Main

Autophagy is an intracellular pathway where cytoplasmic constituents are delivered to the lysosomes for degradation. Defective autophagy can contribute to the age-dependent accumulation of damaged proteins and organelles leading to altered cellular homeostasis and loss of function.1, 2, 3, 4, 5 The metabolic roles of autophagy can be classified into two types, basal and induced. In nutrient-rich conditions, autophagy is suppressed but still occurs at low levels (basal autophagy); however, when cells are subjected to stress (starvation, injury, hypoxia), autophagy is activated immediately (induced autophagy).6 Induced autophagy maintains the amino acid pool inside cells to adapt to starvation while constitutive autophagy has been shown to function as a cell-repair mechanism that is important for long-lived postmitotic cells.7, 8, 9, 10, 11 Defects in autophagy have been associated with neurodegenerative diseases,12, 13, 14, 15 diabetes,16, 17 lysosomal storage disease18 and the loss of vision.19 In addition to macroautophagy, microautophagy and chaperone-mediated autophagy (CMA) have been described. Although little is known about microautophagy in mammalian cells, macroautophagy (hereafter autophagy) is a major pathway for bulk degradation of cytoplasmic components. CMA is a more selective pathway for degradation of cytosolic proteins that can compensate for the loss of macroautophagy.2, 20, 21, 22

Inherited retinal degenerative diseases such as retinitis pigmentosa or Leber’s congenital amaurosis are characterized by premature and progressive death of rod and cone photoreceptor cells.23 These diseases are characterized by the loss of night vision due to the death of rods followed by the loss of cones leading to diminished visual acuity and a reduction in the quality of life for patients. Disease is typically associated with the production of harmful gene products that promote pathology by inhibiting critical pathways resulting in cell death.24, 25, 26 Strategies to prevent photoreceptor death during retinal degenerative disease such as gene replacement therapies or inhibition of cell death pathways have been undertaken with some success;27, 28, 29 however, effective treatments for these blinding disorders are lacking.

Another strategy that could be used in conjunction with other therapies might be to enhance survival by stimulating autophagy. Augmenting autophagy would increase the supply of nutrients to stressed cells and accelerate removal of damaged proteins thereby prolonging cell survival beyond what would be possible by only preventing cell death. Although canonical22, 30, 31, 32, 33 and noncanonical autophagic mechanisms34 have been detected in the eye, our knowledge of basic autophagy functions in this organ is still limited. In order to understand how autophagy maintains retinal homeostasis and function, we undertook studies to examine the consequences of deleting the essential autophagy gene Atg5 in rod photoreceptors.

Results

Deletion of Atg5 in rod photoreceptors

As mice deficient in Atg5 or Atg7 die soon after birth,35, 36 we generated the Atg5ΔRod mouse strain in which basal autophagy was deleted specifically in rod photoreceptors (see Methods for details). Western blot analysis of the neurosensory retina revealed that Atg5/12 expression was reduced beginning at 6 weeks of age with further reduction in this complex at 8 and 16 weeks (Figure 1a). The conversion of microtubule-associated protein 1 light chain 3B (LC3B-I) to its lipidated form (LC3B-II) was almost completely suppressed at these time points suggesting that much of the detectable LC3B processing in the retina takes place in the rods. Importantly, Atg5-deficient retinae accumulated p62/SQSTM1, a hallmark of defective autophagy,37, 38 further substantiating the loss of basal autophagy in the rods. Quantification of the LC3B-II and p62/SQSTM1 (Figures 1b and c, respectively) by determining their average abundance relative to the actin control in three separate experiments confirmed these findings. Analysis of Lamp2A (Figures 1a and d) and Hsc70 (Figures 1a and e) revealed that these proteins were not increased in the whole retina suggesting that CMA was not elevated at these time points.

Figure 1
figure 1

Deletion of Atg5 in rod photoreceptors. (a) Representative western blots of retinae from Atg5ΔRod and control (Atg5f/f) mice examining Atg5/12, p62/SQSTM1 (*), Lamp2A, Hsc70, LC3B-I and LC3B-II. ( b–e) Relative expression levels of specific proteins compared with actin in 6, 8 and 16-week-old Atg5 f/fand Atg5 ΔRod mice averaged from three separate experiments; (b) LC3BII (*P<0.05), (c) p62 (*P<0.05), (d) Lamp2A, (e) Hsc70. ns=not significantly different

