Peroxisome turnover and diurnal modulation of antioxidant activity in retinal pigment epithelia utilizes microtubule-associated protein 1 light chain 3B (LC3B)

Abstract

The retinal pigment epithelium (RPE) supports the outer retina through essential roles in the retinoid cycle, nutrient supply, ion exchange, and waste removal. Each day the RPE removes the oldest ~10% of photoreceptor outer segment (OS) disk membranes through phagocytic uptake, which peaks following light onset. Impaired degradation of phagocytosed OS material by the RPE can lead to toxic accumulation of lipids, oxidative tissue damage, inflammation, and cell death. OSs are rich in very long chain fatty acids, which are preferentially catabolized in peroxisomes. Despite the importance of lipid degradation in RPE function, the regulation of peroxisome number and activity relative to diurnal OS ingestion is relatively unexplored. Using immunohistochemistry, immunoblot analysis, and catalase activity assays, we investigated peroxisome abundance and activity at 6 AM, 7 AM (light onset), 8 AM, and 3 PM, in wild-type (WT) mice and mice lacking microtubule-associated protein 1 light chain 3B (Lc3b), which have impaired phagosome degradation. We found that catalase activity, but not the amount of catalase protein, is 50% higher in the morning compared with 3 PM, in RPE of WT, but not Lc3b^−/−, mice. Surprisingly, we found that peroxisome abundance was stable during the day in RPE of WT mice; however, numbers were elevated overall in Lc3b^−/− mice, implicating LC3B in autophagic organelle turnover in RPE. Our data suggest that RPE peroxisome function is regulated in coordination with phagocytosis, possibly through direct enzyme regulation, and may serve to prepare RPE peroxisomes for daily surges in ingested lipid-rich OS.

INTRODUCTION

The retinal pigment epithelium (RPE) supports photoreceptor activity and outer segment (OS) renewal through the retinoid visual cycle and ingestion of spent photoreceptor OS disk membranes (38, 44, 45, 70, 77). On a daily basis, the RPE ingests ~10% of the lipid- and protein-rich OS, for recycling and degradation, while new OS material is added basally (73, 78). Incomplete phagosome degradation in the highly oxidative environment of the RPE predisposes undigested lipids to oxidative damage and the formation of lipid peroxidation adducts contributing over time to an immune response and tissue damage (3, 20, 29, 37, 74).

A complex mix of saturated and highly unsaturated fatty acids make up 50% (by weight) of ingested OS disk membranes and provide the RPE with substrate for mitochondrial and peroxisomal fatty acid oxidation (1, 24, 65, 68, 72). Within this lipid pool, OS membranes contain ~30% very long chain fatty acids (VLCFAs, C ≥ 20; 24), which are preferentially catabolized in peroxisomes via β-oxidation (68, 72). Therefore, efficient lipid catabolism in the RPE requires coordination of β-oxidation pathways within both mitochondria and peroxisomes (1, 24, 65). Once VLCFAs are shortened in peroxisomes, they are substrates for mitochondrial β-oxidation (1, 65). Excess acetyl-CoA is funneled into the mitochondrial ketogenesis pathway, producing β-hydroxybutyrate (β-HB). β-HB is subsequently transported across the apical membrane of the RPE into the subretinal space and taken up by photoreceptors as a metabolic substrate (1). Given their role in the disposal of VLCFAs from ingested OS, it is not surprising that numerous microperoxisomes can be found in RPE cells, where they are contiguous with endoplasmic reticulum tubules and proximal to mitochondria (48, 50, 66).

Peroxisomes are ubiquitous organelles highly enriched in liver, intestine, adipose tissue, and brain. In addition to their functions in α- and β-oxidation-mediated catabolism of fatty acids (51), peroxisomes play essential roles in lipid synthesis, docosahexaenoic acid (DHA) synthesis, and bile acid synthesis. Peroxisomes are also sites for both the generation and reduction of reactive oxygen and nitrogen species (ROS and RNS, respectively), which serve as mediators for cellular signaling (60, 76). The most abundant of these is hydrogen peroxide, produced as a by-product of the initial desaturation of acyl-CoA esters by acyl-CoA oxidases (ACOXs) in the committed step of peroxisome fatty acid oxidation (68, 72). Hydrogen peroxide generated in the ACOX reaction is rapidly degraded by the antioxidant enzyme catalase. Catalase activity can be regulated through diverse posttranslational modifications (10, 11, 17, 28, 41, 63), which are often sensitive to the redox state of the cell. Peroxisome antioxidant activity declines with cellular age (47). Dysregulation of peroxisome antioxidant capacity with age can have broad cellular impact. Inhibition of catalase activity in fibroblasts resulted in increased mitochondrial ROS levels and mitochondrial dysfunction (47, 71, 75). Catalase activity is also lower in peroxisomes from human primary RPE of aged donors and those with age-related macular degeneration (AMD; 49), and paradoxically, this is accompanied by an overabundance of peroxisomes (7, 8, 22), suggesting a role for impaired peroxisome turnover in declining peroxisome function.

