Regulation of Phagolysosomal Digestion by Caveolin-1 of the Retinal Pigment Epithelium Is Essential for Vision*

Caveolin-1 associates with the endo/lysosomal machinery of cells in culture, suggesting that it functions at these organelles independently of its contribution to cell surface caveolae. Here we explored mice lacking caveolin-1 specifically in the retinal pigment epithelium (RPE). The RPE supports neighboring photoreceptors via diurnal phagocytosis of spent photoreceptor outer segment fragments. Like mice lacking caveolin-1 globally, RPECAV1−/− mice developed a normal RPE and neural retina but showed reduced rod photoreceptor light responses, indicating that lack of caveolin-1 affects photoreceptor function in a non-cell-autonomous manner. RPECAV1−/− RPE in situ showed normal particle engulfment but delayed phagosome clearance and reversed diurnal profiles of levels and activities of lysosomal enzymes. Therefore, eliminating caveolin-1 specifically impairs phagolysosomal degradation by the RPE in vivo. Endogenous caveolin-1 was recruited to maturing phagolysosomes in RPE cells in culture. Consistent with these in vivo data, a moderate increase (to ∼2.5-fold) or decrease (by half) of caveolin-1 protein levels in RPE cells in culture was sufficient to accelerate or impair phagolysosomal digestion, respectively. A mutant form of caveolin-1 that fails to reach the cell surface augmented degradation like wild-type caveolin-1. Acidic lysosomal pH and increased protease activity are essential for digestion. We show that halving caveolin-1 protein levels significantly alkalinized lysosomal pH and decreased lysosomal enzyme activities. Taken together, our results reveal a novel role for intracellular caveolin-1 in modulating phagolysosomal function. Moreover, they show, for the first time, that organellar caveolin-1 significantly affects tissue functionality in vivo.

Support of the neural retina generally and of adjacent photoreceptor neurons specifically by the retinal pigment epithelium (RPE) 5 is essential for vision (1). A major function of the RPE is its contribution to photoreceptor outer segment renewal, a continuous and life-long rejuvenation process that involves the formation of new membrane disks at the proximal end of the outer segment and diurnal shedding of distal spent outer segment tips (2). Outer segment renewal is critical for photoreceptor function and survival, and any abnormality is thought to impair vision. RPE cells participate in outer segment renewal by clearing shed photoreceptor outer segment fragments (POS) by receptor-mediated phagocytosis (3).
Mechanistically, RPE phagocytosis belongs to a family of conserved non-inflammatory clearance phagocytosis pathways that other cell types use to remove apoptotic cells and debris. These pathways have in common that their failure to efficiently clear debris contributes to human disease. However, unlike other forms of phagocytosis, RPE clearance of POS occurs in a strict diurnal rhythm that is regulated by light and circadian mechanisms (4). This is a unique advantage for RPE phagocytosis studies because all steps of the synchronized phagocytic process may be quantified precisely in situ in the intact, undisturbed retinas of experimental animals. Content in the RPE of engulfed rod POS phagosomes peaks shortly after light onset and declines characteristically within several hours as RPE cells complete digestion of their phagocytic load before the next burst of intake (5).
Like other phagocytic pathways, ingested phagosomes in the RPE fuse with lysosomal vesicles to form phagolysosomes. In POS phagolysosomes, degradation of opsin, which constitutes ϳ85% of POS protein, requires the aspartic protease cathepsin D and phagosomal acidification (6,7). Because RPE cells are post-mitotic in the mammalian eye and ingest numerous POS daily, prompt and complete POS engulfment is essential to prevent gradual buildup of undigested debris in the RPE (8). Inefficient RPE lysosomal function causes accumulation of debris in human and experimental animal RPE that can be toxic and contribute to age-related blindness (9 -12). Despite its importance, the molecular control of phagolysosomal digestion by the RPE as well as other phagocytic cells remains poorly understood.
The membrane organizer protein caveolin-1 is expressed by the RPE (13) but also by other retinal cell types and the choroidal vasculature (14 -16), and global knockout of caveolin-1 impairs rod-driven visual function (17). Caveolins regulate cellular processes by recruiting protein complexes either on the inner leaflet of the plasma membrane or on cytoplasmic organelles (18 -20). Interestingly, caveolin-1 on a subset of early endosomes has recently been suggested to influence the fate and signaling of internalized TGF-␤ receptors, suggesting that vesicular caveolin-1 may alter vesicle functionality (21). Here we explore mice manipulated to lack caveolin-1 specifically and solely in the RPE. Strikingly, eliminating caveolin-1 from the RPE alone is sufficient to impair retinal function. Moreover, our studies identify a novel function for caveolin-1 in regulating phagolysosomal acidification and digestive enzyme activity to ensure efficient and complete clearance phagocytosis.
