The Role of Mislocalized Phototransduction in Photoreceptor Cell Death of Retinitis Pigmentosa

Most of inherited retinal diseases such as retinitis pigmentosa (RP) cause photoreceptor cell death resulting in blindness. RP is a large family of diseases in which the photoreceptor cell death can be caused by a number of pathways. Among them, light exposure has been reported to induce photoreceptor cell death. However, the detailed mechanism by which photoreceptor cell death is caused by light exposure is unclear. In this study, we have shown that even a mild light exposure can induce ectopic phototransduction and result in the acceleration of rod photoreceptor cell death in some vertebrate models. In ovl, a zebrafish model of outer segment deficiency, photoreceptor cell death is associated with light exposure. The ovl larvae show ectopic accumulation of rhodopsin and knockdown of ectopic rhodopsin and transducin rescue rod photoreceptor cell death. However, knockdown of phosphodiesterase, the enzyme that mediates the next step of phototransduction, does not. So, ectopic phototransduction activated by light exposure, which leads to rod photoreceptor cell death, is through the action of transducin. Furthermore, we have demonstrated that forced activation of adenylyl cyclase in the inner segment leads to rod photoreceptor cell death. For further confirmation, we have also generated a transgenic fish which possesses a human rhodopsin mutation, Q344X. This fish and rd10 model mice show photoreceptor cell death caused by adenylyl cyclase. In short, our study indicates that in some RP, adenylyl cyclase is involved in photoreceptor cell death pathway; its inhibition is potentially a logical approach for a novel RP therapy.


Introduction
Retinitis pigmentosa (RP, MIM 26800) is a common group of inherited retinal diseases that lead to blindness. RP affects 19-27 out of 100,000 people all over the world [1,2]. Usually, patients first suffer from peripheral visual field loss because of peripheral rod photoreceptor cell death. Photoreceptor cell death gradually but steadily progresses until patients lose central visual function, which degrades quality of life. Such slow progressive photoreceptor cell death is a prominent feature of RP.
This mislocalization may play an important role in human disease because it is observed in human photoreceptor cell death caused by RP, retinal detachment [15,16], and age-related macular degeneration [17]. We have also shown that the mislocalization causes ectopic phototransduction, which accelerates photoreceptor cell death in ovl [14]. However the mechanism of photoreceptor cell death caused by the mislocalization is entirely unknown. Phototransduction is the process of a cell absorbing light and creating a response. In this pathway the photoreceptor-specific G protein, transducin, mediates between the light receptor, rhodopsin, and the effecter enzyme, phosphodiesterase ( Figure 1A). In this article, we demonstrate a detailed mechanism of photoreceptor cell death coupled with mislocalized phototransduction via transducin.

Light exposure accelerates rod photoreceptor cell death
To investigate the relationship between photoreceptor cell death and phototransduction, we assessed whether the photoreceptor cell death in ovl is accelerated by phototransduction. In ovl, rod photoreceptor cells degenerate progressively ( Figure S1). At 108 hours post fertilization (hpf), the degeneration of rod photoreceptors in ovl is obvious ( Figure 1B, C). As human retinitis pigmentosa, the degeneration is predominantly shown in rod photoreceptor and, cone photoreceptors do not degenerate severely in early stage. ( Figure S2). And the ovl retina does not develop an outer segment, resulting in mislocalized photopigments, which mimics many human photoreceptor diseases ( Figure 1D, E) [9,15,16]. We first confirmed whether rod photoreceptor cell death in ovl is light dependent as we previously reported [14]. The number of rod photoreceptor cells showed a  Figure 1F-H). The exposure did not influence the number of wild-type rod photoreceptors (light exposure group; n = 13, average = 65.2 and control group; n = 11, average = 66.1, p = 0.269). In addition we have shown that morpholino knock down of mislocalized rhodopsin rescues ovl photoreceptor cell death [14]. These results strongly suggest that early mislocalized phototransduction leads to photoreceptor cell death.
Transducin a in phototransduction cascade is involved in rod photoreceptor cell death For further investigation of the role of the phototransduction cascade in rod photoreceptor cell death, we shut down phototransduction using a morpholino knock down of transducin in normal light cycle condition. We used anti-transducin a morpholinos which effectively down-regulated transducin a transcription at 108 hpf ( Figure 1I). Suppression of transducin a significantly increased the number of surviving rod photoreceptor cells in the morpholino-treated ovl retina 108 hpf (n = 26, average = 21.3), compared to the control (n = 21, average = 15.1, p,6.68E-4). Over 35% of the rods survived in transducin a morpholino-treated animals ( Figure 1J-L). On the other hand, in wild-type animals the number of surviving rod photoreceptor cells is equivalent (p = 0.265) between morpholino treated (n = 23, average = 32.7) and control (n = 25, average = 33.4) animals (data not shown). TUNEL analysis revealed that apoptosis in outer nuclear layer in ovl is decreased by transducin a morpholino treatment ( Figure S3). These results indicate that transducin a, in addition to rhodopsin, is involved in the rod photoreceptor cell death pathway in ovl.

