Early AMD-like defects in the RPE and retinal degeneration in aged mice with RPE-specific deletion of Atg5 or Atg7.

PURPOSE: To examine the effects of autophagy deficiency induced by RPE-specific deletion of Atg5 or Atg7 in mice as a function of age.
METHODS: Conditional knockout mice with a floxed allele of Atg5 or Atg7 were crossed with inducible VMD2-rtTA/Cre transgenic mice. VMD2-directed RPE-specific Cre recombinase expression was induced with doxycycline feeding in the resulting mice. Cre-mediated deletion of floxed Atg5 or Atg7 resulted in RPE-specific inactivation of the Atg5 or Atg7 gene. Plastic and thin retinal sections were analyzed with light and electron microscopy for histological changes. Photoreceptor outer segment (POS) thickness in plastic sections was measured using the Adobe Photoshop CS4 extended ruler tool. Autophagic adaptor p62/SQSTM1 and markers for oxidatively damaged lipids, proteins, and DNA were examined with immunofluorescence staining of cryosections. Fluorescence signals were quantified using Image J software.
RESULTS: Accumulation of p62/SQSTM1 reflecting autophagy deficiency was observed in the RPE of the Atg5ΔRPE and Atg7ΔRPE mice. 3-nitrotyrosine, advanced glycation end products (AGEs), and 8-hydroxy-2'-deoxyguanosine (8-OHdG), markers for oxidatively damaged proteins and DNA, were also found to accumulate in the RPE of these mice. We observed retinal degeneration in 35% of the Atg5ΔRPE mice and 45% of the Atg7ΔRPE mice at 8 to 24 months old. Degeneration severity and the number of mice with degeneration increased with age. The mean POS thickness of these mice was 25 µm at 8-12 months, 15 µm at 13-18 months, and 3 µm at 19-24 months, compared to 35 µm, 30 µm, and 24 µm in the wild-type mice, respectively. Early age-related macular degeneration (AMD)-like RPE defects were found in all the Atg5ΔRPE and Atg7ΔRPE mice 13 months old or older, including vacuoles, uneven RPE thickness, diminished basal infoldings, RPE hypertrophy/hypotrophy, pigmentary irregularities, and necrosis. The severity of the RPE defects increased with age and in the mice with retinal degeneration. RPE atrophy and choroidal neovascularization (CNV) were occasionally observed in the Atg5ΔRPE and Atg7ΔRPE mice with advanced age.
CONCLUSIONS: Autophagy deficiency induced by RPE-specific deletion of Atg5 or Atg7 predisposes but does not necessarily drive the development of AMD-like phenotypes or retinal degeneration.

Introduction

The RPE plays an essential role in supporting photoreceptor function. One critical role of the RPE is to dispose of waste products from photoreceptors [1,2]. RPE cells are post-mitotic and are subjected to lifelong exposure to high levels of oxidative stress. The lack of RPE cell turnover requires that RPE cells adapt efficient systems to limit environmental damage. This includes the ubiquitin-proteasome system and autophagy as a means of eliminating damaged cellular components. While the proteasome system predominantly degrades misfolded proteins and requires substrates to be unfolded to pass through the narrow pore of the proteasomal barrel, autophagy is a potent lysosomal degradation pathway capable of the disposal of bulk damaged macromolecular complexes and organelles [3,4].

Three forms of autophagy have been identified: macroautophagy, chaperone-mediated autophagy, and microautophagy [5]. Autophagy typically refers to macroautophagy, the major catabolic mechanism used by eukaryotic cells to degrade long-lived proteins, lipids, and organelles [5]. More than 30 ATG (AuTophaGy) genes that execute or regulate autophagy have been identified from yeast to mammals [5]. Autophagy occurs at low basal levels in virtually all cells to perform essential homeostatic functions but is rapidly upregulated when cells need to eliminate damaging cytoplasmic components or intracellular pathogens [6]. In general, it is accepted that autophagy function declines with age since a common characteristic of all aging cells is the accumulation of macromolecules and organelles [6]. This occurs even in the absence of any mutations that predispose the cells to a pathogenic phenotype, such as aggregation-prone mutant proteins [4,6,7]. The accumulation of altered components is particularly detrimental in non-dividing cells, such as neurons. Overwhelming evidence indicates that autophagy is protective against neurodegeneration in a wide range of neuronal degenerative diseases [3,6,8-10].

