Protein Gq Modulates Termination of Phototransduction and Prevents Retinal Degeneration*

Background: Appropriate termination of photoresponse is critical for photoreceptors to achieve high temporal resolution and to prevent excessive Ca2+-induced cell toxicity. Results: We isolated a novel Gαq mutant allele and revealed that metarhodopsin/Gq interaction affects Arr2-Rh1 binding. Conclusion: Gq modulates the termination of phototransduction and prevents retinal degeneration. Significance: Our study revealed the novel role of Gq in phototransduction deactivation and in retinal degeneration. Appropriate termination of the phototransduction cascade is critical for photoreceptors to achieve high temporal resolution and to prevent excessive Ca2+-induced cell toxicity. Using a genetic screen to identify defective photoresponse mutants in Drosophila, we isolated and identified a novel Gαq mutant allele, which has defects in both activation and deactivation. We revealed that Gq modulates the termination of the light response and that metarhodopsin/Gq interaction affects subsequent arrestin-rhodopsin (Arr2-Rh1) binding, which mediates the deactivation of metarhodopsin. We further showed that the Gαq mutant undergoes light-dependent retinal degeneration, which is due to the slow accumulation of stable Arr2-Rh1 complexes. Our study revealed the roles of Gq in mediating photoresponse termination and in preventing retinal degeneration. This pathway may represent a general rapid feedback regulation of G protein-coupled receptor signaling.

Heterotrimeric G proteins play pivotal roles in mediating extracellular signals from hormones, neurotransmitters, peptides, as well as sensory stimuli to intracellular signaling pathways (1,2). In Drosophila photoreceptors, G proteins are essential for the activation of the phototransduction cascade (3,4). Photon absorption leads to the photoisomerization of chromophores, resulting in the formation of activated metarhodopsin. In turn, metarhodopsin activates heterotrimeric G proteins and PLC. 2 The activation of PLC leads to transient receptor potential and transient receptor potential-like channels opening and extracellular Ca 2ϩ influx (5)(6)(7)(8).
It is also critical for each step of the phototransduction cascade to be terminated appropriately, which is essential for the high temporal resolution of fly vision (9,10). The most impor-tant step in phototransduction termination is the deactivation of metarhodopsin. During this step, arrestin (Arr2) plays an important role by displacing the G q ␣ subunit and allowing it to bind with rhodopsin (Rh1) (11,12). Unlike other G proteincoupled receptors (GPCRs), the phosphorylation of fly rhodopsin is not required for its deactivation (13) but is essential for its endocytosis (11). In contrast, the dephosphorylation of rhodopsin by retinal degeneration C (RDGC) is essential for receptor deactivation (14). Ca 2ϩ also plays critical roles in regulating the termination of the photoresponse in Drosophila (8,15,16). Several proteins that mediate this Ca 2ϩ -regulated termination have been identified, such as eye-specific protein kinase C (INAC), calmodulin, INAD, and myosin III (NINAC) (8,(17)(18)(19)(20).
It has been shown that the loss of DG q leads to the slow termination of the light response (4). Here, we provide evidence that the metarhodopsin/G q interaction affects subsequent Arr2-Rh1 binding, which mediates the deactivation of metarhodopsin. Our study has characterized the role of G q in photoresponse termination. This pathway represents a general rapid feedback regulation in GPCR signaling.

EXPERIMENTAL PROCEDURES
Fly Genetics-The genotype of wild-type flies is cn,bw. sast mutants were generated with the chemical mutagen ethyl methanesulfonate (EMS) in cn,bw background, and all flies were put into cn,bw background to eliminate the effects of eye color. The mutant alleles used for each gene in this work are ninaE 8 , G␣ q 1 , and norpA 24 . To avoid age-dependent retinal degeneration, all flies were reared at 22°C in dark and examined when they were 2-3 days old. The G␣ q mutant and p[hs::dG␣ q ] transgenic fly were obtained from the C. Zuker laboratory, and all other flies used in this work were from the Bloomington Stock Center. The deficiency line Df(2R)Exel7121 was obtained from the Bloomington Stock Center and recombined into FRT 42D background. Genetic mosaics were induced by the FLP-FRT p [GMR-hid] technique with an ey-FLP driver to generate mitotic clones of a single genotype in the eye (21).
