The Dependence of Retinal Degeneration Caused by the Rhodopsin P23H Mutation on Light Exposure and Vitamin A Deprivation

Abstract

Purpose.: To characterize the influence of light and vitamin A on retinal degeneration in an animal model of retinitis pigmentosa caused by the rhodopsin P23H mutation.

Methods.: Retinal degeneration was examined in transgenic Xenopus laevis expressing P23H rhodopsin, in which retinal degeneration is completely rescued by preventing light exposure. The sensitivity of this retinal degeneration to varying intensities, wavelengths, and durations of light exposure, and to vitamin A deprivation was characterized.

Results.: Green light was the most effective inducer of retinal degeneration in this model. Retinal degeneration was induced by prolonged exposure to green light and was prevented by filters that block short wavelengths. Reducing the duration of light exposure prevented retinal degeneration, even when the light intensity was proportionally increased. Vitamin A deprivation also induced retinal degeneration associated with defects in P23H rhodopsin biosynthesis. Vitamin A deprivation did not cause retinal degeneration in nontransgenic animals.

Conclusions.: The mechanism of retinal degeneration in this animal model of RP involves the interaction of light with rhodopsin rather than with free chromophore or bleached rhodopsin. These results may explain the clinical benefits of vitamin A for patients with retinitis pigmentosa and may indicate that pharmacological chaperones are a viable approach to RP therapy. Results also suggest strategies for minimizing RD in patients through controlling light exposure duration or wavelengths.

Rhodopsin, the photopigment of rod photoreceptors, is composed of an opsin apoprotein covalently linked to a vitamin A-derived chromophore, 11-cis retinal. Photoisomerization of the chromophore bound to rhodopsin is the initial event of the visual cascade. A missense mutation in the rhodopsin gene, P23H, is the most common cause of autosomal dominant retinitis pigmentosa (RP) in North America. ¹ This mutation causes the degeneration of photoreceptors, initially resulting in night blindness, followed by progressive tunnel vision and blindness. Previous studies indicate that this mutation prevents the exit of newly synthesized mutant rhodopsin from the endoplasmic reticulum (ER). ^2–7 As it does in other disorders associated with protein misfolding, prolonged ER stress likely initiates cell death. ⁸  

Several animal models of RP based on P23H rhodopsin exhibit light-sensitive retinal degeneration (RD), among them mouse, ⁹ rat, ¹⁰ Xenopus laevis, ³ and Drosophila melanogaster. ¹¹ For these models, it has been demonstrated that protection from light ameliorates RD or, conversely, that high light levels promote RD. Some evidence for light sensitivity has also been obtained for human patients carrying this mutation because the lower retina, which is exposed to higher light levels, is preferentially affected, ¹² a phenotype referred to as sector RP (recently reviewed by Paskowitz et al. ¹³ ). Other mutations of the N terminus of rhodopsin have also been associated with light-sensitive RD, including T4R, ¹⁴ T4K, ¹⁵ N15S, ¹⁶ and T17M. ¹⁷ In addition, brief exposures to intense white, ¹⁸ blue, ¹⁹ and green ²⁰ light cause RD in mice and rats, though it is unclear whether these mechanisms are related to those underlying sector RP. 

We have previously demonstrated that RD induced by bovine P23H rhodopsin expressed in X. laevis photoreceptors can be completely rescued by dark rearing ³ and that RD caused by human P23H rhodopsin can be partially rescued. Light-sensitive RD is associated with a decreased propensity for P23H rhodopsin to exit the ER under conditions of light exposure. This rescue of RD requires an intact chromophore binding site; however, the precise mechanism of involvement of chromophore is unclear. Chromophore may promote folding of the P23H opsin; therefore, a decreased supply of chromophore attributed to light exposure would inhibit folding. Alternatively, chromophore binding may stabilize folded rhodopsin, with the supply of chromophore again the limiting factor. Yet another possibility is that light-induced isomerization and the resultant dissociation of all-trans retinal from newly synthesized P23H rhodopsin may destabilize the protein within the biosynthetic pathway (Fig. 1). ²¹ Understanding the underlying mechanism would provide insight into the function of pharmacological chaperones and is therefore of importance for the rational design of drugs for disorders affecting protein folding, including RP. 

Figure 1.