The high oxygen environment of the outer retina results in an enormous amount of oxidized waste that needs to be removed to preserve vision. As an example, the RPE phagocytoses 10% of the volume of the photoreceptor outer segment (POS) every day [1]. Even under normal circumstances, the autophagic capacity of RPE cells is burdened by this immense amount of metabolic waste [11]. Lipofuscin (highly crosslinked aggregates of oxidized proteins and lipids) and other waste products generated through phagocytosis and other physiologic processes accumulate in the RPE early in life and increase with age [11,12]. This accumulation is thought to play a role in age-related macular degeneration (AMD) by imposing an increasing burden on RPE cells. For this reason, it can be hypothesized that any impairment of autophagy could be detrimental to the cellular functions of the RPE. In support of this hypothesis, it has been reported that autophagy dysfunction in the RPE is associated with the pathogenesis of AMD and other forms of retinal degeneration [13-19]. Importantly, markers of autophagy have been found in drusen, a form of sub-RPE deposits associated with AMD, as well as in aged human and mouse RPE, Bruch’s membrane, and the choroid [15,18,20]. RPE cells in AMD are often engorged with lipofuscin and damaged organelles [6,12,21,22]. A common early feature of AMD is the presence of sub-RPE deposits accompanied by RPE hypertrophy [21,23,24]. RPE geographic atrophy and choroidal neovascularization (CNV) define the late stages of AMD [21,24]. Although it is not understood how sub-RPE deposits are formed, all of the deposits contain common components, including oxidized proteins, lipids, DNA, ubiquitin, advanced glycation end products (AGEs), membrane and cellular debris, vacuoles, and inflammation-related proteins [7,25,26]. Proteins modified by products of lipid peroxidation or glucoxidation, such as 4-hydroxynonenal (HNE), carboxyethyl pyrrole (CEP), or AGEs, are found in lipofuscin granules or sub-RPE deposits associated with AMD [8,12,27]. It has been postulated that faulty degradative processes in the RPE may account for their accumulation and the deposits instigate chronic local inflammation [8,21,28,29].

ATG5 and ATG7 are two core components of the autophagy machinery and are so essential for autophagy function that Atg5 and Atg7 mice die shortly after birth [5,30-32]. Conditional knockout of Atg5 in mouse RPE has shown that the interplay of phagocytosis and autophagy in an Atg5-dependent manner is required for POS degradation and the maintenance of retinoid levels to support vision [33]. RPE-specific deletion of Atg7 in mice indicates that autophagy is important in maintaining RPE homeostasis [34,35]. As dysfunction of the RPE becomes progressively worse with age, and retinal degenerative changes progress with age, the goal of this study was to investigate the impact of RPE-specific deletion of Atg5 or Atg7 in the retina as a function of age.

Methods

Mice

Conditional knockout mice with floxed alleles of Atg5 (Atg5) or Atg7 (Atg7) were obtained from the RIKEN BioResource Center (Koyadai, Ibaraki, Japan). The Atg5 mice were generated by Dr. Noboru Mizushima [36]. The Atg7 mice were generated by Dr. Masaaki Komatsu [32]. These mice were crossed with VMD2-rtTA/cre transgenic mice [37] to generate mice with inducible RPE-specific deletion of Atg5 or Atg7. The transgenic mice carry the human vitelliform macular dystrophy 2 (VMD2) promoter-directed reverse tetracycline-dependent transactivator (rtTA), and the tetracycline-responsive element-directed cre gene [37]. VMD2 is an older name for the BEST1 (Gene ID: 7439; OMIM 607854) gene that has been replaced by the Human Genome Organization (HUGO) nomenclature committee. Bestrophin 1, the protein encoded by VMD2/BEST1, has been shown by us to localize specifically to RPE cells [38], and the VMD2 promoter directs RPE-specific gene expression [39]. Strain background can affect the phenotype of genetically modified mice [40], so we crossed these mice with a pigmented (C57BL/6J) or albino (Balb/c) background. The mice were backcrossed eight times to each background. VMD2-directed RPE-specific Cre recombinase expression was induced with doxycycline (Sigma Aldrich, St. Louis, MO) feeding for 1 week after the mice were born. Doxycycline is light sensitive and was administered in the drinking water at a dose of 0.2 mg/ml in amber water battles. Cre-mediated deletion of floxed Atg5 or Atg7 fragments results in RPE-specific inactivation of the Atg5 (Atg5) or Atg7 (Atg7) gene. Neither the transgenes nor the inducing drugs have an adverse effect on the health of the retina [37]. All mice were handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, using protocols approved by the Institutional Animal Care and Use Committee of the University of Arizona or Mayo Clinic. Other than doxycycline induction, the mice were housed in standard conditions and maintained on a 12 h:12 h light-dark cycle with free access to water and food.

