11-cis-Retinal Reduces Constitutive Opsin Phosphorylation and Improves Quantum Catch in Retinoid-deficient Mouse Rod Photoreceptors*

Rpe65 −/− mice produce minimal amounts of 11-cis-retinal, the ligand necessary for the formation of photosensitive visual pigments. Therefore, the apoprotein opsin in these animals has not been exposed to its normal ligand. The Rpe65 −/− mice contain less than 0.1% of wild type levels of rhodopsin. Mass spectrometric analysis of opsin from Rpe65 −/− mice revealed unusually high levels of phosphorylation in dark-adapted mice but no other structural alterations. Single flash and flicker electroretinograms (ERGs) from 1-month-old animals showed trace rod function but no cone response. B-wave kinetics of the single-flash ERG are comparable with those of dark-adapted wild type mice containing a full compliment of rhodopsin. Application (intraperitoneal injection) of 11-cis-retinal to Rpe65 −/−mice increased the rod ERG signal, increased levels of rhodopsin, and decreased opsin phosphorylation. Therefore, exogenous 11-cis-retinal improves photoreceptor function by regenerating rhodopsin and removes constitutive opsin phosphorylation. Our results indicate that opsin, which has not been exposed to 11-cis-retinal, does not generate the activity generally associated with the bleached apoprotein.

RPE65 is a major protein in the retinal pigment epithelium (RPE) 1 (1) and has also been identified in cone photoreceptors (2,3). The photoreceptors of Rpe65 Ϫ/Ϫ mice are almost completely depleted of 11-cis-retinal, the native ligand of cone and rod opsins, resulting in minimal levels of rhodopsin and the deterioration of their photosensitivity (4). Photoreceptor function has been shown to be partially restored by supplying exogenous ligand (5,6). Thus, the Rpe65 Ϫ/Ϫ mouse is an excellent model in which to study the two key factors that control the activity of rhodopsin, the supply of the ligand (7,8) and the level of rhodopsin/opsin phosphorylation. Therefore, we have analyzed opsin phosphorylation levels of Rpe65 Ϫ/Ϫ mice and correlated those levels with electroretinogram (ERG) responses and rhodopsin levels in the absence and presence of exogenous 11-cis-retinal.
Retinal function is controlled by the availability of 11-cisretinal, which can be affected by the absence or dysfunction of any number of participants in the retinal metabolic pathway (e.g. Refs. 9 and 10). RPE65 is essential for production of 11-cis-retinoids (4). RPE65 mutations in humans result in congenital retinal dystrophies ranging from night blindness to loss of vision (11). Although night blindness would suggest loss of rod rather than cone function, the ERG retained in Rpe65 Ϫ/Ϫ mice appears to originate from rod photoreceptors (12).
A second key factor that controls visual sensitivity is opsin/ rhodopsin phosphorylation. The C terminus of activated rhodopsin is multiply phosphorylated by rhodopsin kinase (G protein-coupled receptor kinase 1) (13) in vivo (14,15), and this multiple phosphorylation has been shown to be necessary for the rapid return of sensitivity (16). In moderate light levels, it has been proposed that opsin dephosphorylation and its regeneration to rhodopsin occur simultaneously, leaving no free opsin that could potentially activate the signal transduction cascade (15).
The goal of our study was to elucidate the consequences of the absence of 11-cis-retinal, the native ligand, on opsin structure and function and to determine the effect on these parameters of restoring the native ligand. We demonstrate that opsin phosphorylation is closely correlated with photoreceptor sensitivity, as measured by ERG recordings. 11-cis-Retinal injections restore sensitivity of the rods by increasing quantum catch (regeneration of rhodopsin), which was paralleled by opsin dephosphorylation. Restoration of cone function was not observed. Our results demonstrate that opsin that has never been exposed to its native ligand ("virgin opsin") does not activate the transduction cascade, in contrast to opsin formed by the detachment of all-trans retinal from bleached pigment ("bleached opsin").