Although countless studies have focused on understanding peroxisome proliferator-activated receptor regulation and correlation with disease (53, 64, 67, 80), to our knowledge, little information is available regarding the activity and numbers of peroxisomes in the RPE over a 12-h light-dark cycle. In this study, we investigate peroxisome function and dynamics in the context of phagocytic degradative processes in RPE cells. We find that catalase activity is increased by ~50% during the time of peak phagocytosis, and this diurnal regulation of antioxidant activity is absent in mice lacking microtubule-associated protein 1 light chain 3B (Lc3b). We also find that although LC3B-dependent processes contribute to peroxisome turnover, peroxisome number does not depend on time of day. Our data suggest that peroxisome function in the RPE is regulated dynamically in coordination with phagocytosis, likely through direct regulation of enzyme activity. Elevated catalase activity during the early morning suggests the possibility of a homeostatic mechanism to ready RPE peroxisomes for the daily synchronous surge in ingested lipid-rich outer segments.

MATERIALS AND METHODS

Animals.

C57BL/6J [wild-type (WT)] mice and the Lc3b^−/− mouse line [strain name, B6;129P2-Map1Lc3b^tm1Mrab/J; stock no. 009336 (9)] were purchased from the Jackson Laboratory (Bar Harbor, ME). The Lc3b^−/− mice were backcrossed for at least five generations onto a C57BL/6J background. The Lc3b^−/− and the WT mice were confirmed to be free of the rd8 mutation by genomic DNA PCR using primers as described by Chang et al. (13). Maintenance of mouse colonies and all experiments involving animals were as described previously (25, 26). Mice were housed under standard cyclic light conditions (12:12-h light-dark cycle) and fed ad libitum. Both male and female mice were used in these studies, with an age range of 3–8 mo. All procedures involving animals were approved by the Institutional Animal Care and Use Committees of the University of Pennsylvania and the University of California, Los Angeles, and were performed in accordance with the Association for Research in Vision and Ophthalmology guidelines for use of animals in research.

Microarray analysis for heat map.

The heat map was generated using the publicly available high-throughput Mus musculus gene expression microarray data set GSE10246 (43) available through National Center for Biotechnology Information Gene Expression Omnibus (GNF Mouse GeneAtlas V3). The probe IDs for each gene of interest were obtained from an up-to-date microarray annotation file (Mouse430_2na36annot) obtained from the Affymetrix website and used to search GSE10246. The heat map was made in MultiExperiment Viewer version 4.9.0 using log2 scale for values averaged from two replicates for each tissue.

Immunohistochemistry.

Eyes were removed from euthanized mice and immediately placed in 4% paraformaldehyde. For cryosections, the anterior segment and lens were removed during fixation, leaving an intact posterior eye (eyecup). Eyecups were fixed overnight at 4°C. Following cryoprotection in 30% sucrose and 1X PBS at 4°C overnight, eyecups were embedded in optimum cutting temperature compound (Sakura Finetek, Torrance, CA), frozen, and sectioned with a cryostat (Microm HM 550) at 10–20-µm thickness. Cryosections were rehydrated and washed in 1X PBS followed by incubation with antibodies diluted in blocking buffer (5% BSA, 0.1% Triton X-100, and 1X PBS). Sections were washed three times in 1X PBS, and bound primary antibodies were detected by incubation with Alexa Fluor-conjugated secondary antibodies diluted in blocking buffer (Thermo Fisher/Invitrogen, Eugene, OR) for 1 h at 37°C. For flat mounts of RPE/choroid/sclera, melanin and other pigments were bleached after fixation, using the method of Kim and Assawachananont (40). Briefly, each eyecup was incubated at 55°C in 10% peroxide-1X PBS for 2.5 h and immunolabeled as above after three to four washes in PBS. Immunolabeled tissue was imaged with a Nikon A1 confocal microscope (Nikon Instruments, Melville, NY). We used the following primary antibodies for immunohistochemistry (with dilution and company): rabbit anti-catalase (ab-1877, 1:500; Abcam), rabbit anti-peroxisomal biogenesis factor 14 (anti-Pex14, 10594-1-AP, 1:500; ProteinTech), and mouse anti-ATP-binding cassette transporter subfamily D member 3 (anti-PMP70, SAB-4200181, 1:300; Sigma).

Electron microscopy with 3,3′-diaminobenzidine labeling of peroxisomes.

Peroxisomes were labeled with alkaline 3,3′-diaminobenzidine (DAB) following established methods (21). Briefly, mouse eyes were enucleated, the anterior segment was removed as above, and eyes were immersed in fixative (2.5% glutaraldehyde and 2% paraformaldehyde, in 0.1 M cacodylate buffer) overnight at 4°C. After fixation, small pieces of eyecup were washed two times for 30 min in 0.1 M cacodylate buffer and then washed for 20 min in Tris·HCl, pH 8.5, and then in 0.2% DAB for 0.5 h in darkness in a water bath at 37°C. Control tissue was incubated at 37°C with Tris·HCl buffer alone. DAB solution contained 0.2% DAB (from a 4% stock, cat. no. 13080; Electron Microscopy Sciences), 0.1 M Tris·HCl, and 0.15% H₂O₂. Following DAB incubation, tissue was washed for 15 min in 0.1 M Tris·HCl in darkness and then for 15 min in 0.1 M cacodylate in darkness. The labeled tissue was embedded in Epon. All postfixation, embedding, and ultrathin sectioning were performed by the University of Pennsylvania Electron Microscopy Resource Laboratory. Ultrathin sections were imaged with a Jeol-1010 transmission electron microscope.

Immuno-electron microscopy.