Generation of Transgenic Flies-To generate G q H212A transgenic flies, a mutate dG q cDNA, which mutate His 212 to Ala, was subcloned into pUAST vector and then injected into w 1118 flies. The transgene was subsequently crossed into the G␣ q 1 and G␣ q 961 mutant background, respectively. G q H212A protein was expressed by using eye-specific driver (p[gmr::GAL4]).
Electrophysiological Recordings-Electroretinogram (ERG) recordings were performed as described previously (22). Briefly, fly eyes were stimulated with 5-s light pulses (4000 lux). For each genotype and condition, Ͼ10 flies were examined. To quantitate the speed of response termination, the time required for half-recovery was measured, and the S.E. was calculated. t1 ⁄ 2 was defined as the time required for a half-recovery from the responses upon stimulation cessation. Prolonged depolarization afterpotentials were examined using the same setup except with different color filters and light intensities.
Relative light sensitivity measurement was carried out as described previously (23). For each genotype and condition, 10 flies were examined, and the relative sensitivities were averaged to obtain a mean. The S.E. were calculated and presented as error bars in the figures.
Intracellular recordings were performed as described previously (24). Briefly, a low resistance glass microelectrode filled with 2 M KCl was placed into a small hole on the compound eye. A reference electrode was filled with Ringer's solution and inside the eye at the retina layer. The microelectrode placed on the eye was gradually inserted into the opening until light-induced membrane depolarization was observed. The signal was amplified and recorded using a Warner IE210 Intracellular Electrometer.
EM and Immunolocalizations-Fly heads were immersed in 2.5% glutaraldehyde, 0.1 M sodium cacodylate (pH 7.2) at 4°C for 12 h. After rinsing three times with 0.1 M sodium cacodylate, the fixed tissue was stained with 1% osmium tetroxide for 1 h at room temperature. A standard ethanol dehydration series was performed, and tissue was immersed in two 10-min washes of propylene oxide. The tissue was then embedded using standard procedures. For electron microscopy, thin sections (100 nm) were cut, collected on copper support grids, and stained with uranyl acetate followed by lead citrate staining. Micrographs were taken at 80 kV on Hitachi-7650.
Arr2 Binding Assays-Arr2 binding assays were performed as described previously with a minor modification (26). For each group and condition, eight adults were collected and adapted with food in complete dark for Ͼ2 h. After exposure for 60 s to pure blue light (480 Ϯ 10 nm), fly heads were isolated by liquid nitrogen and homogenized in the dark. After centrifuging at 14,600 ϫ g for 5 min, pellet and supernatant fractions were separated under very dim red light for Western blotting analysis. Arr2 release assays were performed in the same manner, except that the flies were exposed to 60 s of pure blue light followed by 2 min of pure orange light (580 Ϯ 10 nm).
Statistics-For each quantitative analysis, the images of Western blots and immunostaining were quantified using ImageJ software. After averaging three sets of data, the mean values Ϯ S.E. (as error bars) are presented.

Isolation and Identification of a novel G␣ q Mutant Allele-
To characterize the components of the phototransduction machinery, we performed an ERG-based chemical mutagenesis screen for additional genes in the fly. We isolated a mutant fly, sast (small amplify and slow termination), which shows small ERG responses (3.7 Ϯ 0.3 mV versus 11.2 Ϯ 0.4 mV) and slow photoresponse termination (t1 ⁄ 2 ϭ 1.02 Ϯ 0.22 s versus t1 ⁄ 2 ϭ 0.06 Ϯ 0.01 s) compared with wild-type flies (Fig. 1, A-C). In addition, sast mutant displays the reduced ON transient and OFF transient (Fig. 1, A and D). The light sensitivity of this mutant is also significantly reduced (Fig. 1E). Intracellular recording further revealed that the defective light response in sast mutants was due to an abnormality in photoreceptor cells (Fig. 1F).