Suppression of phosphodiesterase 6b does not affect the rod photoreceptor cell death
We then tested the contribution of the rod cGMP-phosphodiesterase b subunit (PDE6b), the step following transducin during phototransduction in rod photoreceptor cell death. Injection of PDE6b morpholino effectively suppressed PED6b expression ( Figure 1M), however, it did not affect the number of surviving rod photoreceptor cells in the morpholino-treated ovl retina (n = 20, average = 19.1, Figure 1N) at 108 hpf, compared to control morpholino-treated mutants (n = 21, average = 18.9; Figure 1O, p = 0.475, Figure 1P). Similarly, in wild-type animals, the number of surviving rod photoreceptor cells is equivalent (p = 0.0603) between morpholino treated (n = 22, average = 45.7) and control animals (n = 31, average = 40.1) at 108 hpf. These results suggest that the pathway of rod photoreceptor cell death passes mainly thought transducin.
Rod photoreceptor cell death mechanism operates through adenylyl cyclase A sequential knockdown experiment suggests that the mislocalized phototransduction signal is connected to rod photoreceptor cell death signal through transducin, but how? Phototransduction is a typical G-protein coupled signaling pathway.
Rhodopsin is a G-protein coupled receptor. Transducin is a Gt type GTP-binding protein, and PDE is its effecter. We hypothesized that an effecter molecule other than PDE miscouples to transducin because of the ectopic expression of photopigments. Adenylyl cyclase (ADCY) is a candidate, because it is reported that ADCY 1, 2, and 4 are expressed in eye [18,19] and we confirmed ADCY2 is expressed in the inner segments of photoreceptor cells, and is not expressed in the outer segments ( Figure 3C). Alfinito et al. has proposed this mechanism using saramander photoreceptors in vitro [20]. To confirm this hypothesis, we applied both the agonist and antagonist of ADCY to ovl. SQ 22536, a cell-permeable adenylyl cyclase inhibitor, reduced rod photoreceptor cell death in ovl in a dose dependent manner. At 10 mM, SQ 22536 increased surviving rod photoreceptors in ovl retina by 38% (Figure 2A-E). On the other hand, in the wild-type retina, the ADCY inhibitor had no effect on photoreceptor survival. These results suggest that ADCY is an entry point to the rod photoreceptor cell death signaling pathway through phototransduction.
Forced expression of adenylyl cyclase 2B in the outer segments leads to rod photoreceptor cell death For further confirmation of the role of ADCY in rod photoreceptor cell death, we ectopically expressed ADCY in wild-type outer segments. Expression of ADCY [21][22][23] in the eye [19] has been reported. We first tested the expression of ADCY subtypes in eye. Among transmembrane ADCY subtypes, ADCY2B, 6, 8 and 9 were expressed in eye (result of RT-PCR of ADCYs is shown in Figure 3A), ADCY2 is also detected in eyes by western blotting ( Figure S4) and immunohistological analysis showed that ADCY was expressed in photoreceptors and localized in the inner segments ( Figure 3C) as reported [24]. To ectopically express ADCY in the outer segments, we used a fusion protein with rhodopsin C-terminus amino acids, because the rhodopsin Cterminus is essential for the apical projection of rhodopsin [25,26], and the sequence had been used in some experiments for ectopic fusion protein projection to the outer segments [27]. We generated a transgenic ADCY2B fish with the rhodopsin C-terminus tail driven by zebrafish rhodopsin promoter, named ''tail (+)'', and an ADCY2B fish without the rhodopsin tail named ''tail (2)'' ( Figure 3D).
Expression of ADCY2B in ADCY transgenic tail (+) and tail (2) fish was confirmed by RT-PCR ( Figure 3B). Sub-cellular localization of ectopically expressed ADCYs was confirmed by immunohistochemistry. In tail (+) fish, positive ADCY staining was detected in rod photoreceptor outer segments as well as in the inner segments by anti-adenylyl cyclase 2 antibody ( Figure 3E, F). On the contrary, in tail (2) fish ( Figure 3E), adenylyl cyclase was not detected in the outer segments, but was detected in the inner segments similar to wild-type expression ( Figure 3C). We also confirmed that even in ADCY tail (+) fish, the transport of rhodopsin is not significantly affected ( Figure S5). Thus, we were able successfully to express ADCY2B in the outer segments using the rhodopsin tail.
The reduction of rod photoreceptor cell number in tail (+) fish was due to rod photoreceptor cell death, as apoptotic cell death was observed in only the tail (+) retina. The number of apoptotic photoreceptor cells in the adenylyl cyclase 2B tail (+) transgenic fish retina significantly increased (n = 8, average = 7.4), compared with that in transgenic rhodopsin tail (2) fish (n = 5, average = 0.40) at 14 dpf (p,1.36E-5, Figure 3J-M). Furthermore, as we expected, this rod photoreceptor cell death is light dependent. Mild light exposure increased the number of rod photoreceptors in ADCY tail (+) retina at 7 dpf ( Figure S6, darkness group, average = 60.6; light group, average = 31.2; p = 4.27E-9). On the contrary, there is no significant difference in the number of cone photoreceptors between wild type and tail (+) fish ( Figure S7, wild type, average = 97.3; tail (+), average = 101.8; p = 0.255). These results indicate activation of adenylyl cyclase coupled with mislocalization of phototransduction accelerates rod photoreceptor cell death.