EXPERIMENTAL PROCEDURES
Animals-Rpe65 Ϫ/Ϫ mice (1 month old) were generated and genotyped as described previously (4,17). Age-matched C57Bl/6 mice were purchased from Jackson Laboratory. The analysis was restricted to 1-month-old animals, since no structural alterations have been observed in the inner retina at this age. 2 For dark adaptation, animals were kept in complete darkness overnight, and light-sensitive procedures were performed under dim red light. Animals were injected intraperitoneally with 0.05 (1ϫ) or 0.25 g/g body weight (5ϫ) 11-cis-retinal (1 mg/ml in 10% ethanol, 10% bovine serum albumin, and 0.9% NaCl) or vehicle and were kept in the dark. All experiments were performed in accordance with the Policy on the Use of Animals in Neuroscience Research and were approved by the University Animal Care and Use Committee.
Mass Spectrometry-Rod outer segments of dark-adapted mice (13 Rpe65 Ϫ/Ϫ or 8 wild type (WT)) were isolated, digested, and analyzed by mass spectrometry (MS) as described previously (18). For phosphorylation measurements, retinas (1-4) were homogenized in 8 mol/liter urea and digested with Asp-N (50 ng/200 l; Sigma) in Tris buffer (10 mmol/liter, pH 7.5). Supernatants were collected by centrifugation (120,000 ϫ g) and analyzed online with a LCQ mass spectrometer (14,19). The MS data were acquired by repetitive scanning and corrected with a factor for the decreased detection efficiency of the C-terminal phosphopeptides, as measured from synthesized standards (the peak heights of phosphopeptides were multiplied by 1.15). For determination of the sites of phosphorylation, MS/MS data were acquired for the mouse rhodopsin ASP-N-cleaved phosphopeptides, and the fragmentation patterns were qualitatively analyzed for b and y ions, characteristic of the specific phosphorylation sites.
Rhodopsin Measurements-Endogenous levels of rhodopsin were determined by extraction of a homogenate of retinas (n ϭ 4) in 1% dodecylmaltoside and measurements of difference spectra (Spectro-Pette, WPI). The total rhodopsin that could be regenerated (rho max ) was determined by incubation with excess 11-cis-retinal (12 h, 4°C). Difference spectra were fitted with a nomogram with peak absorbance at 504 nm.
ERG Analysis-Full field ERGs were recorded from the left eyes as described previously with minor changes (20). The setup was modified to include a dual channel optical bench for light stimulation, providing stimulus and background light. The optical pathways consisted of a single 250-watt halogen lamp, mechanical shutters, manually operated neutral density, and a 500-nm bandpass filter, a filter cube to combine the two pathways, and lenses to focus the light beam to the end of the light guide. Light intensity per 10 ms provided in the stimulus path could be varied in steps of 0.3 log units from 3.0 ϫ 10 5 to 3.0 ϫ 10 11 photons/mm 2 at 510 nm or 1.9 ϫ 10 7 to 1.9 ϫ 10 13 photons/mm 2 using white light conditions, whereas the rod adaptation beam provided a continuous flux of 5.5 ϫ 10 8 photons/mm 2 /s at 510 nm (corresponding to a single flash intensity of 5.5 ϫ 10 10 photons/mm 2 , an intensity at which the rod ERG is no longer a pure rod ERG). Electroretinograms were recorded by two protocols: 1) 10-ms flashes of increasing light intensities under scotopic and photopic conditions, and 2) 10-Hz flicker ERG under photopic conditions. Peak a-wave amplitude was measured from base line to the initial negative-going voltage, whereas peak b-wave amplitude was measured from the trough of the a-wave to the peak of the positive b-wave. Flicker amplitudes were measured from the preceding trough to the peak of the flicker response. b-Wave kinetics were obtained from the time constant of a single exponential function fitted to the leading edge of the scotopic b-wave responses (starting at midopening of the shutter) at threshold (ϳ35 V in amplitude), which is an appropriate alternative to the kinetic model developed by Robson and Frishman (21) in the absence of a pure photoreceptor and bipolar cell response. The time constant provides an indication of the adaptational state of the underlying rod photoreceptor response (22). ERG recordings were stored, displayed, and analyzed with a PC interface and PClamp (Axonlab) and Origin software. ERG recordings were performed on 16 WT, 11 Rpe65 Ϫ/Ϫ , and 8 11-cis-injected Rpe65 Ϫ/Ϫ mice (four at each dose). In previous experiments (20,23), such group sizes were found appropriate to generate statistically reliable results. Data are expressed as means Ϯ S.E. and analyzed by the Student's t test. whereas WT levels were consistently measured at 0.35 Ϯ 0.027 nmol/eye. With a detection limit of 0.225 pmol of rhodopsin, this would set the endogenous amount of rhodopsin in Rpe65 Ϫ/Ϫ mice at less than 0.1% of WT levels.

