Activation of JNK signaling promotes all-trans-retinal–induced photoreceptor apoptosis in mice

Disrupted clearance of all-trans-retinal (atRAL), a component of the visual (retinoid) cycle in the retina, may cause photoreceptor atrophy in autosomal recessive Stargardt disease (STGD1) and dry age-related macular degeneration (AMD). However, the mechanisms underlying atRAL-induced photoreceptor loss remain elusive. Here, we report that atRAL activates c-Jun N-terminal kinase (JNK) signaling at least partially through reactive oxygen species production, which promoted mitochondria-mediated caspase- and DNA damage-dependent apoptosis in photoreceptor cells. Damage to mitochondria in atRAL-exposed photoreceptor cells resulted from JNK activation, leading to decreased expression of Bcl2 apoptosis regulator (Bcl2), increased Bcl2 antagonist/killer (Bak) levels, and cytochrome c (Cyt c) release into the cytosol. Cytosolic Cyt c specifically provoked caspase-9 and caspase-3 activation and thereby initiated apoptosis. Phosphorylation of JNK in atRAL-loaded photoreceptor cells induced the appearance of γH2AX, a sensitive marker for DNA damage, and was also associated with apoptosis onset. Suppression of JNK signaling protected photoreceptor cells against atRAL-induced apoptosis. Moreover, photoreceptor cells lacking Jnk1 and Jnk2 genes were more resistant to atRAL-associated cytotoxicity. The Abca4−/− Rdh8−/− mouse model displays defects in atRAL clearance that are characteristic of STGD1 and dry AMD. We found that JNK signaling was activated in the neural retina of light-exposed Abca4−/−Rdh8−/− mice. Of note, intraperitoneal administration of JNK–IN-8, which inhibits JNK signaling, effectively ameliorated photoreceptor degeneration and apoptosis in light-exposed Abca4−/−Rdh8−/− mice. We propose that pharmacological inhibition of JNK signaling may represent a therapeutic strategy for preventing photoreceptor loss in retinopathies arising from atRAL overload.

Disrupted clearance of all-trans-retinal (atRAL), a component of the visual (retinoid) cycle in the retina, may cause photoreceptor atrophy in autosomal recessive Stargardt disease (STGD1) and dry age-related macular degeneration (AMD). However, the mechanisms underlying atRAL-induced photoreceptor loss remain elusive. Here, we report that atRAL activates c-Jun N-terminal kinase (JNK) signaling at least partially through reactive oxygen species production, which promoted mitochondria-mediated caspase-and DNA damage-dependent apoptosis in photoreceptor cells. Damage to mitochondria in atRAL-exposed photoreceptor cells resulted from JNK activation, leading to decreased expression of Bcl2 apoptosis regulator (Bcl2), increased Bcl2 antagonist/killer (Bak) levels, and cytochrome c (Cyt c) release into the cytosol. Cytosolic Cyt c specifically provoked caspase-9 and caspase-3 activation and thereby initiated apoptosis. Phosphorylation of JNK in atRAL-loaded photoreceptor cells induced the appearance of ␥H2AX, a sensitive marker for DNA damage, and was also associated with apoptosis onset. Suppression of JNK signaling protected photoreceptor cells against atRAL-induced apoptosis. Moreover, photoreceptor cells lacking Jnk1 and Jnk2 genes were more resistant to atRAL-associated cytotoxicity. The Abca4 ؊/؊ Rdh8 ؊/؊ mouse model displays defects in atRAL clearance that are characteristic of STGD1 and dry AMD. We found that JNK signaling was activated in the neural retina of light-exposed Abca4 ؊/؊ Rdh8 ؊/؊ mice. Of note, intraperitoneal administration of JNK-IN-8, which inhibits JNK signaling, effectively ameliorated photoreceptor degeneration and apoptosis in light-exposed Abca4 ؊/؊ Rdh8 ؊/؊ mice. We propose that pharmacological inhibition of JNK signaling may repre-sent a therapeutic strategy for preventing photoreceptor loss in retinopathies arising from atRAL overload.
Timely clearance of all-trans-retinal (atRAL) 2 released from photoactivated rhodopsin is very important for sustaining vision (1,2). Interruptions in the clearing of atRAL in outer segments of photoreceptors result in an accumulation of atRAL in both photoreceptors and the retinal pigment epithelium (RPE), which may cause typical manifestations of retinal dystrophy in patients with autosomal recessive Stargardt disease (STGD1) and dry age-related macular degeneration (AMD) (3,4). Moreover, several lines of evidence demonstrate that transient amassing of atRAL by delayed clearance from the retina is one of the key mechanisms in light-induced retinal degeneration (2,5,6). During the visual (retinoid) cycle, light-activated rhodopsin releases atRAL (1,7). Subsequently, atRAL is transported by ABC transporter 4 (ABCA4) or diffuses from the disk lumen to the cytoplasm of photoreceptor outer segments, followed by reduction of atRAL to all-trans-retinol, also known as vitamin A, by all-trans-retinol dehydrogenase 8 (RDH8) (8 -14). Therefore, the responsibility of ABCA4 and RDH8 in the visual (retinoid) cycle is to clear atRAL in the retina. Mice lacking Abca4 and Rdh8 genes (Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice) exhibit the primary features of STGD1 and dry AMD, such as photoreceptor loss and RPE dystrophy (2,5,6). These findings suggest a link between atRAL toxicity and photoreceptor/RPE degeneration. Previous studies have indicated that atRAL provokes apoptosis in the human RPE cell line ARPE-19 via mito- cro ARTICLE chondrial injury (15,16), Bax activation by phosphorylated p53 (17), DNA damage (17), and endoplasmic reticulum stress (15). Most recently, we also show that atRAL activates the NLRP3 inflammasome to induce cell death and stimulates caspase-3/ GSDME-mediated pyroptosis in ARPE-19 cells (4). In addition, Maeda and co-workers (17) report that atRAL activates Bax and elicits cell death in a murine 661W photoreceptor cell line.
The c-Jun N-terminal kinases (JNKs), which are members of the mitogen-activated protein kinase superfamily, are activated by stress stimuli, inflammatory cytokines, and growth factors (18 -24). Activation of JNK signaling is linked to apoptosis (23), and it involves the development of several kinds of degenerative diseases, including Alzheimer's disease and Parkinson's disease (25,26). However, whether aberrant aggregation of atRAL activates JNK signaling in photoreceptor cells and its potential relationship to photoreceptor loss remain unclarified. In this study, we further elucidated the mechanisms underlying photoreceptor degeneration in retinopathies correlated with disrupted atRAL clearance.

