Differential Modification of Phosducin Protein in Degenerating rd1 Retina Is Associated with Constitutively Active Ca2+/Calmodulin Kinase II in Rod Outer Segments *

Retinitis pigmentosa comprises a heterogeneous group of incurable progressive blinding diseases with unknown pathogenic mechanisms. The retinal degeneration 1 (rd1) mouse is a retinitis pigmentosa model that carries a mutation in a rod photoreceptor-specific phosphodiesterase gene, leading to rapid degeneration of these cells. Elucidation of the molecular differences between rd1 and healthy retinae is crucial for explaining this degeneration and could assist in suggesting novel therapies. Here we used high resolution proteomics to compare the proteomes of the rd1 mouse retina and its congenic, wild-type counterpart at postnatal day 11 when photoreceptor death is profound. Over 3000 protein spots were consistently resolved by two-dimensional gel electrophoresis and subjected to a rigorous filtering procedure involving computer-based spot analyses. Five proteins were accepted as being differentially expressed in the rd1 model and subsequently identified by mass spectrometry. The difference in one such protein, phosducin, related to an altered modification pattern in the rd1 retina rather than to changed expression levels. Additional experiments showed phosducin in healthy retinae to be highly phosphorylated in the dark- but not in the light-adapted phase. In contrast, rd1 phosducin was highly phosphorylated irrespective of light status, indicating a dysfunctional rd1 light/dark response. The increased rd1 phosducin phosphorylation coincided with increased activation of calcium/calmodulin-activated protein kinase II, which is known to utilize phosducin as a substrate. Given the increased rod calcium levels present in the rd1 mutation, calcium-evoked overactivation of this kinase may be an early and long sought for step in events leading to photoreceptor degeneration in the rd1 mouse.

Photoreceptor degeneration resulting from genetic mutation or age is a major cause of progressive vision loss in the western world. Therapeutic prevention of this process, however, is hindered as the molecular pathomechanisms of degeneration are currently not well defined. To aid in their elucidation, several animal models for photoreceptor degeneration exist, including the retinal degeneration 1 (rd1) 1 mouse, which carries a non-sense mutation in the gene coding for the ␤ subunit of the rod photoreceptor-specific cGMP phosphodiesterase 6 (PDE6-␤) (1,2). Because mutations of the same gene have been linked to some forms of the human disease retinitis pigmentosa (3), the rd1 model is a relevant tool for studying various aspects of human retinal degeneration.
As a consequence of the rd1 mutation, PDE6-␤ expression in rods yields a nonfunctional protein leading to cGMP accumulation in the cytoplasm of rod outer segments (4). Normal regulation of rod cGMP-gated cation channels occurs through cGMP fluctuations generated by the phototransduction cascade (5,6), and the cGMP accumulation found in rd1 rods is therefore a likely cause for the abnormally high Ca 2ϩ levels detected in the rd1 retina (7,8). Increased intracellular Ca 2ϩ per se has been correlated with rod cell death (8,9), and it is reasonable that uncontrolled Ca 2ϩ influx into rd1 rods triggers apoptosis, which characterizes the photoreceptor degeneration in this model (10 -12). In accordance, Ca 2ϩ channel blockers have been shown to rescue rd1 photoreceptors (13,14). However, although a consensus on the role of Ca 2ϩ as initiator of degeneration in the rd1 retina may exist, the cellular steps subsequent to this remain unresolved, leading for instance to conflicting views on whether caspase cascades underlie the execution of apoptosis. Whereas some studies have demonstrated caspase activation in rd1 retinae from the 2nd to 3rd postnatal week, i.e. at peak or postpeak of photoreceptor cell apoptosis (15)(16)(17), Doonan et al. (18) were unable to detect increased caspase expression or activation at these and earlier time points in the same tissue, and caspase inhibition or ablation of caspase-3 failed to rescue rd1 photoreceptors from apoptosis (19,20). It has also been shown that rd1 photoreceptors degenerate irrespective of the expression of p53, a tumor suppressor gene involved in several types of apoptosis (21), or c-Fos (22), which by contrast is entirely essential for light-induced photoreceptor degeneration (23). With respect to antiapoptotic components, only minimal improvement of rd1 photoreceptor survival is observed by overexpressing survival-promoting genes, such as bcl-2 (24), which prevents degeneration in other neuronal systems (25). Similarly we have observed an overactivation of the antiapoptotic Akt-kinase pathway in rd1 rods (26), but these cells degenerate nevertheless.
