1% gamma globulin-free horse serum in phosphate-buffered saline (PBS). Wild-type arrestin, 4K→A arrestin, and the PH domain of phospholipase C δ1 fused to glutathione-S-transferase (GST-GRIP; Echelon, Inc) (0.5 μg/mL) were then incubated with the PIP strips for 4 h. After washing with 0.05% Tween-20 in PBS, the blots were immunoprobed to detect binding using anti-arrestin monoclonal SCT-128 for bovine arrestin and anti-GST monoclonal for the GRIP positive control. Binding of the primary antibody was detected using an anti-mouse antibody conjugated to alkaline phosphatase with nitroblue tetrazolium/5-bromo-4-chloro-3-indoyl phosphate as the substrate.
Relative affinity of bovine arrestin and 4K→A arrestin for phytic acid was assessed by competitive elution from heparin agarose. Arrestin and 4K→A (1 mg) was immobilized on a 1 mL heparin sepharose column (Amersham) in 10 mM HEPES/30 mM NaCl, pH 7.0. Aliquots of phytic acid (1 μM – 10 mM) in the same mobile phase were added to the column and the amount of arrestin eluted in each aliquot measured by absorbance at 278 nm.
Homologous mutations were also created in Xenopus arrestin by PCR site-directed mutagenesis, substituting lysines 232, 235, 236, and 267 with glutamine (4K→Q). This substituted arrestin was fused with GFP as previously described and used to create transgenic Xenopus tadpoles. Translocation of the Ar(4K→Q)-GFP was assessed by confocal microscopy using tadpoles that were dark-adapted for 3 days, then exposed to laboratory lighting for 50 min, or for 240 min.
Our previously published work clearly demonstrates that arrestin translocates from the inner segments (RIS) of dark-adapted rod photoreceptors, moving to the rod outer segments (ROS) upon exposure to light (Peterson et al. 2003). In these studies, it also appears that arrestin leaves the rod outer segments and returns to the inner segments in response to either dark adaptation or extended light adaptation. However, another equally tenable explanation is that the arrestin that translocates to the outer segments is proteolyzed and that the arrestin subsequently seen in the inner segment is a result of newly synthesized arrestin (Azarian et al. 1995).
To discriminate between these two mechanisms, transgenic Ar-GFP tadpoles were treated with cycloheximide (CHI) to inhibit protein synthesis (Obrig et al. 1971), and translocation subsequently assessed. Figure 1 shows total homogenates prepared from one eye from each of these tadpoles, separated on 12% SDS-PAGE. Protein staining of total homogenates prepared from eyes treated with 100 μM CHI revealed a similar profile as the untreated animals, although the total protein content extracted was slightly less in the treated animals (Figure 1A). Autoradiography of this same gel indicates that there was substantial incorporation of the radiolabeled cysteine and methionine in the untreated animals, but that new protein synthesis was largely blocked in the tadpoles treated with cycloheximide (Figure 1B).
The contralateral eyes from the same tadpoles used in Figure 1 were fixed in formaldehyde and processed for confocal microscopy to show the distribution of the Ar-GFP fusion protein, using the endogenous fluorescence of GFP (Figure 2). In both the untreated and CHI-treated tadpoles, the Ar-GFP localizes to the RIS and axonemes in dark-adapted animals. In response to 45 min of light exposure, the Ar-GFP almost completely translocates to the outer segments. If the cycloheximide-treated tadpoles are kept in light for 4 h or returned to the dark for 1 h (data not shown), the Ar-GFP again concentrates in the RIS and axonemes for both treated and untreated tadpoles. Because we have demonstrated that treatment of the tadpoles with CHI blocks new protein synthesis (Figure 1), we conclude that arrestin translocates both to and from outer segments, and that at least the vast majority of the redistribution of arrestin is a consequence of translocation and not proteolysis followed by replacement by newly synthesized protein.
These results are consistent with quantitative studies showing that the mass of arrestin-GFP is conserved during translocation from the RIS to the ROS in response to light (Peet et al. 2004). This study did not quantify the translocation from ROS to RIS during dark adaptation. A similar study using cycloheximide to inhibit protein synthesis in mouse retinas shows the light-driven redistribution of transducin is also largely a consequence of transducin translocation and not synthesis of new protein (Sokolov et al. 2002).
When the first observations were made showing that arrestin redistributes to the outer segments during light adaptation, it was assumed that this translocation was a consequence of arrestin binding to light-activated phosphorylated rhodopsin (Wilden et al. 1986; Mangini et al. 1994). However, in mice that are deficient for rhodopsin phosphorylation, either lacking rhodopsin kinase or in which the C-terminus of rhodopsin has been mutated to remove the phosphorylation sites, light-driven translocation of arrestin between the inner segment and outer segments occurs in a manner indistinguishable from that in wild-type mice (Mendez et al. 2003; Zhang et al. 2003). To corroborate and extend these findings, we analyzed the translocation of arrestin in transgenic Xenopus that expressed an arrestin that is unable to bind light-activated phosphorhodopsin (R*P). In previous studies, we demonstrated that binding of arrestin to R*P could be blocked by inserting a ten amino acid myc tag (EQKLISEEDL) in a loop of bovine arrestin between Leu-77 and Ser-78 (Dinculescu et al. 2002). Reasoning that this variant of arrestin could be used to address the necessity of arrestin binding to R*P for translocation, we created a homologous substitution in our Xenopus arrestin-GFP fusion, inserting the myc tag between Leu-76 and Thr-77 (xAr-myc-GFP).