atRAL induces apoptosis in 661W photoreceptor cells via activating the mitochondria-dependent pathway
The health of 661W photoreceptor cells exposed to atRAL for 6 h was examined. atRAL decreased viability of 661W photoreceptor cells in a concentration-dependent manner (Fig.  1A). The IC 50 value for atRAL incubated with 661W photoreceptor cells for 6 h was 5.68 M. Additionally, treatment of 661W photoreceptor cells with atRAL for 6 h at concentrations of 5, 10, and 20 M caused significant decreases in cell viability of ϳ40.5, 80.5, and 92.4%, respectively. Based on the results of these cytotoxicity tests, 661W photoreceptor cells were treated for 6 h with 5 M atRAL in subsequent experiments. TUNEL staining, 6 h after incubating 661W photoreceptor cells with 5 M atRAL, revealed a significant increase in the number of apoptotic cells (Fig. 1, B and C). Imaging of mitochondrial membrane potential (⌬⌿m) using rhodamine-123 showed that the treatment of 661W photoreceptor cells with 5 M atRAL for 6 h significantly reduced the levels of ⌬⌿m (Fig. 1D). Immunoblot analysis indicated that atRAL stimulated a significant decline in the expression of antiapoptotic protein Bcl2 in lysates of 661W photoreceptor cells (Fig. 1, E and F). 661W photoreceptor cells treated for 6 h with 5 M atRAL or an equal volume of dimethyl sulfoxide (DMSO) serving as a vehicle control were fractionated to collect mitochondrial and cytosolic fractions for Western blot analysis of Bak and cytochrome c (Cyt c), respectively. The data manifested significant increases in protein levels of mitochondrial Bak (Fig. 1G) and cytosolic Cyt c (Fig. 1H) in atRAL-loaded 661W photoreceptor cells, suggesting that increased Bak protein may escape suppression by Bcl2 protein (27), leading to Bak activation and Cyt c release from mitochondria to the cytosol. Protein levels of cleaved caspase-9 and cleaved caspase-3 were found be clearly elevated in lysates of atRAL-loaded 661W photoreceptor cells (Fig. 1I). Immunofluorescence staining also showed that 5 M atRAL significantly increased protein levels of cleaved caspase-3 in 661W photoreceptor cells after 6 h of exposure (Fig. 1J). Overexpression of Bcl2 protein in 661W photoreceptor cells was done at a level that was very far above the level in control cells (Fig. 1K), and attenuated protein levels of cleaved caspase-3 in lysates of atRAL-loaded 661W photoreceptor cells (Fig. 1K). However, with such a huge overexpression of Bcl2 protein, we still observed a certain level of cleaved caspase-3 ( Fig. 1K) in atRALloaded 661W photoreceptor cells and only a marginal effect on restoration of cell viability (Fig. 1L), thus suggesting that the role of Bcl-2 decrease may not be crucial in this particular mitochondrial apoptotic process. Moreover, treatment with 20 M caspase inhibitor Z-VAD-FMK decreased protein levels of cleaved caspase-3 by ϳ28% in lysates of atRAL-loaded 661W photoreceptor cells (Fig. 1M) and increased cell viability by ϳ29% (Fig. 1N). Taken together, these findings indicate that atRAL stimulates an apoptotic cell death in photoreceptor cells through the mitochondria-dependent pathway.

atRAL activates JNK signaling in 661W photoreceptor cells
Analyses using Western blotting and confocal microscopy with immunofluorescence staining demonstrated that protein levels of phosphorylated JNK (p-JNK), an active form of JNK, were significantly up-regulated in lysates of 661W photoreceptor cells treated for 6 h with 5 M atRAL (Fig. 2, A and B), indicative of the ability of atRAL to activate JNK signaling in 661W photoreceptor cells. Given that the phosphorylation of c-Jun, a direct substrate of JNK, is an indicator of JNK activation, we also examined c-Jun activation status in 661W photoreceptor cells in response to atRAL by immunoblotting and immunofluorescence staining. As expected, protein levels of phosphorylated c-Jun (p-c-Jun), a downstream transcription factor directly activated by JNK, were clearly elevated in atRALtreated 661W photoreceptor cells (Fig. 2, C and D). Because JNK-IN-8 is considered as an efficient, specific, and irreversible JNK inhibitor that inhibits c-Jun phosphorylation (28), we used JNK-IN-8 to suppress JNK signaling for determining whether JNK activation by atRAL contributed to apoptosis of 661W photoreceptor cells. Immunoblot analysis demonstrated that JNK-IN-8 at a concentration of 1 M clearly decreased protein levels of p-c-Jun and cleaved caspase-3 in lysates of atRAL-loaded 661W photoreceptor cells (Fig. 2E). Notably, the results of the CellTiter 96 AQueous One-Solution Cell Proliferation assay (MTS) showed that treatment with 1 M JNK-IN-8 effectively rescued 661W photoreceptor cells from apoptosis caused by atRAL (Fig. 2F). Collectively, these findings disclose that atRAL evokes photoreceptor apoptosis at least in part via activating JNK signaling.