 

Hypothesized mechanisms for light sensitivity of RD induced by P23H rhodopsin expression. Model demonstrating possible roles for light and chromophore in the biosynthesis of P23H rhodopsin, beginning with cotranslational insertion of the P23H peptide into the ER, folding, stabilization by chromophore binding, and ending with ER exit of the mature P23H rhodopsin. Light may deplete the supply of chromophore (by conversion from the 11-cis to the all-trans form), thereby preventing stabilization of the P23H opsin. This could proceed either through the free chromophore (A) or chromophore bound to rhodopsin (B). Light could also destabilize newly folded P23H rhodopsin before ER exit (C). Yet another possibility (not illustrated) is that the depletion of free chromophore by light directly prevents folding of rhodopsin. The model also predicts that a deficiency in the chromophore precursor vitamin A would also cause a defect in P23H rhodopsin biosynthesis. All these mechanisms would prevent rhodopsin ER exit, eventually resulting in cell death. The values shown are predicted wavelength sensitivities for RD in the X. laevis bovine P23H animal model and in human patients with retinitis pigmentosa (bovine P23H values were arrived at as described; values for human patients were based on the absorbance of human rhodopsin [496 nm] reported by Fulton et al. ²² ). The red shift in the X. laevis versus mammalian values was attributed to differences in the chromophore 11-cis 3,4-dehydro retinal (vitamin A2) in X. laevis tadpoles vs. 11-cis retinal (vitamin A1) in mammalian retina. ERAD, ER-associated degradation.

Because free chromophore and rhodopsin have significantly different absorbance spectra, with rhodopsin absorbing maximally at 495 to 510 nm (depending on species and chromophore variants) and free chromophore absorbing maximally in the UV range (370–400 nm, depending on chromophore variant), these distinct mechanisms predict different spectral sensitivities for the resultant RD (Fig. 1). Furthermore, the hypothesized underlying mechanism of a defect in rhodopsin biosynthesis makes certain predictions. First, brief exposures to light (even if very intense) should be less damaging than longer exposures; second, a deficiency in the supply of the chromophore precursor vitamin A should result in RD very similar to that induced by light. Therefore, in this study, we have analyzed the sensitivity of the RD caused by P23H rhodopsin to different regimens of light exposure and to vitamin A deprivation using the X. laevis model of RP. 

Methods

Transgenic Animals

F1 offspring of a single transgenic male X. laevis expressing bovine P23H rhodopsin were used for this study. Fifty percent of the offspring of this animal carry the transgene, with no indication of multiple phenotypes (i.e., the results indicate a single integration site). Under dark-rearing conditions, the bovine P23H rhodopsin protein typically comprised 12% of total rhodopsin. Nontransgenic embryos were eliminated by G418 selection, as previously described. ^3,23 This study was conducted in adherence to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Housing and Lighting

For standard cyclic light conditions, animals were housed in transparent plastic bins in a low-temperature incubator (Precision, model 818; Thermo-Electron, Pittsburgh, PA) illuminated with standard fluorescent bulbs, which provided an average light intensity in the incubator of 1700 lux (equivalent to shade or a cloudy day). The lighting was cycled on and off at 12-hour intervals, and the temperature was set at 18°C. For experiments involving neutral density filters, the bins were placed inside bags fashioned from the neutral density material (TAP plastics). For experiments involving light-emitting diode (LED) light sources, the clear tanks were painted opaque white and were illuminated from above using LEDs inserted through holes drilled in the lids. To minimize variation in illumination across the tank, a diffuser plate was inserted between the LEDs and the tadpoles. Each tank was wrapped in aluminum foil to eliminate outside illumination. For experiments involving therapeutic tints, clear polymethyl methacrylate panels were tinted using therapeutic tints (BPI, Miami, FL) and used as tank lids. 

Animals were fed powdered frog brittle (Nasco, Fort Atkinson, WI) for the experiments shown in Figures 1 to 7 or powdered specialty diet (TestDiet 5LP3, containing 0.29 ppm carotene and 9 IU/g vitamin A as retinyl acetate, or 5B8V, based on 5LP3 lacking vitamin A and carotene [Purina Mills, St. Louis, MO]) for the experiments shown in Figure 8. 

Photometry

Light intensities (lux) were recorded using a light meter (Extech, Waltham, MA) and were converted to photon flux using a standard photopic curve. For the experiment depicted in Figure 5, photon flux measurements were taken using a radiometer/photometer (ILT1700; Extech) equipped with a detector (SED033/W; Extech) using a detector response curve supplied by the manufacturer. Because X. laevis eyes are directed downward at this stage of development, readings were always taken with the detector directed toward the bottom of the tank. The radiometer/photometer (ILT1700; Extech) was also used to measure the intensity of the 532-nm laser. 