Mouse eyes were enucleated and fixed at 4°C in 4% formaldehyde and 0.2% glutaraldehyde in 0.1 M sodium cacodylate buffer. The anterior segment was removed, and eyecups were dissected along the dorsal-ventral axis. Samples were embedded in LR White resin, and ultrathin sections were immunolabeled as described previously (5, 42). Catalase antibody (ab-15834; Abcam, Cambridge, United Kingdom) was diluted 1:500, followed by anti-rabbit IgG conjugated to 18-nm gold particles (Jackson ImmunoResearch Laboratories, West Grove, PA). Samples were stained with 5% uranyl acetate in ethanol for 5 min. All samples were imaged with a JEM 1200EX transmission electron microscope (JEOL, Peabody, MA) at 80 kV. Labeling density of the peroxisomes was determined by counting the number of gold particles per micrometers squared of sectioned peroxisome. Measurements were performed using Fiji (ImageJ version 2.0.0; image-processing software package; available at https://fiji.sc/). Data were collected from animals fixed at 1 and 6 h after light onset.

RPE/choroid explants: OS and peroxide treatment.

RPE explants were incubated in bicarbonate-buffered Ringer solution with l-carnitine or in Ringer solution with either bovine OS particles (100,000 per chamber) or peroxide (0.5 mM) using published methods (65). Bovine outer segments were obtained from InVision BioResources (Seattle, WA). Stocks of 2 × 10⁷ particles/mL were made and stored in 15% sucrose. OS or peroxide were diluted to their final concentrations in Ringer solution, and explants were incubated for 1–2 h at 37°C with 5% CO₂.

Western blot analysis.

RPE lysates were prepared from mouse eyecups after removal of the cornea, lens, and retina (RPE/choroid eyecups), as described (65). RPE proteins were loaded into precast 4–12% NuPage Bis-Tris gels (Thermo Fisher Scientific/Life Technologies, Carlsbad, CA) and resolved with PAGE. Proteins were transferred to PVDF membranes, and the membranes were blocked for 1 h at room temperature (RT) in 5% dry milk and Tris-buffered saline-Tween 20 (137 mM NaCl, 2.7 mM KCl, and 19 mM Tris base with 0.1% Tween 20). The membranes were probed overnight at 4°C with antibodies dissolved in blocking buffer. We used the following antibodies for Western blot analysis: rabbit anti-catalase (ab-1877, 1:5,000; Abcam), rabbit anti-Pex14 (10594-1-AP, 1:1,000; ProteinTech), mouse anti-PMP70 (SAB-4200181, 1:1,000; Sigma), rabbit anti-β-catenin (D-10AB; cat. no. 8480, 1:1,000; Cell Signaling Technology), mouse anti-β-actin (A-2228, 1:2,500; Sigma), and mouse anti-cyclophilin A (2175S, 1:1,000; Cell Signaling Technology). Following washes, membranes were probed for 1 h at RT with horseradish peroxidase (HRP)-conjugated secondary antibodies used to detect mouse (G-21040, 1:2,500; Thermo Fisher Scientific) and rabbit IgG (cat. no. 31460, 1:2,500; Thermo Fisher Scientific). Chemiluminescence signals were developed using SuperSignal West Dura extended-duration substrate (cat. no. 34075; Thermo Fisher Scientific). One eye each of three WT and three Lc3b^−/− mice was used for Western blot analyses. The companion eye was used for catalase assays.

Catalase assays.

Catalase activity was measured as the amount of peroxide depleted in samples by detection with a probe reacting with H₂O₂ in the presence of HRP to produce the fluorescent Resorufin (Amplex Red Catalase Assay Kit; Thermo Fisher Scientific, Eugene, OR). Fluorescence at 590 nm was measured with a microplate reader (Fluoroskan Ascent microplate fluorometer; Thermo Systems). Catalase activity in experimental samples was determined as change in fluorescence compared with known standards. Lysates for measuring catalase activity were prepared from an eyecup of three different WT or Lc3b^−/− mice at each time point, similarly to Western blot analysis, except that solubilization was performed with manual homogenization in PBS. Triton X-100 was added to a final concentration of 0.5% followed by vigorous vortexing to enhance the extraction of catalase from peroxisomal membranes.

Citrate synthase activity assays.

Citrate synthase activity was measured using MitoCheck citrate synthase activity assay (cat. no. 701040; Cayman Chemical, Ann Arbor, MI) as described (35), from lysates of eyecups of two WT and Lc3b^−/− mice at each time point. This assay is based on the change in absorbance measured with the release of SH-CoA due to the citrate synthase-catalyzed condensation of oxaloacetate and acetyl-CoA to form citrate. The reaction was initiated with the addition of oxaloacetate to reaction wells containing RPE lysates diluted ~50-fold. Absorbance at 412 nm as a function of time was measured for lysate samples, positive controls (using a known amount of enzyme), and negative controls (with either no enzyme or no enzyme and no oxaloacetate). Specific activity was determined by dividing the reaction rate by lysate protein concentration.

Statistical analyses.

All bar graphs represent mean quantities, and error bars denote standard error (SE). Estimates of relative quantities of peroxisomal proteins by Western blot were analyzed by one-way analysis of variance. Statistical significance of fluorescence intensities in immunolabeled fixed tissue was determined using Student’s t test. Catalase and citrate synthase activities for Lc3b^−/− and WT mice, at each time point, were compared using two-way analysis of variance. Catalase activities after treatment of eyecup explants with OS, H₂O₂, or Ringer solution alone were analyzed with Student’s t test.