Animals, Tissue Harvest, and Processing-All procedures involving animals were performed according to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were reviewed and approved by the Institutional Animal Care and Use Committees of the University of Oklahoma Health Sciences Center and Fordham University. RPE CAV1 Ϫ/Ϫ mice were generated by crossing mice expressing Cre in a doxycycline-inducible fashion under the control of the RPE-specific VMD2 promoter (23,24) to mice carrying a floxed CAV1 gene (16,25). Mice were in the C57BL6 background and were screened and found not to carry the rd8 mutation. RPE-specific Cre expression was induced by feeding pregnant dams a doxycycline-supplemented diet (Bio-Serv, Flemington, NJ) ad libitum. The doxycycline diet was provided until weaning, at which point weanlings were transferred to standard mouse chow. Our mating strategy allowed us to generate RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ from the same litters by crossing homozygous floxed, Cre-carrying males with homozygous floxed, Cre-negative females. Therefore, all offspring were exposed to doxycycline, but only Cre expressors displayed recombination of the floxed allele. Mice were housed under a strict 12:12 h light on/off cycle and fed a standard mouse diet (after weaning) and water ad libitum. Mice were euthanized by CO 2 asphyxiation before immediate eye enucleation. Eyes were immersion-fixed in Davidson's fixative (33% ethanol, 22% formalin, and 11.5% acetic acid) or in 4% paraformaldehyde (PFA) for paraffin embedding or in 4% PFA in PBS with 1 mM MgCl 2 and 0.1 mM CaCl 2 (PBS-CM) for Optimal Cutting Temperature compound (OCT) embedding and cryopreservation. Sections from paraffin-embedded eyes were cut on a microtome, mounted on glass slides, and deparaffinized before labeling with hematoxylin and eosin (both from Millipore) for light microscopy or with antibodies for immunofluorescence microscopy. Frozen sections were cut on a cryostat before antibody labeling and immunofluorescence microscopy. Whole mounts of posterior, retina-free eye cups were prepared by dissecting neural retinas from fresh eyes, followed by making radial cuts to flatten the eye cup, fixation in 4% PFA in PBS-CM, and antibody labeling for fluorescence microscopy. For assays requiring tissue extracts, enucleated mouse eyes were opened to remove the lens, the neural retina was dissected from posterior eye cups from select eyes, and the resulting eye tissues were frozen at Ϫ80°C before lysis.
Expression plasmids encoding WT caveolin-1-myc and the scaffolding mutant mut-caveolin-1-myc (F92A, V94A) (29) (a gift from Dr. Patricio Meneses, Fordham University, Bronx, NY) were used to generate replication-defective, recombinant adenoviruses (Welgen, Worcester, MA). The ␤-galactosidase control adenovirus was purchased from Cell Biolabs (San Diego, CA). Adenoviruses were diluted in serum-free DMEM before addition to confluent RPE-J cells 4 days after plating. After overnight infection, cells were incubated in complete growth medium for a further 24 h before experiments. Replication-deficient lentiviral particles encoding either a scrambled shRNA sequence that will not lead to degradation of any known cellular mRNA (catalog no. sc-108080) or a mixture of three to five target-specific 19-to 25-nucleotide (plus hairpin) silencing shRNA sequences targeting caveolin-1 (catalog no. sc-106996-V) were purchased from Santa Cruz Biotechnology. RPE-J cells were infected and selected with puromycin according to the protocol of the manufacturer. Entire populations of cells were selected rather than single cell clones to minimize selection of individual clones that had drifted phenotypically from the parental RPE-J cell line.
Synchronized Cell Culture Phagocytosis Assay-POS were purified from fresh porcine eyes obtained from a local slaughterhouse in accordance with an established protocol (30) and used unlabeled or labeled covalently with Alexa Fluor-647 (Life Technologies). For pulse-chase experiments, cells were challenged with 10 POS/cell in serum-free DMEM with 30 nM MFG-E8 for 1 h at 20°C (pulse, POS binding only) followed by chase for specific time periods with DMEM supplemented with 5% delipidated, heat-inactivated FBS and fixation for microscopy or lysis.
Immunofluorescence Labeling, Lysosome Labeling, and Microscopy-Tissue sections, whole mount eye cup preparations, or fixed cells were labeled with antibodies for fluorescence microscopy. Nuclei were counterstained with DAPI (Life Technologies). For labeling of acidified compartments, RPE-J cells were incubated with 0.1 M LysoTracker Red DND-99 (Life Technologies) for 15 min at 32°C. Cells were either imaged live or after they were fixed with 4% PFA for 15 min at room temperature, permeabilized with 0.2% Triton X-100 in PBS-CM for 15 min at room temperature, and co-stained with antibodies. A TSP5 laser-scanning confocal microscopy system (Leica Microsystems, Wetzlar, Germany) was used for image acquisition. Images were further processed and compiled using Adobe Photoshop CS5. For cathepsin D staining or opsin/cathepsin D co-staining, PFA-fixed tissue sections were used. For phagosome counting, paraffin tissue sections were labeled with the opsin antibody B6-30. Stacks of precisely 5-m thickness were acquired to yield maximal projections showing phagosomes in a constant tissue volume. ImageJ software was used for processing images, and the average number of phagosomes per 100 m of retina was calculated (described in detail in Ref. 5).