Spectra

Spectra were recorded using a spectral radiometer (I1 Pro; X-Rite, Grand Rapids, MI). Information on the spectral output of the UV LEDs was supplied by the manufacturer. 

Dot Blot Assay for RD

At 14 days postfertilization (dpf), animals were killed and one eye from each was solubilized and used for a dot blot assay for rod opsin content, as previously described ³ using the antibody B630N. ²⁴ This antibody detects endogenous X. laevis rod opsin and rhodopsin and the full-length form of bovine P23H rhodopsin but not the cleaved form of P23H rhodopsin. Therefore, it should be noted that lack of detection of cleaved P23H (<12% of total opsin/rhodopsin in dark-reared animals, <5% on light exposure) introduces a small inaccuracy in the reported values for total rhodopsin. Nonetheless, the B630N signal provides a useful estimate of the relative opsin/rhodopsin content of retinas and is a good indicator of rod photoreceptor degeneration. For the experiment shown in Figure 8, transgenic and nontransgenic animals were distinguished using a second dot blot analysis with the antibody 1D4, which is specific for mammalian rhodopsin and detects both full-length and cleaved forms. 

Confocal Microscopy

The contralateral eye from each animal was fixed in 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4. Representative samples were cryosectioned and processed for confocal microscopy, as previously described. ^3,25 Sections were labeled with Alexa 488-conjugated wheat germ agglutinin, Hoechst 33342 dye, and anti-mammalian rhodopsin monoclonal antibody 2B2. Antibody 2B2 was detected by a Cy3-conjugated secondary antibody (Jackson Laboratories, Bar Harbor, ME). 

Results

Effect of Light Intensity on RD

We have previously used white light (1700 lux) to induce RD in X. laevis expressing bovine P23H rhodopsin. To define the threshold level of white light necessary to induce RD, we reared animals under several different light intensities from fertilization to 14 dpf. The various intensities were produced by filtering the white light source using neutral density filters. At 14 dpf, RD was assessed using an antibody-based dot blot assay for rod opsin (i.e., rhodopsin + opsin) using the antibody B630N, and the data were used to construct a dose-response curve (Fig. 2A). We have previously shown that loss of rod opsin correlates well with RD. ³ RD was subsequently confirmed by confocal microscopy of contralateral eyes (Fig. 2B). The results showed that 100 lux (94% reduction of light intensity) was insufficient to cause RD, whereas a half-maximal effect occurred at 270 lux or 16% of the standard cyclic light intensity. The maximal light intensity did not cause RD in wild-type animals (not shown) or in animals expressing wild-type bovine rhodopsin. ³  

Figure 2.

 

The dependence of RD caused by bovine P23H rhodopsin on light intensity. (A) Dose-response curve showing the effects of varied light intensity on rod opsin content, used as an indicator for RD. Light intensity was reduced using neutral density filters and is plotted as the value for the “on” period of the cyclic illumination schedule (12 hours of darkness, 12 hours at maximum of 1700 lux). Error bars are ± SEM; n = 14 for each condition. (B) Histology of representative contralateral eyes corresponding to each point in the dose-response curve confirms that the reduction in rod opsin content is caused by rod photoreceptor degeneration. Cryosections were stained with Texas-red conjugated wheat germ agglutinin and imaged by confocal microscopy. Scale bar, 20 μm.

