Light-induced Oxidation of Photoreceptor Outer Segment Phospholipids Generates Ligands for CD36-mediated Phagocytosis by Retinal Pigment Epithelium

Materials

Na¹²⁵I was supplied by ICN Pharmaceutical, Inc. PLPC, PAPC, DHA-PC, 1,2-ditridecanoyl-sn-glycero-3-phosphatidylcholine (DTPC), 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphatidylcholine (lyso-PC), and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine (POPC), were obtained from Avanti Polar Lipids Inc. (Alabaster, AL). Synthetic oxPCs were prepared as previously described (24, 28, 29). LDL was prepared from human plasma and oxidized by exposure to leukocyte myeloperoxidase in the presence of nitrite as previously described (22). Other chemicals were obtained from Sigma unless specified.

Photo-oxidation of DHA-PC

DHA-PC (1 mg) was hydrated at 37 °C for 1 h in argon-sparged phosphate buffer (1 ml) supplemented with DTPA to minimize transition metal-induced auto-oxidation. The hydrated DHA-PC were then made into unilamellar vesicles by extrusion using an Avanti Mini-Extruder (Avanti Polar Lipids, Inc). The vesicles (400 μl) were kept in the dark under argon blanket at room temperature until use the same day. Photo-oxidation was performed by transfer of vesicles into a quartz tube placed in the center of a Rayonet photochemical reactor with three 80-watt low pressure mercury UV (254 or 350 nm) lamps (Southern New England Ultraviolet, Midtown, CT). Reaction mixtures were incubated in air with temperature maintained at 28 °C. Sampling (30 μl) was performed at various times, and the reaction was stopped by adding catalase (10 μl and 150 μM final) and butylated hydroxytoluene (2 μl and 40 μM final). For quantification purposes, lipid extraction was performed immediately after addition of DTPC (300 ng) using the method of Bligh and Dyer (30). Extracts were dried under N₂ and stored in vials sealed under argon at −80 °C until analysis.

Mass Spectrometric Analysis of Light-irradiated DHA-PC

Mass spectrometric analyses of irradiated DHA-PC was performed on an ABI 365 mass spectrometer equipped with Ionics EP 10+ upgrade (Concord, Ontario, Canada), interfaced with a Waters 2790 (Waters, Milford, MA) HPLC system. Oxidized DHA-PC (30 μg) was dissolved in 1 ml of methanol. The solution (10 μl) was diluted with 140 μl of methanol containing 10% water. This solution (50 μl) was chromatographed on a Prodigy ODS C18 column (150 × 2 mm, 5 μm, Phenomenex, Torrance, CA) with a binary solvent gradient, starting from 100% water for the first 5 min, then a linear gradient to 100% methanol in 15 min, and then 100% methanol for 20 min, at a flow rate of 0.2 ml/min. All mobile phase solvents contained 0.2% formic acid. The total ion current was obtained in the mass range of m/z 200–1000 using a cone energy of 30 V in the positive ion mode. Three kV was applied to the electrospray capillary.

Characterization of DHA-PC Oxidation Products as CD36 Ligands

Specific binding of synthetic oxPC species were determined in competition binding assays employing ¹²⁵I-labeled LDL exposed to the myeloperoxidase/H₂O₂/ ${\text{NO}}_2^{-}$ system ([¹²⁵I]NO₂LDL) and CD36 cDNA-transfected 293 cells, as previously described (25). Presentation of the individual oxPC was normalized by incorporation as a low mol % component (10%) of vesicles generated with POPC as carrier (25). CD36 independent binding of [¹²⁵I]NO₂LDL was assessed in control vector-transfected 293 cells and subtracted from binding to CD36-transfected 293 cells, whose binding was typically <10–20% that of CD36-transfected cells. Each measurement was performed in triplicate, and results shown represent averages from at least three independent experiments.