atRAL causes DNA damage response in 661W photoreceptor cells
Protein levels of ␥H2AX, a marker of DNA double-strand breaks, were significantly increased in 661W photoreceptor cells treated with 5 M atRAL for 6 h, as determined by immunoblotting and immunofluorescence staining (Fig. 3, A and B). Given that ataxia telangiectasia-mutated (ATM), a serine/ threonine protein kinase, is recruited, autophosphorylated, and activated at sites of DNA double-strand breaks (29,30), protein levels of phosphorylated ATM (p-ATM) in atRAL-loaded 661W photoreceptor cells were assayed using immunofluores-JNK activation by all-trans-retinal cence staining. As expected, the p-ATM fluorescence intensity was markedly elevated in 661W photoreceptor cells exposed to 5 M atRAL for 6 h (Fig. 3C). These results suggest that atRAL induces DNA damage response in photoreceptor cells. Furthermore, immunoblot analysis demonstrated that treatment with 1 M JNK-IN-8 significantly reduced protein levels of ␥H2AX in atRAL-loaded 661W photoreceptor cells (Fig. 3D).

Knockout of Jnk1 and Jnk2 genes rescues 661W photoreceptor cells from atRAL-induced apoptosis via blocking JNK signaling
To further ascertain the role of JNK activation in apoptosis caused by atRAL, Jnk1 and Jnk2 genes were knocked out in 661W photoreceptor cells using the CRISPR-CAS9 system. As expected, immunoblot analysis revealed a deficient expression of JNK protein in lysates of 661W photoreceptor cells lacking Jnk1 and Jnk2 genes (Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells) (Fig. 4A). Also by Western blotting, a significant decrease in protein levels of cleaved caspase-3, cleaved poly-ADP-ribose polymerase (PARP), and ␥H2AX (Fig. 4, B and C), and a deficiency of p-JNK, JNK, and p-c-Jun (Fig. 4, B and D) were observed in lysates of atRAL-loaded Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ versus wildtype (WT) 661W photoreceptor cells. Moreover, in contrast to the finding that atRAL stimulated a significant decline in protein levels of Bcl2 in atRAL-loaded 661W photoreceptor cells (Fig. 1, E and F), protein expression of Bcl2 was remarkably elevated in Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ over WT 661W photoreceptor experiments, and statistical analyses were performed by using Student's t test. **, p Ͻ 0.01. G, Western blotting was used to analyze protein levels of mitochondrial Bak in 661W photoreceptor cells incubated with 5 M atRAL for 6 h. Control cells were treated with vehicle (DMSO) alone. COX IV was used as a loading control. H, protein levels of Cyt c in the cytosol of 661W photoreceptor cells exposed to 5 M atRAL for 6 h were detected by Western blotting. Control cells were incubated with vehicle (DMSO) alone. GAPDH was used as a loading control. I, Western blot analysis of pro-caspase-9, cleaved caspase-9, and cleaved caspase-3 in lysates of 661W photoreceptor cells exposed to 5 M atRAL or vehicle (DMSO) alone for 6 h. GAPDH was used as an internal control. J, immunofluorescence staining of cleaved caspase-3 in 661W photoreceptor cells incubated for 6 h with 5 M atRAL or vehicle (DMSO) alone. Nuclei were stained blue with DAPI. Scale bars, 20 m. K, immunoblot analysis of Bcl2, Flag (Bcl2), and cleaved caspase-3 in lysates of Bcl2 overexpressing 661W photoreceptor cells incubated with 5 M atRAL or vehicle (DMSO) alone for 6 h, respectively. Cells transfected with vector pcDNA3.1 served as a control. GAPDH was used as a loading control. L, cell viability, 6 h after exposure of Bcl2 overexpressing 661W photoreceptor cells to 5 M atRAL or vehicle (DMSO) alone, was determined by MTS assay. Cells transfected with vector pcDNA3.1 were used as a control. M, immunoblot analysis of cleaved caspase-3 in 661W photoreceptor cells exposed to 5 M atRAL for 6 h in the absence or presence of 20 M caspase inhibitor Z-VAD-FMK. Note that cells were pretreated with 20 M Z-VAD-FMK for 1 h. Cells treated with Z-VAD-FMK or vehicle (DMSO) served as controls. GAPDH was utilized as an internal control. N, cell viability was probed by MTS assay. 661W photoreceptor cells were pretreated with caspase-3 inhibitor Z-VAD-FMK (20 M) for 1 h, followed by incubation with or without 5 M atRAL for 6 h. Control cells were cultured with atRAL or vehicle (DMSO) alone. Each value represents mean Ϯ S.D. (error bars) of three independent experiments, and statistical analyses were performed by using one-way ANOVA with Tukey's post-test. cells regardless of atRAL exposure (Fig. 4E). Conversely, exposure of Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells to 5 M atRAL for 6 h resulted in significant decreases in protein levels of mitochondrial Bak and cytosolic Cyt c (Fig. 4, F and G), revealing that Bak is the downstream factor that damages mitochondria after JNK activation. More importantly, the results of MTS assay showed that the viability of Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells, 6 h after exposure to 5 M atRAL, was clearly increased compared with that of WT 661W photoreceptor cells (Fig. 4H). These findings indicate that, after atRAL loading, inactivation of JNK signaling in 661W photoreceptor cells attenuates mitochondria-mediated caspase-/DNA damage-dependent apoptosis.

atRAL provokes reactive oxygen species (ROS) generation in 661W photoreceptor cells
Considering that ROS can induce apoptosis by activating JNK signaling (31), intracellular ROS generation was measured by flow cytometry and confocal laser-scanning microscopy after atRAL-loaded 661W photoreceptor cells were incubated with the fluorescent probe H 2 DCFDA. Exposure of 5 M atRAL to 661W photoreceptor cells for 6 h dramatically stimulated the production of intracellular ROS (Fig. 5, A-C), with an ϳ7.8fold increase over control cells (Fig. 5B). MitoSOX Red is a specific dye for mitochondrial ROS (32). After incubating atRAL-loaded 661W photoreceptor cells with MitoSOX Red, confocal microscopy was used to examine fluorescentlystained cells and revealed that ROS induced by atRAL were at least in part localized to mitochondria (Fig. 5D).