Spectral Sensitivity of Bovine P23H-Induced RD

Our previous studies indicated that light absorption by either rhodopsin or free chromophore is responsible for RD in this model. ³ To verify this, we exposed transgenic tadpoles expressing bovine P23H rhodopsin to various wavelengths of light. The extinction coefficient of free 11-cis retinal (A1) is maximal at 375 nm, whereas the extinction coefficient of bovine rhodopsin was maximal at 498 nm, with the value for bovine P23H rhodopsin blue-shifted from wild-type by 8 nm². X. laevis tadpoles use an alternate chromophore, 11-cis 3,4-dehydroretinal (A2), which absorbs maximally at 392 nm. This difference is responsible for a 13-nm shift in absorbance from 511 nm (A1) to 524 nm (A2) for X. laevis rhodopsin, ^26–28 whereas bovine opsin reconstituted with A2 chromophore absorbs at 516 nm. ²⁹ Therefore, we expected the spectral sensitivity of RD in X. laevis-expressing bovine P23H rhodopsin to be maximal at 392 nm, 524 nm, or 508 nm (508 = 516 − 8), depending on the underlying mechanism. To determine which visible wavelengths are responsible for RD, we reared F1 transgenic X. laevis under blue, green, orange, and red LED light sources with maximal emissions of 470, 510, 610, and 630 nm, respectively (Fig. 3A). The lighting conditions were calibrated to produce equal photon flux (2.7 × 10¹² photons/s/cm²). At 14 dpf, we analyzed the animals by dot blot for rod opsin and histology. Under these conditions, 470 nm and 510 nm light caused significant losses of rod opsin relative to 610- and 630-nm light (Fig. 3B; one-way ANOVA followed by Tukey's multiple comparisons), which was subsequently confirmed as RD by histology (not shown). The effects of 470-nm and 510-nm light were not significantly different from each other in this experiment. When the dark-reared control group was included in the statistical analysis, there was no significant difference between 610-nm light, 630-nm light, and dark rearing. At this intensity or at significantly higher intensities, 510-nm light had no effect on wild-type animals (not shown). These results are consistent with the hypothesis that light absorption by either rhodopsin or free chromophore was responsible for the RD. 

Figure 3.

 

Dependence of RD on wavelength. (A) Emission spectra of LEDs used in this study, normalized for peak output. The LEDs had emission maxima of 470, 510, 610, and 635 nm, respectively. (B) Tadpoles expressing bovine P23H rhodopsin were reared under cyclic light using LEDs as the sole light sources. Light intensity was adjusted such that each group was exposed to a photon flux of 2.7 × 10¹² photons/s/cm², corresponding to 7, 37, 30, and 16 lux for blue, green, orange, and red LEDs, respectively (right). Eyes were analyzed for rod opsin content as an indicator of RD, as in Figure 2. Samples from siblings concurrently reared in dark and standard cyclic light conditions were also analyzed but were not included in the same statistical analysis (left). ***P < 0.001 by t-test (left) or ANOVA followed by Tukey's multiple comparisons (right). Differences were significant between blue/green and orange/red but not between blue and green or between orange and red. RD was confirmed by histology (not shown). Error bars are ± SEM; n = 14 for each condition except constant darkness (n = 10).

Involvement of Rhodopsin versus Free Chromophore in P23H-Induced RD

To distinguish between possible mechanisms of RD mediated by free chromophore and mechanisms mediated by chromophore bound to opsin, we more rigorously defined the relative effects of 470 nm and 510 nm light. We reared transgenic tadpoles under several different intensities of each wavelength of light and constructed dose-response curves using the same techniques shown in Figure 2. The results indicated that green light of 510 nm was more toxic than blue light of 470 nm (Fig. 4), demonstrating that absorption of photons by rhodopsin is principally responsible for RD. However, this still did not rule out the possibility that absorption of photons by free chromophore also plays a role because both molecules absorb significantly at 470 nm. Therefore, we exposed tadpoles expressing bovine P23H rhodopsin to 375-nm and 510-nm light, which approximated the absorbance maxima of free chromophore and rhodopsin, respectively (Fig. 5); 375-nm light was dramatically less effective at causing RD than 510-nm light for an equivalent photon flux, indicating that absorption of photons by free chromophore had little or no role in influencing P23H-induced RD. Similarly, these results demonstrate that photoconversion of bleached rhodopsin by short-wavelength light, reported by Grimm et al. ¹⁹ as responsible for RD caused by intense blue light, is unlikely to play a role in this RD. 

Figure 4.

 

Differential effect of blue and green light on RD. Tadpoles expressing bovine P23H rhodopsin were reared under different intensities of 470-nm and 510-nm light, their eyes were assayed for rod opsin content, and a dose-response curve was constructed. Light intensities are plotted as photon flux, and the points correspond to 1, 3, and 6 lux (470 nm) and 4, 8, 14, and 27 lux (510 nm). For a given photon flux, blue light was less toxic than green light, but both were capable of causing a reduction in rod opsin content (confirmed as RD by histology; not shown). Error bars are ± SEM; n = 11 for each condition

Figure 5.

 

Differential effects of UV and green light on RD. Dose-response curve showing the effects of 375-nm and 510-nm light on rod opsin content of transgenic eyes expressing bovine P23H rhodopsin. Light intensities are plotted as photon flux. For the 510-nm light data, the points correspond to 3, 7, 10, and 18 lux. Although 375-nm light had minimal effects on rod opsin content, 510-nm light exposure again caused dramatic loss of rod opsin content (confirmed as RD by histology; not shown). Error bars are ± SEM; n = 10 for each condition.