cAMP-related rod photoreceptor cell death
Activation of adenylyl cyclase generates cAMP at the inner segments of rods. Our next experiment investigated whether cAMP is related to rod cell death by using 8-Bromo-cAMP. 8-Bromo-cAMP is a cell permeable cAMP analog that we applied to ovl embryos. As we expected, the number of rod photoreceptor cells showed significant reduction (p,4.09E-5) between the 10 mM cAMP treated group (n = 22, average = 14.5) and a control group of ovl mutants (n = 36, average = 25.4) at 5 dpf ( Figure 4A-F). Because the thickness of the inner nuclear layer (INL) was not affected significantly by cAMP treatment ( Figure S8, control group, n = 5, average = 30.2 mm; cAMP treated group, n = 5, average = 29.8 mm; p = 0.298), so the decreased number of rod photoreceptors was not due to delay caused by the drug. On the other hand, 8-Bromo-cGMP did not show any effect on rod photoreceptor cell numbers in ovl mutants (p = 0.282; Figure 4G-I). These results show that the elevation of intracellular cAMP levels induces rod photoreceptor cell death.
PKA activation is an important downstream effect of cAMP induction. KT5720, an inhibitor of PKA, reduces rod photoreceptor cell death in ovl. At 0.1 mM, KT5720 increased surviving rod photoreceptors by 32% (p,0.0326) in the ovl retina ( Figure 4J-L). These results indicate that cAMP plays a role in rod photoreceptor cell death. We also tested CREB phosphorylation, one of the main targets of PKA in ovl animals. We collected 80 eyes of ovl and wt animals and performed western blot by anti phosphorylated CREB (pCREB) and b-actin (We have tried blot of anti-CREB with two different antibodies, however, they did not work. So we used b-actin as a control). The pCREB level was increased 6.37% in ovl eye with normalized by b-actin blot level ( Figure S9).