The unresolved issues and conflicting views on rd1 photoreceptor degeneration illustrate the necessity for clarifying the critical differences between the rd1 mutation and its congenic, wild-type (WT) counterpart. Such differences have been described at the transcriptome level using microarray techniques covering from ϳ600 to 12,000 genes and expression sequence tags (27)(28)(29). However, alterations in mRNA expression are not necessarily reflected in the altered expression of corresponding proteins or vice versa. It is therefore of interest that potential degeneration-related differences at the protein level between rd1 and WT retinae have been addressed recently by Cavusoglu et al. (30). Their study resolved ϳ250 protein spots by two-dimensional electrophoresis (2DE) and identified these by mass spectrometry, revealing differential levels of crystallin at a late developmental stage when degeneration of photoreceptors is complete. This underscores the potential of proteomic techniques to disclose protein alterations consequent to degeneration. It is likely, therefore, that by increasing the resolving power of the proteomic approach applied at earlier stages of degeneration the detection of subtler differences directly related to or even preceding the death process should be possible.
In the present study we used a high resolution 2DE system consistently able to resolve ϳ3000 protein spots. The separations were coupled with computer-assisted gel analysis to select differentially regulated protein spots for subsequent identification by mass spectrometry. Postnatal day 11 (PN11) was selected as the time point for rd1/WT comparisons because the number of photoreceptor cells is then still stable. Together with the use of whole retina samples, PN11 analysis reduces the risk of false positive protein differences related to the biased contribution of a particular cell type (photorecep-tors) rather than to disease-associated changes. Relevant comparisons are therefore possible because active mechanisms committing cells to die are intense at PN11 (see e.g. Ref. 18). With this paradigm we identified previously unreported changes in the proteome of the young rd1 retina, including a differential post-translational modification of phosducin.

EXPERIMENTAL PROCEDURES
Animals and Tissue Preparation-All animal experiments were approved by the Swedish ethics committee (permits M9-02 and M213-03). Normal (C3Hϩ/ϩ, hereafter referred to as WT) and rd1 mutant (C3H rd1/rd1) mice were obtained from in-house breeding colonies. Mice (PN11) were sacrificed by asphyxiation on dry ice. For 2DE analyses, eyes were generally collected from light-adapted animals. For a subset of analyses and immunostaining studies, eyes were collected both from light-and dark-adapted mice. For these studies, animals from one litter were divided into two equal groups; one group was light-adapted, and the other was dark-adapted for a period of 6 h prior to enucleation of the eyes. Dark-adapted eyes were enucleated under red light and processed in the dark. For immunostaining studies, the eyes were removed and immediately immersed in 4% paraformaldehyde (PFA) in PBS on ice for 4 h and then transferred to 20% sucrose in Sö rensen's phosphate buffer. After embedding, sections (cryostat; 8 m) were collected on chrome alum-gelatinated slides (slides briefly dipped in warm solutions of 0.5% gelatin (Sigma) and 0.05% potassium chrome(III)-sulfate (Merck) in H 2 O and allowed to dry at 60°C prior to use). Tissue sections were stored at Ϫ20°C until used for immunostaining.
Sample Preparation-For preparation of 2DE samples, dissected eyes were immediately immersed in ice-cold dissection buffer (10 mM Tris, 1 mM EDTA, 150 mM NaCl, 1 mM freshly activated Na 3 VO 4 , 50 nM okadaic acid, Complete protease inhibitor mixture, pH 7.5 with HCl). Retinae were dissected on ice and stored at Ϫ80°C until use.
For each experiment, fresh WT and rd1 samples were prepared as follows. Fourteen retinae were pooled and homogenized with a Teflon glass homogenizer (Braun Biotech International) in 2 ml of ice-cold nanopure water containing a protease inhibitor mixture (Roche Applied Science). The homogenate was lyophilized and stored at Ϫ80°C. Prior to 2DE, proteins were solubilized in denaturing lysis buffer (9 M urea, 2 M thiourea, 4% CHAPS, 1% dithioerythritol, 2.5 mM EDTA, and 2.5 mM EGTA) for 4 h at room temperature. Samples were cleared by centrifugation (50,000 ϫ g for 50 min), and protein concentrations were determined by Bradford assay (Bio-Rad). Protein resolution was done by loading 150 and 500 g of total proteins from each sample onto analytical and preparative two-dimensional gels, respectively Preparation of 2DE samples from dark-and light-adapted eyes involved enucleation, immediate freezing, and storage at Ϫ80°C until further use. The complete eyes were then homogenized and processed as described above.