Lysosomal pH Measurement-Lysosomal pH values in cells grown to confluence in 96-well plates were measured using LysoSensor Yellow/Blue DND-160 (Life Technologies) following a published protocol (31). Briefly, cells were loaded with 7.5 M LysoSensor Yellow/Blue DND-160 for 3 min at 32°C in DMEM, followed by washing with 20 mM HEPES buffer containing 145 mM KCl, 10 mM dextrose, and 1 mM MgCl 2 . Cells were incubated further for 5 min before fluorescence measurements using a Spectramax M2E scanner (Molecular Dynamics, Sunnyvale, CA). The excitation/dual emission parameters used were 380 nm and 525/436 nm. The ratio of light emission at 525/436 nm was converted to absolute pH measured according to Guha et al. (31). At least three independent experiments were performed with triplicate samples each.
SDS-PAGE and Immunoblotting-Chilled samples were lysed in HNTG buffer (50 mM HEPES, 150 mM NaCl, 10% glycerol, and 1% Triton X-100) with 1ϫ protease inhibitor mixture and 1 mM PMSF (both from Sigma-Aldrich). Clarified lysates were separated using standard protocols for SDS-PAGE, followed by immunoblotting and enhanced chemiluminescence detection (PerkinElmer Life Sciences). Densitometric quantification of immunoblot bands on scanned x-ray films was carried out using ImageJ software (National Institutes of Health, Bethesda, MD).
Quantification of Constitutive Transferrin Receptor Lysosomal Degradation-Transferrin receptor levels were established following exactly the detailed protocol by Matsui and Fukuda (32). Briefly, transferrin receptor levels were quantified by immunoblotting and densitometry, as described above, of lysates obtained from cells treated with 50 g/ml cycloheximide to stop protein synthesis for various periods of time. Addition of 100 nM bafilomycin A eliminated lysosomal function during treatment in select samples. Decreases in transferrin receptor levels in cycloheximide-treated cells, but not in cells treated with both cycloheximide and bafilomycin A, were interpreted as lysosomal degradation.
Enzyme Assays-For in vivo cathepsin D and ␤-N-acetylglucosaminidase activity assessment, mice were euthanized 0.5 or 8 h after light onset. Cathepsin D and ␤-N-acetylglucosaminidase activity was measured in dissected tissue fractions or cells using specific activity assay kits on the basis of synthetic cathepsin D-specific or ␤-N-acetylglucosaminidase-specific fluorogenic substrates following the instructions of the manufacturer (Sigma). Cathepsin D activity was established by directly comparing sample activity to the active cathepsin D standard (Abcam). Enzyme activities were calculated as unit activity per total sample protein content measured using Bradford colorimetric protein quantification.
Data Analysis-All in vivo experiments were performed comparing at least four age-and background-matched RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ mice each per assay and per time point where applicable. For cell culture experiments, at least three independent experiments with duplicates or triplicates in each experiment were performed. Student's t test or one-way ANOVA with Tukey's post hoc test was used to compare a control sample to test samples, with data presented as mean Ϯ S.E. Differences with p Ͻ 0.05 were considered significant. Data were analyzed using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA).

Lack of Caveolin-1 Solely in the RPE Impairs Rod-dependent
Visual Function-To investigate the contribution of caveolin-1 in the RPE to retinal functionality, we crossed mice expressing doxycycline-inducible Cre recombinase specifically in RPE (24) with mice carrying floxed alleles of caveolin-1 (Fig. 1A). Supplying doxycycline to females during pregnancy and until 7 days postnatally resulted in RPE CAV1 Ϫ/Ϫ offspring in which deletion of the floxed allele could be detected in eye cup tissue containing the RPE but not in the retinas from these mice (Fig. 1B). Furthermore, reduced levels of caveolin-1 protein specifically in the RPE (Fig. 1, compare C and D) were observed by immunofluorescence microscopy. Quantification of caveolin-1 in dissected RPE tissue revealed a reduction by 65% on average ( Fig. 1, G and H) in comparison with control mice not carrying the RPE-specific Cre. Control RPE CAV1 ϩ/ϩ and induced RPE -CAV1 Ϫ/Ϫ retinas were morphologically indistinguishable, indicating that doxycycline, Cre, or lack of caveolin-1 did not affect

Caveolin-1 Regulates Phagolysosomal Digestion
retinal development (Fig. 1, compare I and J). To test whether lack of caveolin-1 in RPE alone affected retinal function, we performed electroretinography. We observed significant reductions in rod photoreceptor responses, as measured by scotopic electroretinography from mice that were darkadapted overnight (Fig. 1, K and L). Maximum a-wave amplitudes, a direct measure of rod photoreceptor responses, were reduced significantly in RPE CAV1 Ϫ/Ϫ mice compared with controls from 307 Ϯ 15 V to 252 Ϯ 19 V (mean Ϯ S.E., p Ͻ 0.05, unpaired t test). b-Wave amplitudes, a measure of second-or-der retinal neuronal responses that depend on rod input, were reduced similarly reduced in RPE CAV1 Ϫ/Ϫ mice (Fig. 1, K and L), from 620 Ϯ 35 V in RPE CAV1 Ϫ/Ϫ to 481 Ϯ 41 V (mean Ϯ S.E., p Ͻ 0.05, unpaired t test). Therefore, eliminating caveolin-1 from the RPE has no obvious effect on cell viability and retinal morphology but is sufficient to reduce rod photoreceptor responses to light.