Full size image

Progressive rod degeneration in Atg5ΔRod mice

Photoreceptors are non-replicating, postmitotic cells that must survive the life of an organism without the possibility of renewal suggesting autophagy may be an essential process for cellular repair and energy.7 Consequently, we examined the long-term consequences of Atg5 deletion in the Atg5ΔRod mouse strain. Morphometric analysis demonstrated that beginning at 8 weeks of age the density of the outer nuclear layer (ONL) was significantly decreased compared with control. Over the next 36 weeks, ONL thickness steadily declined such that by 44 weeks, >90% of the ONL nuclei were depleted in both the inferior and superior regions of the retina (Figure 2a). These observations are represented in the micrographs shown in Figures 2b and c where aging had no effect on the ONL of control mice (Atg5f/+;Cre+) but in the Atg5ΔRod strain the ONL was gradually reduced to the thickness of a few nuclei (Figure 2c). We also detected increased numbers of terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL+) nuclei in the ONL of Atg5ΔRod mice (Figures 3a and b) as well as nuclei with apoptotic morphology (Figure 3c) suggesting that rod photoreceptor death was by apoptosis. We also noted that other layers of the retina such as the inner nuclear layer, ganglion cell layer and retinal pigment epithelium (RPE) were not overtly affected by rod deterioration.

Figure 2
figure 2

Morphological changes in the retina of Atg5 ΔRodmice. (a) ONL counts in Atg5ΔRod and control (Atg5f/+;Cre+) mice at 8, 16, 33 and 44 weeks of age. Representative hematoxylin and eosin sections of (b) control (Atg5f/+;Cre+) and (c) Atg5ΔRod mouse retinae from 8, 20 and 44 weeks of age. IS, inner segments; INL, inner nuclear layer; G, ganglion cell layer

Full size image
Figure 3
figure 3

TUNEL-positive cells present in the retina of Atg5ΔRod mice. TUNEL staining was performed on the retinae of 23-week-old Atg5ΔRod mice and littermate controls (Atg5f/+;Cre+). (a) Enumeration of TUNEL+ cells in the Atg5ΔRod and littermate control (Atg5f/+;Cre+) mice (n=4 for each group). * denotes statistically different from control (P<0.05). (b) Representative cryosections of Atg5ΔRod and control Atg5f/+;Cre+ mice. Arrowheads indicate TUNEL+ cells in the Atg5ΔRod mice that co-label with nuclear stain (PI, propidium iodide). (c) Representative transmission electron microscopy of mouse retina showing condensed nuclear chromatin (right panel, arrow) in the Atg5ΔRod retina that was not observed in the control retina (left panel). IS, inner segments; INL, inner nuclear layer

Full size image

Rod degeneration in Atg5ΔRod mice is not dependent on light exposure

It has been suggested that autophagy is a mechanism by which photoreceptors control the level of opsins in response to changes in lighting levels.39, 40 Controlling opsin levels is critical to suppressing light-dependent retinal degeneration that could occur following the absorption of too many photons. Thus, one prediction would be that light exposure of the Atg5ΔRod mice would promote rod degeneration because of the inability of the rods to regulate rhodopsin levels, and that in the absence of light, degeneration would be inhibited. We tested this hypothesis by comparing rod degeneration in mice maintained on the standard light/dark (L/D) cycle in our animal facility to mice that were maintained in continuous darkness. These results show that whether Atg5ΔRod mice were maintained in the L/D cycle or dark-adapted similar numbers of ONL nuclei were lost from all regions of the retina (Figure 4a).

Figure 4
figure 4

ONL counts in mice following dark adaptation or light stress. (a) Atg5ΔRod and control (Atg5f/+;Cre+) mice were dark-adapted from 8 to 40 weeks of age and ONL counts performed. Values were compared with age-matched Atg5ΔRod and control (Atg5f/+;Cre+) littermates that were maintained under normal L/D conditions for the experimental period (ns=not significantly different). (b) Atg5ΔRod and Atg5f/+;Cre+ control mice and light-sensitive 129/SvlmJ mice (13 weeks of age) were exposed to 13 000 lux for 7 h and then returned to the normal L/D cycle for 1 week. ONL counts were performed and compared with age-matched Atg5ΔRod and Atg5f/+;Cre+ that were maintained on the normal L/D cycle for the same experimental period (n=3 for each group). * denotes that there is a significant difference in ONL numbers in the 129/SvImJ mice following light stress (P<0.05); (ns= not significantly different)