Image Analysis-An experiment comprised six gels each from WT and rd1 retinae processed in parallel. In each experiment, three to four gels from either condition were selected for quality of focusing and scanned on a transmission scanner (Epson GT-9600) with 12 bit/300 dpi resolution, and resultant gel images were then imported to a 2DE analysis software program (Proteom Weaver, release 2.1.; Definiens, Germany). The following parameters for protein spot detection were used: minimum spot radius of 4, minimum spot intensity (volume above base level) of 2000, and minimum contrast (height above base level) of 10. Gels from each experiment were processed by the pair-match-based normalization, which erases intensity differences of similar spots in different gels not due to regulation but experimental variability of the method (e.g. protein load or silver stain intensity). Subsequently respective WT and rd1 protein spots were matched and filtered to find significant differences within the detection limit as follows. Only spots matched in at least two-thirds of the gel images were considered, and only these filtered spots exceeding an intensity threshold of 0.1 were taken for further analysis. The remaining spots were sorted according to a regulation factor (quotient of rd1 and WT intensities), and for each experimental set, the threshold regulation factor for the significance level p Ͻ 0.05 was determined by the Proteom Weaver software. Only those spots regulated more than the factor required for significance were further considered as candidate spots and subsequently subjected to manual verification for matching accuracy to avoid assigning false positives. The entire experiment was performed in triplicate.
In-gel Proteolysis and Identification by Mass Spectrometry-Regulated spots identified by image analysis were selected for identification and either excised from experimental 2DE gels (150-g protein load) or the respective spots were excised from preparative gels (500-g protein load). Spots were washed for 30 min in 100 l of nanopure water, destained (33), and dehydrated in 100 l of 40% acetonitrile (3 ϫ 15 min). Samples were subjected to tryptic proteolysis in 5-10 l of 1 mM Tris-HCl, pH 7.5, containing 0.01 g/l trypsin (sequencing grade modified trypsin, Promega) overnight at 37°C.
MALDI-TOF peptide mass fingerprints were obtained on a Bruker Reflex III mass spectrometer (Bruker Daltonics, Bremen, Germany) as described previously (31). Peptide sequence information was obtained by LC-coupled MS/MS analysis on a Q-TOF2 system (Micromass) coupled with a CapLC system (Micromass) as described previously (31).
Database Searching-Database searches were performed using the Mascot software (34) at the following parameter settings: one miscleavage allowed, search restricted to database entries from metazoa; MALDI-TOF, 100 ppm mass accuracy; Q-TOF, 0.8-Da peptide tolerances; 0.2-Da MS/MS tolerance. Peptide masses of the tryptic digests were compared with the virtually generated tryptic peptide masses of the National Center for Biotechnology Information non-redundant (NCBInr) protein database and the Mass Spectrometry Protein Sequence Database (MSDB).
Western Blotting-Protein patterns from small 2DE gels comprising all experimental conditions (WT, dark and light; rd1, dark and light) were blotted semidry onto one PVDF membrane. Unspecific binding was blocked with 5% BSA in TBS-T (50 mM Tris, pH 7.4, 150 mM NaCl, 2 mM EDTA, and 0.1% Tween 20) for 1 h and incubated in primary antibody (rabbit anti-phosducin, 1:1000, a kind gift from M. Castro, University of Wü rzburg, Wü rzburg, Germany) at 4°C overnight and then followed by the horseradish peroxidase-coupled secondary antibody (anti-rabbit IgG, Jackson Laboratories, 1:15,000). Signal was developed by ECLϩ kit (Amersham Biosciences) according to the manufacturer's instructions and detected on Hyperfilm (Amersham Biosciences).
Terminal Deoxynucleotidyltransferase-mediated dUTP Nick End Labeling (TUNEL) Staining-Sections as above were washed 4 ϫ 5 min in PBS and incubated with 10% goat serum. TUNEL staining for apoptotic nuclei was done using an in situ cell death detection kit (Roche Diagnostics) conjugated with FITC.
Hematoxylin-Eosin Staining-For general light microscopic analysis, tissue sections of paraffin-embedded, PFA-fixed retinae were stained with hematoxylin-eosin (HE) according to standard protocols.

Retinal Morphology Is Unaltered in rd1 Retina at PN11, but
Photoreceptor Apoptosis Is Abundant-The rd1 mutation presents with a very early and rapid photoreceptor degeneration (10 -12). To define the postnatal time point at which diseaseinduced apoptosis is most abundant while tissue morphology is still unaltered, sections from rd1 and WT retinae at PN7-13 were stained with HE and TUNEL. We found that although PN11 represents a developmental stage involving abundant photoreceptor apoptosis in the rd1 outer nuclear layer (ONL) ( Fig. 1, C and D), the cellular integrity and thickness of this structure was comparable to WT (Fig. 1, A and B). Thus PN11 offers the advantage of reduced bias risk in comparisons of protein expression patterns. Conversely the high incidence of disease-related apoptosis provides an opportunity to detect differential expression of important proteins. Consequently PN11 retinae from rd1 and WT mice were subjected to 2DEbased comparisons to identify proteins potentially involved in early processes of retinal degeneration.