PCR analysis of isolated RPE cells

PCR was performed with DNA extracted from isolated RPE cells to confirm the floxed Atg5 or Atg7 fragment is deleted in the RPE of Atg5 or Atg7 mice. The neuroretina was used as a positive control. To isolate the RPE cells, the mice were euthanized with CO2 asphyxiation, and the eyes were enucleated. A circumferential incision was made posterior to the limbus and the anterior segments removed. The neurosensory retina was carefully peeled from the eyecup. The RPE was exposed after the neurosensory retina was removed. With the optic nerve head at the center of the eyecup, four radial cuts were made in the eyecup to flatten it. RPE sheets were gently scraped from Bruch’s membrane with a #15 scalpel blade. For the Atg5 mice, PCR primers (A, 5′-GAA TAT GAA GGC ACA CCC CTG AAA TG-3′; B, 5′-GTA CTG CAT AAT GGT TTA ACT CTT GC-3′; C, 5′-ACA ACG TCG AGC ACA GCT GCG CAA GG-3′; D, 5′-CAG GGA ATG GTG TCT CCC AC-3′; E, 5′-AGG TTC GT TCA CTC ATG GA-3′; F, 5′-TCG ACC AGT TTA GTT ACC C-3′) described by Hara et al. were used [36]. For the Atg7 mice, PCR was performed using the primers (1, 5′-TGG CTG CTA CTT CTG CAA TGA TGT-3′; 2, 5′-GAA GGG ACT GGC TGC TAT TGG GCG AAG TGC-3′; and 3, 5′-TTA GCA CAG GGA ACA GCG CTC ATG G-3′) described by Komatsu et al. [32].

Reverse transcription polymerase chain reaction (RT-PCR) was performed (1 µl of RT samples was used for a 25-µl PCR reaction) as previously described [41] to confirm the absence of Atg5 or Atg7 mRNA expression in the RPE of the Atg5 or Atg7 mice. The neuroretina was used as a positive control. For the Atg5 mice, the primers for RT–PCR were 5′-GAT GTG CTT CGA GAT GTG TG-3′ and 5′-CTG GGT AGC TCA GAT GCT CG-3′. For the Atg7 mice, the primers for RT–PCR were 5′-CAA CAT GAG CAT CCC CAT GC-3′ and 5′-AGG TGA ATC CTT CTC GCT CG-3′. Efemp1 primers [40] were used in the control.

Immunofluorescence

The mice were euthanized, and the eyes were dissected and fixed in 4% paraformaldehyde (pH 7.4) in 0.1 M phosphate buffer at 4 °C for 4 h. The fixed eyes were cryoprotected in 20% sucrose and embedded in optimum cutting temperature (OCT; Thermo Fisher Scientific, Waltham, MA) at −20 °C. Immunofluorescence staining of the 10 µm frozen sections was performed as previously described [42] using a rabbit polyclonal antibody against HNE (Abcam, Cambridge, MA), malondialdehyde (MDA; Abcam), AGEs (Abcam), or p62/SQSTM1 (MBL International, Woburn, MA), or a mouse monoclonal antibody against 3-nitrotyrosine (39B6, Abcam), or 8-OHdG (15A3, Santa Cruz, Dallas, TX). A goat anti-rabbit immunoglobulin G (IgG) or anti-mouse IgG Alexa Fluor 488 conjugate was used as a secondary antibody (Invitrogen, Carlsbad, CA). Cell nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). The sections were examined and photographed using a Nikon E600 (Melville, NY) microscope equipped with a charge coupled device (CCD) camera.

Quantification of fluorescence intensities was performed using ImageJ software (National Institutes of Health, Bethesda, MD). Sections from three mice (n = 3) per genotype were used. Three images taken from the three randomly selected areas in each stained section were measured. In each image, the outline of the RPE layer was drawn, and the area, mean fluorescence, and background readings were measured. The total RPE fluorescence was calculated as integrated density − (area of selected sections × mean fluorescence of background readings). Data are reported as mean ± standard deviation (SD; p < 0.05).