RPE Cells

An SV40-transformed cell line derived from rat RPE, RPE-J cells, has been previously characterized and is frequently used to study phagocytic function (31). These cells were routinely maintained at 32 °C in DMEM supplemented with 4% fetal calf serum. Cells were seeded at 50% confluence on 5-mm glass coverslips in 96-well plates and grown for 6–7 days before use at a density of 3.3 × 10⁵ cells/well. Under these conditions, RPE-J cells possess epithelial morphology and apical expression of phagocytic proteins αvβ5 and MerTK (32). Where indicated, primary RPE cells were used. These were obtained from both CD36 null mice and wild-type C57Bl/6 controls. CD36 nulls were backcrossed seven times onto a C57Bl background. Mice were sacrificed by CO₂ asphyxiation at 12–14 days of age. Following eyeball enucleation and removal of cornea, iris, lens, and vitreous, eyecups were incubated in 1 mg/ml hyaluronidase in Ca²⁺ and Mg²⁺-free Hanks' buffered saline solution to detach neural retinas from the underlying RPE. A circumferential cut was made just below the ora serrata to remove neural retinas from eyecups. Eyecups with exposed RPE apical surface were further incubated in 2 mg/ml trypsin in Hanks' buffered saline solution containing Ca²⁺ and Mg²⁺ for 45 min at 37 °C. Sheets of RPE were peeled off Bruch's membrane, trypsinized, pelleted, resuspended, and seeded as patches in DMEM supplemented with 10% fetal calf serum. Primary RPE cells were routinely used 4–7 days after preparation. To characterize epithelial morphology, we fixed samples with ice-cold methanol and labeled nuclei with 1 μg/ml propidium iodide and tight junctions with ZO-1 antibodies (Zymed Laboratories Inc.) and secondary alexa488-conjugated antibodies (Molecular Probes). Laser scanning confocal microscopy using a Leica TSP2 system revealed no significant difference in yield, attachment, spreading, or pigmentation of RPE derived from CD36 null compared with wild-type controls.

OS Phagocytosis Assays by RPE Cells

OSs were isolated on 25–60% sucrose gradients from fresh bovine retinas according to established protocols (32) and labeled with fluorescein isothiocyanate (FITC, Molecular Probes, Eugene, OR) as previously described (7). RPE cells on 5-mm coverslips in 96 well plates were incubated with FITC-labeled OSs (5 × 10⁶ OSs in 40 μl/well, in excess of 10 OSs/RPE cell) in DMEM with 5% fetal bovine serum in the presence of lipids or ethanol solvent at a concentration of 1%. At the end of the incubation time, excess OSs were removed and cells were washed. To quench external OSs, samples were incubated with 0.4% trypan blue for 10 min, before washing and fixation in ice-cold methanol. Nuclei were labeled with 1 μg/ml propidium iodide. FITC and propidium iodide fluorescence of coverslips mounted on slides was quantified by fluorescence scanning on a STORM 860 scanner (Molecular Dynamics). Data were evaluated using Molecular Dynamics ImageQuaNT version 1.2 software. Samples were also evaluated by laser scanning confocal microscopy. Individual X-Y scans were acquired from representative fields. All studies included parallel control cytotoxicity and apoptosis assays to evaluate if exposures reduced cell viability. We did not observe toxicity of compounds or solvents on any of the RPE preparations when testing cell viability and signs of apoptosis by trypan blue dye exclusion and nuclear DNA fragmentation, respectively.

Light Exposure of Sprague-Dawley Rats and Retina Harvest

Weanling male Sprague-Dawley rats were obtained from Harlan Inc. (Indianapolis, IN) and maintained in a darkened room for 60 days. Animals were exposed to green light (∼1200 lux; 200 microwatts/cm²) for periods of either 1 or 4 h, as described previously (33), at which time animals were euthanized and retinas rapidly harvested as described (34). Excised retinas were immediately placed into argon-sparged saline (0.9%), supplemented with antioxidant mixture (100 μM butylated hydroxytoluene and 2 mM DTPA, pH 7) and rinsed of any free blood. Individual retinas were then rapidly submerged in 1-ml cryo vials filled with 0.75 ml of argon-sparged anti-oxidant mixture-containing saline, headspacepurged with argon, capped, snap frozen in liquid N₂, and then stored at −80 °C until analysis.

Separation of Phospholipids from Retina

To prevent artificial oxidation of retina, all procedures were performed rapidly within a dark room with dim red light. All glassware tubes and pipette tips were washed with nitric acid to remove trace transition metals, extensively rinsed with Chelex-treated water containing 1 μM DTPA, and then rinsed with pure Chelex-treated water. Plastic tips were further rinsed in methanol and air-dried prior to use. All glassware was baked at 500 °C for 24 h to remove residual organics. Retinas were removed from the −80 °C freezer and thawed at 4 °C. Tissue was homogenized with a Teflon pestle in a form-fitting plastic vial with argon blanket. Vial contents were transferred to 10-ml threaded glass test tubes equipped with teflonlined caps, and then lipids within samples were extracted using the method of Bligh and Dyer (30) under argon after addition of 5 ng of DTPC as internal standard, followed by immediate HPLC with on-line electrospray ionization tandem mass spectrometry (LC/ESI/MS/MS) analysis. Extraction efficiencies of each of the oxPCs indicated (Fig. 1A) were determined using synthetic authentic oxPC species and the corresponding precursor (PA-PC, PL-PC, and DHA-PC) and DTPC as internal standard. After extraction, the amounts recovered were determined by LC/ESI/MS/MS.