Differentially Expressed Proteins at PN 11: 2DE Analysis of Wild-type Versus rd1 Retina-An average of 3348 Ϯ 352 protein spots were detected on each gel separation. One typical 2DE separation is shown in Fig. 2. Statistically significant changes within one experiment were filtered with the 2DE analysis software as described in detail under "Experimental Procedures." To eliminate false positive candidates, experimental variability was controlled on several levels as follows. (a) All gels from one experiment were processed in parallel. (b) Gel-to-gel variations due to experimental procedure were eliminated by pair-match-based normalization mode using an algorithm in the 2DE image analysis software. (c) The statistically significant regulation factor threshold was determined separately for each experimental set of 2DE gels (p Ͻ 0.05). (d) rd1 and WT retina comparisons at PN11 were performed in triplicate.
Because of this high stringency, only five of all possible detected proteins remained as candidates for differential expression in rd1 to WT comparisons ( Fig. 3 and Table I). Two proteins in the rd1 retina, COP9 subunit 8 and 14-3-3 , were up-regulated by 2.6 and 1.9-fold, respectively, as compared with WT. Conversely two proteins, ␤-adaptin and guanylate kinase, were down-regulated by 0.4 and 0.5-fold, respectively. The fifth candidate protein, phosducin, was found to be both up-and down-regulated in rd1 retina. Phosducin resolved on the 2DE gels as a group of spots (labeled 1-7 in Fig.  4A), along the acidic-to-basic axis (approximate pI, 4.32-4.60), with a concomitant but limited alteration in apparent molecular weight. According to mass spectrometric identifications, five of these spots (spots 2-6) contained the same protein, phosducin (Table I), suggesting post-translational modifications of the protein. The remaining two spots (spots 1 and 7) were identified by immunoblotting with a phosducinspecific antibody (see below). In the rd1 condition, an upregulation of the acidic isoforms was found together with a down-regulation of the basic isoforms (ϳ50% reduction; see regulation factor list, Fig. 4A). Total retinal phosducin for rd1 and WT on 2DE gels was compared by obtaining the cumulative average intensities for all phosducin spots. No significant difference in total phosducin was found between rd1 and WT (Fig. 4B), indicating that detected differences were indeed related to basic-to-acidic shifts of the molecule instead of alterations in expression levels. This was further supported by phosducin immunofluorescence of WT and rd1 where no immediate differences in staining patterns could be detected (see Fig. 5, A and B) as well as on one-dimensional Western blot experiments (data not shown).
The rd1 phenotype is caused by a mutation at the PDE6-␤ subunit resulting in a significant decrease of both mRNA level and PDE6-␤ activity (35,36). We would therefore expect differential levels in PDE6-␤ protein expression between rd1 and WT. Indeed PDE6-␤ protein levels were high on onedimensional Western blots of WT retina lysates but below detection levels in rd1 samples (data not shown). However, the protein could not be detected on 2DE blots of either rd1 or WT retina (not shown); this likely explains why PDE6-␤ failed to appear among the significantly regulated spots. This corroborates our previous experimental observations 2 in which PDE was excluded from 2DE gels but could be unequivocally detected by two-dimensional electrophoresis with cationic detergent benzyldimethyl-n-hexadecylammonium chloride  each added phosphate molecule leads to an increment in acidity of 0.04 pI units. The phosducin isoforms observed for rd1 retina were shifted toward the acidic side, which would be compatible with increased phosducin phosphorylation in the mutated state. Because phosducin is a well known retinal phosphoprotein whose increased phosphorylation has been linked to dim light adaptation in healthy phenotypes (37,38), the question arose whether differences between rd1 and WT phosducin phosphorylation were related to light status. Hence WT and rd1 eyes at PN11 sampled during the light phase were compared with eyes collected in prolonged dark phase. Protein samples were resolved on 2DE gels, blotted, and probed for phosducin. Fig. 4C shows that immunoreactions could be separated into two categories. The full set of isoforms, including the acidic hyperphosphorylated forms of phosducin, was observed for WT only during the dark phase, whereas during the light phase WT displayed a clear reduction in the number of phosducin spots (lower and upper left panel, respectively), indicating reduced phosphorylation under light exposure. The rd1 retina, however, showed acidic, hyperphosphorylated forms of phosducin in both light and dark phases. When compared with WT retina, the phosphorylation states of rd1 phosducin in both light and dark