Full size image

Light-induced retinal degeneration has been observed in certain strains of mice and this is dependent on a polymorphism in the RPE65 protein expressed in the RPE.41, 42 Mice with a leucine at position 450 (Leu-450) (e.g., BALB/c and 129) are susceptible to light stress, whereas mice with a methionine at position 450 (Met-450) (e.g., C57BL/6) are resistant. The Leu-450 amino acid results in an accelerated RPE visual cycle and leads to degeneration of the photoreceptors because of the rapid accumulation of toxic visual cycle products in the RPE. In strains with Met-450, the slower visual cycle does not lead to toxic product accumulation or degeneration. The Atg5ΔRod mice are on the C57BL/6J background permitting us to test the effect of intense light on Atg5-deficient photoreceptors without the confounding effects of the RPE-visual cycle. First, we exposed the light-stress-susceptible strain 129/SvImJ to 13 000 lux white light for 7 h and observed a significant loss of ONL nuclei in the inferior and superior regions of the retina demonstrating that this light intensity can induce severe retinal damage in a susceptible strain (Figure 4b). We then exposed Atg5ΔRod to light under the same conditions and observed that rod degeneration was not accelerated (Figure 4b) as mice exposed to intense light had a similar ONL thickness to Atg5ΔRod mice maintained in the normal L/D cycle. Thus, without the influence of the RPE visual cycle, autophagy in rod photoreceptors is not a protective mechanism under conditions of intense bright light.

Cones are preserved structurally but not functionally in Atg5ΔRod mice

In many retinal degenerative diseases, rod cell loss is often accompanied by degeneration of the cone photoreceptors.30 In Atg5ΔRod mice, rods are decreased by >90% by 38 weeks of age; however, when cone numbers were assessed at this time, only 10% of the cones were lost (Figure 5a, Supplementary Figure 1). Examination of cone morphology by co-staining with PNA (green) and specific antibodies to M-opsin (red) and S-opsin (blue) revealed that the remaining cones had significantly shorter outer segments (OSs) (Figure 5b, arrowheads). This was confirmed by morphometric measurements of M- and S-cones OS lengths (Figure 5c). Interestingly, cone function was significantly diminished as shown by the light-adapted b-wave amplitudes (Figure 5f) where cone responses were significantly reduced beginning at 8 weeks age. Not surprisingly, scotopic electroretinogram (ERG) responses were diminished over the same time periods reflecting the progressive loss of rod photoreceptors in the Atg5ΔRod mice (Figures 5d and e).

Figure 5
figure 5

Long-term impact of Atg5-deficient rods on cone photoreceptors. (a) Cone cell counts were performed on the nasal, dorsal, temporal and ventral regions of Atg5ΔRod and control (Atg5f/+;Cre+) mice at 38 weeks of age. * denotes significantly different from littermate control (P<0.05) (n=3 for all groups). Representative PNA-stained images for these data are available in Supplementary Figure 1. (b) Representative immunofluorescent images of the dorsal and ventral region of the retina of 38-week-old mice stained for PNA (green), S-opsin (blue) and M-opsin (red). Arrowheads denote cone OSs in the Atg5ΔRod retina. (c) OS length of M- and S- cones in the mid-dorsal region of the retina from 38-week-old mice. * denotes significantly different from littermate control (P<0.05) (n=3 for all groups). (d–f) ERG responses in the retina of Atg5ΔRod and control Atg5f/+;Cre+ mice performed at 6, 8, 18 and 43 weeks of age. (d) Dark-adapted a-wave; (e) Dark-adapted b-wave; (f) Light-adapted b-wave. * denotes statistically different from control (P<0.05); n.s.=not significantly different (n=5 for each group)

Full size image

Light-dependent arrestin and transducin translocation in Atg5ΔRod mice

Light-dependent redistribution of phototransduction proteins such as arrestin and transducin enable photoreceptors to modulate signaling capabilities by changing the concentration of these proteins in the OS.43, 44, 45 In the dark-adapted state, transducin is localized to the OS and arrestin distributes to the cell body. Light exposure redistributes transducin from the OS to the cell body while arrestin moves to the OS. Defects in the light-dependent translocation of these proteins can result in increased light-induced retinal degeneration.46 That light exposure did not accelerate retinal degeneration (Figure 4b) suggests that Atg5-deficient rods may not have defects in phototransduction protein translocation and this was confirmed by examining the light-dependent redistribution of arrestin and transducin (as detected by antibodies to the transducin alpha subunit, Tα) in control (Atg5f/+:Cre+) and Atg5ΔRod mice (Figure 6). In the dark-adapted state, rod arrestin (red) was distributed throughout the rod cell body and was not co-localized with rhodopsin (green) in the OS of either strain (Figure 6a, left panels). In the light, arrestin moved to OS co-localizing with rhodopsin in both strains (Figure 6a, right panels). Tα (red) localized to the OS in the dark (co-localizing with rhodopsin) (Figure 6b, left panels), but was translocated upon light exposure in both strains (Figure 6b, right panels). Single channel images for these micrographs are presented in Supplementary Figure 2.