Lack of Caveolin-1 Solely in the RPE Leads to Delayed Phagosome Clearance in Vivo-We hypothesized that photoreceptors may lose function because RPE cells devoid of caveolin-1 fail to  MARCH 18, 2016 • VOLUME 291 • NUMBER 12 fully support outer segment renewal given its requirement for extensive membrane dynamics. The diurnal rhythm of retinal phagocytosis allows quantification of the engulfment and digestion capacity of the RPE by quantifying phagosomes in the RPE in situ at specific times in relation to light onset (5). Here we compared phagosome frequency in RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ RPE 0.5, 2, 4, 8, and 12 h after light onset and 1 h before light onset. A similar content of POS marker-positive phagosomes 1 h before and 0.5 h after light onset in these mice suggested that POS shedding and synchronized engulfment take place normally in RPE CAV1 Ϫ/Ϫ mice (Fig. 2, A and B). However, late in the day, when phagosome numbers in control RPE are low after digestion of the morning phagocytic load, we found phagosome numbers to be abnormally high in RPE CAV1 Ϫ/Ϫ mice (Fig. 2, A and B). This suggested a defect in POS-opsin digestion rather than engulfment. To assess whether CAV1 ablation leads to digestion defects for additional POS proteins, we double-labeled RPE whole mount preparations with transducin antibody and with the same opsin antibody as used initially for sections, clone B6-30 raised against the stable rhodopsin N terminus (22). Phagosomes were co-stained for both proteins at both early and late time points, and, as in sections, the phagosome load after light onset was similar in RPE CAV1 Ϫ/Ϫ and RPE CAV1 ϩ/ϩ RPE but elevated late in the day ( Fig. 2C). We also labeled phagosomes in whole mounts with antibody against the less stable opsin C terminus (clone ID4 (33,34)) and found consistent results (Fig. 2E). Therefore, regardless of phagosome protein marker, RPE CAV1 Ϫ/Ϫ RPE retained more undigested phagosomes at late time points (Fig. 2, D-F). The fraction of transducin and opsin N terminus double-positive phagosomes was similar irrespective of time point or genotype, suggesting an overall defect in phagolysosomal protein digestion (Fig. 2G). These in vivo data imply that caveolin-1 is not required for phagocytic recognition, binding, or internalization but important for phagolysosomal degradation.

RPE CAV1 Ϫ/Ϫ RPE in Vivo Exhibits Altered Diurnal Regulation of Lysosomal Enzymes but Normal Protease Recruitment to
Phagosomes-The activity of the aspartic protease cathepsin D is essential for digestion of POS-opsin (10,35,36). To test whether persistent phagosomes in RPE CAV1 Ϫ/Ϫ RPE were due to impaired protease activity, we compared the levels, activity, and recruitment of cathepsin D to phagolysosomes in RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ RPE in situ. Immunofluorescence microscopy of cathepsin D in tissue sections revealed higher levels of cathepsin D protein in RPE CAV1 ϩ/ϩ RPE during the early phases of phagocytosis, at 0.5 and 2 h, than 4 and 8 h after light onset (Fig. 3, A and B). The opposite temporal profile was observed for RPE CAV1 Ϫ/Ϫ RPE, with cathepsin D immuno-reactivity low at light onset and increasing by 8 h after light onset (Fig. 3, A and B). Therefore, lack of caveolin-1 leads to a reversal of the diurnal regulation of cathepsin D, likely through a form of organelle load feedback. The molecular mechanisms of this regulation will be a subject of future studies.