To identify the specific mutation in sast flies, we first mapped the mutation to the 49B5-49B12 chromosomal region based on the ERG phenotype covered by the deficient chromosome Df(2R)Exel7121 (missing 49B5-10 to 49B12) ( Fig. 2A). This genomic region contains 26 genes, including G␣49B, which encodes the G q protein. Because the mutant flies showed an ERG phenotype similar to that of the G␣ q 1 mutants, the sast mutant might harbor a G␣ q mutant allele. The sast/G␣ q 1 flies displayed an ERG phenotype similar to that of either the sast or G␣ q 1 mutant (Fig. 2B), supporting this conclusion. A previous study showed that the G␣49B gene encodes several DGq splice variants (27) and that DG q 1 plays a predominant role in phototransduction in Drosophila (4). Using an antibody that recognizes DG q 1 , we revealed that the DG q 1 protein is absent in sast mutants ( Fig. 2C and supplemental Fig. 1). In contrast, other phototransduction proteins were present at normal levels (Fig. 2D). Subsequent DNA sequencing revealed that the sast mutant contains a C961T mutation in the G␣49B gene (Fig. 2E). This mutation corresponds to a nonsense mutation (Arg 117 to stop codon) in exon 4 (amino acids 153-196 of DG q 1 ; Fig. 2E), which is not included in the DG q 3 and DG q 4 variants (27). To examine whether the sast mutant retains any truncated DG q 1 protein, we performed Western blotting with an antibody against the N terminus of DG q 1 . We could not detect any truncated DG q 1 protein, suggesting that the sast mutant is likely to possess one G␣ q 1 -null allele (Fig. 2F). Therefore, we named the sast mutant G␣ q 961 .
Because G␣ q 961 is a G␣ q 1 -null mutant and DG q 1 plays a predominant role in phototransduction in Drosophila, we worried whether the residual light response observed in the G␣ q 961 mutant was evoked by either DG q 3 or DG q 4 variants. To detect DG q 3 and DG q 4 protein in both G q 1 and G␣ q 961 mutants, we performed Western blotting by using the anti-G q ␣ subunit antibody (371754; Calbiochem) that also recognizes DG q 3 and DG q 4 variants. Unfortunately, we did not detect any DG q 3 or DG q 4 in either G q 1 and G␣ q 961 mutants even loading with 32 isolated retinas (data not shown). This observation implied that the expressions of DG q 3 or DG q 4 in the retina were extremely low and the residual light response in the G␣ q 961 mutant might be attributable to other G protein genes, but not the Gq49B gene. To further support this hypothesis, we generated clones of G␣49B-nulls in the retina through ey-FLP-induced FRT recombination in deficiency line, Df(2R)Exel7121, which deletes the whole G␣49B gene. This fly also showed small light response and slow termination ERG phenotype (supplemental Fig. 2). This observation indicates that the residual light response observed in the G␣ q 961 mutant is contributed by other G protein, but not DG q 3 or DG q 4 . To further confirm that the absence of the DG q 1 protein is indeed responsible for the G␣ q 961 mutant phenotype, we performed rescue experiments by expressing DG q 1 protein in a G␣ q 961 mutant background. After induction of heat shock, the levels of the DG q 1 protein were recovered (Fig. 2G), and ERG responses were restored in G␣ q 961 ;p[hs-dG␣ q ] flies (Fig. 2H) Fig. 3A), accompanied by an increase in the termination speed (t1 ⁄ 2 ϭ 0.47 Ϯ 0.08, Fig. 3B). Extended heat shock induced further increases in the levels of DG q protein (22.4 Ϯ 1.7%, Fig. 3A) and triggered rapid termination (t1 ⁄ 2 ϭ 0.11 Ϯ 0.01, Fig. 3B). We excluded the possibility that the increased termination speed was due to the heat-shock treatment, because the heat-shock treatment did not change the protein level of DG q and the termination speed in wild-type flies (Fig. 3B, right, and supplemental Fig. 3). These data suggest that termination speed depends on the available amount of the DG q protein.
Previous work has shown that G␣ q 1 mutants generate quantum bumps with prolonged latency (4). It is possible that the slow termination phenotype reflects a slow activation process. To test this hypothesis, we generated a variant DG q H212A transgenic fly with a mutated His 212 to Ala, resulting in a loss of capacity of G q to activate PLC (28). In cn,G␣ q 961 ,bw; gmrϾDG q H212A flies, the levels of variant DG q H212A protein were comparable with the levels of G q protein in wild-type flies (supplemental Fig. 4). Interestingly, deactivation kinetics, but not ERG amplitude, was nearly restored in this fly (Fig. 3C). These results indicate that the slow termination phenotype in G␣ q mutants is not due to delayed bump generation.