Figure 6
figure 6

Arrestin (Arr) and Tα expression in rod cells. Twenty-three-week-old Atg5ΔRod mice and littermate control (Atg5f/+;Cre+) mice were either dark-adapted overnight (Dark) or were exposed to 10 000 lux (Light) for 90 min after overnight dark adaptation. (a) Representative immunofluorescent images for rod arrestin (Arr). Sections were co-stained for rhodopsin (green) and Arr (red). (b) Representative immunofluorescent images for rod Tα. Sections were co-stained for rhodopsin (green) and Tα (red). Single channel images are shown in Supplementary Figure 2. IS, inner segments; INL, inner nuclear layer

Full size image

Autophagy regulates the level of rod Tα

The α-subunit of Tα mediates phototransduction between light-activated rhodopsin and the effector enzyme cGMP phosphodiesterase leading to depolarization in rod cells and the initiation of the visual cascade.44, 47 Translocation of transducin from the OS reduces transducin activation in bright light diminishing sensitivity. Although the translocation of transducin was unaffected by the loss of autophagy (Figure 6b), when we examined the relative level of this protein in Atg5ΔRod mice at 6, 8 and 16 weeks, we noted that as the amount of rhodopsin steadily declined (reflecting the degeneration of rods), the level of Tα remained constant over these time points (Figure 7a). This suggested that Tα was accumulating as it was not decreasing as the rod photoreceptors died. We confirmed this by quantifying the ratio of Tα to rhodopsin in four separate experiments (Figure 7b) where the Tα:rhodopsin ratio increased as the rods degenerate. Thus, another possible effect of lost autophagy is inability of the rods to regulate the level of the critical phototransduction protein Tα.

Figure 7
figure 7

Expression of Tα in the retina of Atg5ΔRod mice. (a) Representative western blots for retinal samples from Atg5ΔRod and control (Atg5f/+;Cre+) mice performed at 6, 10 and 16 weeks of age. Analysis was performed for rhodopsin, Tα and actin. (b) Densitometric analysis of the relative expression levels of Tα compared with rhodopsin in Atg5ΔRod mice and control (Atg5f/+;Cre+) mice. Bars represent the average ratio of four separate experiments. n.s.=not significantly different, * denotes statistically different from control (P<0.05), ** denotes statistically different from control (P<0.01)

Full size image

We sought more definitive proof of this possibility by performing additional experiments. It is known that level of transducin is optimal during dark adaptation and that light exposure reduces its levels.48, 49 Therefore, we asked whether the light-induced reduction in transducin might be related to autophagic degradation by measuring the level of Tα in dark-adapted and light-adapted retinae from control and Atg5ΔRod mice. In control mice (Atg5f/+;cre+), Tα levels decreased when the mice are exposed to light in relation to the levels observed after dark-adaptation (Figure 8a), confirming published observations.48, 49 However, in the Atg5ΔRod retina, Tα levels did not vary between light and dark conditions suggesting that Atg5 is necessary to reduce the level of rod transducin in the light-exposed retinae. Quantification of data from four separate experiments is presented in Figure 8b confirmed these findings.

Figure 8
figure 8

Regulation of Tα in the retina by autophagy. (a) Representative western blots for retinal samples from 12-week-old Atg5ΔRod and control (Atg5f/+;Cre+) mice that were dark-adapted overnight or exposed to 600 lux room light for 5 h. (b) Densitometric analysis of the relative expression levels of Tα to actin. Bars represent the average ratio of three separate experiments. (c) Representative western blots from 12-week-old Atg5ΔRod and control (Atg5f/+;Cre+) retinal explants obtained after overnight culture with or without chloroquine (50 μm). Blots were probed for actin, rhodopsin, Tα and p62. (d) Densitometric analysis of the relative expression levels of Tα compared with rhodopsin in treated and non-treated samples averaged for three separate experiments. * denotes statistically different from control (P<0.05), ns= not significantly different