RESULTS

Mouse RPE contains abundant peroxisomes.

Fatty acids composed of 20 carbons and longer are catabolized in peroxisomes (68, 72), where the rate-limiting step is desaturation of a fatty acyl-CoA by the enzyme ACOX1, producing the by-product hydrogen peroxide (H₂O₂; Fig. 1A). Catalase, the most abundant peroxisomal antioxidant, reduces H₂O₂ produced in peroxisomes during this reaction, preventing its toxic accumulation. Using a publicly available microarray data set (GSE10246; Fig. 1B), we analyzed expression profiles of peroxisome lipid-associated genes in mouse tissues. RPE expressed fatty acid transporters and peroxisome β-oxidation enzymes. Three members of the ATP-binding cassette (ABC) lipid transporter superfamily reside in peroxisomes; these include ATP-binding cassette transporter subfamily D member 1 [Abcd1; adrenoleukodystrophy protein (ALDP)], Abcd2 [adrenoleukodystrophy-related protein (ALDR)], and Abcd3 [70-kDa peroxisomal membrane protein (PMP70)]. Of these, Abcd1 is expressed in relative abundance within mouse RPE, with levels of Abcd3 in RPE similar to that in liver (Fig. 1B). Transcripts for Acox1 and catalase as well as 3-ketoacyl-CoA thiolase (Acaa1/thiolase) are highly expressed in liver but are also enriched in the RPE relative to retina, cornea, and neural tissues (Fig. 1, A and B). Two thiolase genes are expressed in mouse, Acaa1a and Acaa1b, with the expression of thiolase b restricted to liver, with lower expression in kidney, intestine, and adipose tissue (14). The complete β-oxidation of polyunsaturated fatty acids (FAs) within peroxisomes requires additional enzymes, because of the presence of cis and trans double bonds (72), including peroxisomal 2,4-dienoyl-CoA reductase (Decr2). We find that transcripts for these enzymes are also enriched in RPE relative to retina and brain (Supplemental Fig. S1; see https://doi.org/10.6084/m9.figshare.8144894).

Fig. 1.

Enrichment of β-oxidation genes in peroxisomes of mouse retinal pigment epithelium (RPE) and their identification with transmission electron microscopy. A: schematic depicting key enzymes involved in fatty acid β-oxidation within peroxisomes. The initial committed step, desaturation of acyl-CoA esters by acyl-CoA oxidase 1 (ACOX1), produces H₂O₂ as a by-product, which is then neutralized by catalase. ABCD, ATP-binding cassette transporter subfamily D; FA, fatty acid; FFA, free fatty acid; OS, photoreceptor outer segment disk material. B: microarray expression analysis from GSE10246 showing relative enrichment of peroxisomal β-oxidation enzymes, including catalase, in mouse RPE. Acaa1/PTHIO, peroxisomal 3-ketoacyl-CoA thiolase; ALDP, adrenoleukodystrophy protein; ALDR, adrenoleukodystrophy-related protein; Hsd17b4, hydroxysteroid 17-β dehydrogenase 4; PMP70, 70-kDa peroxisomal membrane protein. C: transmission electron microscopy image of wild-type (WT) RPE labeled with diaminobenzidine. BM, RPE basement membrane; M, mitochondria; MG, melanin granule; MV, microvilli. D: boxed field (in C) at higher magnification. Yellow arrowheads in C and D indicate peroxisomes. E: example images of immuno-electron microscopy in WT mouse RPE showing gold particle labeling of catalase.

We next examined peroxisome localization and abundance in the RPE, using DAB reactivity by transmission electron microscopy (EM) complemented with immuno-EM analysis. Catalase reacts with DAB in the presence of excess H₂O₂ to form a dense precipitate localized to catalase-positive peroxisomes (16). DAB was localized to small oval-shaped organelles with ~100-nm diameter in ultrathin sections of RPE (Fig. 1, C and D). When immuno-EM analysis was used to detect catalase, gold particles labeled small rounded structures adjacent to mitochondria (Fig. 1E), morphologically consistent with our DAB labeling. In RPE whole mounts labeled with PMP70 and Pex14, we observe numerous peroxisomes throughout the cytoplasm, some with a tubular morphology (Supplemental Fig. S2; see https://doi.org/10.6084/m9.figshare.8145473) as was recently observed for peroxisomes of mouse fibroblasts using stimulated emission depletion microscopy (69). Catalase immunolabeling appeared most intense in the RPE and ganglion cell layer relative to other retina layers in cryosections of mouse eye (Fig. 2A). In RPE, catalase-positive puncta were enriched in the peroxisome-specific membrane protein, PMP70, a fatty acid ABC transporter used as a marker of peroxisome number (23, 55; Fig. 2, B and C). Linear intensity profiles show that peaks in catalase signal are spatially correlated with that of PMP70 (Fig. 2C), and codistribution analysis of catalase and PMP70 has a Pearson’s correlation coefficient of 0.89 ± 0.006 (n = 58 regions of interest). We also examined catalase expression in RPE of the Lc3b^−/− mouse, a model of defective phagosome maturation and lipid dysregulation (19, 20). Catalase is enriched in Lc3b^−/− mouse RPE relative to retina, similar to that observed in WT (Fig. 2D), with linear intensity profiles of PMP70 and catalase well correlated (Fig. 2, E and F). Our data show that relative to other cells within the eye, peroxisomes are abundant in RPE, with catalase codistributing with PMP70-positive peroxisomes in both Lc3b^−/− and WT mice.