Histological analysis

Mouse eyes were fixed in 4% paraformaldehyde/2% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2) overnight. The tissues were processed as previously described [42]. One micron plastic sections were stained with toluidine blue. Sixty nanometer thin sections were examined and photographed using a JEM1400 electron microscope (JEOL USA, Peabody, MA) equipped with a Gatan Ultrascan 1000XP CCD camera (Gatan, Pleasanton, CA). Basal laminar deposit (BLamD) severity and frequency in mice were graded based on a semiquantitative grading system described previously [43].

The thickness of the POS in the toluidine blue–stained sections was measured using the Adobe Photoshop CS4 extended ruler tool at 250 μm interval distances from the optic nerve head. Results of n =3 mice were plotted. POS thickness measurements from the Atg5 and Atg7 mice were compared with those in the wild-type control mice using the Student t test.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay was performed for the in situ detection of apoptosis on 10 µm cryosections of mouse eyes using a TACS.XL-Blue Label In Situ Apoptosis Detection Kit (Trevigen, Gaithersburg, MD) according to the manufacturer’s instructions.

Results

The gross appearance of the Atg5 and Atg7 mice was similar to that of wild-type mice. To determine whether RPE-specific deletion of Atg5 or Atg7 causes defects in the RPE or the outer retina and whether age is a factor, we examined the ocular phenotype of 20 wild-type, 43 Atg5, and 49 Atg7 mice from 8 to 24 months of age (Table 1). Because the mouse strain background has been reported to affect some phenotypes of genetically modified mice [40], pigmented (C57BL/6J) and albino (Balb/c) knockout mice (Table 1) were examined. Different background did not affect the RPE or retinal phenotypes we observed.

Table 1

Retinal degeneration in Atg5 and Atg7 mice.

Genotype		+/+ (phenotypic/total)		Atg5ΔRPE (phenotypic/total)		Atg7ΔRPE (phenotypic/total)
Background		Pigmented	Albino	Pigmented	Albino	Pigmented	Albino
Retinal degeneration	8–12 months	0/3	0/3	1/6	2/6	2/9	1/6
	13–18 months	0/3	0/3	2/8	2/7	4/10	3/7
	19–24 months	0/4	0/4	5/9	3/7	7/10	5/7

The number of mice with retinal degeneration out of the total number of mice in each genotype, background, and age group was indicated. Note that more Atg5 and Atg7 mice showed retinal degeneration as they age. +/+, wild-type controls.

RPE-specific deletion of Atg5 or Atg7

PCR performed with genomic DNA extracted from isolated RPE cells or the neuroretina confirmed the floxed Atg5 or Atg7 fragment is deleted in the RPE but not the neuroretina of the Atg5 or Atg7 mice. To further validate that there was no Atg5 or Atg7 expression in the RPE of these mice, RT–PCR was performed with RNA extracted from isolated RPE cells or the neuroretina. The results showed that no Atg5 or Atg7 was expressed in the RPE cells, but both were expressed in the neuroretina isolated from the Atg5 or Atg7 mice (Figure 1A).

Figure 1

RPE-specific deletion of Atg5 or Atg7 and p62/SQSTM1 accumulation in the RPE of Atg5 and Atg7 mice. A: Reverse transcription polymerase chain reaction (RT-PCR) showed that Atg5 or Atg7 was not expressed in RPE cells isolated from Atg5 or Atg7 mice but was expressed in the neuroretina isolated from these mice. RT–PCR using Efemp1 primers confirmed the integrity of the RNA from all the samples. B: Frozen sections from 8-month-old wild-type, Atg5, and Atg7 mice were stained with an antibody (green signal) against p62/SQSTM1. Note the bright green staining in the RPE of the Atg5 and Atg7 mice. The nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI; blue signal). C: The p62 staining in the RPE was quantified using Image J software. n = 3 mice per genotype. Error bars indicate the mean ± standard deviation (SD). WT, wild-type; ONL, outer nuclear layer; IS, photoreceptor inner segment; OS, photoreceptor outer segment.

p62/SQSTM1 accumulation in the RPE of Atg5 and Atg7 mice

p62/SQSTM1 is a cytosolic autophagic adaptor that selectively recognizes autophagic cargo and mediates its engulfment into autophagosomes [44]. The accumulation of p62/SQSTM1 is an indicator of autophagy impairment [44-46]. To assess whether autophagy function was affected in the RPE of the Atg5 and Atg7 mice, we performed immunofluorescence using frozen sections from 8-month-old mice. Bright fluorescent punctate was observed in the RPE of the Atg5 and Atg7 mice but not in the wild-type controls (Figure 1B). The p62 fluorescence signal in the RPE was quantified using Image J software. The signal intensity was similar in the RPE of the Atg5 and Atg7 mice and was nearly double that of the wild-type controls (Figure 1C). This indicates that autophagy in the RPE of the Atg5 and Atg7 mice was deficient.