Full size image

We further examined this phenomenon in explant cultures using retinae taken from Atg5f/+;Cre+ and Atg5ΔRod mice. Whole retinae were removed and placed in culture for 18 h in the presence or absence of chloroquine, a known inhibitor of lysosomal function and autophagy. Following harvest, western blots for Tα were performed and we monitored p62 as an indicator of the level of autophagic flux. We also tracked the level of rhodopsin in the cultured retinae to be certain that the levels of this protein were not affected by the culture conditions. In untreated cultures of retinae from control mice, Tα levels diminished whereas p62 was not detected indicating a high rate of autophagic flux. However, in the presence of chloroquine, Tα accumulated in control mice, as did p62 (Figure 8c) demonstrating an effective block in autophagy and transducin degradation by the addition of chloroquine. In Atg5-deficient rod photoreceptors, the level of Tα was not affected by chloroquine treatment, nor was the level of p62 protein. Figure 8d is the quantification of these results averaged from three separate experiments comparing Tα with rhodopsin demonstrating that blocking autophagy prevented degradation of Tα. It should also be noted that chloroquine did not affect rhodopsin levels further suggesting that autophagy may not regulate the level of rod opsin.

Discussion

Inherited retinal diseases are often associated with genetic mutations that generate harmful gene products leading to retinal degenerative disorders.26, 30, 50, 51 Our results demonstrate that the lack of autophagy is associated with retinal degeneration even in the absence of such toxic gene products. Our results also highlight the importance of basal autophagy to the long-term survival of the rods suggesting that the recycling functions of this pathway may be providing essential building blocks and energy to support the normal functions of the rods. Autophagy then represents an important pathway that should be considered as a point of therapeutic intervention where survival can be reinforced to inhibit cell death and preserve vision.

Our examination of the Atg5ΔRod mice revealed three critical findings concerning the importance of autophagy in the light responses of the retina. First, rod degeneration in these mice was independent of daily light exposure as we detected retinal degeneration in the retina of Atg5ΔRod mice even when the animals were dark-adapted for a long period of time. This further emphasizes the survival functions of autophagy in rods and demonstrates that it is independent of the light-sensing functions of the retina.

Second, autophagy does not appear to have a role in the resistance of the retina to the damaging effects of intense light. We had hypothesized that if autophagy were a critical process for resisting light-induced damage that inhibiting this pathway in the light-stress-resistant C57BL/6J strain41, 42 would make them more susceptible to damage. This was not the case as we could not accelerate rod degeneration by intense light exposure of the Atg5ΔRod mice. This finding contrasts a recent study by Chen et al.,33 who reported that deletion of the essential autophagy gene Atg7 increased the sensitivity of rods to light-induced damage. We utilized nearly three times the light intensity used in that study (5000 lux versus 13 000 lux in our studies) and exposed the animals for a longer period of time (2 h versus 7 h in our studies) and degeneration was still not observed. It is unlikely that the difference lies in the autophagy gene targeted (Atg7 versus Atg5); however, these authors did use a different Cre-expressing strain (LMOP-Cre52 versus iCre-75 stain53) so we cannot rule out effects of differential Cre-expression. Another important difference may lie in the background of the mice utilized as these authors used mice that had an accelerated visual cycle because of the presence of Leu-450 the RPE protein,33 which can increase light-induced degeneration.41 It is also noteworthy that the rod-specific deletion of Atg7 in that study did not lead to retinal degeneration in the absence of light stress33 as did the rod-specific deletion of Atg5 presented here.

A third insight into the role of autophagy in the light-sensing functions of the rods comes from our study of rhodopsin and the visual transduction proteins arrestin and transducin. Autophagy is well known to regulate the turnover of cytosolic proteins by removing damaged or misfolded proteins.14, 54 In the retina, autophagy has been proposed to regulate the level of rhodopsin;39, 40 however, we did not observe accumulation of rhodopsin in the rods of Atg5ΔRod mice nor did we observe rhodopsin accumulation when autophagy was blocked in retinal explant cultures (Figures 8c and d). These results show that autophagy may not directly regulate rhodopsin levels in rods.

Although we did not observe an effect on arrestin and transducin translocation in light and dark, we did observe that rod Tα accumulated in Atg5-deficient rods as they degenerated suggesting that autophagy is involved in the turnover of this protein. Light sensitivity in rod photoreceptors is set by the rate of transducin activation and depends on the concentration of transducin in the photoreceptor OS. Our results demonstrate that autophagic degradation can regulate the concentration of transducin and indicate that this process may be an additional mechanism whereby rod photoreceptors regulate their sensitivity to light stimuli.