Inhibiting ROS generation protects 661W photoreceptor cells from apoptosis induced by atRAL via attenuating activation of JNK signaling
Treatment with antioxidant N-acetyl-L-cysteine (NAC) at a concentration of 2 mM effectively reduced ROS production in 661W photoreceptor cells after 6 h of exposure to 5 M atRAL (Fig. 6, A and B), and it promoted cellular survival (Fig. 6C) against apoptosis caused by atRAL. Western blot analysis indicated that 2 mM NAC significantly decreased the ratio of the p-JNK/JNK protein level and protein levels of cleaved   h, respectively. GAPDH was used as an internal control. C, protein levels of cleaved caspase-3, cleaved PARP, and ␥H2AX were normalized to those of GAPDH and presented as fold changes relative to vehicle (DMSO) controls. Ctrl, control. Data are presented as mean Ϯ S.D. (error bars) of three independent experiments, and statistical analyses were performed by using two-way ANOVA with Sidak's post-test. ***, p Ͻ 0.001. D, immunoblot analysis of p-c-Jun and c-Jun in lysates of WT and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells exposed to 5 M atRAL or vehicle (DMSO) for 6 h, respectively. GAPDH was used as an internal control. E, Western blot analysis of Bcl2 in lysates of WT and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells exposed to 5 M atRAL or vehicle (DMSO) for 6 h, respectively. GAPDH was used as an internal control. F, Western blotting was used to analyze protein expression of mitochondrial Bak in WT and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells treated with 5 M atRAL or vehicle (DMSO) for 6 h, respectively. GAPDH was used as an internal control. G, immunoblot analysis of Cyt c in the cytosol of WT and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells exposed to 5 M atRAL or vehicle (DMSO) for 6 h, respectively. GAPDH was used as an internal control. caspase-9, cleaved caspase-3, cleaved PARP, ␥H2AX, and p-c-Jun in lysates of atRAL-loaded 661W photoreceptor cells (Fig. 6, D-F). Moreover, treatment with 2 mM NAC resulted in a significant decrease in protein levels of Cyt c in the cytosol of atRAL-exposed 661W photoreceptor cells (Fig.  6G). These findings suggest that attenuation of JNK signaling in atRAL-exposed 661W photoreceptor cells by suppressing the ROS generation alleviates mitochondriamediated caspase-/DNA damage-dependent apoptosis and thereby increases cell viability.

Light-induced photoreceptor degeneration and apoptosis in Abca4 ؊/؊ Rdh8 ؊/؊ mice involves activation of JNK signaling
Previous studies have indicated that exposure of Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice featuring defective atRAL clearance to intense light increases atRAL levels in the retina and leads to severe photoreceptor degeneration (5,6). In this investigation, 2day dark-adapted 4-week-old C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice were exposed to light-emitting diode (LED) light with an intensity of 10,000 lx for 2 h and then were raised in the dark for 1, 3, and 5 days. Control C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice were normally raised in the dark for 7 days in the absence of light exposure. As expected, hematoxylin and eosin (H&E) staining showed that the morphology of the neural retina of light-exposed Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice was altered obviously and damaged severely, and manifested marked thinning of the photoreceptor layer, particularly the outer segment (OS) layer and outer nuclear layer (ONL) (Fig. 7A). Interestingly, TUNEL staining of the neural retina of Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice exposed to light displayed a significant increase in the number of apoptotic photoreceptor cells (Fig. 7B). Immunoblot analysis revealed a clear elevation of p-JNK protein levels in extracts from the neural retina of light-exposed Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice (Fig. 7C). In addition, protein levels of ␥H2AX and cleaved caspase-3, which function downstream of JNK signaling (Fig. 4B), were notably up-regulated in extracts from the neural retina of light-exposed Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice (Fig.  7C). On the contrary, C57BL/6J WT mice illuminated at 10,000 lx for 2 h exhibited no obvious photoreceptor degeneration and apoptosis (Fig. 7, A and B). Likewise, protein levels of p-JNK, ␥H2AX, and cleaved caspase-3 remain almost intact in extracts from the neural retina of light-exposed C57BL/6J WT mice (Fig. 7C). The results imply that photoinduced photoreceptor injury in Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice is at least in part associated with atRAL overload and JNK activation.

Discussion
AMD is broadly divided into two types: dry AMD and wet AMD (33,34). Wet AMD is characterized by the formation of choroidal neovascularization mediated by vascular endothelial growth factor (VEGF) (35,36). Previous studies indicate that suppressing JNK signaling reduces choroidal VEGF expression and neovascularization in a murine model of wet AMD (37), and hypoxia-induced JNK activation promotes retinal VEGF production and pathological angiogenesis in a murine model of retinopathy of prematurity (38). Dry AMD affects up to 90% of AMD cases, for which there is no effective cure. Photoreceptor degeneration is a common feature of STGD1 and dry AMD (39,40). However, the correlation of JNK activation with photore-