Rhodopsin Levels and Opsin
Opsins ( Fig. 1) from WT and Rpe65 Ϫ/Ϫ mice were mapped using MS (14). No differences were found between the opsins (Table I). Phosphorylation levels were analyzed by digestion of whole retinas with the Asp-N enzyme, which cleaves the opsin C terminus (Fig. 1), and the measurement of the soluble fraction by MS. One eye of the dark-adapted mice was covered, and the fellow eye was exposed for ERG analysis. Only background level rhodopsin phosphorylation was detected in the darkadapted WT eye ( Fig. 2A), whereas opsin from the light-exposed eye showed substantial, multiple (single to triple) phosphorylation (Fig. 2B). This phosphorylation pattern was similar to that of mice exposed to continuous bright light (data not shown). The Rpe65 Ϫ/Ϫ opsin, independent of exposure to light, was monophosphorylated ( Fig. 2, C and D). The tandem mass spectral data showed complex mixtures of several singly phosphorylated peptides, as indicated by the presence of collision-induced dissociation ions characteristic of phosphorylation at specific sites. The presence of both the phosphorylated b ϩ 14 (m/z ϭ 1460) and unphosphorylated b ϩ 13 (m/z ϭ 1293) ions indicated that the site of phosphorylation was Ser 343 . Similarly, the presence of the phosphorylated y ϩ 15 (m/z ϭ 1528) together with the unphosphorylated y ϩ 14 or y ϩ 13 (m/z ϭ 1361 or 1290, respectively) ions implied that Ser 334 as the site of phosphorylation. Phosphorylation at one of these two sites was always present in the peptides obtained from the Rpe65 Ϫ/Ϫ opsin. However, phosphorylations at other sites, confirmed by their respective ion pairs (e.g. m/z ϭ 1031 (y ϩ 10 ) and 1198 (y ϩ 11 ) for Ser 338 and m/z ϭ 1192 (b ϩ 12 ) and 1373 (b ϩ 13 ) for Thr 342 ), were also detected. Our data are supportive of Ser 334 and Ser 343 being sites of phosphorylation, but more experiments are needed with the appropriate reference peptides to quantitate the distribution of phosphorylation between these and the other four possible sites. At this point, we cannot comment on any possible changes in the phosphorylation patterns due to the different treatment conditions.
Retinal Function-The effect of the absence of 11-cis-retinal (i.e. Rpe65 Ϫ/Ϫ mice) on visual function was analyzed in vivo using ERGs. Single, scotopic flash ERGs (to test presumed rod function) and photopic single flash and flicker ERGs (to test presumed cone function) were employed. Rod function of Rpe65 Ϫ/Ϫ mice was severely impaired, with b-wave thresholds elevated by ϳ5 log units (Fig. 3, A and B). At maximum light intensity, WT rod a-waves reached 437 Ϯ 91 V in amplitude (Table II), whereas in the Rpe65 Ϫ/Ϫ mice, rod a-waves could not be elicited under our recording conditions (Fig. 3B, Table II). b-Waves, on the other hand, reached less than 10% of the WT amplitudes (i.e. 60 Ϯ 13 V, as compared with 913 Ϯ 113 V for WT) (Fig. 3, A and B; Table II). b-Wave time constants (Fig. 4, A and B) were slightly but insignificantly shorter in Rpe65 Ϫ/Ϫ (38.14 Ϯ 2.9 ms) in comparison with WT mice (39.14 Ϯ 3.2 ms; p Ͻ 0.9).