Fig. 2.

Mouse retinal pigment epithelium (RPE) peroxisomes are enriched in catalase. A and D: single-plane confocal micrographs of cryosections of eyes from 4-mo-old wild-type (WT) and microtubule-associated protein 1 light chain 3B knockout (Lc3b^−/−) mice harvested and fixed at 10 AM and immunolabeled with antibodies to 70-kDa peroxisomal membrane protein (PMP70, green) and catalase (red). GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; IS, photoreceptor inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, photoreceptor outer segments. B and E: higher-magnification views of the boxed regions in A and D, respectively. C and F: intensity profiles for PMP70 (green), catalase (red), and Hoechst nuclear dye (blue) across the lines depicted in B and E, respectively. AU, arbitrary units. The degree of overlap of catalase and PMP70 was estimated with a Pearson’s correlation coefficient of 0.89 ± 0.006 (mean ± SE; n = 58 regions of interest) for WT and 0.89 ± 0.005 (n = 55) for Lc3b^−/− measured from circular regions of interest (5-μm diameter).

Peroxisome numbers do not depend on time of day.

Peroxisomes respond to changes in the cellular environment by adapting their number, morphology, and metabolic functions. In RPE, the level of VLCFA substrate for peroxisome oxidation is regulated in a synchronous diurnal manner through phagocytosis of OSs (30, 44, 45). Phagocytic ingestion of rod photoreceptor OS peaks following light onset (45). Because peroxisome abundance can be dynamically regulated to adapt to changes in cellular metabolic needs and environmental conditions, we wanted to determine whether OS disk phagocytosis influences peroxisome abundance and function in the RPE. We analyzed peroxisome abundance in WT mouse RPE at various times of day and compared this to peroxisome abundance in the RPE of the Lc3b^−/− mouse, having defective phagosome maturation (19, 20). RPE/choroid was isolated at 6 AM (1 h before lights on), 7 AM (at lights on), 8 AM (an hour after lights on), and 3 PM (8 h after light onset). We compared the levels of two membrane proteins widely used as markers for peroxisomes: Pex14, an essential transmembrane component of the peroxisome translocation machinery that imports cytosol-translated enzymes into the peroxisome lumen (81), and PMP70, one of three ATP-binding cassette family members that transport FAs into peroxisomes (23, 55). Pex14 (Fig. 3, A and C) and PMP70 levels (Fig. 3, B and D) remained relatively stable throughout the day in RPE/choroid isolated from WT mice. The levels of Pex14 in the Lc3b^−/− mouse RPE were elevated by 20% at 7 AM and 12% at 8 AM compared with WT mice, whereas PMP70 expression was elevated above WT by an average of 32% at all time points (Fig. 3, E and F). Peroxisome abundance was next assessed in WT and Lc3b^−/− mouse RPE by confocal imaging (Fig. 4, A and B). Pex14 immunoreactivity increased by at least 50% in Lc3b^−/− versus WT (Fig. 4C). PMP70 immunoreactivity was increased by 50% in one data set (P < 0.02), but in another data set the mean PMP70 intensity was only 20% higher in Lc3b^−/− compared with WT (P > 0.05). A third data set showed no difference between Lc3b^−/− and WT immunoreactivities. Collectively, these studies suggest that peroxisome degradation is reduced in RPE of Lc3b^−/− mice.

Fig. 3.

Expression of peroxisomal biogenesis factor 14 (Pex14) and 70-kDa peroxisomal membrane protein (PMP70) is elevated in microtubule-associated protein 1 light chain 3B knockout (Lc3b^−/−) mice during the day but does not vary diurnally. A and B: Western blots of retinal pigment epithelium (RPE) lysates made at indicated times of day. Immunoblot analysis was performed with antibodies to peroxisomal membrane proteins Pex14 (A) and PMP70 (B) and loading control actin. C and D: quantification of mean intensities (± SE) of Pex14 (C) or PMP70 (D) relative to loading control band intensities as a function of time of day for wild-type (WT; bars at left) and Lc3b^−/− lysates (bars at right). Values were not significantly different for Lc3b^−/− or WT across time of day in C and D (1-way ANOVA). Ratio of immunoreactivity of Pex14 (E) or PMP70 (F) in Lc3b^−/− vs. WT estimated from the same data in C and D, respectively. Values were not significantly different across each time point in E and F (1-way ANOVA).

Fig. 4.

Peroxisomes are more numerous in microtubule-associated protein 1 light chain 3B knockout (Lc3b^−/−) mice. A and B: cryosections of eyes from wild-type (WT) and Lc3b^−/− mice harvested at 10 AM showing the retinal pigment epithelium (RPE) layer, immunolabeled with antibodies to peroxisome-specific markers peroxisomal biogenesis factor 14 (Pex14, red) and ATP-binding cassette transporter subfamily D member 3 (PMP70, green), and stained with Hoechst nuclear dye (cyan). C: example ratios of fluorescence intensities for Pex14 or PMP70 in Lc3b^−/− vs. WT from a data set of 3–4 image fields per each genotype. Error bars denote SE. Statistical significance between WT and Lc3b^−/− intensities: *P < 0.02, **P < 0.01 (Student’s t test).