phases were identical to the WT hyperphosphorylation state in darkness. Differential Phosducin Phosphorylation Is Correlated with Differential Phosphorylation of CaMKII-Because of the differences in phosducin phosphorylation between rd1 and WT at the light-adapted state, we hypothesized that a kinase FIG. 4. Analysis of differentially modified phosducin. A, phosducin resolved on 2DE gels as a row of distinct spots (1-7, acidic to basic, respectively). The more acidic isoforms (1, 2, 3, and 4) were consistently up-regulated in the rd1 genotype. The most acidic phosducin isoform (equivalent to spot 1) was below the detection limit of silver stain in WT. The basic isoforms of phosducin (5, 6, and 7) were down-regulated in rd1 (mean regulation factor from three independent experiments is indicated). B, absolute intensities from spots 1-7 were added (WT, open bars; rd1, filled bars) within different experiments, and the cumulative phosducin intensity was compared between WT and rd1. No differences were found in phosducin levels between the genotypes. C, total eye extracts from light-adapted and dark-adapted WT and rd1 mice were resolved by 2DE, blotted, and probed with anti-phosducin antibody. WT dark-adapted eyes and rd1 dark-and light-adapted eyes showed similar phosducin spot separation patterns with seven different isoforms. In contrast, WT light-adapted eyes showed a phosducin separation pattern having only three isoforms. Blot image overlay revealed that these WT isoforms were the basic-most phosducin spots. responsible for the differential phosphorylation of phosducin may be differentially activated. Three distinct kinases are known to phosphorylate phosducin. Protein kinase A (PKA) and G protein-coupled receptor kinase-2 (GRK-2) have both been shown to contribute to phosducin modification on one phosphorylation site each (39,40). In contrast, CaMKII was able to phosphorylate on five separate amino acids (Ser-6, Ser-36, Ser-54, Ser-73, and Ser-106) during in vitro experiments (41). Because we detected seven phosducin isoforms in the 2DE experiments that were all different between rd1 and WT (Fig. 4A), CaMKII is a likely candidate for contributing to the observed modification. Although CaMKII may not phos-phorylate phosducin at more than the Ser-54 site, when analyzed under in vivo conditions (42), the phosphorylation of this site is particularly important for the interactions between phosducin and other retinal proteins, notably the 14-3-3 group of proteins (of which 14-3-3 was found to be upregulated here) and the photoreceptor-specific G-protein transducin (37,38,(41)(42)(43). In light of this we immunostained retinal sections to assess the presence of possible differential CaMKII expression and/or activation in rd1 and WT.
Because CaMKII is dependent on Ca 2ϩ and CaM for activation (44), we stained rd1 and WT PN11 retinae for CaM. Similar staining patterns were found for both genotypes (Fig.   FIG. 5. Immunofluorescent stainings  of WT (A, C, E, and G) and rd1 (B, D, F,  and H) retina on PN11. Murine PN11 retina sections were stained with either anti-phosducin antibody (A and B), anticalmodulin antibody (C and D), anti-total CaMKII antibody (E and F), or anti-pCaMKII antibody (G and H). Phosducin (A and B) was expressed exclusively in the photoreceptor layer. No differences were found between WT and rd1 in agreement with the unchanged total expression of total phosducin obtained by 2DE. Both calmodulin (C and D) and CaMKII (E and F) were expressed predominantly in inner retinal layers. CaMKII staining was also seen in the outer retina, including in the photoreceptor segment portion (see magnified insets in E and F). For both CaM and CaMKII no differences between WT and rd1 retinae could be detected. Similarly no staining differences between WT and rd1 were seen for pCaMKII at the inner retina (G and H). However, many more photoreceptor segments showed clear immunostaining for the phosphorylated protein in rd1 (H) compared with WT retinae (G). See Fig. 6 for further depiction of this phenomenon. OPL, outer plexiform layer; IS/OS, inner and outer segments of photoreceptors. Large boxed insets represent magnifications of selected areas of the photoreceptor segments indicated by the smaller boxes. Scale bar is 100 m except for the magnified insets in E and F where it equals 50 m. 5, C and D) that correlated well with that for adult mouse retina reported by Pochet et al. (45). Marked cellular staining was found in the outer and inner part of the inner nuclear layer (INL) as well as in some radial processes running through the INL. Cells in the ganglion cell layer (GCL) were also labeled, and punctate staining was observed in the inner plexiform layer (IPL), reminiscent of synaptic structures. There was only faint CaM immunoreactivity in structures related to photoreceptors, i.e. the ONL or the inner and outer segments.