Although our data suggest that the death of the rods in the Atg5ΔRod mice is via apoptosis, we do not yet know the mechanism by which death is initiated. The importance of autophagy in maintaining nutrient and energy supplies in cells7 suggests that the rods may be starving to death. However, it is notable that degeneration took a number of months to be completed, implying a compensatory mechanism. It was recently reported that the age-dependent decrease in macroautophagy in the retina was accompanied by a concomitant increase in CMA to help preserve the retina.22 We found no evidence for increased CMA in retina when Atg5 was deleted in rods as determined by measuring the levels of Lamp2A and Hsc70. It should be noted that in this previous study involving retinal aging, Atg5 was deleted using a Nestin-Cre that deletes Atg5 in all neural precursors beginning at embryonic day 10.5.54 Neuronal cell deletion of essential autophagy genes causes many neurological and developmental defects in the mice.54 Thus, in the retina, Atg5 deletion would occur in rods, Mueller cells, RPE cells, cones, ganglion cells, bipolar cells, etc. making it difficult to determine whether the degenerative effects on the retina or the increase in CMA observed by these authors22 were the result rod photoreceptor-specific deletion of Atg5 or the cumulative effects of Atg5 deficiency in the other cell types.

The loss of autophagy led to the near complete loss of rods; however, we noted that cones were mostly preserved, but their function was significantly impaired. Although we do not know the reason for this observation, our results imply that the death of the rods may not be sufficient to induce cone death as reported in other models of retinal degeneration.26, 30, 50, 51 However, our results do suggest that function of the cones may be intimately tied to that of the rods and that methods to boost cone function would significantly improve visual acuity in patients with rod-specific disease-causing mutations.

Materials and Methods

Mice

Rhodopsin-iCre-75 transgenic mice53 (provided by Dr Jason Chen, Baylor College of Medicine) were crossed to the Atg5flox/flox (provided by Dr Noboru Mizushima, Tokyo Medical and Dental University)54 to generate the Atg5ΔRod mouse line in which Atg5 was specifically deleted in rod cells. Atg5ΔRod offspring are born viable, in reasonable numbers, and do not show any developmental defects in the eye or other organs at birth (data not shown). All transgenic and conditional knockout lines were crossed to the C57BL/6J (strain #000664, Jackson Laboratories, Bar Harbor, ME, USA) and verified as congenic to C57BL/6J by microsatellite analysis (Research Animal Diagnostic and Investigative Laboratory, RADIL, University of Missouri, Columbia, MO, USA). In addition, they were free of the rd8 mutation by PCR analysis.55 129/SvImJ mice (strain # 002448) were purchased from Jackson Laboratories. Mice were housed in a barrier facility operated under standard 12 h light/12 h dark cycle (referred to as normal L/D) with the luminance ranging from 60 to 100 lux during the daily/light cycle. Facilities are maintained by the Division of Comparative Medicine of Washington University School of Medicine. Experiments were carried out in strict accordance with recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH), in accordance with the ARVO Resolution on the Use of Animals for Ophthalmic Research, and approved by the Washington University Animal Studies Committee. In all experiments, littermates were used (Atg5f/+;Cre+ or Atg5f/f) as controls and experiments contained at least three mice per group. Each experiment was repeated a minimum of three times.

Antibodies

For western immunoblot analysis, the following antibodies were used: actin-HRP (1 : 3000, sc-1815), anti-SQSTM1 (1 : 250, sc-25575), anti-Gnat1 (Tα) (1 : 2000, sc-389) from Santa Cruz Biotechnology (Paso Robles, CA, USA); anti-Rhodopsin (1 : 3000- AB5356) from Millipore (Danvers, MA, USA); anti-LC3B (1 : 2000, L7543), anti-ATG5/N-terminal (1 : 1000, A0851), from Sigma Aldrich (St Louis, MO, USA); anti-Lamp2A (1 : 1000, ab18528 ) from Abcam (Cambridge, MA, USA); anti-Hsc70 (1 : 2000, ADI-SPA-820) from Enzo Life Sciences (Farmingdale, NY, USA) anti-rabbit-HRP (1 : 3000, #7074S) from Cell Signaling Technologies (Danvers, MA, USA); anti-mouse-HRP (1 : 5000, #554002) from BD Biosciences (Torrance, CA, USA). The following antibodies were used for immunofluorescence staining: anti-M opsin (1 : 200, AB5405) and anti-rhodopsin (1 : 200, AB5356) from Millipore; anti-S opsin (1 : 100, sc-14363), anti-Gnat1 (Tα) (1:200,sc-389), anti-arrestin (1 : 100,sc-34547) from Santa Cruz Biotechnology; anti-rabbit AlexaFluor 594 (1 : 300, A21207), anti-goat AlexaFluor 647 (1 : 300, A21447) from Life Technologies (Grand Island, NY, USA).