In human patients with putative Rpe65 null mutations (5), light sensitivities are similar in dark-and light-adapted conditions, arguing that the same photoreceptor system is operational. We used photopic ERGs to determine whether light adaptation changes light sensitivity in Rpe65 Ϫ/Ϫ mice. Consistent with the human data, we observed that in Rpe65 Ϫ/Ϫ mice the b-wave thresholds for single flash scotopic and photopic ERGs are identical at ϳ1.9 ϫ 10 11 photons/mm 2 (data not shown). In Rpe65 Ϫ/Ϫ mice, the overall waveform of photopic flicker ERGs resembles WT rod rather than cone flicker ERGs (Fig. 5A), which is in accordance with a previous report (12). The major difference between the WT rod and the Rpe65 Ϫ/Ϫ flicker ERG is the implicit time. The Rpe65 Ϫ/Ϫ responses peak faster, yet the rise and decay follow similar kinetics as the WT responses (Fig. 5A).
Dose-dependent Increases of Retinal Responses by 11-cis-Retinal-The previous observation that exogenous retinal improves photoreceptor function in Rpe65 Ϫ/Ϫ mice (5, 6) was confirmed and extended by using 11-cis-retinal (the native ligand). Rpe65 Ϫ/Ϫ mice were injected intraperitoneally with either vehicle alone or one of two concentrations of 11-cisretinal 24 h before the ERG analysis: 0.05 g (1ϫ) and 0.25 g (5ϫ) of 11-cis-retinal/g of body weight, respectively. Vehicle-injected mice showed no changes in their ERGs, but mice receiving 11-cis-retinal had ERGs consisting of both aand b-waves (Fig. 3C). At the 1ϫ dose, maximal rod a-wave amplitudes recovered to 3.8 Ϯ 0.5% of the age-matched WT levels, whereas maximal rod b-wave amplitudes reached 30 Ϯ 2.4%. The 5ϫ dose yielded an a-wave recovery of 11.9 Ϯ 2.6% and a b-wave recovery of 38.9 Ϯ 3.5% (Table II). The b-wave responses of the 11-cis-retinal-injected Rpe65 Ϫ/Ϫ mice had significantly shorter time constants (24.15 Ϯ 3.8 ms; Fig. 4, A and B) when compared with either uninjected Rpe65 Ϫ/Ϫ (38.29 Ϯ 1.46 ms) or WT mice (39.14 Ϯ 3.2 ms; p Ͻ 0.05 for all comparisons). Shorter b-wave time constants in rod ERGs are an indication of light adaptation (22). In support of the hypothesis that 11-cis-retinal-treated rods behave like light-adapted rods is the observation that the rising phase of the b-wave of the 11-cis-retinal-treated mice exhibits multiple small oscillations (Fig. 3C), which in WT mice are only present in light-adapted rod ERGs or photopic ERGs (cone ERGs).
Since RPE65 has been identified in mouse cones (2,3), it was of interest to determine whether cone function could be restored by supplying the native ligand. 11-cis-Retinal improved the photopic threshold and shifted the maximal flicker ampli-tude in a dose-dependent manner by a total of 1.3 log units from 7.7 ϫ 10 12 photons/mm 2 (vehicle injected) to 1.9 ϫ 10 12 (1ϫ) and 3.1 ϫ 10 11 photons/mm 2 (5ϫ), respectively (Fig. 5B). Although the overall waveform of the flicker ERG (Fig. 5A) was unaffected, the rise and decay kinetics appear to be altered. The lack of a cone-specific waveform in the photopic flicker ERG suggests that the exogenous 11-cis-retinal did not benefit the remaining cones (3) in the Rpe65 Ϫ/Ϫ retina and that we were recording the activity of light-adapted rods. Alternatively, the recording conditions may be inappropriate to elicit cone responses in the presence of highly desensitized rods.