Effect of Light Filtration on RD

The finding that only blue and green light caused significant RD suggested the possibility of preventing RD by filtering out shorter wavelength light. To test this hypothesis, we used long-pass filters based on commercially available dyes designed for tinting eyeglass lenses. We reared transgenic tadpoles for 14 days in opaque white tanks with clear polymethyl methacrylate lids tinted to filter out wavelengths below 500, 580, and 600 nm and compared them with siblings reared in tanks with nontinted lids. Filtering wavelengths shorter than 580 nm significantly ameliorated RD in the transgenic tadpoles, whereas filtering wavelengths below 500 nm did not have an appreciable effect on rod opsin levels (Fig. 6). This further confirmed that short-wavelength light is responsible for RD in our transgenic model. 

Figure 6.

 

Effects of long-pass filters on RD. Tadpoles expressing bovine P23H rhodopsin were reared under standard cyclic light conditions or standard conditions filtered through long-pass filters with cutoff values of 500, 580, and 600 nm (right) and were analyzed for rod opsin content. Samples from concurrently dark-reared siblings were also analyzed but were not included in the statistical analysis (left). Filters with the highest long-pass values protected animals against decreased rod opsin content (confirmed as RD by histology; not shown). The degree of protection was similar to that afforded by dark rearing. ***P < 0.001 by ANOVA followed by Tukey's multiple comparisons. Differences were significant between light/500 and 580/600 but not between light and 500 or between 580 and 600. Error bars are ± SEM; n = 11 for each condition.

Effect of Duration of Light Exposure on RD

The proposed mechanism of RD in this animal model involves trapping of newly synthesized P23H rhodopsin in the ER, resulting in prolonged ER stress. Because only a limited amount of P23H rhodopsin is synthesized during a given time period, we hypothesized that a sufficiently brief period of illumination should not induce RD, even under conditions of very intense illumination. Therefore, we reared transgenic X. laevis under 510-nm light in conditions in which cumulative light exposure was held constant, but the duration of light exposure in each 24-hour cycle was varied. Our results confirmed this hypothesis because reducing the period of illumination fivefold from 4 hours to 45 minutes resulted in the elimination of RD, even though the intensity of illumination was simultaneously increased fivefold from 30 to 150 lux (Fig. 7; note that maximal levels of RD were induced by 18 to 27 lux of 12:12 cyclic 510-nm light in Figs. 4, 5). Subsequently, we also attempted to induce RD in anesthetized animals using a 5-minute exposure to very intense illumination (estimated at 1 × 10⁸ lux) from a 532-nm laser. This resulted in a statistically insignificant reduction in rod opsin levels and no observable RD (not shown). These results indicate that prolonged exposure to green light is required to cause RD, supporting the hypothesis that the pathogenic mechanism involves P23H rhodopsin transiting the biosynthetic pathway. 

Figure 7.

 

The dependence of RD on the duration of light exposure. Tadpoles expressing bovine P23H rhodopsin were reared under cyclic 510-nm illumination in which the period of illumination was varied between 12 hours and 45 minutes per day for 2 weeks. Decreased time of exposure was compensated by increased intensity of exposure, such that total light exposure was constant. Retinal extracts were analyzed for rod opsin. Decreasing exposure time to 45 minutes protected against reduced rod opsin content (confirmed as RD by histology; not shown). Samples from siblings concurrently reared in dark and standard cyclic light conditions were also analyzed but were not included in the same statistical analysis (left). The degree of protection was similar to that afforded by dark rearing. ***P < 0.001 and **P < 0.01 by t-test (left) or ANOVA followed by Tukey's multiple comparisons (right). (Comparisons shown in the right are with the 45-minute data point.) Error bars are ± SEM; n = 17 for each condition.