Electroretinogram

ERGs on Atg5ΔRod and age-matched littermate control mice were performed as previously described.34 Briefly, mice were dark-adapted overnight and anesthetized by intraperitoneal injection of mouse cocktail (86.9 mg/kg ketamine and 13.4 mg/kg xylazine diluted in PBS). After pupil dilation (10% phenylephrine hydrochloride and 1% tropicamide), full-field scotopic ERGs were recorded using a Tucker–Davis System 3 Complete ABR/OAE Workstation (Tucker–Davis Technologies, Alachua, FL, USA). Photopic ERGs were recorded after light adaptation for 10 min.

Histological analysis

Eyes were removed and fixed in 4% paraformaldehyde overnight at 4 °C. Tissues were then embedded in paraffin and sagittal sections of 5 μm thickness were cut and stained with hematoxylin and eosin.

Transmission electron microscopy

Mice were deeply anesthetized with Ketamine/Xylazine and fixed by intracardiac perfusion with a quick rinse of 0.1% Heparin in 0.1M sodium phosphate buffer followed by 2% paraformaldehyde/2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.35). Eyes were removed and immersion-fixed overnight in the same fixative, post-fixed in 1% osmium tetroxide for 1 h and stained en bloc with 1% uranyl acetate in 0.1 M acetate buffer for 2 h. Blocks were then dehydrated in a graded series of ethanol followed by propylene oxide, and embedded in Araldite 6005/EMbed 812 resin (Electron Microscopy Sciences, Hatfield, PA, USA). Semi-thin sections (0.5–1 μm) were cut through the entire retina at the level of the optic nerve and stained with toluidine blue. Ultra-thin sections were taken from a 600–800 μm length of retina where the majority of inner segments were oriented longitudinally; samples from the same areas were matched between control and experimental animals. Ultra-thin sections were then post-stained with uranyl acetate and lead citrate, viewed on a Hitachi H7500 electron microscope (Wallingford, CT, USA) and documented in digital images.

Immunofluorescence staining

Eyes were removed and fixed in 4% paraformaldehyde overnight at 4 °C. They were then transferred to 30% sucrose/PBS at 4 °C overnight and then embedded in O.C.T. Compound (TissueTek, Torrance, CA, USA). Sagittal cryosections of 7 μm thickness were cut. Sections were placed in blocking buffer (10% donkey serum, 2% BSA, 0.2% Triton X-100 in PBS) for 2 h at RT and then incubated with primary antibodies in dilution buffer (2% donkey serum, 0.2% Triton X-100 in PBS) overnight at 4 °C followed by secondary fluorescent-labelled antibodies in dilution buffer for 2 h at RT. Visualization of cone cells was achieved by incubating sections with a mix of primary antibodies and AlexaFluor 488-conjugated PNA (Life Technologies) overnight at 4 °C followed by secondary antibodies as described. Nuclei were stained with DAPI present in the mounting media SlowFade Gold (Life Technologies). Images were acquired using Olympus Fluoroview FV1000 confocal microscope (Olympus America Inc., Center Valley, PA, USA).

Light stress and dark adaptation

Light stress experiments were performed on mice that were dark-adapted overnight. Cyclopentolate hydrochloride (1%; Alcon, Fort Worth, TX, USA) was used to temporarily stop the eye from focusing and to dilate the pupil. Pupils were then further dilated with 2.5% phenylephrine hydrochloride and the mice were exposed to 13 000 lux of white light for 7 h, with reapplication of phenylephrine every 2 h. The mice were then returned to the normal L/D cycle for 1 week. In transducin and arrestin translocation studies, mice were dark-adapted over night and eyes were harvested in dim red light to localize the molecules. Light adaptation was performed by placing dark adapted animals in 10 000 lux of white light for 90 min prior to harvest. Following dark and light adaptation, the mice were anesthetized by i.p. injection of a mixture of 86.9 mg/kg ketamine and 13.4 mg/kg xylazine diluted in PBS. Eyes were enucleated and processed for histological analysis.