Effects of 11-cis-Retinal on Rhodopsin Levels and Opsin Phosphorylation-Dose-dependent rhodopsin regeneration with 11-cis-retinal was observed, the 1ϫ dose resulting in 1.67 Ϯ 0.77% of rho max and the 5ϫ dose yielding 8.8 Ϯ 0.81% (Fig. 6). At the 5ϫ dose, ϳ8.6 ϫ 10 18 molecules were injected intraperitoneally, resulting in the formation of only ϳ1.4 ϫ 10 13 molecules of functional rhodopsin per retina, demonstrating that this process is relatively inefficient. The opsin phosphorylation was affected in a dose-dependent manner by the administration of 11-cis-retinal. At 24 h, opsin phosphorylation was significantly decreased in both the light-exposed and the dark-adapted retinas with no obvious change in sites of phosphorylation (24 Ϯ 2% (vehicle injected) to 13 Ϯ 3% (1ϫ) and 8 Ϯ 2% (5ϫ), respectively) (Fig. 6).

DISCUSSION
Opsin Structure in Rpe65 Ϫ/Ϫ Mice-The opsins from WT and Rpe65 Ϫ/Ϫ mice were found to be identical with the exception of phosphorylation. Therefore, the possibility of truncation or unusual post-translational modifications of opsin from a mouse deprived of its native ligand affecting the rate or level of rhodopsin regeneration and visual sensitivity can be excluded. Although it was reported previously that the opsin from the Rpe65 Ϫ/Ϫ mouse did not regenerate properly in vitro (4), no such problems were encountered here, a discrepancy that can be attributed to the use of frozen retinas with inadequate protease protection in the original experiments.
C-terminal opsin phosphorylation in dark-adapted Rpe65 Ϫ/Ϫ mice was 7-fold higher than in opsin from WT mice, resulting in one-fifth of the Rpe65 Ϫ/Ϫ opsin protein being monophosphorylated. This confirms and expands upon the previous qualitative results of Van Hooser et al. (6), who reported that dark-adapted Rpe65 Ϫ/Ϫ rod outer segments are immunopositive for phosphorylated opsin. Constitutive phosphorylation has also been reported qualitatively in an opsin mutant, where the ligand binding site was removed with a point mutation (24). These data support our conclusion that opsin that does not have its native ligand available, due to a dysfunctional retinoid metabolism (e.g. Rpe65 Ϫ/Ϫ mouse) or an inability to bind the ligand (e.g. the K296E mutant), can be constitutively phosphorylated.
ERG Analysis in Rpe65 Ϫ/Ϫ Mice-Previously published ERG studies in Rpe65 Ϫ/Ϫ mice have confirmed that the light-evoked signals are generated by rods (12) and not, as previously as- , whereas if the Rpe65 Ϫ/Ϫ mice were treated with 11-cis-retinal, both a-and b-waves could be recorded (C). Light intensities in A ranged from 1.6 ϫ 10 6 to 3.1 ϫ 10 11 photons/mm 2 in steps of 0.3 log units unless otherwise noted. Light intensities in B and C ranged from 3.1 ϫ 10 10 to 3.1 ϫ 10 11 and 7.7 ϫ 10 9 to 3.1 ϫ 10 11 photons/mm 2 , respectively. a-wave (D) and b-wave (E) amplitudes of Rpe65 Ϫ/Ϫ mice treated with 1ϫ (circles) and 5ϫ (triangles) retinal increased in a dose-dependent way with treatment; however, larger percentage changes were recorded in the a-waves.