Mass Spectrometric Analysis of Phospholipids

Lipid extracts were re-dissolved in methanol:water (90:10) and LC/ESI/MS/MS performed using a Prodigy ODS C18 column for separation of lipids and the HPLC solvent system described above. Mass spectrometric analyses were performed online using electrospray ionization tandem mass spectrometry in the positive ion, multiple reaction monitoring (MRM) mode (cone energy 30 V/collision energy 25 eV). The MRM transitions used to detect individual oxidized phospholipids were the mass to charge ratio (m/z) for the molecular cation [M+H]⁺ and their daughter ion 184, the phosphatidylcholine group. Calibration curves for quantitative analyses of individual oxPC molecular species were constructed by mixing a fixed amount of internal standard (DTPC) into various amounts of authentic oxPC samples (35).

Nitrotyrosine Quantification

Protein-bound nitrotyrosine in retina was quantified by stable isotope dilution LC/MS/MS as previously described (36) on a triple quadrupole mass spectrometer (API 365, Applied Biosystems, Foster City, CA) with Ionics EP 10+ upgrade (Concord, Ontario, CA) interfaced to a Cohesive Technologies Aria LX Series HPLC multiplexing system (Franklin, MA). Synthetic ¹³C₆-labeled standard was added to delipidated tissue homogenates and used as internal standards for quantification of natural abundance analytes. Simultaneously, a universally labeled precursor amino acid, [¹³C₉,¹⁵N₁]tyrosine, was added. Proteins were hydrolyzed under argon atmosphere in methane sulfonic acid, and then samples were passed over mini solid-phase C18 extraction columns (Supelclean LC-C18-SPE minicolumn, 3 ml, Supelco, Inc., Bellefone, PA) prior to mass spectrometry analysis. Results are normalized to the content of the precursor amino acid tyrosine, which was monitored within the same injection. Intrapreparative formation of nitro[¹³C₉,¹⁵N]tyrosine was routinely monitored and negligible (i.e. ≪5% of the level of the natural abundance product observed) under the conditions employed.

Evolution Profiles of Structurally Defined oxPCs from Light-exposed DHA-PC

We previously characterized a conserved high affinity CD36 binding motif in a family of oxidized choline glycerophospholipids derived from PAPC and PLPC (oxPC_CD36) consisting of an oxidatively truncated sn-2 acyl group with a terminal γ-hydroxy(or oxo)-α,β-unsaturated carbonyl (Fig. 1A) (25). According to mechanistic predictions, the oxidation of DHA-PC should produce compounds structurally analogous to these ligands (28). Photo-induced oxidation products of DHA-PC, however, have not yet been described. We synthesized the presumptive truncated peroxidation products of DHA-PC that might serve as CD36 ligands (see Fig. 1A for structures) and confirmed their structures by NMR and mass spectrometry (24, 28, 35). LC/ESI/MS/MS-based methods were then developed for their simultaneous quantification (Fig. 1B). To test the hypothesis that these species are formed by photo-oxidation of DHA-PC, small unilamellar vesicles comprised of DHA-PC were either exposed to light (350 nm) or incubated in the dark and the evolution profiles of specific oxPC quantified by LC/ESI/ MS/MS (Fig. 1C). In buffer free of transition metal ions, DHA-PC vesicles maintained in the dark failed to generate significant levels of oxidation products. In marked contrast, irradiation with 350 nm light resulted in immediate oxPC production (Fig. 1C). HOHA-PC reached a maximum observed yield (1.3%) after 4 h and then started to decrease. Production of other species such as KOHA-PC, HHdiA-PC, KHdiA-PC, and OB-PC increased quickly during the first 4 h and subsequently slowed. Production of S-PC displayed an almost linear increase with time, and reached 1.7% yield after 28 h, suggesting that S-PC may be a final unhydrolyzed truncated product, and other oxPCs from DHA-PC apparently are further fragmented or oxidized into S-PC. Consistent with this hypothesis, light exposure to individual preparations of POPC vesicles containing 25 mol % of other DHA-PC oxidation products (e.g. HOHA-PC, KOHA-PC, HHdiA-PC, and KHdiA-PC) resulted in production of S-PC (data not shown).