DG q Mediates Deactivation of Metarhodopsin-During fly photoresponse termination, metarhodopsin deactivation is the most important step (29). Thus, we first investigated whether DG q was required for the deactivation of metarhodopsin. Prolonged depolarization afterpotential (PDA) is induced when the number of activated rhodopsin molecules exceeds the number of available regulators (29). In mutants lacking rhodopsin regulators, because lower amounts of rhodopsin need to be activated, the light intensity required to induce PDA is much lower than that in wild-type flies (14). The minimum light intensity   (Fig. 4, A and B). Therefore, this result suggests that the deactivation of metarhodopsin is impaired and that there is a shortage of rhodopsin regulator molecules in G␣ q 961 mutant. Because the deactivation of metarhodopsin is impaired in G␣ q 961 mutants, we attempted to increase the termination speed of G␣ q 961 photoresponses by reducing the level of Rh1, the major rhodopsin in all outer (R1-R6) photoreceptor cells. We genetically reduced Rh1 levels via the introduction of ninaE 8 , which contains Ͻ1% Rh1 protein (30), into a G␣ q 961 background. Because much less Rh1 need to be deactivated, the impaired regulating machinery might be sufficient for the deactivation of rhodopsin, and the termination speed of the photo-response should be faster in G␣ q 961 ;ninaE 8 double mutants compared with that in the G␣ q 961 single mutant. Indeed, in G␣ q 961 ;ninaE 8 double mutants, the deactivation of the photoresponse was much faster than that in the G␣ q 961 single mutant (Fig. 4, C and D), suggesting that DG q is required for the deactivation of metarhodopsin.
Metarhodopsin/G q Interaction Affects Arr2-Rh1 Binding-Because Arr2 is the primary deactivator of rhodopsin through its binding with rhodopsin (11, 12), we next assessed whether G q mediates the deactivation of metarhodopsin by affecting Arr2-Rh1 binding. To characterize the Arr2/Rh1 interaction, we used an arrestin pelleting assay (31,32). Upon 30 s of blue light stimulation, Ͼ80% Arr2 binds to metarhodopsin in wildtype flies, whereas only 52% Arr2 binds to metarhodopsin in the G␣ q 961 mutant (80.1 Ϯ 3.2% versus 52.3 Ϯ 4.3%, Fig. 5A, middle panel). Immunostaining results were consistent with those of  . DG q is required for termination of light response. A, G q protein levels in different treatments. cn,G␣ q 961 ,bw; P[hs::dG␣ q ] adults reared at 21°C were exposed to 30 min of heat shock at 37°C (30 min) or 1 h of heat shock at 37°C (60 min) or no heat shock (no hs). All flies were collected and analyzed at 24 h after heat shock. INAD was used as the loading control. Note that the upper bands are nonspecific bands. Quantification of relative DG q protein levels is shown on the right. B, ERG recordings showing termination kinetics of light responses in each treatment. G q protein was induced as described in A. For all ERG traces, event markers represent 5-s orange light pulses, and the treatment conditions for each trace are marked. Quantification of the time required for half-recovery in each treatment, including wildtype controls, is shown on the right. C, representative ERG traces of cn,G␣ q 961 ,bw and cn,G␣ q 961 ,bw;gmrϾG q H212A flies. Quantification of the time required for half-recovery in each treatment is shown on the right. Error bars indicate S.E. the arrestin pelleting assay (Fig. 5B). These observations suggest that DG q indeed affects Arr2-Rh1 binding.
Because G q functions as a molecular switch on phototransduction cascade, we next investigated whether G q affects Arr2-Rh1 binding by activating the phototransduction cascades. In cn,G␣ q 961 ,bw;grmϾG q H212A flies, phototransduction cascade cannot be activated by the variant DG q H212A (Fig. 3C). Interestingly, the minimum light intensity needed to induce PDA in this fly (2200 Ϯ 300 lux) was comparable with that in wild-type flies (2600 Ϯ 200 lux) (Fig. 6A), reflecting the normal Arr2-Rh1 binding. An Arr2-binding assay further showed that Arr2-Rh1 binding is normal in these flies (Fig. 6B). Furthermore, Arr2-Rh1 binding is normal in norpA mutants, which also show defects in the activation of phototransduction (Fig. 6C) (33). Taken together, these observations suggest that the metarhodopsin/G q interaction, but not the activation of the phototransduction cascade, affects Arr2-Rh1 binding.