LC3B contributes to early morning peak in catalase activity in RPE.

Although peroxisome numbers are relatively constant over the time period tested, because RPE ingests a peak quantity of lipid-rich OS at light onset, we sought to determine whether peroxisome enzymatic activity depends on substrate availability. We evaluated catalase activity in WT and Lc3b^−/− mice as a readout of ACOX1 function as these are coupled. Therefore, we compared peroxisomal catalase activity at 6 AM, 7 AM (at lights on), 8 AM [at which time the OS-derived long-chain fatty acid (LCFA) and VLCFA β-oxidation substrates are maximal], and 3 PM (8 h after lights on). At the 8-h time point, virtually no phagosomes were observed in WT mice, whereas over 50% of the phagosomes remain undegraded in the Lc3b^−/− mouse RPE (19, 20). Catalase activity was elevated ~1.5-fold during the early morning compared with the afternoon in WT mice (Fig. 5A), but not in Lc3b^−/− mice, suggesting that an LC3B-dependent process influences peroxisome catalase activity. To determine whether elevated catalase activity was due to increased enzyme levels, we investigated catalase protein amounts at 6 AM, 7 AM (at lights on), 8 AM, and 3 PM, by immunoblot analysis from the same mice used to determine catalase activity. Catalase protein levels remained relatively constant over the time periods studied (Fig. 5, B and C). In contrast to our catalase activity data, we observed an increase in relative catalase protein levels between 8 AM and 3 PM in WT (P < 0.02, Student’s t test); however, catalase protein levels at 6 and 7 AM were not significantly different from those at 3 PM. We quantified catalase immunoreactivity from our immuno-EM analysis of WT mouse eyes removed at 1 and 6 h after light onset. At 1 h after light onset, we found 467 ± 30 gold particles/µm² (n = 41 fields), which was comparable to our finding of 432 ± 20 particles/µm² (n = 39 fields) at 6 h after light onset. Taken together, our data suggest that changes in catalase protein levels do not account for the elevated activity during the morning. We also tested whether diurnal modulation of metabolic activity was unique to peroxisomes, and so we investigated whether mitochondrial activity was also regulated diurnally by assessing activity of citrate synthase, the first enzyme of the tricarboxylic acid cycle (Fig. 5D), and found no difference in citrate synthase activity as a function of time of day. We directly tested whether catalase is activated by increased VLCFA and LCFA β-oxidation substrate after uptake of OSs or by elevated H₂O₂, by measuring catalase activity in RPE/choroid explants treated with OSs or H₂O₂. Catalase activity was increased by ~50% above Ringer solution control (P < 0.05; Student’s t test) in RPE explants incubated with 0.5 mM H₂O₂, but not OSs (Fig. 5E). We found no statistically significant differences in catalase activity in RPE explants from Lc3b^−/− mice for either treatment.

Fig. 5.

Elevated catalase activity during early morning in retinal pigment epithelium (RPE) of wild-type (WT) mice, but not microtubule-associated protein 1 light chain 3B knockout (Lc3b^−/−) mice. A: catalase activity measured from RPE/choroid lysates made at different times relative to light onset (7 AM). Average catalase activity is expressed relative to milligrams of protein. Eight data sets from 3 mice of each genotype were averaged. Error bars denote SE. *P < 0.02, **P < 0.0006, ***P < 0.0001 (Student’s t test). Values were not significantly different across time points for Lc3b^−/− mice (1-way ANOVA). Differences in catalase activities between WT and Lc3b^−/− were statistically significant (P < 0.009; 2-way ANOVA), but at 3 PM, WT and Lc3b^−/− catalase activity differences did not reach statistical significance (P < 0.1; Student’s t test). B: RPE/choroid lysates from WT and Lc3b^−/− mice made at the indicated times of day were immunoblotted with antibodies to catalase and cyclophilin A (cycl. A, as a loading control). C: relative intensities (means ± SE) of catalase vs. cyclophilin A in immunolabeled bands from Western blots of RPE/choroid lysates made at different times of day. D: average citrate synthase activity in WT and Lc3b^−/− lysates of posterior eyecup after removal of neural retina (RPE/choroid). Error bars denote SE. Values were not significantly different (2-way ANOVA). E: catalase activity (means ± SE) in lysates of RPE/choroid explants incubated with Ringer solution or with added bovine outer segments and 5 mM glucose (OS + glucose) or 0.5 mM H₂O₂. Catalase activity was increased by 55% in WT after H₂O₂ treatment compared with Ringer solution control (P < 0.05; Student’s t test).

DISCUSSION

RPE peroxisomes are enriched in β-oxidation genes.