Accumulation of oxidatively damaged proteins and DNA in Atg5 and Atg7 mice

Oxidatively damaged proteins, lipids, and DNA accumulate in the RPE of patients with AMD [12,27,47], trigger complement activation, and contribute to the pathogenesis of macular degeneration [47]. Impaired degradation in the RPE has been suggested to account for the accumulation of oxidative products [11]. Thus, we assessed whether autophagy deficiency induced by RPE-specific deletion of Atg5 or Atg7 affects the turnover of oxidized products in the RPE by performing immunofluorescence for markers of oxidation and lipid peroxidation. We found increased levels of 8-OHdG, 3-nitrotyrosine, or AGE fluorescence in the RPE of the Atg5 (Figure 2) and Atg7 mice (Figure 3). 8-OHdG is one of the predominant forms of free radical-induced oxidative lesions and has been used as a marker for the measurement of endogenous oxidative DNA damage [48]. 3-nitrotyrosine is a product of tyrosine nitration of proteins and is considered a marker of nitric oxide–dependent, reactive nitrogen species–induced oxidative stress [49]. AGEs are proteins or lipids that become glycated by glycation, oxidation, and/or carbonylation [50]. The formation and accumulation of AGEs have been implicated in the progression of age-related diseases [50]. We did not find a difference in the level of HNE or MDA, two markers of lipid peroxidation [51], between the wild-type and the Atg5 or Atg7 mice. Quantification of marker fluorescence in the RPE showed a greater than 51% increase in 8-OHdG, 3-nitrotyrosine, or AGE in the Atg5 and Atg7 mice compared with those in the wild-type controls (Figure 4). Increased levels of these markers reflect the accumulation of oxidatively damaged proteins and DNA and indicate that the ability of the RPE to eliminate oxidized wastes was reduced in the Atg5 and Atg7 mice.

Figure 2

Increased levels of oxidized proteins and DNA in the RPE of aged Atg5 mice. Frozen sections from 8-month-old wild-type (+/+; A, C, E) and Atg5 (Atg5−/−; B, D, F) mice were stained with antibodies (green signal) against 8-hydroxy-2’-deoxyguanosine (8-OHdG; A and B), 3-nitrotyrosine (C and D), or advanced glycation end products (AGEs; E and F). Note the bright green punctate staining in the RPE of the Atg5 mice. The nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI; blue signal).

Figure 3

Increased levels of oxidized proteins and DNA in the RPE of aged Atg7 mice. Frozen sections from 8-month-old wild-type (+/+; A, C, E) and Atg7 (Atg7−/−; B, D, F) mice were stained with antibodies (green signals) against 8-hydroxy-2’-deoxyguanosine (8-OHdG; A and B), 3-nitrotyrosine (C and D), or advanced glycation end products (AGEs; E and F). Note the bright green staining in the RPE of Atg7 mice. The nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI; blue signal).

Figure 4

Quantification of 8-OHdG, 3-nitrotyrosine, and AGEs in the RPE of Atg5 and Atg7 mice. The immunofluorescence staining of 8-OHdG, 3-nitrotyrosine, and advanced glycation end products (AGEs) in the RPE was quantified using Image J software. n = 3 mice per genotype. Error bars indicate the mean ± standard deviation (SD). WT, wild-type control.