Human rhodopsin mutant transgenic fish
In ovl, ADCY plays a role in rod photoreceptor cell death via the phototransduction cascade. The essential questions are what is the photoreceptor cell death mechanism in ovl, and is it at work in human retinitis pigmentosa. To address these questions, we generated a transgenic fish with the human rhodopsin Q344X mutation. Q344X causes severe human RP and its transgenic rodents show similar photoreceptor degeneration [28][29][30]. As with the ADCY transgenic, we generated human rhodopsintransgenic fish with and without the Q344X mutation, under the control of the zebrafish rhodopsin promoter. Because we induced human rhodopsin, we could confirm the ectopic expression of human rhodopsin in the transgenic fish by RT-PCR ( Figure 5A). The transgenic fish expressed human rhodopsin only in the eye, whereas the wild-type did not express it at all. The Q344X mutation was confirmed by direct sequencing ( Figure 5B). Immunohistochemistry revealed that in the Q344X transgenic fish, rhodopsin was not only detected in the outer segments but also in the entire cell membrane of the rod photoreceptor cell, as reported in mice and humans [30] ( Figure 5J, K).
The number of rod photoreceptor cells was not significantly different in the wild-type human rhodopsin (no mutations) transgenic fish and rhodopsin Q344X mutation transgenic fish by 3 dpf (no mutations group; n = 23, average = 40.8, Q344X; n = 33, average = 46.5, p = 0.076, Figure 5C, D). However as the days passed, the Q344X retina showed photoreceptor degeneration. The number of rod photoreceptor cells in the Q344X retina at 5 (n = 9, average = 47.0, Figure 5F) and 7 dpf (n = 25, average = 27.4, Figure 5H) were significantly reduced compared with that in the unmutated rhodopsin transgenic fish at 5 (n = 16, average = 62.6, p,0.0280, Figure 5E) and 7 dpf (n = 14, aver-  Figure 5G, I). We confirmed by TUNEL assay that the degeneration in Q344X transgenic fish was due to apoptosis. The number of apoptotic photoreceptor cells in the rhodopsin Q344X transgenic fish retina (n = 8, average = 2.6) was significantly larger, compared with that in the wild-type retina (n = 8, average = 0.6) 5 dpf (p,0.0193, Figure 5L-N). Like ovl cones do not decrease in Q344X transgenic fish at 5 (p = 0.390), 7 dpf (p = 0.206) and even 1.5 mpf, ( Figure S10, 11). These results demonstrate that the Q344X transgenic fish mimics human RP with the rhodopsin mutation.
Then, we tested the mechanism of rod photoreceptor cell death in the Q344X mutant. As shown in ovl, the number of rod photoreceptor cells showed a significant difference (p,1.13E-4) between the continuous light exposure (n = 21, average = 7.6) and control groups (n = 13, average = 20.5) in Q344X transgenic fish at 5 dpf ( Figure 6A-C).
Morpholino knock down of transducin a significantly (p,0.0159) increased the number of surviving rod photoreceptor cells in the transducin morpholino-treated group (n = 15, average = 34.7) at 5 dpf, compared to the control morpholino-treated group (n = 14, average = 25.9) in the Q344X transgenic fish ( Figure 6D-F). Over 33% of rods survived in transducin a morpholino-treated animals.
As in ovl, suppression of PDE6b did not rescue rod photoreceptor cell death in the Q344X transgenic. Injection of a PDE6b morpholino reduced the number of surviving rod photoreceptor cells in the Q344X mutant. The number of rod photoreceptor cells in the PDE6b morpholino-treated Q344X transgenic retina at 5 dpf was 30.5 (n = 15), whereas it was 50.6 (n = 14) in the control morpholino-treated Q344X transgenic retina (p = 7.9E-4; Figure 6G-I).
Rod photoreceptor cell death in the Q344X transgenic was also suppressed by ADCY antagonist in a dose dependent manner. Application of 100 mM of SQ22536 significantly (p,0.0179) increased the number of surviving rod photoreceptor cells (n = 19, average = 44.1) in the Q344X retina at 5 dpf, compared to controls (n = 13, average = 34.5). Concentrations of 10 and 20 mM of SQ 22536 did not significantly increase the  Figure 6J-L). These results suggest that ADCY plays a role in rod photoreceptor cell death in Q344X rhodopsin mutant, like ovl.