Peroxisomal β-oxidation preferentially breaks down very long chain fatty acids that are abundant in OS membranes. The chain-shortened FAs can then be utilized by mitochondria for further rounds of β-oxidation. In this study, we show that genes encoding key enzymes of fatty acid β-oxidation within peroxisomes, such as Acox1, as well as the peroxisomal antioxidant catalase, are highly expressed in mouse RPE relative to other eye tissues. Peroxisome β-oxidation genes are most highly expressed in liver, consistent with the liver’s principal role in fatty acid metabolism and regulation of systemic fatty acid availability (51, 57). Peroxisomes in the liver, unlike adipocytes, brain, and intestinal mucosa, are large and contain a matrix enriched in luminal enzymes, including catalase (51). β-Oxidation pathways of the liver play multifaceted roles in fatty acid metabolism, including the breakdown of dicarboxylic acids and dietary pristanoyl-CoA. Single rounds of β-oxidation in liver are also critical for synthesis of bile acids from cholesterol and DHA from essential fatty acids (6, 72). Previous studies have shown that RPE, like the liver, is ketogenic, with the ability to catabolize FAs derived from phagocytosed photoreceptor OS membranes through mitochondrial β-oxidation pathways that produce acetyl-CoA and the secreted ketone metabolite β-hydroxybutyrate (1, 65). Peroxisomes of the RPE are found adjacent to mitochondria, and their proximity supports their coordination of metabolic pathways. In addition to metabolic coordination, mitochondria and peroxisomes overlap in redox sensing and ROS/RNS signaling and regulation. Mitochondria and peroxisomes both produce significant amounts of ROS through their different metabolic pathways. Each round of β-oxidation in peroxisomes produces H₂O₂. We show that the antioxidant catalase, the primary enzyme responsible for the removal of H₂O₂, is enriched in peroxisomes within RPE, relative to most other retinal cells. Peroxisomal antioxidant activity plays a key role in redox homeostasis, and disrupted catalase activity has serious consequences for mitochondrial health (47, 71, 75).

Catalase activity may be regulated posttranslationally.

We show that catalase has highest activity during the early morning, when OS-derived VLCFAs are expected to reach their maximum in RPE cells and β-oxidation reaches its peak (65). The elevated catalase activity is not due to increased expression of catalase protein, suggesting that diurnal modulation of catalase activity occurs at the posttranslational level. Catalase activity is regulated by a diverse array of posttranslational modifications depending on cell context, including phosphorylation, S-nitrosylation, oxidation, and ubiquitination (10, 11, 17, 28, 41, 63), as well as regulation of its multimerization (63). Catalase is thought to exist mainly in a tetrameric, enzymatically active form, but higher-order oligomer and dimer forms have been encountered inside cells (2, 62). The enzymatic activities of catalase in H₂O₂ removal, as well as its peroxidase activity, require its heme group, and heme binding is required for tetramerization and maturation (12). During small elevations in cellular H₂O₂ in cultured mouse fibroblasts, the redox-sensitive kinases Abelson tyrosine-protein kinase 1 (c-Abl) and Abelson-related gene (Arg) increase catalase activity by phosphorylation of key tyrosines and directly binding, protecting catalase from dephosphorylation (10, 11, 63). Catalase is an important antioxidant whose activity declines with age and in AMD-afflicted RPE (49), and the regulation of its activity in RPE is largely unexplored.

Diurnal modulation of catalase activity depends on LC3B.

In contrast to WT RPE, RPE of Lc3b^−/− mice did not exhibit a boost in catalase activity during the morning. RPE of Lc3b^−/− mice have impaired degradation of phagocytosed OS-derived material, depressed fatty acid oxidation, eventual accumulation of lipid deposits, and low levels of esterified and free DHA, due in part to the role of LC3B in LC3-associated phagocytosis (20). Lower catalase activity in RPE of Lc3b^−/− mice during the morning is consistent with impaired β-oxidation of OS-derived FAs in Lc3b^−/− (20). Elevated catalase activity in RPE may reflect higher peroxisomal β-oxidation activity during the morning when VLCFA availability may be maximal. Similarly, alveolar macrophages show increased catalase activity upon phagocytosis (27). The mechanisms governing diurnal regulation of phagocytosis and peroxisome activity in RPE are likely to be intricately intertwined. A recent study illuminates a complex relationship between phagocytosis and RPE circadian cycles, suggesting that phagocytosis of disk membranes can reset the cyclic expression of various RPE functional genes (54). Our own results suggest that substrate availability from ingested outer segments regulates catalase activity. Moreover, these studies also suggest that the regulators of catalase diurnal rhythm likely mimic that shown for disk membrane phagocytosis.

The synchronous burst in OS membrane uptake by phagocytosis in mammalian RPE occurring ~1 h after light onset is necessary for the maintenance of photoreceptor health (56). We hypothesized that catalase activity is modulated in step with this burst, through increased H₂O₂ produced when availability of OS-derived LCFA and VLCFA substrate reaches its peak. If this were true, RPE explants made from WT mice fed OS might respond with a boost in catalase activity, and this might be impaired in RPE explants from Lc3b^−/− mice. However, increasing H₂O₂ independently of substrate availability might override the impairment and act more directly on a modifier of catalase activity (e.g., an H₂O₂-sensitive kinase) resulting in increased catalase activity in both genotypes. Our results show that H₂O₂ resulted in ~50% increase in catalase activity for both WT and Lc3b^−/− RPE explants, making H₂O₂-sensitive kinase-mediated phosphorylation a plausible mechanism for modulating catalase activity in RPE cells. However, elevated catalase activity was not observed after treatment of RPE explants of either WT or Lc3b^−/− mice with OS, leaving open the possibility of catalase activity modulation through another pathway that is impaired in the absence of LC3B. Future studies are needed to decipher the role of LC3B in modulating catalase activity.