In subsequent experiments, antibodies recognizing total CaMKII were used (Fig. 5, E and F). We found kinase expression within GCL and INL cell groups of the PN11 retina in a pattern generally compatible with previous reports in adult mouse retina (46). Total CaMKII staining was also observed in the outer retina, including in the photoreceptor segments (see Fig. 5, E and F, insets). Furthermore no staining differences were observed between rd1 and WT retinae in agreement with our lack of differential CaMKII protein detection on 2DE comparisons. However, alterations in enzyme activity may not necessarily reflect differences in its expression pattern. To address the question of differential CaMKII activation despite similar retinal expression patterns in the two genotypes, we used antibodies specifically recognizing CaMKII phosphorylated at Thr-286. Addition of a phosphate group at this amino acid occurs by autophosphorylation following stimulation of the enzyme by Ca 2ϩ and CaM, thus activating the kinase (44). Thus, the phosphorylation of CaMKII at Thr-286 marks the active state of this kinase (see e.g. Ref. 47). Fig. 5, G and H, shows that staining with an anti-Thr-286 pCaMKII antibody produced similar immunoreactive patterns for rd1 and WT in IPL and outer plexiform layer as well as cytoplasmic staining in certain cells of the innermost INL and GCL. These results generally agree with those of Liu et al. (46) for the adult mouse retina, and both the IPL and outer plexiform layer staining would be compatible with the suggested presence and function for CaMKII in synapses (48). However, a higher optical resolution suggested a significant difference in pCaMKII staining in photoreceptors (not clearly resolved in Fig. 5, G and H) whereby fewer pCaMKII-positive OS were found for WT than rd1. The decreased number and the spacing suggested that pCaMKII-positive OS in WT could be allocated to cones. Consequently co-stainings with PNA were done to discriminate cone segments (49), and the analyses were extended to include sections of dark-adapted WT and rd1 retinae analyzed FIG. 6. Confocal images of immunofluorescence detection of phosphorylated CaMKII (green) and PNA labeling (red). Light-adapted (A and B) and dark-adapted (C and D) retinae from WT (A and C) and rd1 (B and D) at PN11 were stained with antibody against pCaMKII (green fluorescence) and labeled with fluorophore-coupled PNA to reveal the position of cone outer segments (red fluorescence). In the WT light-adapted phase, pCaMKII labeling was restricted to PNA-positive structures (A), whereas in all other cases pCaMKII was clearly also located in photoreceptor segments between the PNA-labeled structures, i. e. rod segments (B, C, and D). The differential staining between rd1 and WT in the light-adapted phase was seen in all examined retinae (at least five independent animals of each genotype) as were the similarities in the dark-adapted state (four eyes from two independent animals of each genotype). The retinal pigment epithelium is located at the superior aspect of each panel. Scale bar, 10 m. by confocal microscopy. The resulting higher resolution revealed that the pattern of pCaMKII activity visualized by pCaMKII staining strictly correlates with that in phosducin blotting experiments of rd1 and WT retinae under light/dark conditions (see Fig. 4C). Again the phosphorylation states of CaMKII at Thr-286, indicating active kinase in rd1 rod photoreceptors in both light and dark phases, were equal to the WT hyperphosphorylation state in darkness (Fig. 6, B-D). WT light phase specimens showed a restricted number of positive structures at regularly spaced intervals in the photoreceptor segment layer co-labeled with PNA, suggesting that CaMKII in this situation was active only in cone segments, particularly at their very distal outer ends (Fig. 6A). Upon dark adaptation most, if not all, of the remaining photoreceptor segments were positive for activated CaMKII, including those not labeled by PNA, and also here staining was observed at the distal most aspects of the segments (Fig. 6C). The WT structures that were immunopositive for activated CaMKII in the dark greatly outnumbered the PNA co-labeled cones. Furthermore these stained structures were located closer to the retinal pigment epithelium face compared with PNA-co-labeled cones, clearly indicating the activation of CaMKII in rod outer segments (Fig.  6C). Together these findings suggested that CaMKII was active in WT rods solely in the dark. This situation was clearly different from that of the rd1 retina in which activated CaMKII was seen in both rods and cones of both light and dark phases (Fig. 6, B and D). The pCaMKII staining thus suggests that the calcium-dependent kinase CaMKII that could be responsible for phosducin hyperphosphorylation is constitutively activated in rd1 rods irrespective of the presence of light. It further suggests that, in yet unresolved molecular terms, the rd1 rods operate as rods do in a normal retina in the dark-adapted state. The above stainings were performed with the Thr-286 pCaMKII antibody from Cell Signaling Technology. Similar staining patterns could be produced with a different antibody (Thr-286 pCaMKII antibody, Promega; data not shown).