JNK activation by all-trans-retinal
ceptor loss in STGD1 and dry AMD is still unclear. Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice, a model of STGD1 and dry AMD characterized by disrupted clearance of atRAL, develop severe photoreceptor dystrophy (2). Exposure of Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice to 10,000 lx of fluorescent light leads to rapid accumulation of atRAL in the retina (6), and it provokes retinal damage and photoreceptor cell death (5). In this study, we showed that JNK signaling was activated in neural retina of Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice versus age-matched C57BL/6J WT mice after 10,000-lx LED light exposure for 2 h (Fig. 7C). Nevertheless, it is reported that JNK activation is associated with apoptosis (31). Evidently, these lines of investigation disclose that atRAL accumulating in photoreceptor cells may activate JNK signaling to cause apoptosis. With the use of in vitro cell-based assays, we confirmed that atRAL at the concentration of 5 M significantly induced apoptosis of 661W photoreceptor cells via activating JNK signaling (Figs. 1, 2, and 4). Moreover, JNK activation by atRAL stimulated mitochondria-mediated caspase-/DNA damage-depend- alone. Alternatively, 661W photoreceptor cells were pretreated with 2 mM NAC for 2 h and incubated with 5 M atRAL for 6 h. Subsequently, the cells were stained by 10 M H 2 DCFDA for 10 min at 37°C, and the levels of intracellular ROS were determined by flow cytometry. B, quantity of H 2 DCFDA-positive cells was assayed by flow cytometry and expressed as a percentage of the total cell population. Each value represents mean Ϯ S.D. (error bars) of three independent experiments, and statistical analyses were performed by using one-way ANOVA with Tukey's post-test. **, p Ͻ 0.01; ***, p Ͻ 0.001. C, cell viability was examined by MTS assay. 661W photoreceptor cells were pretreated with antioxidant NAC (2 mM) for 2 h, followed by incubation with or without 5 M atRAL for 6 h. Control cells were treated with atRAL or vehicle (DMSO) alone. Each value represents mean Ϯ S.D. (error bars) of three independent experiments, and statistical analyses were performed by using one-way ANOVA with Tukey's post-test. D, immunoblot analysis of p-JNK, JNK, pro-caspase-9, cleaved caspase-9, cleaved caspase-3, PARP, cleaved PARP, and ␥H2AX in lysates of 661W photoreceptor cells treated for 6 h with 5 M atRAL in the absence or presence of 2 mM NAC. Note that cells were pretreated with 2 mM NAC for 2 h. Cells treated with NAC or vehicle (DMSO) alone served as controls. GAPDH was utilized as an internal control. Molecular mass markers (kDa) are indicated to the left of the blots. E, ratios of p-JNK/JNK protein level and protein levels of cleaved caspase-9, cleaved caspase-3, cleaved PARP, and ␥H2AX were shown as fold changes relative to vehicle (DMSO) controls. Levels of each protein were normalized to those of GAPDH. Note that the normalization for band intensity of p-JNK or JNK against that of GAPDH was individually made before calculating the p-JNK/JNK ratio. Each value represents mean Ϯ S.D. (error bars) of three independent experiments, and statistical analyses were performed by using one-way ANOVA with Tukey's post-test. **, p Ͻ 0.01; ***, p Ͻ 0.001. F, immunoblot analysis of p-c-Jun and c-Jun in lysates of 661W photoreceptor cells incubated with 5 M atRAL for 6 h in the presence or absence of 2 mM NAC. Note that cells were pretreated with 2 mM NAC for 2 h. Control cells were treated with NAC or vehicle (DMSO) alone. GAPDH served as an internal control. G, Western blot analysis of cytosolic Cyt c in 661W photoreceptor cells exposed to 5 M atRAL for 6 h with or without 2 mM NAC. Note that cells were pretreated with 2 mM NAC for 2 h. Cells treated with NAC or vehicle (DMSO) alone served as controls. GAPDH was used as an internal control.

JNK activation by all-trans-retinal ent apoptotic cell death in 661W photoreceptor cells (Figs. 1-4).
We found that protein levels of p-JNK were almost intact in Bcl2-overexpressing 661W photoreceptor cells in response to atRAL versus atRAL-loaded vector-control cells (Fig. S1), suggesting that expression changes of Bcl2 protein do not affect atRAL-mediated JNK activation. Moreover, treatment with JNK-IN-8 increased protein expression of Bcl2 by ϳ24% in atRAL-loaded 661W photoreceptor cells (Fig. S2), implying that suppressing JNK signaling attenuates Bcl2 decrease. In Fig.  4E, we provided evidence that protein expression of Bcl2 was significantly elevated in atRAL-loaded Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ over WT 661W photoreceptor cells. Taken together, these findings reveal that Bcl2 functions downstream of JNK. In atRAL-loaded 661W photoreceptor cells, increased Bak protein (Fig. 1G) likely caused Bak activation and damage to mitochondria, thereby eliciting the release of Cyt c into the cytosol (Fig. 1H). Previous evidence has revealed that cytosolic Cyt c functions as a trigger for caspase-9 activation, and the active form of caspase-9 (cleaved caspase-9) cleaves/activates caspase-3, followed by activated caspase-3-mediated cleavage of PARP (41)(42)(43). As expected, we observed that protein levels of cleaved caspase-9, cleaved caspase-3, and cleaved PARP were significantly increased in lysates of atRAL-loaded 661W photoreceptor cells (Figs. 1, I and J, 4B, and 6D).
Phosphorylation of H2AX at the sites of DNA damage is a very early step in the cellular response to DNA double-strand breaks (44,45). Either JNK or ATM involves the generation of phosphorylated H2AX (␥H2AX) (46,47). From the results shown in Figs. 2 and 3, increased protein levels of p-JNK, ␥H2AX, and p-ATM were clearly detected in atRAL-loaded 661W photoreceptor cells, implying that activation of JNK and ATM contributes to DNA damage. Moreover, protein levels of ␥H2AX were significantly decreased in atRAL-loaded Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ over WT 661W photoreceptor cells (Fig. 4, B and C), further corroborating that JNK activation by atRAL evokes DNA damage response.
After incubating with 661W photoreceptor cells, atRAL induced a marked increase in levels of ROS (Fig. 5, A-C), partially generated inside mitochondria (Fig. 5D). As reported previously, ROS are involved in activation of JNK signaling (48). Indeed, we showed that treatment with NAC, a general ROS inhibitor, attenuated JNK activation in atRAL-loaded 661W photoreceptor cells (Fig. 6, D and E). It was worth noting that Four-week-old C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice were exposed to LED light with an intensity of 10,000 lx for 2 h after their pupils were dilated with 1% tropicamide. Eyeballs, 1, 3, and 5 days after light exposure, were harvested for subsequent studies. C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice raised normally in the dark for 7 days in the absence of 10,000 lx LED light served as controls. A, histology of mouse retina was examined by H&E staining. Scale bars, 20 m. B, apoptotic cells in mouse retina were detected by TUNEL staining. Scale bars, 20 m. C, immunoblot analysis of p-JNK, JNK, ␥H2AX and cleaved caspase-3 in extracts from mouse neural retina. GAPDH was used as an internal control. Molecular mass markers (kDa) are indicated to the left of the blots. D, genotyping of Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice on a C57BL/6J genetic background was carried out by PCR amplification of tail DNAs with specific primers. Abca4 deletion in Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice was identified by producing a 455-bp amplicon with the A0 and N1 primers, yet C57BL/6J WT mice generated a 619-bp amplicon with the ABCR1 and ABCR2 primers. Rdh8 deletion in Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice was corroborated by giving rise to a 450-bp amplicon with the Neo1 and DMR11 primers, whereas C57BL/6J WT mice produced an 800-bp amplicon with the DMR5 and DMR6 primers. To further confirm the absence of Rd8 mutation of the Crb1 gene in C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice used in the study, DNA samples extracted from mouse tail biopsies were amplified for Wt allele and mutant Rd8 allele using primers mCrb1 mF1, mCrb1 mF2, and mCrb1 mR. C57BL/6N mice carrying Rd8 mutation in Crb1 gene served as a control. Amplicon sizes were 220 bp for Wt allele and 244 bp for Rd8 allele. A 100-bp DNA ladder was used as a size standard. OS, outer segment; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer.