TABLE II
Scotopic electroretinogram amplitudes a-and b-wave amplitudes of 1-month-old WT and Rpe65 Ϫ/Ϫ animals (with and without 11-cis-retinal) were recorded and expressed in both absolute (V) and relative (%) terms. p values are for Rpe65 Ϫ/Ϫ control versus 1ϫ and 1ϫ versus 5ϫ data. sumed, by cones (25). Our experiments using flicker ERGs support this notion. Due to light limitations, no rod-elicited a-waves could be recorded from Rpe65 Ϫ/Ϫ mice in our study. The remaining b-waves were small and had time constants similar to those recorded from WT mice. Administration of 11-cis-retinal to Rpe65 Ϫ/Ϫ mice increased the amount of rhodopsin and thereby improved photoreceptor function and output. In the 11-cis-retinal-treated group, the b-wave kinetics were significantly faster when compared with both the WT and the untreated Rpe65 Ϫ/Ϫ mice. A cone response was still not recorded. Van Hooser et al. (5) reported that the small a-waves that could be recorded in the Rpe65 Ϫ/Ϫ mice under bright light conditions were characterized by low photoreceptor sensitivity with no obvious change in gain of the photoreceptor transduction cascade or temporal characteristics. In a follow-up study using single-cell recordings (6), they demonstrated that Rpe65 Ϫ/Ϫ rods have responses greatly reduced in sensitivity and amplitude, whereas kinetics were similar to, but slightly faster than, WT kinetics. Taken together, the results suggest that rods in the Rpe65 Ϫ/Ϫ mouse retina are dark-adapted. Thus, the slow b-wave kinetics reported here for the Rpe65 Ϫ/Ϫ mice are indicative of dark-adapted retinas.
Using single-cell recordings, Van Hooser et al. (6) went on to demonstrate that rods of Rpe65 Ϫ/Ϫ mice treated orally with small amounts of 9-cis-retinal respond with increased sensitivity but also faster kinetics to single light flashes. The same 9-cis-retinal treatments, on the other hand, did not affect the gain and kinetics of the a-wave (5). Faster photoreceptor response kinetics are a hallmark of light adaptation (26). Thus, the faster kinetics that we observed in our study in the b-wave of 11-cis-retinal-treated mice correlate with the faster photoreceptor responses recorded in the isolated photoreceptors of 9-cis-retinal-treated mice. Together, these results suggest that the retina in the 11-cis-retinal-injected Rpe65 Ϫ/Ϫ mice is lightadapted. This suggestion is supported by experiments demonstrating faster b-wave kinetics in light-adapted rat rod ERGs (22) and isolated bipolar cell response (PII) in the light-adapted cat eyecup preparation. 3 In addition, Hood and Birch (27) demonstrated that b-wave implicit times (another measure of bwave kinetics) can be used as a measure of receptor sensitivity. In summary, we suggest that the observed light adaptation in 3 L. Frishman, personal communication.

FIG. 4. b-wave kinetics of scotopic b-wave responses at threshold.
Time constant of a single exponential function fitted to the leading edge of the scotopic b-wave responses at threshold (ϳ35 V in amplitude) were obtained. Threshold amplitudes were obtained at the following light intensities (in photons/mm 2 ): 1.5 ϫ 10 6 for WT, 7.7 ϫ 10 10 for the untreated Rpe65 Ϫ/Ϫ mice, 1.9 ϫ 10 10 for the 1ϫ, and 7.7 ϫ 10 9 for the 5ϫ 11-cis-retinal-injected Rpe65 Ϫ/Ϫ mice (A). Time constants did not differ between WT and Rpe65 Ϫ/Ϫ mice (p Ͼ 0.5) but shortened significantly upon 11-cis-retinal treatment (p Ͻ 0.05 for all comparisons) (B). Please note that time constants were not significantly different between the 1ϫ and 5ϫ 11-cis-retinal-injected animals, and data were therefore combined for this analysis.