Retinal degeneration in Atg5 and Atg7 mice

We observed retinal degeneration in 15 Atg5 and 22 Atg7 mice (Table 1). These mice comprise 35% and 45% of the Atg5 and Atg7 mice examined, respectively. The number of mice with retinal degeneration increased with age (Table 1). In the Atg5 and Atg7 mice, the outer plexiform layer (OPL), the outer nuclear layer (ONL), the photoreceptor inner segment (IS), and the POS became thinner with age. These layers were almost completely diminished in some of the mice in the 13- to 18- and 19- to 24-month age groups (Figure 5). Quantification of the POS thickness at 8–12, 13–18, and 19–24 months old showed that degeneration progressed with age (Figure 6). While the POS of the wild-type mice had a mean thickness of about 35 µm at 8–12 months, 30 µm at 13–18 months, and 24 µm at 19–24 months, the POS of the Atg5 mice had a mean thickness of about 25 µm, 15 µm, and 3 to 4 µm, respectively. The POS thickness of the Atg7 mice was similar to that of the Atg5 mice (Figure 6) indicating that the effect of the Atg5 or Atg7 deletion was similar. Degeneration of the ONL mirrored that of the POS in terms of age and genotype.

Figure 5

Retinal degeneration in aged Atg5 and Atg7 mice. Representative images of toluidine blue–stained sections showing the retina layers of 17-month-old wild-type, Atg5, and Atg7 mice. Note the retina layers of the Atg5 and Atg7 mice without retinal degeneration (no rd) were similar to those of the wild-type mice but the outer plexiform layer (OPL), ONL, IS, and OS were diminished in the Atg5 and Atg7 mice with retinal degeneration (rd). The image for the Atg7 mice without retinal degeneration was from a pigmented background, and the other images were from albino mice. Arrows indicate engorged RPE cells. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer.

Figure 6

POS thickness in Atg5 and Atg7 mice with retinal degeneration. The thickness of the photoreceptor outer segment (POS) was measured in the retinas of the mice at 8–12, 13–18, and 19–24 months of age. Each graph represents data from 1 µm toluidine blue–stained sections of three mice (n = 3) per genotype and age group. On each section, nine data points with 250 µm intervals starting from the optic nerve head were measured, and each data point was the mean of three measurements. The value represents the mean of the data points from three mice ± standard deviation (SD). wt, wild-type control mice.

However, we found that the retinal degeneration phenotype was not fully penetrant. Twenty-eight Atg5 and 29 Atg7 mice examined had normal-appearing retinas and showed no histological signs of degeneration (Figure 5). These results suggest that RPE-specific deletion of Atg5 or Atg7 contributes to retinal degeneration but on its own is not sufficient to cause retinal degeneration.

RPE defects in Atg5 and Atg7 mice

RPE abnormalities similar to the early morphologic features of AMD were found in all the Atg5 and Atg7 mice with or without retinal degeneration and regardless of the background. The abnormalities ranged from the presence of vacuoles, uneven RPE thickness, diminished basal infoldings, hypertrophy/hypotrophy, and pigmentary irregularities to necrosis (Table 2). The severity of the RPE abnormalities correlated with retinal degeneration and increased with age. In the 8- to 12-month-old Atg5 and Atg7 mice without retinal degeneration, the only abnormality in the RPE was the presence of vacuoles (Table 2). However, in all age groups of Atg5 and Atg7 mice with retinal degeneration, all these categories of RPE abnormalities were observed. All these abnormalities were also present in 13- to 18- and 19- to 24-month-old Atg5 and Atg7 mice without retinal degeneration (Table 2 and Figure 5). When present, the abnormalities were pan retinal.

Table 2

RPE abnormalities in Atg5 and Atg7 mice.

Genotype		Atg5ΔRPE		Atg7ΔRPE
Retinal degeneration		-	+	-	+
Vacuoles	8–12 months	+	+	+	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+
Uneven RPE thickness	8–12 months	-	+	-	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+
Loss of basal infoldings	8–12 months	-	+	-	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+
Hypertrophy / hypotrophy	8–12 months	-	+	-	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+
Pigmentary irregularities	8–12 months	-	+	-	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+
Necrosis	8–12 months	-	+	-	+
	13–18 months	+	+	+	+
	19–24 months	+	+	+	+

RPE abnormalities observed in Atg5 and Atg7 mice aging from 8 to 24 months were classified into six categories. “-“ indicates no difference between Atg5/Atg7 and age-matched wild-type control mice; and “+” indicates abnormalities in Atg5 or Atg7 mice