Prolonged dark adaption studies were performed by placing 8-week-old Atg5f/+;Cre+ and Atg5ΔRod mice in a dark room (no white light) until they were 40 weeks old. Cages, water bottles and food were changed under dim red light.

SDS-PAGE and western blot analysis

Briefly, retinae were removed from enucleated eyes and placed in ice-cold lysis buffer (10 mM Tris (pH7), 30 mM sodium pyrophosphate, 1 mM sodium fluoride, 1 mM sodium vanadate, 50 mM sodium chloride, 1% Triton X-100) containing 1 × protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN, USA) then homogenized and insoluble material removed by centrifugation for 10 min at 12 000 × g at 4 °C. Protein concentration was determined using a Pierce BCA assay kit (ThermosScientific, Rockford, IL, USA). Retina samples (either 25 μg or 15 μg) were separated on a 12% Bis-Tris NuPage gel (Life Technologies) transferred to Immobulon FL membranes (Millipore), blocked for 1 h at RT in 5% non-fat milk powder in TBS pH 7.4 with 0.1% Tween 20 (TBST), incubated with primary antibodies in 5% BSA-TBST overnight at 4 °C followed by incubation with corresponding secondary peroxidase-conjugated antibodies in 5% BSA-TBST for 1 h at RT. Immunoreactive bands were visualized using SuperSignal West Pico (ThermoScientific, Rockford, IL, USA) chemiluminescence and exposing to autoradiographic film. Alpha-Imager software was used to quantify the density of bands relative to actin loading control. Tα levels were normalized relative to rhodopsin expression levels in Figure 7 and to actin in Figure 8.

TUNEL staining

TUNEL assays were performed on cryosections prepared as above and stained using a fluorescent TUNEL kit (#17-141, Millipore) per the manufacturer's protocol. Enumeration of TUNEL-positive cells was performed by capturing four to five images per section (two sections per sample) using an Olympus BX51 fluorescence microscope and counting the number of positive cells per field.

Enumeration of the ONL and cone counts

Photoreceptor nuclei in the ONL were counted in sagittal cut cryosections that were stained with the nuclear stain DAPI. Images were captured using an Olympus BX51 fluorescence microscope, and cell nuclei counted using the Image J software in an area of 50 μm × 50 μm at 200 μm intervals from the optic nerve head in the superior and inferior regions. Data are expressed as the nucleus number per 50 μm2.

Enumeration of cone cells was performed on retinal tissue that was carefully dissected to separate it from the underlying RPE, choroid and sclera. Retinae were stained with AlexaFluor 488-conjugated PNA overnight at 4 °C and radial cuts (average, four) were made from the edge of the retinal tissue towards the center of the retina, then flat-mounted onto slides with the photoreceptor OSs facing up. Cone cells were then counted in three non-overlapping fields with an area of 0.0616 mm2, each field at a location midway between the optic nerve head and the retinal margin, using images collected with an Olympus BX51 fluorescent microscope. Data are expressed as average cell number/mm2.

Cone cell length was determined by staining cryosections with AlexaFluor 488-conjugated PNA (Life Technologies) to identify the cones and anti-M and -S opsin antibodies to locate the OSs. Individual cone OS lengths were then measured using images collected with an Olympus BX51 fluorescent microscope using the MetaMorph software (Molecular Devices, Sunnyvale, CA, USA). Data are expressed in micrometers.

Explant cultures

Enucleated eyes were placed in Dulbecco’s PBS containing 100 U/ml penicillin and 100 μg/ml streptomycin for 10 min on ice. The cornea was then carefully cut away and lens removed before the retinae were gently pealed away from the underlying RPE and choroid. Retinae were then placed in RPMI-1640 medium (Life Technologies) containing 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin with or without 50 μM chloroquine (Sigma) and cultured overnight at 37 °C with 5% CO2. Retinae were then rinsed with ice-cold PBS, and placed in fresh ice-cold PBS for 5 min before being placed in ice-cold lysis buffer and prepared for western blot analysis as described above.

Statistical analysis

All data were analyzed using a two-tailed, unpaired Student’s t test. Statistical significance was considered as P<0.05 and indicated by an asterisk. n.s. denotes not significant. Error bars represent mean±standard error (S.E.).