Cathepsin D is synthesized as a proform whose enzymatic cleavage in a low pH environment is required for the generation of enzymatically active cathepsin D. Because our antibody staining could not distinguish between inactive and active cathepsin D, we next measured specific enzymatic activity in tissue extracts. To discriminate cathepsin D activity in RPE/ choroid and neural retinas, we dissected neural retinas from posterior eye cups before tissue extraction. We found a similar trend for cathepsin D activity as seen for protein levels. Levels of active cathepsin D were elevated in RPE CAV1 ϩ/ϩ RPE early after light onset compared with 8 h later, whereas, in RPE CAV1 Ϫ/Ϫ RPE, cathepsin D levels were elevated at 8 h but not immediately after light onset (Fig. 3C). Changes in cathepsin D activity occurred specifically in the RPE/choroid because we did not detect differences in cathepsin D activity with time of day in neural retina extracts from either RPE CAV1 ϩ/ϩ or RPE CAV1 Ϫ/Ϫ mice (Fig. 3C). In addition, we tested the activity profile of another lysosomal enzyme, ␤-N-acetylglucosaminidase, which is essential for deglycosylation of phagocytosed POS-opsin (37, 38). The enzyme activity profiles of ␤-N-acetyl-glucosaminidase 0.5 and 8 h post-light onset mirrored that of cathepsin D for RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ RPE/choroid tissues fractions (Fig. 3D). Similar to cathepsin D, there was no change in enzymatic activity in neural retina fractions (Fig. 3D). Activities of both enzymes were ϳ10-fold lower in neural retina fractions than the respective activities in RPE/choroid fractions (Fig. 3, C  and D). Despite high enzyme activity in RPE CAV1 Ϫ/Ϫ RPE, undigested POS phagosomes remained at the 8-h time point tested. We conclude that the 8-h time point represents early or just prior to onset of phagosome digestion, which is accomplished by RPE CAV1 Ϫ/Ϫ RPE in time for the next POS challenge (Fig. 2B, Ϫ1 h time point). Next, we tested whether recruitment of cathepsin D to POS-opsin phagosomes may require caveolin-1 by double-labeling phagosomes with opsin N terminus and cathepsin D antibodies. In this experiment, we established the fraction of cathepsin D-positive phagosomes in each sample, a value that is independent of absolute phagosome numbers, which vary among the samples, as shown in Fig. 2. We found that the percentage of opsin/cathepsin D double-positive phagosomes was similar for RPE CAV1 ϩ/ϩ and RPE CAV1 Ϫ/Ϫ RPE at both early and late time points. This suggests that cathepsin D recruitment was not affected by caveolin-1 deficiency (Fig. 3, E and F).
Caveolin-1 Localizes to Phagolysosomes in RPE Cells-To directly test whether caveolin-1 contributes to POS digestion, we examined the role of caveolin-1 in phagocytic RPE cells in culture. The rat RPE-derived RPE-J cell line possesses and employs the same phagocytic pathway via ␣v␤5 integrin and Mer tyrosine kinase receptors as RPE in situ, and it has been used extensively for POS uptake studies previously (39 -42).
Here we used a two-phase phagocytosis assay to synchronize POS engulfment and processing (28), followed by triple-labeling of opsin, LysoTracker (to specifically label acidified phagolysosomes), and caveolin-1. We challenged polarized RPE cells with POS particles for 1 h at 20°C (pulse), a temperature that allows normal POS binding by the RPE-J cell line (via ␣v␤5 integrin) but does not allow internalization (43). Caveolin-1 did not colocalize with bound POS immediately after the pulse (Fig. 4A1). Caveolin-1 labeling was abundant in the cell cytoplasm, where, as expected, there were no engulfed POS (Fig. 4A2). After 2 h of further incubation at a permissive temperature (chase), we found that some (but not all) engulfed POS were acidified, but these did not co-stain with caveolin-1 antibody (Fig. 4B). After 6 h of chase, most POS-opsin-positive phagosomes were acidified, and caveolin-1, in part, co-localized with these phagolysosomes (Fig. 4C). Quantification of LysoTracker and caveolin-1 labeling of opsin-positive phagosomes confirmed that caveolin-1, but not LysoTracker, was largely absent from early engulfed phagosomes (Fig. 4D). In contrast, cathepsin D and the lysosomal organelle protein LAMP-1 localized to early as well as late engulfed POS (Fig. 4,  E-H). Therefore, caveolin-1 resides only on engulfed phagosomes at a specific stage but not on early internalized phagosomes. Notably, this suggests that engulfed phagosomes acquire caveolin-1 decoration de novo rather than carrying it with them from the internalized plasma membrane.