The efficiency of regeneration of pigment achieved in this study using 11-cis-retinal is below that found in earlier experiments using 9-cis-retinal (5, 6). We have not attempted to maximize the retinal incorporation in our studies. There are various differences in the routes of administration, which may well affect the efficiency with which these retinals reach the retina. The affinity of the various binding proteins for the two isomers as well as the ability of these compounds to cross the membranes are among the numerous parameters that need to be assessed. Further investigations are needed to resolve these issues.
Virgin Opsin Does Not Activate Transducin-Visual sensitivity is dependent on both the rhodopsin, which can absorb the stimulating light, and the level of "bleached" opsin, which has been shown to activate the signal transduction cascade (28,29). In the dark-adapted WT photoreceptor, the ligand 11-cis-retinal is covalently bound to the apoprotein opsin, acting as a reverse agonist and locking rhodopsin in its inactive form. Light absorbed by rhodopsin triggers an isomerization of 11cis-to all-trans-retinal, which now acts as an agonist, activating the signal transduction cascade. Signaling is turned off by a complex process, involving rhodopsin phosphorylation on multiple sites by G protein-coupled receptor kinase 1. Alltrans-retinal is subsequently removed, and the free opsin with its empty ligand site has been shown to have weak activity (28 -30). Rhodopsin is unphosphorylated in the dark, and lightinduced phosphorylation regulates the activity of the visual signal transduction cascade. Kennedy et al. (15) demonstrated that light-induced phosphorylation and subsequent dephosphorylation and regeneration of rhodopsin followed a similar time course as loss and recovery of visual sensitivity after a light flash.
With the Rpe65 Ϫ/Ϫ model, we have a very interesting case of opsin that has never been exposed to its ligand. We find that these retinas have minimal levels of rhodopsin, resulting in very small b-wave responses. As previously mentioned, using single-cell recordings, Van Hooser et al. (6) demonstrated that Rpe65 Ϫ/Ϫ rods have responses greatly reduced in sensitivity and amplitude. The surprising result from the data by Van Hooser et al., however, is that the kinetics of the Rpe65 Ϫ/Ϫ rod (see Fig. 4C of Ref. 6) did not resemble those expected of a fully bleached but rather those of a dark-adapted rod. We argue that when recording from a single rod, if the opsin was to act like "bleached" opsin, activating transducin (28), one should see a reduced but very fast response as shown by several laboratories (e.g. Refs. 26, 31, and 32). We suggest therefore that the results in single Rpe65 Ϫ/Ϫ rods observed by van Hooser et al. (6) support our hypothesis that the opsin, which has never been exposed to its native ligand, has the kinetics of the receptor containing its inverse agonist, such as in newly regenerated rhodopsin. If this hypothesis were to be true, then photoreceptors containing opsin not exposed to its ligand (i.e. created by other means such as by vitamin A deficiency or early in development) should also behave like dark-adapted photoreceptors. Data in the literature suggest that to be the case. First, Witkovsky et al. (33) demonstrated that in vitamin A-deprived tadpoles, the free opsin did not produce the expected decrease in sensitivity as measured in dark-adapted eyecup ERGs. Sec-FIG. 7. Activity model for opsin/rhodopsin. Newly synthesized opsin, which has never been exposed to its natural ligand, is termed "virgin opsin" and is inactive with respect to its ability to activate transducin. Upon 11-cis-retinal binding, virgin opsin enters the rhodopsin cycle. Opsin resulting from the photoactivation of rhodopsin, called "bleached opsin," has some weak basal activity, which is removed upon reexposure to its ligand.