JNK activation by all-trans-retinal
NAC only partially rescued 661W photoreceptor cells from apoptosis caused by atRAL (Fig. 6C), probably because NAC limitedly reduces the ROS production (Fig. 6, A and B), and intracellular generation of ROS by atRAL may not be the only way for JNK activation, DNA damage, and apoptosis. Moreover, nonapoptotic processes may also involve the death of 661W photoreceptor cells induced by atRAL. Treatment of atRALloaded 661W photoreceptor cells with NAC prevented the release of Cyt c into the cytosol (Fig. 6G), indicating that the ROS production is an upstream factor of the mitochondria-dependent apoptotic pathway.
Double-knockout of Jnk1 and Jnk2 genes in 661W photoreceptor cells also partially alleviated apoptosis caused by atRAL through reduction in protein levels of cleaved caspase-3, cleaved PARP, and ␥H2AX (Fig. 4, B, C and H), reflecting that JNK activation functions upstream of caspase-3 activation and PARP cleavage and serves as an initiator for DNA damage. In addition, increased expression of Bcl2 protein and reduced protein levels of mitochondrial Bak and cytosolic Cyt c were observed in atRAL-loaded Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells (Fig. 4, E-G), which reveal that the deletion of JNK protein inhibits the mitochondrial events linked to apoptotic cell death.
Each murine retina is estimated to contain 4.62 mM rhodopsin in disk membrane and 8.23 mM with respect to the rod outer segment cytoplasm (49). Because bleaching of rhodopsin by light generates an equivalent amount of atRAL (50), the concentration of atRAL used in our experiments could be physiologically significant.
Treating atRAL-loaded 661W photoreceptor cells with JNKspecific inhibitor JNK-IN-8 significantly enhanced cell survival via inhibiting JNK activation and its downstream effectors caspase-3 (Fig. 2, E and F) and ␥H2AX (Fig. 3D). More importantly, the results of animal experiments showed that intraperitoneal injection of JNK-IN-8 effectively alleviated photoreceptor degeneration and apoptosis by suppressing JNK signaling in Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice after light exposure (Fig. 8). Therefore, pharmacological inhibition of JNK signaling may ameliorate atRAL-related retinopathies, such as STGD1 and dry AMD, through interference with apoptotic photoreceptor cell death.
Considered collectively, our data suggest that atRAL promotes mitochondria-mediated caspase-/DNA damage-dependent apoptosis by triggering activation of JNK signaling in photoreceptor cells. A plausible mechanism for these findings is depicted in Fig. 9. atRAL accumulating in photoreceptor cells activates JNK signaling at least partially by ROS (including mitochondrial ROS) and reduces ⌬⌿m. JNK activation by atRAL decreases Bcl2 protein expression and produces increased Bak protein that may cause Bak activation, thereby adversely affecting mitochondrial integrity. Cyt c is liberated from ruptured mitochondria into the cytoplasm where it stimulates the formation of apoptosomes consisting of Cyt c, caspase-9, and Apaf-1 (51,52). This complex initiates caspase-9 and caspase-3 cascade activation to provoke apoptotic cell death. Alternatively, the active form of JNK (p-JNK) in the cytosol partially translocates into the nucleus where it phosphorylates H2AX to produce ␥H2AX, a marker of DNA doublestrand breaks, thus leading to apoptosis induced by DNA damage (53). This is the first report describing a relationship between atRAL-mediated apoptosis of photoreceptor cells and activation of JNK signaling. The results of this study suggest a

JNK activation by all-trans-retinal
key role of JNK signaling in photoreceptor dystrophy arising from atRAL accumulation beyond a threshold level and expand the understanding of atRAL-associated toxicity in photoreceptor cells.

