Apolipoprotein Localization in Isolated Drusen and Retinal Apolipoprotein Gene Expression

Abstract

purpose. To evaluate apolipoprotein (Apo) gene expression in native human retinal pigment epithelium (RPE) and neurosensory retina and to detect apolipoproteins within age-related, extramacular drusen.

method. Drusen were isolated manually from 10 eyes of 10 donors (age range, 58–93 years) with grossly normal maculas that were preserved in 4% paraformaldehyde within 6 hours of death. In cryosections of druse-enriched pellets (6–57 drusen per eye), the Apos A-I, A-II, B, C-I, C-II, C-III, E, and J were detected by indirect immunofluorescence. Two graders assessed the prevalence and pattern of immunoreactivity. mRNA transcripts were detected by reverse-transcription polymerase chain reaction (RT-PCR), with human hepatoma HepG2 cells as the positive control.

results. Extramacular drusen were classified in two groups on gross appearance: transparent with a reflective shell and cloudy. The proportion of the latter increased significantly with age. All Apos examined were detectable, in descending order of prevalence: ApoE (99.5%), J (99.5%), C-I (93.1%), B (80.4%), A-I (61.0%), A-II (59.2%), C-II (57.7%), and C-III (16.6%). Immunoreactivity was either diffusely distributed throughout the drusen (56.7%), confined to the druse rim (16.0%), or both (21.2%). Six percent displayed evidence of organized substructure reminiscent of active remodeling. The proportion of diffusely labeled drusen decreased significantly with age for ApoE (P = 0.034) and ApoE/C-I combined (P = 0.027). RT-PCR products for Apos C-I, C-II, E, and J were found in retina and RPE; for ApoA-II in the retina only. The ApoC-III message was undetectable.

conclusions. To an emerging model of an RPE-secreted large lipoprotein particle implied by previous work, this study adds ApoC-I and ApoC-II, major modulators of lipoprotein lipase activity, and confirms previously demonstrated Apos A-I, B-100, and E. It is possible that a neutral lipid-rich druse shell containing Apos will be visible in the living fundus.

Age-related maculopathy (ARM), the major cause of new, untreatable vision loss among the elderly of industrialized nations, is an obscure degeneration involving the retinal pigment epithelium (RPE), Bruch’s membrane (BrM), and the choriocapillaris, with secondary sight-threatening effects on the photoreceptors. Although the ingrowth of leaky choriocapillary endothelial cells through BrM occurs in a minority of patients, it accounts for most vision loss. Events setting the stage for choroidal neovascularization are poorly understood. Without new information, preventions rather than treatments of complications are difficult to envision. 

A prominent clinical and histopathologic sign of ARM is the presence of drusen, histologically defined as focal deposits of heterogeneous debris external to the RPE basal lamina and internal to the inner collagenous layer of BrM. ^{1 2} Drusen increase with age ^{3 4} and in patients with ARM, drusen of large size, sloping sides (“soft”), and confluence are predictors of advancement to neovascular ARM. ⁵ Prophylactic laser photocoagulation of drusen has been advanced as a method to regress these lesions, although trial results have not yet borne out this expectation. ⁶ From a mechanistic perspective, an established research paradigm in ARM pathobiology is the detection of proteins within drusen accompanied by reverse transcription–polymerase chain reaction (RT-PCR) of the RPE and neurosensory retina, to evaluate the potential for intraocular biosynthesis of these proteins, many of which are secreted in abundance by liver or other organs. This approach, validated by proteomics, has identified in drusen acute-phase reactants (vitronectin), markers of inflammation (C-reactive protein), and members of the complement family of innate immunity proteins (e.g., C3 ^{7 8 9 10 11 12} ). As variants in the gene encoding complement factor H, a druse-associated molecule, also confer increased risk for ARM in US populations, ^{7 13 14 15} this reductionist approach has represented a fruitful avenue for identifying ARM-perturbed pathways with potential as therapeutic routes. 

Recent work has spotlighted a role for lipoproteins and neutral lipids in the formation of these characteristic lesions. Lipoproteins are multimolecular assemblies composed of lipid and protein bound by noncovalent forces. Their general structure is that of a microemulsion formed from an outer layer of phospholipids, unesterified cholesterol (UC), and apolipoproteins (Apos), with a core of neutral lipids, predominately esterified cholesterol (EC) and triglycerides (TG). Drusen contain neutral lipid that binds oil red O, EC that binds filipin after extraction with ethanol and hydrolysis with cholesterol esterase, and the Apos A-I, B, and E. ^{9 16 17 18 19 20 21 22 23} Further, RPE expresses mRNA transcripts for genes encoding these Apos, and importantly, also for microsomal triglyceride transfer protein ²⁴ (the abetalipoproteinemia gene product), required for the assembly of an ApoB-containing lipoprotein. ²⁵ Histochemical and ultrastructural signatures of lipoprotein particles are found in BrM of elderly humans and macaques, ^{19 26 27} suggesting a major constitutive pathway for neutral lipid deposition. Animal models with genetically or diet-induced hypercholesterolemia do not yet share these BrM signatures, ^{28 29 30 31} adding to the circumstantial evidence that a large, possibly novel lipoprotein is produced within the eye. The functional significance of such a particle remains to be determined, but it is a plausible pathway for RPE-mediated release of fatty acids derived from phagocytosed outer segment phospholipids as TG. 