ADCY suppression rescues photoreceptor cell death in mammalian RP model
The results of Q344X suggest that the ADCY plays a role in photoreceptor cell death in human RP. However, we still doubted that the phenomenon occurs only in fish. So we used a more common RP rodent model, rd10 mice [31]. Mislocalized rhodopsin is also reported in rd10 mice [31]. Mice homozygous for the rd10 mutation show histological changes at postnatal day 16. Rd10 mice have a point mutation in exon 13 of beta PDE [32,33] and ectopic rhodopsin is found in the IS in photoreceptor cells. Rd10 mice have less PDE protein than the wild-type in the retina [34]. We suspected that mislocalized phototransduction easily occurs in the Rd10 retina, but actually, photoreceptor degeneration depends on light exposure [34]. We applied SQ22536, an adenylyl cyclase inhibitor, to rd10 mice. A 10 mM/ 2 mL solution of SQ22536 was injected into the vitreous of one eye. We injected 2 ml PBS into the other eye as a control. The thickness of the retinal outer nuclear layer (ONL) was measured at 6 retinal points in each eye, and we calculate average of 6 points thicknesses. We injected SQ22536 or PBS at P18 and sectioned at P28. At 10 mM, SQ 22536 increased the ONL thickness by 29% in the rd10 mice retina. The thickness of the ONL in the rd10 mouse retina of the SQ22536 treated group significantly increased (n = 5, average = 16.2 mm, p,0.0314, Figure 7A), compared with the control group (n = 5, average = 12.6 mm, Figure 7B, C).  Figure 7D). The results using ovl, Q344X transgenic, and rd10 mice indicate that adenylyl cyclase may be commonly involved the photoreceptor cell death pathway in human retinitis pigmentosa.