Effect of Vitamin A Deprivation on RD

If chromophore binding stabilizes rhodopsin in the biosynthetic pathway and loss of stabilization on light exposure is responsible for RD, it follows that a reduction in the supply of the chromophore precursor vitamin A should have an effect identical with that of light exposure. To investigate this hypothesis, we reared tadpoles on a vitamin A-restricted diet. Rearing X. laevis in the absence of vitamin A prevents chromophore production. ^30,31 We reared F1 offspring on defined diets with or without carotene and vitamin A (supplied as retinyl acetate) (TestDiet 5LP3 or 5B8V; Purina Mills), in complete darkness. G418 selection for transgenic embryos was not used; hence, both wild-type and transgenic F1 siblings were present in the two groups. Animals were killed at 25 dpf and were analyzed for rod opsin content by dot blot, as in previous experiments. Transgenic and nontransgenic animals were subsequently distinguished using a second dot blot with monoclonal antibody 1D4, which specifically recognizes the mammalian rhodopsin transgene product. ³²  

The results (Fig. 8A) were analyzed by two-way ANOVA (i.e., effects of diet and genotype on rod opsin levels) and indicated a highly significant interaction between diet and genotype (P = 9.0 × 10⁻⁷) such that when reared on a vitamin A-restricted diet, RD was greatly exacerbated in animals expressing P23H rhodopsin. One-way ANOVA analysis was also highly significant (P = 9.5 × 10⁻¹⁵), and post hoc tests (Tukey) showed significant differences between P23H animals reared on the vitamin A-deficient diet and all other groups (P < 2 × 10⁻¹⁰), whereas all other comparisons were not significant. Histologic analysis confirmed RD in P23H animals raised on the vitamin A-deficient diet (Fig. 8B). 

Figure 8.

 

Dependence of RD on dietary vitamin A. (A) Transgenic and nontransgenic F1 offspring of the founder expressing bovine P23H rhodopsin were reared on control and vitamin A-deficient diets. Vitamin A deprivation caused a highly significant decrease in rod opsin content in P23H transgenic animals but not in nontransgenic siblings. ***P < 2 × 10⁻¹⁰ by ANOVA followed by Tukey's multiple comparisons. All possible comparisons with this group were highly significant, but all comparisons between other groups were not statistically significant. Error bars are ± SEM; n = 19–25 for each condition. (B) Histology of representative contralateral eyes corresponding to each group of animals confirms that the reduction in rod opsin content seen in vitamin A-deprived animals expressing P23H rhodopsin was caused by rod photoreceptor degeneration. Color panels show labeling of P23H rhodopsin (green) in vitamin A-deprived and control groups. The markedly lower level of labeling in the rod outer segments of vitamin A-deprived animals is consistent with a reduction in the efficiency of P23H rhodopsin biosynthesis. Cryosections were stained with wheat germ agglutinin (grayscale/red), Hoechst 33342 nuclear stain (blue), and antimammalian rhodopsin antibody 2B2 (green) and were imaged by confocal microscopy. ROS, rod outer segments; ONL, outer nuclear layer; IS, inner segment; N, nucleus. Scale bars: 20 μm (upper), 5 μm (lower).

We examined the distribution of P23H rhodopsin within the photoreceptors of these animals by confocal microscopy. Labeling with antibody 2B2, which binds to mammalian rhodopsins but not to the endogenous X. laevis rhodopsin, showed that less P23H rhodopsin was delivered to the outer segments of animals reared on the vitamin A-deficient diet. In contrast, we did not observe significant changes in the labeling pattern of endogenous rhodopsin in nontransgenic animals using B630N labeling (not shown). 

Discussion

Many disorders are associated with ER retention, instability, or misfolding of transmembrane proteins, including cystic fibrosis and RP, which are caused by mutations in cystic fibrosis transmembrane conductance regulator (CFTR) and rhodopsin, respectively. ⁸ In some cases, a phenotype can be rescued in vitro or in vivo using small molecules or so-called pharmacological chaperones. For example, CFTR mutants can be rescued by corr-2b, ³³ and the rhodopsin mutant P23H can be rescued by 11-cis retinal chromophore in cultured cells. ^2,34 Similar rescue of P23H rhodopsin by chromophore binding occurs in vivo in photoreceptors, ³ but since the chromophore is photolabile, rescue is dependent on protection of the retina from light exposure. Light exposure could prevent interaction of chromophore and opsin by several possible mechanisms. One possibility is that the supply of 11-cis retinal is dramatically reduced on light exposure because of the bleaching of rhodopsin in the rod outer segments. Continuous bleaching of rhodopsin in outer segments would deplete the supply of 11-cis retinal in the biosynthetic pathway. However, it is also conceivable that light could directly photoisomerize free chromophore or directly photoisomerize chromophore bound to P23H opsin in the biosynthetic pathway (causing destabilization of folded protein and subsequent misfolding or delayed ER exit) (Fig. 1). In addition, previous studies of rats have demonstrated highly toxic effects of intense blue light, most likely because of the photoreversal of rhodopsin, ¹⁹ a mechanism that greatly increases the photon catch rate of the retina. 