Animals
C57BL/6J WT mice were purchased from the Xiamen University Laboratory Animal Center. Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice on a C57BL/6N genetic background were provided from the Krzysztof Palczewski Laboratory (Case Western Reserve University, Cleveland, OH) and contained the rd8 mutation in the Crb1 gene (54). However, Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice used in this study were maintained on a C57BL/6J genetic background (Fig.  8D), which were obtained by crossing Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice (C57BL/6N genetic background) with C57BL/6J WT mice. C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice (C57BL/6J genetic background) were free of the rd8 mutation in the Crb1 gene compared with C57BL/6N mice that carried the rd8 mutation of the Crb1 gene (Fig. 7D). Experiments with animals were approved by the Institutional Animal Care and Use Committee of Xiamen University School of Medicine. Mice were handled in strict accordance with good animal practice as defined by the In photoreceptor cells, atRAL evokes collapsed ⌬⌿m and JNK activation. The latter is at least partially mediated by the ROS (including mitochondrial ROS) production. Phosphorylation of JNK results in decreased protein expression of Bcl2 and mitochondrial damage through increased Bak protein that may cause Bak activation. Cyt c is released from ruptured mitochondria into the cytosol where it stimulates the formation of apoptosomes consisting of Cyt c, caspase-9, and Apaf-1 (51,52). Caspase-9 activation cleaves and activates caspase-3 to induce apoptosis. However, p-JNK, the active form of JNK, partially translocates into the nucleus where it phosphorylates the histone H2AX to form ␥H2AX, a sensitive marker for DNA double-strand breaks. Generation of ␥H2AX, which localizes to sites of DNA strand breaks, accounts for the onset of irreversible DNA damage correlated with apoptotic cell death (53). It should be mentioned that the ROS production caused by atRAL is one of the initiators to activate JNK signaling, and JNK activation is also not the only way for atRAL-stimulated DNA damage.

JNK activation by all-trans-retinal
Chinese Association for Research in Vision and Ophthalmology (CARVO), and maintained and bred under specific pathogen-free conditions with a 12-h light/12-h dark schedule at the Xiamen University Laboratory Animal Center. 48-h darkadapted C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice at 4 weeks of age were exposed to LED light with an intensity of 10,000 lx for 2 h after their pupils were dilated with 1% tropicamide. Finally, the mice, 1, 3, and 5 days after light exposure, were euthanized, and their eyeballs were collected for subsequent studies. C57BL/6J WT and Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice raised normally in the dark for 7 days in the absence of 10,000 lx LED light served as controls. Alternatively, 48-h dark-adapted Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice at 4 weeks of age were administered intraperitoneally JNK-IN-8 or vehicle (DMSO) at a dose of 4 mg/kg body weight. 1-h later, pupils of mice were dilated with 1% tropicamide followed by exposure of JNK-IN-8-or vehicle (DMSO)treated mice to LED light with the intensity of 10,000 lx for 2 h. After once-daily treatment with JNK-IN-8 or vehicle (DMSO) (4 mg/kg body weight) for 4 days, the mice, at day 5 after light exposure, were euthanized, and their eyeballs were collected for subsequent studies. Control Abca4 Ϫ/Ϫ Rdh8 Ϫ/Ϫ mice were injected intraperitoneally with JNK-IN-8 or vehicle (DMSO) (4 mg/kg body weight) in the absence of light exposure.

Treatment with atRAL
661W photoreceptor cells were seeded into 96-well plates at a density of 1.5 ϫ 10 4 cells per well and cultured overnight. Cells were then incubated with serial concentrations of atRAL (2.5, 5, 10, and 20 M) for 6 h.

Treatment with NAC, JNK-IN-8, and Z-VAD-FMK
661W photoreceptor cells seeded into 96-well plates at the density of 1.5 ϫ 10 4 cells per well or 6-well plates at a density of 3 ϫ 10 5 cells per well were cultured overnight, and pretreated with 2 mM NAC for 2 h, 1 M JNK-IN-8 for 1 h, or 20 M Z-VAD-FMK for 1 h. Cells were then treated for 6 h with or without 5 M atRAL.

Construction of Jnk1 ؊/؊ Jnk2 ؊/؊ 661W photoreceptor cell line
To generate the Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cell line, target sequences in the third exon of mouse Jnk1 (ACCG-AGAACTAGTTCTTATG) and target sequences in the second exon of mouse Jnk2 (CAACTGAAACCGATCGGCTC) were designed at Massachusetts Institute of Technology (56). The oligonucleotides for specific guide RNAs (gRNAs) are displayed in Table 1 and synthesized by Sangon Biotech (Shanghai, China). The synthesized gRNA primers were annealed to form a gRNA duplex and cloned into the BsmbI site of pL-CRISP-R.EFS.GFP vector purchased from Addgene (Watertown, MA). 661W photoreceptor cells seeded into 6-well plates were transfected with the targeting vector using Lipofectamine LTX & PLUS TM reagent. 24 h hours post-transfection, GFP-positive single cells were sorted into 96-well plates by a MoFlo Astrios flow cytometry (Beckman Coulter; Brea, CA). Two weeks later, single colonies were expanded into 6-well plates. Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells were identified by Western blotting and genome sequencing. The primers for amplification of the genomic areas designed for Jnk1 and Jnk2 are shown in Table 2. Each PCR product was cloned with pMD TM 19-T vector cloning kit, and transformed into E. coli DH5␣. The colonies were sequenced using M13 reverse primers. WT and Jnk1 Ϫ/Ϫ Jnk2 Ϫ/Ϫ 661W photoreceptor cells seeded into 96-or 6-well plates were cultured overnight and exposed to 5 M atRAL for 6 h, respectively.