G␣ q 961 Mutant Undergoes Light-dependent Retinal Degeneration-In Drosophila, mutations in almost any gene involved in the phototransduction cascade lead to light-dependent retinal degeneration (34). Thus, we investigated whether the G␣ q 961 mutant and G␣ q 1 mutant also undergo retinal degeneration. An EM study showed obvious retinal degeneration in older mutants but not in 6-day-old mutants (Fig. 7, A, B,  D, E, G, and H). Additionally, 21-day-old dark-reared mutants did not undergo retinal degeneration (Fig. 7, C, F, and I). These observations indicate that G␣ q 961 mutants and the G␣ q 1 mutant undergo slow light-dependent retinal degeneration. PLC ␤ , encoded by norpA, is the downstream effector of G q (33). A loss of PLC results in the formation of stable Arr2-Rh1 complexes, which trigger retinal degeneration (32). Hence, we examined whether the same mechanism causes retinal degeneration in the G␣ q 961 mutant. With 480-nm blue light exposure, Rh1 is photoconverted to metarhodopsin and induces binding with Arr2. Metarhodopsin can be photoconverted to inacti-  . Lack of G q leads to the impaired Arr2-Rh1 binding. A, assessment of deficits in Arr2-Rh1 binding. Dark-adapted flies were exposed to blue light for 30, 60, or 90 s, respectively. Supernatant (S) and membrane pellet (P) fractions of fly heads were subjected to Western blotting. D, kept in dark; B, blue light exposure. Quantification of Arr2-Rh1 binding is shown on the right. B, immunostaining of Arr2 and Rh1 in different genotypes and under different conditions. Sections were prepared as described under "Experimental Procedures." Quantification of Arr2 in rhabdomeres is shown on the right. The averaged data from three independent sections are shown. Error bars, S.E. vated rhodopsin, leading to the release of Arr2 (32). In wildtype flies, blue light exposure caused Ͼ80% binding between Arr2 and Rh1 and Ͼ80% release of Arr2 from Rh1 following exposure to orange light (Fig. 7, J and K). In contrast, in the mutant, blue light exposure only triggered approximate 48.5% binding between Arr2 and Rh1, and Ͻ60% release of Arr2 from Rh1 following exposure to orange light (Fig. 7, J and K). These data imply that the formation of stable Arr2-Rh1 complexes might trigger retinal degeneration in the G␣ q mutant.

DISCUSSION
Our work isolated and identified a novel G␣ q mutant allele and demonstrated that metarhodopsin/G q interaction affects subsequent Arr2-Rh1 binding, which might mediate the deactivation of metarhodopsin. Our study revealed the involvement of a general rapid feedback regulation in GPCR signaling.
G q Mediates Termination of Light Response-Rapid termination of the light response is critical for fly vision to achieve high temporal resolution and to prevent excessive calcium-induced cell toxicity (9,10). Using an ERG-based chemical mutagenesis screen, we isolated a novel G␣ q mutant allele, G␣ q 961 , which shows defects in the termination of the light response. A similar phenotype has been reported in another G␣ q mutant allele, G␣ q 1 (4). Compared with the G␣ q 1 mutant, G␣ q 961 showed even more severe defects in photoresponse deactivation. One explanation is that G␣ q 961 mutants contain an even lower amount of G q protein than that found in G␣ q 1 mutants. Western blotting revealed that the G␣ q 961 mutant is likely a null mutant. In contrast, G␣ q 1 mutants produce approximately 1% of wild-type levels of the DG q protein (4). This explanation is further supported by rescue experiments, which showed that the termination kinetics depend on the amount of available G q protein.