RPE peroxisome number does not vary with time of day.

Peroxisome number is dynamically regulated to adapt to the changing cellular milieu. The number of peroxisomes at any given time emerges from the dynamic balance of biogenesis, replication, and degradation (15, 32). In this study, we explored the possibility that peroxisome number is dynamically regulated after the synchronous peak in outer segment phagocytosis, when availability of ingested very long chain fatty acids derived from photoreceptor outer segments is maximal. However, our immunoblot evidence suggests that RPE peroxisome mass is relatively stable over time. This was a surprise as previous studies in mice show two brief peaks in RPE peroxisome number 3 h before and 3 h after light onset (50), suggesting rapid expansion followed by rapid removal of peroxisomes in RPE. The chronology of the peak in peroxisome numbers observed by EM roughly corresponds to our observations of peak catalase activity in the morning spanning an hour before and after light onset. However, we saw no evidence for rapid removal of peroxisome material. It is likely that peroxisome removal and replacement constitute a relatively slow process, as pulse-chase studies of fluorescently tagged or isotope-labeled peroxisomal proteins indicated a half-life of 1–3 days for liver peroxisomes (33, 61). The observations of a steep decline in peroxisome numbers 1–2 h after light onset track a peak in numbers of phagosomes in this study, suggesting the possibility of false negatives or misclassification of structures, when phagosomes accumulate and traffic more basally.

Peroxisome degradation depends on LC3B: implications for RPE homeostasis.

Turnover is important in maintaining healthy functional organelles. Degradation of dysfunctional and damaged peroxisomes and other organelles is principally achieved through regulated macroautophagy (or pexophagy), a lysosomal degradative process whereby a double-membrane autophagosome engulfs the entire organelle before fusion with lysosomes (4, 36). The signals that select old dysfunctional peroxisomes for degradation are beginning to be elucidated, and selective ubiquitination plays an important role (39). Prolonged residency of monoubuqitinated Pex5 at the peroxisomal membrane is a signal for binding of pexophagy adapters sequestosome 1/p62 protein (P62) and neighbor of BRCA1 gene 1 (NBR1; 18, 59). P62 contains an LC3 interaction motif and ubiquitin-binding domains allowing its corecruitment of peroxisome cargo and autophagosome machinery. In a distinct pexophagy pathway, under conditions of elevated ROS, ataxia-telangiectasia-mutated kinase phosphorylates Pex5, allowing monoubiquitination at a lysine residue, leading to recruitment of P62 and autophagosomal membranes (79). Our results show that peroxisomes are more numerous in RPE of Lc3b^−/− mice, which likely reflects sluggish degradation and turnover of peroxisomes, given the well-established role of LC3B in autophagosome formation and autophagosome-lysosome fusion (58). LC3B is one of six different members of the mammalian autophagy-related 8 (Atg8) family [LC3/GABA type A receptor-associated protein (GABARAP) proteins] involved in autophagy (46). Of the Lc3 genes, only Lc3a and Lc3b are expressed in mouse RPE (19). Our data suggest that the LC3B isoform is necessary for efficient turnover of peroxisomes within the RPE. This is supported by evidence of elevated peroxisome membrane proteins (Pex14 and PMP70) in RPE of Lc3b^−/− mice from Western blot analysis and immunohistochemistry. Our data are consistent with the recent finding that LC3B was necessary for autophagic degradation of p62, normally degraded along with its cargo, in HEK293T cells (52).

In this study, we found that RPE of mice lacking LC3B expression have elevated peroxisome numbers but lack the ability to increase activity of peroxisome antioxidant catalase during the early morning. Elevated peroxisome numbers with lower antioxidant function is reminiscent of aged and diseased RPE (7, 22). Reduced autophagic flux, accompanied by accumulation of autophagosomes and lysosomes, in RPE derived from patients with AMD was proposed to contribute to poor mitophagy and mitochondrial dysfunction (31). The role for autophagic degradation in mitochondrial quality control in AMD has been the subject of more intense focus (34). Mitochondria and peroxisomes intimately cooperate in metabolic pathways and regulation of ROS. Impairments in autophagy in aging and diseased RPE may have more far-reaching consequences when the health and function of peroxisomes are also considered. Peroxisome turnover defects may impact the resilience of RPE to oxidative stress in aging and contribute to disease pathogenesis.

GRANTS

This work was supported in whole or in part by NIH Grants EY-010420-21 (to K. Boesze-Battaglia), EY-026525-02 (to K. Boesze-Battaglia and N. J. Philp), R01-EY-027442 and P30-EY-000331 (to D. S. Williams), and P30-EY-001583 (University of Pennsylvania Core Grant).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.L.D., J.C., D.S.W., N.J.P., and K.B.-B. conceived and designed research; L.L.D., J.C., S.V., R.C.S., and A.D. performed experiments; L.L.D., J.C., S.V., R.C.S., A.D., and N.J.P. analyzed data; L.L.D., J.C., S.V., A.D., D.S.W., N.J.P., and K.B.-B. interpreted results of experiments; L.L.D. and S.V. prepared figures; L.L.D. drafted manuscript; S.V., R.C.S., A.D., D.S.W., N.J.P., and K.B.-B. edited and revised manuscript; L.L.D., J.C., S.V., R.C.S., A.D., D.S.W., N.J.P., and K.B.-B. approved final version of manuscript.