Fractionation of mitochondria and cytosol
Cell fraction was performed as described previously (55). Briefly, 661W photoreceptor cells were scratched and harvested in cold PBS. After washing twice with PBS, cells were resuspended in digitonin lysis buffer (75 mM NaCl, 1 mM NaH 2 PO 4 , 8 mM Na 2 HPO 4 , 250 mM sucrose, 190 g/ml digitonin). After 5 min on ice, cells were centrifuged for 5 min at 14,000 rpm at 4°C. The supernatants were used for detecting protein levels of Cyt c by immunoblotting. The pellets were resuspended in Triton lysis buffer (25 mM Tris-HCl (pH 8.0), 0.1% (v/v) Triton X-100), followed by analysis of mitochondrial Bak via Western blotting.

JNK activation by all-trans-retinal Cell viability
Cytotoxicity was assessed using MTs Assay, which was performed by adding 20 l of the CellTiter 96 AQueous One Solution Reagent (Promega; Madison, WI) directly to 96-well plates, incubating for 1 h, and then recording absorbance at 490 nm using a 1510 Multiskan GO spectrophotometer (Thermo Fisher Scientific; Vantaa, Finland). Cells treated with DMSO served as vehicle controls.

H&E staining
Mouse eyes were harvested and fixed in 4% paraformaldehyde at 4°C overnight prior to embedding in paraffin. Tissues were cut at a thickness of 8 m. After drying at 60°C for 1 h, sections were deparaffinized, rehydrated, and stained with H&E and visualized by a DM2500 microscope (Leica; Wetzlar, Germany).

Immunofluorescence
Cells were fixed with 4% paraformaldehyde for 15 min at 4°C and permeabilized by 0.2% Triton X-100 in PBS for 20 min at room temperature. Subsequently, cells were blocked with 2% BSA and incubated with primary antibodies (1:100 dilution) overnight at 4°C, followed by incubation with Alexa Fluor 594conjugated secondary antibodies (1: 200 dilution) for 1 h at room temperature. Slides were mounted in Vectashield with DAPI (Vector Labs; Burlingame, CA). The epifluorescence images were performed using an Olympus FV1000 confocal microscope (Tochigi, Japan).

TUNEL assay
Tissue sections or cells were fixed with 4% paraformaldehyde at 4°C for 15 min. Apoptosis was evaluated by TUNEL assay (Promega; Madison, WI). Slides were mounted in Vectashield with DAPI. Images were acquired using the confocal microscope, followed by counting and analysis of apoptotic nuclei.

Measurement of intracellular ROS
661W photoreceptor cells were treated with 5 M atRAL for 6 h, incubated with 10 M H 2 DCFDA for 10 min at 37°C, and then measured using a CytoFLEX flow cytometer (Beckman Coulter; Brea, CA). As an alternative, 661W photoreceptor cells were incubated with 5 M atRAL for 6 h, and stained with 10 M H 2 DCFDA or 5 M MitoSOX Red, a mitochondrial superoxide indicator, at 37°C for 10 min. Nuclei were labeled by Hoechst 33342. After washing with PBS, the cells were examined using the confocal microscope. 661W photoreceptor cells were pretreated with antioxidant NAC (2 mM) for 2 h, followed by incubation with or without 5 M atRAL for 6 h. The levels of intracellular ROS were detected by the Beckman Coulter flow cytometer. Control cells were treated with DMSO alone.

Rhodamine-123 staining
The fluorescent dye rhodamine-123, which accumulates within energized mitochondria, was used to assay ⌬⌿m in an apoptotic cell population. Live cells were stained with rhodamine-123 (10 g/ml) for 20 min at 37°C. After washing with PBS, the cells were imaged using the confocal microscope (Tochigi, Japan).

Western blotting
Cells and tissues were lysed in RIPA buffer supplemented with protease and phosphatase inhibitor mixture (Thermo Fisher Scientific; Rockford, IL). Protein concentrations were determined with BCA protein assay kit (Thermo Fisher Scientific; Rockford, IL). Equal quantities of protein were subjected to electrophoresis in 12% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Roche Applied Science, Mannheim, Germany). After blocking with 5% skim milk for 1 h at room temperature, the membranes were incubated with primary antibodies (1:1000 dilution) at 4°C overnight, followed by incubation with the corresponding secondary antibodies (1:5000 dilution) for 1 h at room temperature. Membranes were developed using ECL Western blotting detection reagents (Advansta; Menlo Park, CA), and blots were analyzed using a ChemiDoc XRSϩ Imaging system (Bio-Rad). Immunoreactive bands were quantified by densitometry using ImageJ software.

Genotyping
Genomic DNAs were isolated from mouse tail biopsies using TIANamp Genomic DNA kit (TIANGEN, Beijing, China) according to the manufacturer's instructions. DNA samples

JNK activation by all-trans-retinal
were amplified by PCR using specific primers (Table 4). To corroborate Abca4 deletion and Wt allele, samples 1 and 2 were amplified with the ABCR1 and ABCR2 primers, and samples 3 and 4 were amplified with the A0 and N1 primers. For the identification of Rdh8 deletion and Wt allele, samples 5 and 6 were amplified with the DMR5 and DMR6 primers, and samples 7 and 8 were amplified with the Neo1 and DMR11 primers. In contrast, genomic DNAs extracted from mouse tail biopsies were amplified for identifying Wt allele and rd8 mutation of the Crb1 gene using primers mCrb1 mF1, mCrb1 mF2, and mCrb1 mR.

Statistical analyses
All data were analyzed using GraphPad Prism software (Version 5.0; La Jolla, CA). The results were averaged from at least three independent experiments to calculate the mean Ϯ S.D. Statistical analyses were performed using one-way or two-way analyses of variance (ANOVA) followed by Tukey's or Sidak's multiple comparison test and Student's t test, as indicated in the corresponding figure legends. Significant levels retained were *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001.

Data availability
The data supporting the findings of this study are available within the article and the supporting information.