A previous study showed that the absence of G q leads to the generation of quantum bumps with prolonged latency (4) because it takes longer for rare residual G q to encounter metarhodopsin. In this study, we provided evidence that the slow termination phenotype of the G␣ q mutants was not due to slow activation. Expression of DG q H212A , which can bind metarhodopsin but cannot activate PLC, is nearly able to restore the FIGURE 6. Metarhodopsin/G q interactions affect Arr2-Rh1 binding. A, minimum light intensities required for PDA production in cn,G␣ q 961 ,bw; gmrϾG q H212A flies, comparable with that in wild-type flies. Representative recorded traces were induced by 2000 lux blue light. All flies were darkreared. o, orange light; b, blue light. Quantification of minimum light intensities required for PDA production in each genotype is shown on the right. B, Arr2-Rh1 binding in cn,G␣ q 961 ,bw;gmrϾG q H212A flies. Arr2-Rh1 binding and Arr2 release assays were performed as described under "Experimental Procedures." Quantification of relative Arr2-Rh1 binding and release is shown on the right. C, Arr2-Rh1 binding in norpA mutants. Arr2-Rh1 binding assays were performed as described under "Experimental Procedures." Quantification of the relative Arr2-Rh1 binding is shown on the right. Error bars, S.E. defects of photoresponse termination in a G␣ q 961 mutant background. In G␣ q 961 ;gmrϾG q H212A flies, the termination speed of the photoresponse is slower than that in wild-type flies, which might be attributed to slow activation and impaired calciumdependent receptor deactivation mechanisms (35).
Metarhodopsin/G q Interaction Affects Arr2-Rh1 Binding-Combined with genetic, biochemical, and electrophysiological approaches, we showed that metarhodopsin/G q interaction affects subsequent Arr2-Rh1 binding. Using the arrestin pelleting assay, we showed that more Arr2 binds to metarhodopsin in wild-type flies than that in the G␣ q 961 mutant upon blue light stimulation (80.1 Ϯ 3.2% versus 52.3 Ϯ 4.3%). Although some of the Arr2 can still bind with Rh1 in the G␣ q 961 mutant, the metarhodopsin/G q interaction clearly affects subsequent Arr2-Rh1 binding, which mediates the deactivation of photoresponse. This conclusion is further supported by the observations in cn,G␣ q 961 ,bw;grmϾG q H212A flies. We showed the rapid deactivation kinetics of ERG, normal Arr2-Rh1 binding, and comparable minimum light intensity needed to induce PDA in this fly. This rapid feedback regulation does not need to activate the entire phototransduction cascade and allows the termination of photoreceptors with extremely fast kinetics, permitting adaptation in a huge dynamic range of light intensities (10,34,36). Such regulation also prevents excessive Ca 2ϩ influx and Ca 2ϩ -dependent excitotoxicity during intensive stimulation (37)(38)(39)(40).
It is possible that the metarhodopsin/G q interaction exposes the binding site in metarhodopsin or shifts metarhodopsin into an intermediate metarhodopsin status, which facilitates Arr2-Rh1 binding. Using fluorescence pump probe and time-resolved fluorescence depolarization measurement approaches, it has been shown that helix 8 of rhodopsin undergoes dynamic changes at the receptor surface, which modulates arrestin-rhodopsin binding (41). In vertebrates, the phosphorylation of rhodopsin is important for ␤-arrestin/rhodopsin binding. GPCR kinase 2 has been shown to undergo selective binding with the G q ␣ subunit (42). These results indicate that the rhodopsin/G q interaction might mediate arrestin-rhodopsin binding via the promotion of rhodopsin phosphorylation. However, this is not the case in the deactivation of the photoresponse in the fly because the phosphorylation of metarhodopsin is not required for Arr2-Rh1 binding in fly photoreceptors (14,32,43).
G␣ q Mutant Undergoes Slow Retinal Degeneration-Mutations in almost any gene that functions during phototransduction trigger rapid retinal degeneration, except the G␣ q 1 mutant, which undergoes slow retinal degeneration (34,43,44). In this study, we showed that the G␣ q 961 mutant undergoes slow lightdependent retinal degeneration, similar to that observed in the G␣ q 1 mutant (44). We provided evidence that mild retinal degeneration in the G␣ q 961 mutant was due to the slow accumulation of stable Arr2-Rh1 complexes, which trigger retinal degeneration (32). Light stimulation triggers Ca 2ϩ influx and activates CaM kinase II, which phosphorylates Arr2 and results in Arr2 release from Rh1 (26,45). In the G␣ q 961 mutant, impaired Ca 2ϩ influx leads to the slow release of Arr2 from Rh1. On the other hand, lack of G q partially inhibits basal Rh1 endocytosis (24) and inhibits the translocation of Arr2 to rhabdomeres (46). These two opposite effects lead to the slow accumu-lation of stable Arr2-Rh1 complexes and trigger mild retinal degeneration.