Increasing or Decreasing Caveolin-1 in RPE Cells in Culture Is
Sufficient to Accelerate or Slow Down Phagolysosomal Digestion, Respectively-To confirm our in vivo results and further specify the direct role of caveolin-1 in the phagocytic process, we next tested POS uptake and digestion by RPE cells in culture after manipulating caveolin-1 expression. First, we transiently overexpressed either WT or a mutant form of caveolin-1 containing F92A and V94A point mutations in the scaffolding domain or ␤-galactosidase as a control by recombinant adenovirus transduction (Fig. 5A). Mut-caveolin-1 is defective in plasma membrane localization and prevents caveola internalization of ligands but retains functionality at cytoplasmic sites (29,44,45). Exogenous WT-and mut-caveolin-1 were both tagged with the myc epitope and expressed at similar levels, increasing total caveolin-1 protein levels to ϳ2.4-fold (Fig. 5B). In synchronized POS phagocytosis assays, we observed similar initial POS binding immediately regardless of expression of WT-or mut-caveolin-1 (Fig. 5, C and D, lanes and columns,  respectively, 0 h chase). This confirms and extends our in vivo phagocytosis quantification, indicating that caveolin-1 is not involved in POS binding. However, quantification of POS-opsin after different periods of chase showed that cells overexpressing either WT-or mut-caveolin-1 eliminated engulfed POS significantly faster than control cells, specifically later in the chase (Fig. 5, C and D, lanes and columns, respectively, 5 and  8 h chase). Differences between caveolin-1-overexpressing and control cells were most pronounced at the late 8-h time point, indicating that increasing levels of either WT or mutant caveolin-1 is sufficient to accelerate phagosome digestion. These results suggest that the caveolin scaffolding domain is not involved in the ability of caveolin-1 to modulate phagocytosis. Next we examined the effect of silencing caveolin-1 expression on RPE phagocytosis. Infection of RPE cells with a caveolin-1silencing lentivirus reduced caveolin-1 protein levels by ϳ50% compared with control cells that received a non-targeting lentivirus (Fig. 5, E and F). Reducing the levels of caveolin-1 did not alter POS binding (Fig. 5, G and H, 0 h lanes and columns, respectively) but caused persistent POS-opsin in cells at later times of chase (Fig. 5, G and H, 5 and 8 h lanes and columns,  respectively). We conclude that decreasing the levels of caveolin-1 slows down phagosome digestion. Together, our in vivo and cell culture results are in complete agreement and indicate that caveolin-1 contributes to the digestion phase of phagocytosis. Moreover, its contribution does not require an intact scaffolding domain that promotes caveolin-1 function at the cell surface.
Silencing Caveolin-1 Is Sufficient to Alter Cathepsin D and ␤-N-acetylglucosaminidase Activity and Alkalinize Lysosomal pH in Resting RPE Cells and to Preclude Maturation and Activation of Cathepsin D during POS Phagocytosis-To determine how caveolin-1 affects phagolysosomal digestion, we next compared the lysosomal characteristics of caveolin-1-silenced and -competent RPE cells in culture. As observed for RPE in vivo, we found that decreasing caveolin-1 in cultured RPE caused moderate but significant decreases in the activity of the phagolysosomal enzymes cathepsin D and ␤-N-acetylglucosaminidase (Fig. 6, A and B). Moreover, lysosomes in caveolin-1-silenced cells were significantly less acidic than lysosomes in control Caveolin-1 Regulates Phagolysosomal Digestion FIGURE 4. Endogenous caveolin-1 localizes to POS phagosomes in RPE cells in culture during the digestion phase. Polarized RPE-J cells were challenged at 20°C with POS for 1 h (pulse) before washing and continued incubation at a permissive temperature for 2 or 6 h (chase, as indicated). All fields show representative single x-y planes from image stacks of labels, as indicated, of at least three independent experimental repeats. Insets show magnified areas. Arrowheads point out example POS. For merged images, x-z views are also provided. Scale bars ϭ 10 m. Scale bars for magnified insets ϭ 2 m. A, bound POS localize to the apical surface (A1) but not to the cell interior (A2) immediately after pulse. Caveolin localizes to the surface and cytoplasm but does not colocalize with POS (merge). The merged panel also shows nuclei (blue). B, after 2 h of chase, engulfed POS co-localize with LysoTracker (green, top panels) but not with caveolin-1 (green, bottom panels). Both LysoTracker and caveolin-1 are shown in green for the ease of showing co-localization. They were imaged in red and green channels, respectively, whereas POS were imaged in far-red. C, after 6 h of chase, POS (red) co-localize with both LysoTracker (green, top panels) and with caveolin-1 (green, bottom panels). Imaging was as in B. D, quantification of engulfed POS colocalizing with caveolin-1 (black columns), LysoTracker (light gray columns), or both (dark gray columns) after 2, 4, and 6 h of chase. Error bars show mean Ϯ S.E. (n ϭ 3). E and F, engulfed POS (red) co-localize with cathepsin D (green) after 2 h (E) and 6 h of chase (F). G and H, engulfed POS (red) co-localize with Lamp-1 (green) after 2 h (G) and 6 h of chase (H). cells (Fig. 6C). In contrast, lysosome numbers, vesicle area, and lysosomal subcellular distribution were not affected by caveolin-1 silencing (Fig. 6, D-F). We conclude that decreasing caveolin-1 is sufficient to impair lysosomal activity in resting RPE cells without phagocytic load. We also found that POS challenge and degradation did not change lysosomal pH in either control or caveolin-1-silenced cells, which remained different throughout the process (Fig. 6G). However, cathepsin D proteolytic processing to the mature form and a rise in cathepsin D enzymatic activity occurred in control RPE cells but not in RPE cells with reduced levels of caveolin-1 (Fig. 6, H-J). These data show that RPE cells up-regulate cathepsin D activity in response to POS phagolysosomal load using a mechanism that depends on caveolin-1. They suggest that the activity increase after light onset we observed in RPE CAV1 ϩ/ϩ in situ is a direct response by RPE cells to POS uptake rather than an entrained diurnal mechanism.