Discussion
In the eye, the photoreceptor is the light detector, and the relationship between light and photoreceptor death has been discussed [35][36][37]. RP is a slow progressive disease and most patients are concerned about light exposure accelerating the progression of the disease.
Hao et al. [38] revealed that, in a mutant mouse model, phototransduction has an apparent relationship to photoreceptor cell death. They also reported that photoreceptor cell death is caused by even relatively low-level light exposure, which is not harmful for wild-type photoreceptors. The cell death mechanism is different from that caused by strong light exposure in the RP retina, but the details are not yet known.
In photoreceptor cell death in rhodopsin mutant, some investigators suggest that transducin is involved [39], and the others suggest not [40]. Our results of genetic manipulation and antagonist experiments prefer the former hypothesis in rod photoreceptor cell death caused by relatively low-level light exposure. In our experiment, PDE does not affect the rod photoreceptor cell death in ovl. However, it is possible that the knockdown of PDE causes rod generation in mammals. So we could not conclude that PDE is not involved in photoreceptor cell death caused by light, however, our results suggest that transducin is included in rod photoreceptor cell death caused by light.
Alfinito [20] first proposed that ADCY is related to photoreceptor cell death, however, they used digested and unhealthy salamander photoreceptors in vitro, and only used antagonists in their experiments, so it had been still unknown whether the same holds true for human RP. In this report we used several lines of investigation that shows ADCY is involved in rod photoreceptor cell death. This adenylyl cyclase mechanism is seen not only in ovl, but also in the human rhodopsin mutation transgenic Q344X, and a mouse model. So it might be shown in human RP.
Cyclic AMP is a major second messenger with various biological effects including apoptosis, which it can in some cases induce and in others reduce. Some reports demonstrated that in diffuse large B-cell lymphoma cells, cAMP induces cell death via the phosphatidylinositol 3-kinase (PI3K)/AKT pathway [41], and in WT S49 T-lymphoma cells, cAMP induces cell death via Bim [42]. The pathway from cAMP to PKA is also known as a cell death factor in coronary endothelial cells via mitochondrial Bax and caspase3 [43]. In the retina of rhodopsin mutant transgenic mice, elevation of the cAMP level was reported [44]. In our results, cell permeable cAMP, not cGMP, could induce rod photoreceptor cell death. These results strongly suggest that elevation cAMP seems to play an important role in some rod photoreceptor cell death pathways. We also found that PKA inhibitor KT5720 partially rescued rod photoreceptor cell death in ovl, and the level of pCREB is increased in ovl eye. However, we are still wondering that the role of PKA, especially CREB phosphorylation level in rod photoreceptor degeneration. First, total CREB level is unknown, because the antibodies did not work in zebrafish. So it is possible that if the CREB translation level itself is increased, the pCREB level is also increased without PKA activation. Second, the down-regulation of pCREB level in photoreceptor degeneration is reported. In rds mice retina cAMP level is abnormally high [45], but CREB activation and expression is downregulated [46]. Interestingly, in rd1 not only PKA but also other CREB kinases (Akt [47], PKC [48], CaMKII [49], ERK1/2 [50]) are activated. Increased activity of CREB kinases in spite of downregulation of CREB activity might indicate that CREB activity might be strongly suppressed under the kinases activation. Inducible cAMP early repressor (ICER) is thought to be a candidate modifier, because it works as endogenous CREB antagonist, which might be dependent on PKA activity [51]. The role of PKA and CREB in photoreceptor cell death should be examined further more.
Here, we demonstrated a mechanism of rod photoreceptor cell death caused by light in vivo ( Figure 7E). In this mechanism, ectopic phototransduction is connected to ADCY via transducin leading rod photoreceptor cell death. It was observed in zebrafish mutants, human rhodopsin mutant transgenic fish, and rd10 mice. However, RP is a group of diseases, and response to light exposure is quite different among models. For example, our mammalian model, rd10 is the pde6e mutant mouse. The mutation can cause photoreceptor cell death but the mechanism might be independent to ADCY pathway we have demonstrated. This is the reason why photoreceptor rescue is weak in rd10 mice by shutdown of ADCY pathway. On the contrary, even in ovl, ADCY pathway is not only the cause of photoreceptor cell death, because complete darkness treatment could not show the perfect rescue. So, the cell death mechanism presented here cannot explain all mechanisms of photoreceptor cell death. It is possible that the mechanism is not a main pathway in some patients. However, we still think it is an important mechanism. Because RP is a late onset and slowly progressive disease, patients usually have no obvious symptoms in childhood with no complains. In our research, the average age when patients were diagnosed with RP is 35 years old [52], and visual loss progresses very slowly leading to blindness. So prevention of any pathway of rod photoreceptor cell death would be a great help in maintaining a patient's quality of life. However, to date, there are no effective cell protection therapies. The cell death pathway that we have demonstrated includes novel target molecules, adenylyl cyclase and PKA, and may be helpful in developing an effective drug treatment for RP.

Fish Strains
The maintenance and breeding of zebrafish strains and staging of embryonic development were performed as described [53,54]. The ovl allele, originally recovered in a large-scale mutagenesis screen [55], was initially characterized in a previous study [56]. RH1upEGFP fish is a gift from Dr. Kawamura [57]. In this line, rod photoreceptors were visualized with EGFP driven by zebrafish rhodopsin promoter (RH1).

Rd10 mice
The P18 inbred rd10 mice were maintained humanely, with proper institutional approval, and in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All animal experiments were carried out under approved protocols and in accordance with the recommendations for the proper care and use of laboratory animals by the Committee for Animals, and Infectious Pathogens Experiments at Kyushu University (Approval ID: A22-070-0) and according to The Law (No. 105) and Notification (No. 6) of the Japanese Government. In the sections below, n is used to indicate the number of treated eyes.