FIG. 6. Levels of rhodopsin and opsin phosphorylation are dependent on 11-cis-retinal. Rhodopsin levels (black bars) and opsin phosphorylation data (gray bars) are expressed as the maximal amount. In vivo, an inverse relationship between constitutive opsin phosphorylation and rhodopsin levels in relation to supplied 11-cis-retinal was demonstrated. In vitro, excess 11-cis-retinal completely regenerated all available opsin to rhodopsin and removed phosphorylation to the level detected in vivo after the addition of the 5ϫ dose (p Ͻ 0.05 for all comparisons). ond, this interpretation is supported by data from Ratto et al. (34), who showed in studies examining the appearance of rhodopsin in the rat that at postnatal day 13 these animals appear to have only 10% of their opsin in the form of rhodopsin, most likely due to the slower development of the systems needed for the generation of 11-cis-retinal. By postnatal day 13, rat pups still have their eyes closed, and their photoreceptors are ϳ50fold less sensitive to light than adult photoreceptors, arguing that in vivo, no photoisomerizations will have yet occurred. Therefore, most likely, in this model, 90% of the opsin has never been exposed to the ligand. The single photon responses recorded from dark-adapted neonates containing only 10% rhodopsin were indistinguishable from those of adult animals containing 100% rhodopsin. If the free opsin were contributing to bleaching adaptation, as in light-adapted rods, the single photon response would have been expected to be reduced in amplitude and duration (26).
We propose that the conformation of opsin, which has not previously bound 11-cis-retinal, is different from that of the opsin resulting from the photoactivation of rhodopsin. The latter, termed "bleached opsin," is induced by light exposure of rhodopsin and is known to have some basal activity. This basal activity is removed upon exposure to its ligand. Opsin that has never bound the native ligand is termed "virgin opsin." This opsin is inactive, showing none of the basal activity noted for "bleached" opsin in physiological studies (Fig. 7).
The role of the constitutive phosphorylation observed for this virgin opsin is not understood, nor has the kinase been identified. Agonist-independent phosphorylation has been reported in a number of other G protein-coupled receptors (for a review, see Bohm et al. (35)). In recent studies on the bradykinin B 2 receptor, Blaukat et al. (36) have discriminated between three types of phosphorylation: 1) a constitutive, basal phosphorylation; 2) phosphorylation mediated by agonist binding; and 3) an agonist-independent site that is proposed to "prime" the receptor for desensitization. In this system, G protein-coupled receptor kinases 2-6 are proposed to only be involved in the phosphorylation mediated by agonist binding to the bradykinin receptor. However, G protein-coupled receptor kinase 3 was shown to phosphorylate the ␣ 2 adrenergic receptor in an agonist-independent manner (37), in contrast to its role in the -opioid receptor (38), providing further evidence that the G protein-coupled receptor kinases may have different roles in regulating various receptors.
The mechanism of phosphorylation of opsin in the Rpe65 Ϫ/Ϫ mouse remains to be elucidated. There may be some parallel with the agonist-independent desensitization such as suggested for other G protein-coupled receptors. Protein kinase C is known to phosphorylate rhodopsin (39), but an unequivocal explanation of the role of this kinase in rhodopsin phosphorylation in vivo has not been elucidated. Understanding the level of 20% opsin phosphorylation requires further experiments. It is interesting to note in this context that a similar amount of phosphorylation was observed in the CRALBP Ϫ/Ϫ mouse after 100% bleach and 0% regeneration (15).
Thus, in the Rpe65 Ϫ/Ϫ mouse model of vitamin A deprivation, the opsin is partially phosphorylated and inactive until exposed to its ligand 11-cis-retinal (or other retinal isomers). The slow degeneration of the Rpe65 Ϫ/Ϫ mouse retina (4) could be due to a number of causes. Impairment of RPE transport due to the oil droplets resulting from excess unprocessed retinylesters in the RPE may affect a number of processes. Alternatively, the inactive rod photoreceptors may undergo slow death. Finally, although in the mouse the number of cones is small (ϳ3% (40,41)), the role of RPE65 is not understood in these cells, and the lack of this protein may trigger degeneration, which in turn affects the remainder of the retina.