oxPCs Formed during Photo-oxidation of DHA-PC Are High Affinity CD36 Ligands

The ability of DHA-PC derived oxPC to bind CD36 was first evaluated by competitive binding assays using the known CD36 ligand [¹²⁵I]NO₂-LDL and CD36 cDNA transfected 293 cells. Each synthetic lipid was mixed at 10 mol % within carrier POPC unilamellar vesicles. In contrast to vesicles containing native DHA-PC, γ-hydroxy-(oxo)-α,β-unsaturated oxPCs dose-dependently inhibited the binding of [¹²⁵I]NO₂-LDL to CD36 (Fig. 2A). Oxidized species possessing a terminal carboxylic acid moiety inhibited more effectively than those possessing a terminal aldehyde (HHdiA-PC versus HOHA-PC), and γ-keto compounds were more potent than γ-hydroxy compounds (KOHA-PC versus HOHA-PC), similar to prior structure/function studies observed with oxPC_CD36 derived from PLPC and PAPC (25, 26). The more terminally oxidized and truncated oxidation products of DHA-PC, S-PC and OB-PC, failed to significantly inhibit when presented at 10 mol % (Fig. 2A) but showed CD36 binding activity at higher levels (100 mol %, 20 μM lipids dissolved in phosphate-buffered saline buffer containing 10% ethanol), with S-PC binding more avidly than OB-PC relative to DHA-PC (Fig. 2B).

oxPCs Inhibit Uptake of OSs by RPE

Multiple oxPCs derived from DHA-PC, PLPC, and PAPC were tested for their ability to inhibit binding and internalization of isolated OSs by RPE cells in culture. The relatively non-oxidizable phospholipid, POPC, was used as a negative control. Because we suspected that some oxPCs might be enzymatically hydrolyzed into lyso-PC in cell culture, the activity of lyso-PC was also tested. RPE-J cells were incubated with FITC-labeled OSs with or without the presence of the indicated phospholipids. At 1.5 h of OSs incubation RPE-J cells bound FITC-OS, but less than 10% were internalized. As expected, none of the oxPC_CD36 examined had an appreciable effect on binding of OSs to RPE-J cells (Fig. 3, p > 0.05 for all lipids versus control). At 1.5 h OB-PC, S-PC, lyso-PC, and POPC showed no effect on uptake of OS, whereas all γ-hydroxy(oxo)-α,β-unsaturated oxPCs evaluated, including the ketoaldehydes (KODA-PC, KOOA-PC, and KOHA-PC), ketoacids (KDdiA-PC, KOdiA-PC, KHdiA-PC), and hydroxy acid, inhibited uptake by 10–20% (p < 0.05, Fig. 3, middle panel). After longer incubation (3.5 h) the most active CD36 ligands, such as KDdiA-PC, inhibited uptake by ∼50% (p < 0.05), whereas POPC and lyso-PC, which may be a hydrolysis product of oxPCs, had little effect. Parallel control studies demonstrated no toxicity of oxPC exposure to RPE-J cells as assessed by trypan blue dye exclusion and nuclear DNA fragmentation (data not shown). Collectively, these results confirm that the functional groups at the sn-2 position of the oxPCs are important for RPE-mediated internalization of OS.

CD36 Recognition of Structurally Specific Oxidized Phospholipids and RPE Phagocytosis of OS

To evaluate the role of CD36 in RPE recognition of oxidized PC species within isolated OS, primary cultures of RPE were prepared from wild-type and CD36 null mice. First, we used our established phagocytosis assay to compare FITC-OSs uptake by wild-type and CD36 null RPE. While at the end point CD36 null RPE cells internalized similar numbers of OSs as wild-type RPE (Fig. 4, A and B), there was a difference in the lag period prior to phagocytosis in wild-type versus CD36 null cells (CD36 null cells demonstrated longer lag period, data not shown). CD36 null and WT RPE demonstrated no detectable differences in either expression or localization of the tight junction marker ZO-1, as observed by fluorescence microscopy (Fig. 4, C and D). Similarly, comparable cell morphology and pigmentation were observed in wild-type versus CD36 null RPE by transmission white light microscopy (Fig. 4, E and F). The overlay of tight junctions (green), cell nuclei (red), and transmitted light (blue) in Fig. 4(G and H) further confirms that RPE cells from both genotypes had similar size, morphology, pigmentation, and epithelial organization.