In this study, we characterized the wavelengths of light responsible for light-sensitive RD induced by the bovine P23H rhodopsin mutant. Our experiments demonstrate that this form of RD is caused by photon absorption of chromophore bound to rhodopsin because green light of 510 nm, corresponding to the absorption maximum of rhodopsin, was most effective at inducing RD. This indicates that the mechanism of RD involves either chromophore depletion by outer segment rhodopsin or direct destabilization of P23H rhodopsin during biosynthesis. The intensity of white light that was necessary to produce the half-maximal effect (270 lux) was similar to the intensity of light previously shown to produce sustained bleaching of 50% of the rhodopsin in an amphibian retina (400 lux), ³⁵ which would represent the half-maximal rate of 11-cis retinal isomerization. Moreover, prolonged light adaptation can result in dramatic reductions in the quantity of 11-cis retinal that can be isolated from retina. ³⁶ Thus, our data are entirely consistent with the first hypothesis. However, we cannot eliminate a possible mechanism involving direct destabilization of P23H rhodopsin by photon absorption in the biosynthetic pathway, which would not be caused by a limitation in the supply of chromophore. This possibility could be examined by constructing a transgenic model in which the absorbance spectrum of P23H rhodopsin has been “tuned” by additional mutations to be distinct from that of endogenous rhodopsin. ³⁷  

Pharmacological chaperones can promote the ER export of destabilized proteins (i.e., proteins that fold correctly but tend to revert to an unfolded conformation) but not of proteins with a high rate of misfolding (i.e., proteins that do not initially fold correctly but instead adopt an alternately stable but nonfunctional conformation). Our results are consistent with 11-cis dehydroretinal acting as a pharmacological chaperone for destabilized rhodopsin and predict that pharmacological chaperones (for example, non–light-sensitive derivatives of 11-cis retinal) are likely to be effective treatments for RD that is exacerbated by light; however, approaches designed to alter ER properties (such as molecular chaperone inducers ³⁴ ) may be more appropriate for RDs that are not exacerbated by light. Our experiment in which the presence of vitamin A in the diet of dark-reared animals rescues retinal degeneration effectively models this concept. Notably, there is prior evidence that high doses of vitamin A (which hypothetically could raise levels of free 11-cis retinal) are beneficial for patients with retinitis pigmentosa ³⁸ and transgenic animal models, ³⁹ though large doses of vitamin A are not recommended by many ophthalmologists because of limited demonstrated efficacy and potential toxicity. ⁴⁰  

For this study, we used an animal model of P23H rhodopsin-induced RP in which RD is entirely associated with light exposure. However, this is not the case in other animal models (including X. laevis models based on human P23H rhodopsin), in which RD is only partially rescued by light deprivation, and is likely not the case in patients with retinitis pigmentosa. In other animal models (and humans), the mechanism described here may contribute to only a portion of the total RD. In these model systems, a smaller proportion of newly translated P23H rhodopsin is stabilized by chromophore; it is likely that these P23H rhodopsin mutants are significantly less stable or are destabilized and misfold more rapidly. In other words, there may be a greater tendency for human P23H rhodopsin to adopt a nonnative structure before chromophore binding that is “off pathway” (misfolding) or a greater tendency after adopting a native conformation to revert to a disordered structure that is “on pathway” (instability). Interestingly, previous studies demonstrate that frog rhodopsin is less thermally stable than bovine rhodopsin, ⁴¹ and our own results suggest that human rhodopsin is intermediate between bovine and frog. ³  

The expression level of P23H rhodopsin in human patients with retinitis pigmentosa has not been reported and, therefore, cannot be directly compared to the expression level of mutant rhodopsin in our frog models of the disease. Although differences in expression levels may influence the severity of the phenotype, differences in expression levels should not affect the wavelength of light that induces maximal retinal degeneration or the dependence of degeneration on the availability of vitamin A. However, it should be noted that the intensity and duration of light exposures that induced retinal degeneration in this study cannot be directly extrapolated to human patients with retinitis pigmentosa because expression level, size, and geometry of the eye and the rate of regeneration of 11-cis retinal could affect these values. 