DISCUSSION
Five functionally unrelated proteins, COP9 subunit 8 (Csn8), 14-3-3 , phosducin, ␤-adaptin, and guanylate kinase, were unambiguously differentially expressed in the rd1 retina at PN11. Of these, phosducin is generally regarded as confined to retinal photoreceptors (50), i.e. the cells directly affected in the rd1 model. There is at present no detailed information on the retinal expression of Csn8 and ␤-adaptin. Guanylate kinase and 14-3-3 are also expressed in retinal cells other than photoreceptors. 3 We found an up-regulation of Csn8, an evolutionary conserved integral part of the COP9 signalosome complex. COP9 is a regulatory component of the ubiquitin-proteasome path-way for regulated protein degradation (51,52). Although the presence and function of COP9 in mouse retina have not been studied, its involvement in Drosophila eye development has been suggested (53). An association between Csn8 and retinal degeneration remains unknown. We also found the upregulation of another protein in rd1, 14-3-3 , which is abundantly expressed in retina (50,54) and which has been demonstrated to bind phosducin upon phosphorylation by CaMKII (41).
Two proteins were down-regulated in rd1: ␤-adaptin and guanylate kinase. ␤-Adaptin is part of the AP2 coat assembly protein complex (55) involved in clathrin-mediated endocytosis of receptors (56,57). In other tissues, ␤-adaptin combined with non-visual arrestins takes part in internalization of G protein-coupled receptors (GPCRs) (58) and thus contributes to GPCR desensitization and GPCR-induced signaling. However, the function of retinal ␤-adaptin is currently unknown, and although its down-regulation in the rd1 retina may indicate an impairment of a yet undefined interaction with rhodopsin, any relation to disease mechanisms must remain speculative. The other down-regulated protein identified in this study was guanylate kinase (GK), the first enzyme in the metabolic pathway that generates cGMP. GK phosphorylates 5Ј-GMP to GDP, which subsequently is phosphorylated by nucleoside-diphosphate kinase to GTP. GTP is then transformed into cGMP by calcium-dependent guanylate cyclase. Thus, GK is crucial for maintaining a certain cGMP level in the retina, e.g. photoreceptors (59), a prerequisite facilitating Ca 2ϩ influx through the cGMP-gated cation channel. As mentioned, the rd1 mutation renders PDE6-␤ inactive, so GK down-regulation may reflect a feedback response from continually elevated cGMP levels.
Phosducin is exclusively expressed in photoreceptors, making this protein a particularly interesting candidate for further investigation especially because the detected regulation on 2DE analysis depended upon a post-translational modification of the protein. Phosducin is a bona fide retinal phosphoprotein (39,41), and both our 2DE results and Western blot data were compatible with increased phosducin phosphorylation in the rd1 retina. Although phosphorylation of phosducin was thought to occur primarily by a cAMP-dependent protein kinase (PKA) and to lesser extent by GRK-2, phosphorylating phosducin at one position each (39,40), recent results indicate that phosducin is additionally phosphorylated by CaMKII (41,42). All seven phosducin spots detected here by 2DE were quantitatively different between rd1 and WT (Fig. 4A), indicating that although PKA (and GRK-2) may have been involved in altering one (or two) of the phosphorylation variants, other kinases, including calciumdependent kinases, could also have contributed to the differential phosphorylation. The localization of activated CaMKII to the outer segment of the rd1 rod photoreceptors, where relevant interactions with phosducin are likely to occur, makes this kinase an attractive candidate for at least some of the altered phosducin phosphorylation.
In photoreceptors, phosducin binds to the ␤-␥ subunits of transducin to facilitate their translocation from outer to inner segments (43). The binding requires a dephosphorylated state of phosducin, which in turn depends on the light status. In the dark, high photoreceptor Ca 2ϩ levels (60) are accompanied by high phosducin phosphorylation, and with light adaptation phosducin becomes dephosphorylated as Ca 2ϩ decreases (61,62). Thus, in the light state dephosphorylated phosducin binds transducin ␤-␥ with high efficiency and helps to move the latter to the inner segment of the photoreceptors (37,38,43). As a consequence, the outer segment signal flow from rhodopsin to cGMP-PDE utilizing transducin is dampened, and the sensitivity of light-adapted rods is therefore reduced (43).