Silencing Caveolin-1 Affects the Constitutive Lysosomal Degradation of Transferrin Receptors without Affecting Steadystate Transferrin Receptor Levels-Finally, we tested whether a lysosomal degradation process unrelated to POS phagocytosis was also affected by the moderate changes in lysosomal func-tion observed in caveolin-1-silenced RPE cells in culture. We chose to study the lysosomal turnover of transferrin receptors, a recently described slow, constitutive degradation pathway (46). First, we established the time course of lysosomal degradation by RPE cells with normal levels of caveolin-1 (Fig. 7, A  and B). Upon treatment with cycloheximide to halt protein synthesis, cells showed normal levels of transferrin receptor for 1 and 2 h but significantly reduced levels after 3 and 5 h (Fig. 7B). At the 3-h (but not the 5-h) time point, the levels of caveolin-1 were normal, indicating that the treatment was not inducing dramatic cell damage (Fig. 7A). Therefore, in a second experiment, we chose a 3-h treatment time to compare the effects of cycloheximide treatment on transferrin receptor levels in control cells and cells with reduced levels of caveolin-1 (Fig. 7, C  and D). We found that cycloheximide treatment did not reduce transferrin receptor levels in caveolin-1-silenced cells like in control cells (Fig. 7D). Decreasing lysosomal activity by treatment with bafilomycin A eliminated the effect of cycloheximide on control cells, confirming the lysosomal dependence of transferrin receptor turnover. Bafilomycin A alone had no effect on receptor levels, implying that cells may adjust receptor synthesis to maintain normal receptor levels over the 3-h time period.

Caveolin-1 Regulates Phagolysosomal Digestion
This agrees well with the finding that solvent-treated cells with reduced levels of caveolin-1 maintain normal transferrin receptor levels (Fig. 7D, n.s.). These data suggest that RPE cells can compensate for the moderate lysosomal defect caused by reducing caveolin-1 expression to maintain normal transferrin receptor levels.

Discussion
Caveolin-1 is a ubiquitous protein that contributes to numerous cell and tissue functions. Global caveolin-1 Ϫ/Ϫ mice exhibit decreased rod-mediated function. However, caveolin-1 Ϫ/Ϫ rod photoreceptors retain the capacity to detect photons and stimulate phototransduction in principle (17). Here we show significant impairment in rod photoreceptor function in vivo in mice with normal caveolin-1 expression in rods and other neuroretinal cells but engineered to lack caveolin-1 specifically in the neighboring RPE. To our knowledge, our study is the first to demonstrate a non-cell autonomous effect of tissue-specific knockout of caveolin-1 and to illustrate the utility of such animal models. Although our data clarify that RPE caveolin-1 is essential for rod function, future studies quantifying vision of mice lacking caveolin-1 only in mature rods will need to examine whether or not rod caveolin-1 is also important for rod phototransduction.
Probing putative roles for caveolin-1 in support functions of the RPE for rods, we scrutinized diurnal outer segment clear-ance because it is critical for retinal homeostasis and relies on dynamics in membrane properties and signaling caveolin-1 is known to be able to facilitate (19,47,48). Our in vivo and cell culture experiments agree that caveolin-1 regulates the efficiency of phagolysosomal digestion by RPE cells. We detected caveolin-1 on phagolysosomes and found that expression of mutant caveolin-1 that does not reach the plasma membrane is effective in accelerating digestion. We conclude that a cytoplasmic pool rather than a plasma membrane pool of caveolin-1 is relevant to phagolysosomal digestion. To our knowledge, these are the first data implicating caveolin-1 in digestion in any phagocytic mechanism.
We found that the initial steps of clearance phagocytosis, recognition/binding and engulfment, are independent of caveolin-1 both in RPE in vivo and in culture. Earlier work has shown a reduction in phagocytosis of apoptotic cells or Escherichia coli by about 20% in thioglycollate-elicited macrophages from caveolin-1 Ϫ/Ϫ mice compared with macrophages from caveolin-1 ϩ/ϩ mice (49). This study did not probe distinct steps of phagocytosis but reported total particle uptake after shortterm particle incubation. The effects seen were therefore likely due to altered particle binding or engulfment. Therefore, although both RPE and elicited macrophages are avid phagocytes, only macrophages seem to employ caveolin-1 in the uptake phase of phagocytosis. These cell types also employ different phagocytic receptors. RPE cells like non-elicited macrophages and immature dendritic cells utilize a pathway dependent on ␣v␤5 integrin whereas elicited macrophages do not. We have shown previously that the particle recognition mechanism used by RPE cells requires activation of the particle recognition/binding receptor ␣v␤5 integrin by interaction with the tetraspanin CD81 (50). We speculate that tetraspanin and caveolin-1 may play similar roles in particle recognition/binding/engulfment in RPE cells and macrophages, respectively.