Our most complete dose-response curves (Figs. 2, 5) have very steep slopes; that is, they have Hill coefficients of 3.5 and 2.5, respectively, with the complete range of effects occurring over approximately one order of magnitude of light intensities. This might be expected if multiple photons must interact with each rhodopsin molecule to initiate cell death (as in a photoreversal mechanism), but such a mechanism is not supported by the spectral sensitivity of the RD. Another possible explanation for the very steep curves is that the underlying mechanism involves the interaction of multiple destabilized P23H rhodopsin molecules to form a protein aggregate. Several in vitro studies suggest that P23H rhodopsin has a tendency to aggregate in cultured cells, leading to the formation of aggresomes. ^7,42 Although we have not observed large aggregates or aggresomes by microscopy in photoreceptors expressing P23H rhodopsins, aggregate forms are apparent by gel electrophoresis. ^3,4 It is possible that small aggregates or complexes of destabilized rhodopsin are responsible for initiating RD. Regardless of the mechanism, the steep dose-response curves imply that even small perturbations of the system (whether pharmaceutical, dietary, or environmental) in favor of stabilization of the mutant rhodopsin are likely to have large benefits for patients. 

Our results have implications for the design of pharmacological chaperones to prevent RP. To mimic the protective effects of decreased light exposure, such compounds must be nonisomerizable, similar to previously described “locked” variants of 11-cis retinal. ^43,44 Such compounds would appear self-defeating because they would prevent signal transduction, but they may be beneficial at intermediate doses such that ER exit would achieve a critical level. Alternatively, a nonisomerizable 11-cis retinal analogue could be created that preferentially binds rod opsin over cone opsin, thus primarily affecting low light vision. Depending on the underlying mechanism of RD (Fig. 1), efficacy would be highly dependent on the ability of such compounds to compete with endogenous 11-cis retinal for binding sites. As previous studies indicate, 11-cis retinal is markedly depleted on light adaptation. ³⁶ This may not be an issue: a pharmacological chaperone can compete effectively during bright light exposure (when it is needed for biosynthesis and when rod-mediated vision is largely irrelevant) and less effectively in dim light, when 11-cis retinal is available for biosynthesis and required for rod vision. 

Pharmacological chaperones for P23H rhodopsin that are not chromophore analogues are also possible, as evidenced by the observations that monoclonal antibodies can stabilize metarhodopsin-1, ⁴⁵ and disulfide bonds can stabilize rhodopsin against denaturation. ⁴⁶ However, the necessary increase in stability may be difficult to achieve, as previous measurements indicate that binding of 11-cis retinal increases the activation energy and ΔH for unfolding of wild-type opsin because of thermal denaturation by 30% to 60%. ^41,47,48  

An obvious therapeutic approach suggested by our results is to minimize patient exposure to light by reducing the time of exposure and by the use of orange- or red-tinted lenses or dark sunglasses. Such lenses are readily available, are used by patients with achromatopsia, and even recommended for patients with retinitis pigmentosa to reduce glare. However, these lenses would also restrict patients' normal vision. It is not yet clear to what extent human patients with the P23H mutation could be rescued by preventing photoexcitation of rhodopsin. However, the use of tinted lenses may represent a relatively innocuous mechanism (compared with eye patching ⁴⁹ ) for examining this question. Our data also suggest that patients with the P23H mutation are unlikely to experience permanent vision loss from brief intense light exposures (such as those occurring during an ophthalmologic examination) and that patients should ensure that they receive at least adequate vitamin A in their diets. However, we would caution that our current results are applicable only to P23H rhodopsin and that light may exacerbate RD caused by other rhodopsin mutations (such as T17M ⁵⁰ and T4K ¹⁴ ) by alternative pathways in which brief light exposures are detrimental. Therefore, it is critical to perform genotyping on patients with retinitis pigmentosa and to carry out additional studies to determine whether other forms of RD respond similarly to light exposure and vitamin A deprivation. 

Footnotes

 Supported by funding from the Canadian Institutes for Health Research, the Foundation Fighting Blindness (Canada), and a summer student fellowship from the University of British Columbia Faculty of Medicine (AQ). OLM is a CIHR New Investigator and FFB-(Ca) W.K. Stell Scholar.

Footnotes

 Disclosure: B.M. Tam, None; A. Qazalbash, None; H.-C. Lee, None; O.L. Moritz, None

The authors thank Paul Hargrave for providing the antibody B630N used in this study, Robert Molday for providing the antibody 2B2, Robert Douglas for the loan of the spectral radiometer, Eduardo Solessio for supplying the 5LP3 and 5B8V diets, and Jenny Wong for technical assistance.