The thickness of the RPE layer in the Atg5 and Atg7 mice varied significantly. Mice without retinal degeneration tended to have uneven RPE thickness (Figure 5). Mice with retinal degeneration had an attenuated RPE layer (Figure 7). RPE cells in older mice appeared to be swollen and engorged with pigmented granules (hypertrophy/hyperpigmentation; Figure 5). There were also pale-looking RPE cells with little pigmentation (hypotrophy/hypopigmentation; Figure 7). RPE basal infoldings were scarce and disorganized in the aged Atg5 and Atg7 mice (Figure 8). RPE cell degeneration reflected by electron dense necrotic cells was observed in some areas of the RPE. We did not detect apoptosis with the TUNEL assays in the RPE in either the Atg5 or Atg7 mice. In addition to these RPE abnormalities, we found CNV, a late stage complication of AMD, in two Atg5 mice (17 and 24 months old) and two Atg7 mice (17 and 22 months old) through electron microscopy examination of thin retinal sections (Figure 9). All these mice also had retinal degeneration.

Figure 7

Thinned RPE in aged Atg5 and Atg7 mice. Representative electronic micrographs showing the RPE region of the 17-month-old wild-type (+/+), Atg5 (Atg5−/−), and Atg7 (Atg7−/−) mice. Note that the RPE thicknesses of the Atg5 and Atg7 mice are approximately one third to one half that of the wild-type mice. The wild-type and Atg5 mice shown were pigmented, and the Atg7 mouse was albino. Arrow indicates a hypotrophic RPE cell. Scale bar = 10 µm.

Figure 8

Diminished RPE basal infoldings in aged Atg5 and Atg7 mice. Representative electronic micrographs showing the RPE basal infolding area of 17-month-old wild-type (+/+), Atg5 (Atg5−/−), and Atg7 (Atg7−/−) mice. Note the scarce basal infoldings (arrow) in the Atg5 and Atg7 mice. Scale bar = 2 µm.

Figure 9

CNV in aged Atg5 and Atg7 mice. Representative electronic micrographs showing the RPE area of 17-month-old wild-type (+/+), Atg5 (Atg5−/−), and Atg7 (Atg7−/−) mice. Note the blood vessels (thin arrows) in the RPE layer of the Atg5 or Atg7 mice. Thick arrow indicates a capillary endothelial cell (Atg5−/−) or a platelet (Atg7−/−). Scale bar = 5 µm.

No sub-RPE deposits in Atg5 and Atg7 mice

Sub-RPE deposits are strongly associated with AMD. In the mouse, BLamD is the main form of sub-RPE deposits. We thus examined whether BLamDs were present in Atg5 and Atg7 mice. We adapted a semiquantitative grading system to grade BLamD severity and frequency [43,52]. Based on this system, mild BLamD referred to the presence of any discrete focal nodule of homogenous deposit between the RPE cell membrane and its basement membrane in at least one micrograph (of at least ten) within a section from an individual specimen [43]. Only small isolated BLamDs were seen occasionally in the Atg5 and Atg7 mice at 19–24 months old. The severity and frequency of BLamDs in these mice were similar to those of the age-matched wild-type controls and did not reach the “mild BLamD” category. This finding indicates that BLamD was not induced by RPE-specific deletion of Atg5 or Atg7.

Discussion

RPE dysfunction and aging are two main risk factors for AMD and other forms of retinal degeneration. In this study, we analyzed the effect of autophagy deficiency in the RPE that resulted from murine Atg5 or Atg7 deletion with advanced age. We found that RPE-specific deletion of Atg5 or Atg7 impairs the RPE’s waste-removing capacity, accelerates the accumulation of oxidized waste, and leads to early AMD-like RPE defects and partially penetrant retinal degeneration. RPE cells are non-dividing cells with an innate capacity to clear the large amount of metabolic waste generated by photoreceptors in the outer retina. The effect of autophagy deficiency appears to take time to manifest and is most evident in aged Atg5 and Atg7 mice. Mice younger than 6 months did not exhibit obvious defects in the RPE or the outer retina. This suggests that RPE cells can function for at least a limited time with deficient autophagy.

The RPE abnormalities found in the aged Atg5 and Atg7 mice are similar to the RPE cellular features of early AMD [11,12,21]. However, we did not find sub-RPE deposits, a characteristic early feature of typical AMD [21,23,24], in these mice. Some patients with AMD do not have sub-RPE deposits [53]. The present data suggest that autophagy deficiency contributes to the pathogenesis of AMD by affecting the biochemical pathway(s) that are not involved in the formation of sub-RPE deposits. Although it is still not clear how sub-RPE deposits form, a plausible hypothesis is that chronic inflammation triggered by oxidatively modified lipids in Bruch’s membrane leads to deposit formation [47]. Interestingly, we did not find increased levels of lipid peroxidation marker HNE or MDA in either the Atg5 or Atg7 mice. Accumulation of 3-nitrotyrosine, AGEs, and 8-OHdG in the RPE of these mice suggests that autophagy deficiency affects their turnover in the RPE but not their accumulation in sub-RPE deposits.