Molecular biology (PCR and RT-PCR)
PCR was carried out in a 50 ml reaction mixture containing 100 ng of genomic DNA, 50 pmol of each primer, 2.0 mM MgCl2, 16 reaction buffer (Takara), 200 mM of each dNTP, and 1.0 U of Ex TaqH polymerase (Takara). Samples were amplified for 30 cycles, 30 s at 94uC, 30 s at 60uC, and 30 s at 72uC.
Total RNA was extracted from embryos using ISOGEN (Wako) and prepared for cDNA synthesis using SuperScriptHII transcriptase (Invitrogen). Reverse transcription was performed in a 20 ml reaction volume, containing ,1 mg total RNA, 500 ng Oligo(dT)12-18, 500 mM of each dNTP, 10 mM DTT, and 50 units of SuperScriptHII (Invitrogen) at 42uC for 50 min. PCR conditions were essentially the same as in the genomic PCR described above except the annealing temperature was changed from 58uC to 62uC, and the extension time was changed from 30 s to 5 min. PCR was performed with gene-specific primers. PCR products were purified using the QIAquickH PCR Purification Kit (Qiagen) following the user manual and sequenced.

Morpholino Knock Down
Morpholino oligonucleotide knock downs were performed as described previously [54].

Histology
For histological analysis of zebrafish, embryos were fixed in 4% paraformaldehyde (PFA, w/v, pH 7.4) in PBST overnight at 4uC. Embedding, sections (with 16 mm thickness along the lens/optic nerve axis, only one section from only one side of eyes per embryo), and staining were performed as described previously [58,59]. So Ns appeared in text and figures mean the number of animals. The phenotype of the embryos was observed using a Zeiss Axioscope microscope or Bio-Rad Confocal Microscopy Radiance 2100 system. Images were recorded using digital cameras and processed with AxioVision (Carl Zeiss), LaserSharp2000 (Carl Zeiss) or Adobe Photoshop software (Adobe, Inc.). The number of rod photoreceptor cells was counted as fluorescent positive cells in each photo.

Transgenic Animals
Human rhodopsin Q344X transgenic fish: The tol2 transposon system was used to produce transgenic zebrafish with the rhodopsin Q344X mutation associated with autosomal dominant Retinitis Pigmentosa in humans.

Evaluation of Light Damage
Embryos, in light exposure experiments ( Figure 1H, 6C and S6), were reared on a light cycle of 14 hr of light exposure (300 Lux) and 10 hr of darkness (0 Lux) at 28uC until 72 hpf and subsequently subdivided into 2 groups that were treated with the following light exposure regimen until 120 hpf: constant light of 800 Lux or constant darkness (0 Lux). Other experiments were done in light cycle condition. Embryos were collected at appropriate time points, fixed, cryosectioned, and stained with anti-rod opsin and adenylyl cyclase 2 antibodies as above. Light intensity was measured using Model LX-1108 digital light meter (Lutron). A nonpaired t test was performed to determine the statistical significance of cell number changes.

Apoptosis assays
TUNEL (TdT-mediated dUTP Nick-End Labeling) assay: Apoptotic cell death was detected using the ApopTagH Fluorescein In Situ Apoptosis Detection Kit (Millipore) standard protocols.

Other Protocols
Retinal sections of rd10 mice The mice were sacrificed at P18, and the eyes were enucleated and fixed with ice-cold 4% paraformaldehyde in PBS. Twentyfour hours later, the samples were embedded in paraffin, and 5 mm thick sections along the pupil/optic nerve axis were examined under a light microscope. Neuroprotective effects were quantified by the thickness of ONL which was measured at three points (center and 1/4 distances from the ora serrata and the optic nerve head) in each of the nasal and temporal hemispheres of the eye. The sections were observed using OLYMPUS BX50 microscope system. Images were recorded using digital cameras and processed with AxioVision (Carl Zeiss) or Adobe Photoshop software (Adobe, Inc.).