To determine if this insertion in Xenopus arrestin has a similar effect as that documented for bovine arrestin, both xAr-GFP and xAr-myc-GFP were heterologously expressed, purified, and used in a centrifugation binding assay with rhodopsin in ROS disk membranes obtained from bullfrog retinas. As previously documented, xAr-GFP retains its selectivity for light activated phosphorhodopsin (Figure 3). In contrast, xAr-myc-GFP has essentially no binding to the rhodopsin-containing disc membranes for any of the four forms of rhodopsin.
In transgenic tadpoles expressing this same transgene, xAr-myc-GFP translocation is indistinguishable from xAr-GFP (Figure 4). In these animals, the xAr-myc-GFP clearly translocates to the ROS in response to light in a manner that is qualitatively and quantitatively identical to xAr-GFP (compare to untreated animals in Figure 2). Clearly the translocation of arrestin is independent of binding to rhodopsin.
Translocation of arrestin in Drosophila photoreceptors has been reported to be dependent upon an association with inositol phospholipids, particularly phosphoinositol trisphosphate (PIP3) (Lee et al. 2003). Mutation of three lysines in Drosophila arrestin (Lys-228, Lys-231, and Lys-257) disrupts this association and significantly slows the translocation of arrestin from the cell body to the rhabdomere in response to light through a NINAC myosin III-dependent mechanism (Lee and Montell 2004). Accordingly, we investigated if a similar mechanism was involved in the translocation of vertebrate arrestin, particularly since bovine arrestin has a demonstrated affinity for inositol phosphates (Palczewski et al. 1991). Alignment of bovine and Drosophila arrestins shows that two of the three mutated lysines are conserved. From an earlier study we had prepared a mutant of arrestin that coincidentally substituted these two lysines with alanines along with the two nearby lysines (K232A/K235A/K236A/K238A, used herein as 4K→A). Although the sequences of Drosophila and bovine arrestins are not perfectly conserved, the placement of the amino acids in three-dimensional projections based on the crystal structure of bovine arrestin (Hirsch et al. 1999) allows us to assign conformational homology. In addition, there are no other lysine residues that project near this grouping of lysines, giving us confidence that we have identified any lysines that might potentially bind inositol phospholipids in a similar location in vertebrate arrestins as identified in Drosophila arrestin.
Utilizing the procedure used in Drosophila (Lee et al. 2003), we assessed the binding of wild-type arrestin and 4K→A arrestin to different types of phospholipids spotted on nitrocellulose strips (Figure 5A). Unlike Drosophila arrestin, which showed significant binding to many forms of phosphoinositol bisphosphate and trisphosphate in this assay, we were unable to detect any binding of wild-type bovine arrestin or 4K→A arrestin to the phospholipid spots. Positive controls using the PH domain of phospholipase C-δ1 (GST-GRIP) or arrestin from house flies (Musca domestica, not shown), show easily detectable binding. These results suggest that the affinity of bovine arrestin for inositol phospholipids is significantly lower than that of Drosophila arrestin. This conclusion is supported by previous studies, showing an IC50 of 10 μM for phytic acid (inositol hexaphosphate) for bovine arrestin (Palczewski et al. 1991), compared to 0.6 μM for Drosophila arrestin (Lee et al. 2003).
To assess whether the four lysine substitutions in bovine arrestin affected its affinity for phytic acid, WT and 4K→A arrestins were immobilized on heparin agarose, and competed off with increasing concentrations of phytic acid. Figure 5B shows that the 4K→A arrestin required a 50-fold lower concentration of phytic acid to elute the bound protein. This result indicates that the lysine substitutions had the predicted effect of reducing the affinity for inositol phospholipids and that we had correctly identified at least some of the lysines that participate in the association of arrestin with inositol hexaphosphate. These substitutions did not affect the specificity of 4K→A arrestin for R*P (data not shown).
Using these data, we prepared homologous substitutions of lysines in Xenopus arrestin to assess the impact of this reduced affinity for inositol phospholipids on arrestin translocation in vivo. Lysines 232, 235, 236, and 267 were substituted with glutamine (4K→Q), the same substitution used in Drosophila arrestin to disrupt translocation of Drosophila arrestin. In tadpoles expressing 4K→Q arrestin fused to the N-terminus of GFP, the translocation is indistinguishable from that of wild-type arrestin/GFP (compare to untreated tadpoles in Figure 2). In the dark-adapted state, 4K→Q arrestin localizes to the inner segments and axonemes of the rod photoreceptors. In response to 50 minutes of light adaptation, 4K→Q arrestin translocates to the outer segments qualitatively and quantitatively equal to wild-type arrestin. Like the native arrestin, 4K→Q arrestin also returns to the inner segments and axonemes during extended light adaptation, or if the tadpoles are subsequently dark adapted (data not shown). The similarity of these translocation results and the lower affinity of vertebrate arrestins for inositol phospholipids lead us to conclude that inositol phospholipids do not appear to play a role in the translocation of vertebrate arrestins, at least not through a direct interaction with arrestin as has been demonstrated for Drosophila.
These studies on selected aspects of arrestin translocation reveal three important findings. First, the light-driven redistribution of arrestin is a consequence of arrestin translocation and not a result of arrestin proteolysis with subsequent replacement by newly synthesized protein. Second, our results further substantiate that the translocation of arrestin to the ROS following light is not a result of arrestin’s affinity for photoactivated, phosphorylated rhodopsin. And finally, we show that unlike Drosophila arrestin, vertebrate visual arrestins do not appear to rely on an affinity for phospholipids to promote translocation. Future studies will focus on identifying the mechanism of arrestin translocation and identify how arrestin interacts with this translocation machinery.
This research was supported by grants from the National Eye Institute (EY08571 and EY06225), the Howard Hughes Medical Institute, and Research to Prevent Blindness.