RPE thinning to atrophy was found in aged Atg5 and Atg7 mice with retinal degeneration. When an RPE cell dies, the nearby cells clear the body and stretch to cover the space. This results in thinned, hypopigmented cells. When the cells can no longer stretch to fill the gap, atrophy occurs [21]. RPE atrophy is a feature of late stage dry AMD [21]. CNV was occasionally found in aged Atg5 and Atg7 mice with retinal degeneration. In wet AMD, CNV is a late stage feature [24]. The present findings suggest that autophagy deficiency contributes to disease progression in AMD.

AMD and retinal degeneration are complex multifactorial diseases. Variation in the severity of RPE defects and partially penetrant retinal degeneration in Atg5 and Atg7 mice suggest that Atg5- or Atg7-dependent autophagy is only one contributing factor in the pathogenesis of these diseases. Manifestation of disease may require the collapse of several RPE cellular functions simultaneously, especially in the case of retinal degeneration. More than half of the Atg5 and Atg7 mice at 8 to 24 months old did not show any histological sign of retinal degeneration. This suggests that RPE-specific deletion of Atg5 or Atg7 in these mice did not tip over the balance maintained by other stress and antistress factors. RPE defects were more severe in the Atg5 and Atg7 mice with retinal degeneration suggesting that these mice are dealing with more overall stress. The number of Atg5 and Atg7 mice showing retinal degeneration increased with age, and the severity of degeneration also increased with age. It is likely that Atg5 or Atg7 deletion adds stress to the RPE, but disease occurs only when the overall stress from other factors increases above a certain threshold as the mouse ages. In the Atg5 and Atg7 mice without retinal degeneration, the overall stress never exceeded the disease threshold.

Several other studies involving the RPE-specific deletion of murine Atg5 or Atg7 did not report retinal degeneration [33-35]. Kim et al. found that Atg5 loss in the RPE resulted in disrupted lysosomal processing, decreased photoreceptor responses to light stimuli, decreased chromophore levels, but no detectable decrease in the numbers of photoreceptors up to 1.5 years of age [33]. Mice with tyrosinase-cre-mediated deletion of Atg7 had accumulation of autophagy substrates and the RPE65 variant M450 accompanied by increased expression of other regulators of the visual cycle but showed no signs of visual impairment up to 2 years of age [35]. Perusek et al. reported that mice with the Atg7 deletion in the RPE had abnormal RPE morphology with RPE hypertrophy, cellular debris, and vacuole formation [34]. These mice had trouble coping with stress caused by di-retinoid-pyridinium-ethanolamine (A2E) accumulation but did not show higher amounts of A2E in the RPE or retinal degeneration at 6 and 12 months old [34]. These studies are consistent with our conclusion that RPE-specific deletion of Atg5 or Atg7 alone is not sufficient to induce retinal degeneration. In their studies, the overall stress in the Atg5 or Atg7 mice did not exceed the disease threshold. Because we examined a large number of aged mice at different time points for retinal degeneration, we were able to observe retinal degeneration in a portion of mice, especially with advanced age.

In this study, the autophagy deficiency in the RPE is induced by the deletion of Atg5 or Atg7, two core components of the conventional autophagy pathway. Recently, an Atg5/Atg7-independent alternative pathway has been reported [54]. This alternative pathway is regulated by some components of the conventional autophagy pathway, including Unc-51-like kinase 1 (Ulk1) and beclin 1, but is also dependent on some proteins that had previously not been associated with autophagy, such as Rab9 [54]. Although conventional and alternative processes lead to the bulk degradation of subcellular constituents, they may be activated by different stimuli in different cell types and may have different physiologic functions. It is possible that in the absence of Atg5 or Atg7, an alternative autophagy pathway is upregulated and prevents retinal degeneration in some Atg5 or Atg7 mice.

In summary, our data show that defects in autophagy predispose to but do not necessarily drive the development of AMD-like phenotypes or retinal degeneration. Other risk factors, such as aging, augment the effect of autophagy deficiency in the pathogenesis of these diseases.