We next compared sensitivity of OSs phagocytosis by wild-type and CD36 null RPE to competition by CD36 phospholipid ligands. Although POPC and lyso-PC showed little effect on OS uptake by RPE, all oxPC_CD36 examined (KDdiA-PC, KODA-PC, KOdiA-PC, KOOA-PC, KHdiA-PC, and KOHA-PC) significantly inhibited OS uptake by RPE from wild-type mice but not in RPE isolated from CD36 null mice (Fig. 5). These results are consistent with the hypothesis that CD36 partially contributes to OS phagocytosis through recognition of structurally specific oxPCs on the surface of OSs.

oxPC_CD36s Are Present in Rat Retina, and Their Production Is Promoted by Physiological Light Exposure

It has been suggested that light exposure may initiate the peroxidation of polyunsaturated fatty acids such as arachidonic acid (37) and DHA in retina (38), although we are unaware of any direct demonstration of structurally defined phospholipid oxidation products formed within retina in vivo following light exposure. We hypothesized that the light-triggered peroxidation of the phosphocholine esters of polyunsaturated fatty acids (PAPC, PLPC, and DHA-PC) in retina would produce oxPCs, including some that have already been detected in vitro or in other biological systems such as atherosclerotic lesions (26). To examine if oxPCs are present in intact retinas and if light exposure promotes their production in vivo, retinas harvested from albino rats with or without light treatment were analyzed by LC/ESI/MS/MS. Three groups of rats (eight in each group) were raised in the dark for 60 days, and two groups were then treated with green light (550 ± 75 nm) for 1 and 4 h, respectively. The intensity of light used, ∼200 microwatts/cm² corneal irradiation, represents ∼2-to 3-fold higher level than that of typical ambient laboratory conditions. To prevent ex vivo lipid peroxidation retinas were harvested immediately in the dark under continuously argon-sparged buffer supplemented with anti-oxidants, and lipids were extracted and preserved in amber-walled vials under inert atmosphere with anti-oxidants present and then rapidly analyzed utilizing LC/ESI/MS/MS. Eighteen distinct oxPCs (see Fig. 1A for structures) and their corresponding precursors, PLPC, PAPC, and DHA-PC, were monitored simultaneously. Eight oxPCs species were reproducibly detected in dark-adapted rat retinas. Among the oxPCs derived from PLPC, ON-PC and A-PC were detected (Fig. 6); among the oxPCs derived from PAPC, OV-PC and G-PC were detected (Fig. 6A); and among the oxPCs derived from DHA-PC, OB-PC, S-PC, KOHA-PC, and HOHA-PC were detected (Fig. 6B). Retention time and characteristic parent ion → daughter ion transitions of each compound were collectively used to identify the oxPC species.

Quantitative data for each detected oxPC within individual retina are presented in Fig. 7. The content of each oxPC is reported as the ratio of oxPC to its precursor (e.g. HOHA-PC/DHA-PC, millimole/mol). Light exposure significantly increased the levels of most of the oxPC monitored, including OB-PC, HOHA-PC, KOHA-PC, OV-PC, G-PC, KOOA-PC, A-PC, and ON-PC (p < 0.05 each). The failure to detect some oxPCs may be a consequence of either their low abundances, further oxidation into alternative species, or covalent adduction to nucleophilic groups such as retinal proteins. As observed with the in vitro experiments, the more terminal oxidation cleavage products, including simple aldehydes (ON-PC, OV-PC, and OB-PC) and acids (A-PC, G-PC, and S-PC), were the most abundant species detected, and they collectively accounted for in excess of 15–20 mol % of their corresponding unoxidized precursors.

In a final series of studies, the role of nitric oxide (NO)-derived oxidants as a potential participant in retinal oxidant stress following light exposure was explored, because recent studies suggest that enhanced nitrative stress may participate in retinal pathologies, such as age-related macular degeneration (39). Stable isotope dilution tandem mass spectrometry-based analyses of retinal proteins confirmed time-dependent increases in retinal protein nitrotyrosine content following light exposure (Fig. 8).