Mechanistically, our experiments show a role for caveolin-1 in regulating lysosomal function by promoting lysosomal acidification. Reduced lysosomal enzyme activities and more alkaline lysosomal pH even in resting RPE cells suggests that caveolin-1 does not only act on the lysosomal system in actively phagocytic cells. Indeed, we show that cells manipulated to lack protein synthesis do not degrade transferrin receptors, a constitutive lysosomal turnover process. However, unmanipulated cells maintain normal transferrin receptor levels. We conclude that RPE cells with reduced levels of caveolin-1 are able to maintain normal degradation processes despite moderate impairment of lysosomal organelles. In contrast, the high and acute load of protein and lipids in POS phagolysosomes may be especially sensitive to moderate changes in lysosomal function and pH and, therefore, allowed us to detect changes in the rate of digestion both in culture and in vivo. In our RPE cell culture systems, even a reduction by 50% of caveolin-1 was sufficient to significantly delay phagosome digestion. RPE phagosomes mature in two steps: initial weak to moderate acidification and a stronger secondary acidification that is required for digestion.
Step 1 occurs in a kiss-and-run type mechanism that transfers soluble enzymes like cathepsin D, whereas step 2 requires bona fide membrane fusion of lysosomes with phagosomes (6, 7). Our data show that caveolin-1 is not involved in recruitment of Stably selected populations of RPE-J cells expressing scrambled, non-silencing shRNA as control (scr sh, gray columns) or shRNA specific to caveolin-1 (cav-1 sh, black columns) were grown to post-confluence before treatment with cycloheximide (ϩ), with cycloheximide plus bafilomycin A (Bϩ), with solvent (Ϫ), or with solvent plus bafilomycin A (BϪ) before lysis and immunoblotting analysis of transferrin receptor (TfR), caveolin-1, tubulin, or actin as indicated. Blots show representative membranes of three independent experiments probed sequentially for transferrin receptor, caveolin-1, and a loading control. All error bars show mean Ϯ S.D. A, control cells were lysed after 1-5 h of treatment with (ϩ) or without (Ϫ) cycloheximide. B, quantification of transferrin receptor protein levels in cycloheximide-treated cells relative to untreated cells. Cells were treated as in A. Transferrin receptor levels were reduced significantly at 3 and 5 h of treatment, as established by Student's t test. **, p Ͻ 0.01; **, p Ͻ 0.001. C, comparison of transferrin receptor, caveolin-1, and actin protein content in cells with control or reduced caveolin-1 levels after 3-h treatment as indicated and described above. D, quantification of transferrin receptor protein levels in samples obtained as in C. All error bars show relative levels compared with levels in control cells treated with solvent, which was set as 1. Comparing treatment effects with solvent for each cell population, ANOVA shows a significant reduction only for cycloheximide treatment of control cells (**, p Ͻ 0.01; n.s., not significant). All other treatments did not differ from solvent treatment. Cycloheximide treatment of control cells also differed significantly from bafilomycin A alone and cycloheximide plus bafilomycin A (not indicated in the figure). Comparing cell populations, transferrin receptor levels did not differ between solvent-treated control and caveolin-1-silenced cells (n.s.), indicating that cells with reduced levels of caveolin-1 maintain normal steady-state transferrin receptor levels.
cathepsin D, which is acquired in step 1 by early phagosomes. Similar decreases in both cathepsin D and ␤-N-acetylglucosaminidase activity by lack or reduction of caveolin-1 in vivo or in cell culture further suggest that enzyme activity is not specifically altered. Phagosome maturation in step 2 is accelerated by caveolin-1, and decreasing levels of caveolin-1 even by only 50% increases lysosomal pH values. Changes in cathepsin D maturation and activation in response to a single burst of POS phagocytosis by RPE cells in culture show that RPE cells respond directly to uptake of POS into early lysosomal compartments with enzyme activity regulation compared with being entrained to a diurnal cycle of enzyme activity. Together, although the changes in lysosomal activity caused by loss of caveolin-1 in the RPE are modest, they have significant physiological consequences because of the very high and life-long phagocytic burden of post-mitotic RPE cells.
Our results contribute novel mechanistic insights to the growing body of evidence regarding the importance of intracellular caveolin-1 in endo/lysosomal organelle traffic and degradation. Fibroblasts and breast cancer cells in culture lacking caveolin-1 increase autophagy markers and LysoTracker-positive compartments in a response to increased oxidative stress (51,52). Notably, non-canonical autophagy shares components with clearance phagocytosis pathways (53,54). Recent studies have demonstrated the importance of caveolin-1 in multifunctional endosomal and lysosomal compartments (21,55). Our results extend these observations, showing that phagosomes are regulated by caveolin-1 as well. Moreover, we provide the first direct evidence that caveolin-1 is of physiological significance for intracellular degradative pathways in vivo.
Author Contributions-S. S. designed experiments, performed experiments, analyzed results, and wrote the paper. T. C. and X. G. performed experiments and analyzed results. G. C. and T. C. T. provided critical models and tools. M. H. E. designed experiments, performed experiments, and interpreted data. S. C. F. designed experiments, performed experiments, interpreted data, and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.