For the WT situation, our studies of dark-adapted retinae demonstrated increased phosducin phosphorylation as well as CaMKII activation in rod photoreceptor segments when compared with the light phase. This corroborates the view that CaMKII participates in phosducin phosphorylation and that the latter is counteracted by light (41,42). However, in the rd1 retina, the light-dependent regulation of CaMKII activation as well as phosducin phosphorylation was found to be impaired. There is consensus on increased Ca 2ϩ levels in rd1 photoreceptors (7,8,13,14), and high Ca 2ϩ has likely contributed to the constitutive activation of CaMKII, which thus in turn could have contributed to the increased phosducin phosphorylation in rd1 independently of light conditions. However, it should be noted that an involvement of other Ca 2ϩ -dependent kinases cannot be ruled out in this context. Does increased phosducin phosphorylation by e.g. CaMKII promote or participate in rd1 photoreceptor degeneration, or is it solely consequential to events leading to photoreceptor death? It seems unlikely that continuous phosducin phosphorylation on its own drives rod degeneration as retinal damage should then be expected from light deprivation alone in WT mice. However, although experiments demonstrate the importance of light in the maturation of synaptic function in inner retina, no indications exist for light deprivation-induced degenerative processes in normal animals (63). It is noteworthy that CaMKII expression can be influenced by light exposure. Increased CaMKII mRNA and protein levels have been found in the retina of dark-reared PN12 rats (64), suggesting that elevated CaMKII in dark-reared animals phosphorylates the GluR1 subunit of the ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid glutamate receptors and in consequence affects synaptic maturation as already demonstrated for CaMKII in brain (65,66). Although total CaMKII protein levels were unchanged in our studies, the observed strong difference in rod phospho-CaMKII immunostaining between lightand dark-adapted WT PN11 retinae confirms a link between CaMKII activity and light/dark status. The disease-induced and constitutive CaMKII activation may therefore affect the normal synaptic development in the rd1 retina.
CaMKII activity requires a Ca 2ϩ -binding protein (CaBP) like CaM for activation (44), but the presence of CaM in photoreceptors has been difficult to assess. Pochet et al. (45) reported lack of immunoreactivity in mouse photoreceptors, which fits well with the very weak CaM signal in ONL and inner and outer segments in this study. On the other hand, CaM was detected in feline and bovine OS by ultrastructural immunostaining and biochemical methods, respectively (67,68). Furthermore using in situ hybridization, Kovacs and Gulya (69) observed low levels of CaM mRNA in the rat retina at the myoid (inner segment) portion of the photoreceptors where protein synthesis occurs (70). Other CaBPs may augment or even substitute for CaM function (71). Several of these are able to stimulate CaMKII (71), and at least one, CaBP4, is expressed by photoreceptors (72). The presence of CaBPs may therefore compensate for the relative paucity of CaM in photoreceptors.
Although CaMKII activity in light-adapted rd1 rods seems to resemble the dark-adapted WT state, the consequences of continually elevated CaMKII activity combined with constitutively raised Ca 2ϩ levels are likely to create a distinct pathological phenotype in the rd1 mouse that differs from darkadapted WT photoreceptors: Ca 2ϩ modulation plays a crucial function in the development of the neuronal connectivity of the visual system and a regulatory role within mature photoreceptors in the conversion of the light signal received by photoreceptors into an electrical signal transmitted to the brain (73). Recent studies on guanylate cyclase (GC)-activating proteins (GCAP1-3), specific Ca 2ϩ -sensitive regulators of retinal GC (for a review, see Ref. 74), have correlated these regulators and receptors for Ca 2ϩ homeostasis to retinal pathological states, including genotypes associated with autosomal dominant cone-rod dystrophy (75,76). Diseaselinked GCAP1 mutations thus far identified are inherited in a dominant fashion, leading to augmented cGMP levels in the cytoplasm of rod and cone outer segments through increased GC activity, which results in constitutively higher intracellular Ca 2ϩ levels in photoreceptors as the cGMP-gated channels remain open (77). As an upstream component of the same molecular network regulating intracellular Ca 2ϩ , a stop mutation within the ␤ subunit of PDE6 leads to retinal degeneration in the rd1 mouse. PDE6 removes cGMP, the product of guanylate cyclase activity, and can therefore be regarded as an indirect antagonist to GC. In the absence of functional PDE, photoreceptors suffer from a constitutive Ca 2ϩ overload (7,8). Consequently mutations in GCAP1 and PDE6 likely cause a similar pathological phenotype with respect to Ca 2ϩ homeostasis that differs from the normal, dark-adapted state and that eventually leads to degeneration. Therefore and as suggested by for instance Frasson et al. (13) and Takano et al. (14), Ca 2ϩ blockers may represent rational therapeutic agents for such forms of retinal degenerations. Within this context CaMKII activity as well as the differential phosphorylation of phosducin can be regarded as molecular markers of the de-generating rd1 mouse retina. Future mechanistic and comparative studies in other models for genetically inherited retinal diseases are required to determine whether CaMKII activity represents a distinct surrogate marker for retinal degenerations or whether it is causatively linked to specific forms of this disease.