In the present study, we used the druse-associated molecule approach to frame a more detailed picture of a postulated lipoprotein that can inform future studies designed to elicit its secretion in culture. Specifically, we used immunofluorescence to label drusen for previously unexamined Apos A-II, C-I, and C-II, as well as already demonstrated Apos A-I, B, E, and J, directing attention to prevalence and staining pattern. To ensure a druse-enriched sample, drusen were manually isolated and pelleted. We sampled extramacular retina (i.e., peripheral to sites used by clinical macular grading systems), where these lesions are abundant ^{32 33} and exhibit the same proteins, if not the same lipid composition and autofluorescence properties, as macula. ^{21 33} We used RT-PCR to examine Apo gene expression in native human RPE, neurosensory retina, the ARPE-19 cell line, and human hepatoma HepG2 cells. We found evidence of local biosynthesis of ApoC-I and C-II, strengthening the inferred evidence of an intraocular lipoprotein. The within-druse distribution of Apos may have consequences for druse visibility in the living fundus and for understanding the different stages of druse formation. 

Methods

Human Tissues and Preservation of Eyes

The age, gender, and race of 10 eye bank donors with grossly normal maculas are shown in Table 1 . Eyes obtained from donors within 6 hours of death were preserved by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer (PB), for 6 to 16 hours after removal of the cornea, and stored in 1% paraformaldehyde at 4°C until used (median preservation to experiment interval, 18.2 months; range, 10.8–34.6). Use of human tissues in this study was approved by Institutional Review at the University of Alabama at Birmingham and conformed to the guidelines in the Declaration of Helsinki for research involving human tissue. 

Isolation and Sectioning of Drusen

Under stereomicroscopic guidance at 50× to 63× magnification, RPE-encased drusen were removed from BrM with an insect pin attached to a dowel and collected with a borosilicate glass pipette heat-stretched to an inner diameter of ≤50 μm and attached to a PC-S3 salt-bridge electrode (E. W. Wright, Guilford, CT). The electrode was in turn attached to polyethylene intramedic tubing (PE-200) and a 5-mL syringe, which provided mild suction to retrieve drusen and place them within 600-μL Beem capsules (Electron Microscopy Sciences, Hatfield, PA). Quadrants of peripheral retina-choroid-sclera with abundant drusen were pinned to a prepared wax surface in a Petri dish. The retina was removed with forceps, and the RPE-choroid–lined scleral cup was flushed and refilled with PB saline (PBS). With the insect pin tool, drusen were cleaved from the RPE/BrM interface. Usually drusen were easily dislodged when lightly pushed with the pin tip. Cleaved drusen were counted, gathered in small groups on the RPE surface (see Fig. 1A ), and carefully drawn into the end of a 0.58-mm-inner-diameter glass pipette (no. 6010; AM Systems, Everett, WA), keeping them from entering the salt bridge. Then, under reverse suction, pipette contents were allowed to fall slowly to the capsule bottom through PBS (Fig. 1B) . From a swath of peripheral RPE-BrM, drusen selected only on the basis of visibility and detachability from BrM were thus removed. Starting sample sizes are indicated in Table 1 . The median number of drusen collected from individual eyes was 181 (range, 82–880). 

Drusen clusters were washed in PBS twice, soaked in 1% sodium borohydrate (5 mL PBS, 5 mL deionized H₂O, 0.1 g NaBH) for 1 hour to enhance antigenicity, ^{34 35} covered by slow addition of 400 μL of a 0.75% agar/5% sucrose solution, then set aside to congeal. The Beem capsule was slit open, and the agar pellet was trimmed with a razor blade, placed in a rectangular cryomold, infiltrated with ascending concentrations of sucrose in 0.1 M PB (10%, 20%, 30%), 4:1 30% sucrose-mountaing compound (Histoprep SH7512D; Fisher Scientific, Pittsburgh, PA), 2:1 30% sucrose-mounting compound (Histoprep; Fisher Scientific), and frozen in liquid nitrogen. Ten-micrometer-thick sections were collected on slides (SuperFrost Plus; Fisher Scientific) and stored at 20°C until used. The number of drusen visible in a section ranged from 6 to 57 in individual eyes. 

Indirect Immunofluorescence

Cryosections were removed from −20°C storage, heated for 30 minutes at 50°C to 55°C, and rinsed briefly with PBS. After 10 minutes’ incubation with 0.2% Triton-100/PBS, slides were incubated with primary antibody against Apos overnight at 4°C. Goat anti-human ApoA-I (1 mg/mL, 1:200), ApoA-II (1 mg/mL, 1:100), ApoB (1 mg/mL, 1:250), ApoC-I (1 mg/mL, 1:100), and ApoJ (1 mg/mL, 1:200) were purchased from Biodesign (Saco, ME). Goat anti-human ApoC-II (1 mg/mL, 1:50) and ApoC-III (1 mg/mL, 1:100) were purchased from Chemicon (Temecula, CA). Goat anti-human ApoE (92 mg/mL, 1:500) was purchased from Calbiochem (San Diego, CA). According to the manufacturers, cross-reactions with other Apos or human serum proteins, as determined by Western blot analysis or ELISA, were negligible. Identical concentrations of goat IgG were used on negative control sections. After three 10-minute washes with 0.1% Tween-20/PBS (PBST), slides were incubated with biotinylated anti-goat IgG 1:500 for 2 hours at room temperature. After three PBST washes, sections were incubated with rhodamine Red-X-conjugated Streptavidin (1:500; Jackson ImmunoResearch, West Grove, PA) for 30 to 60 minutes. After three PBS washes, coverslips were mounted with 1,4-diaza bicyclo[2.2.2]octane in glycerol (Slow Fade Light Antifade Kit; Invitrogen, Eugene, OR). 

Immunofluorescence: Processing, Imaging, and Evaluation

Sections were examined by microscope (Eclipse 80i; Nikon, Melville, NY), a 10× planapo or 40× plan fluor objective, and two filter systems (in nanometers, excitation-dichroic-barrier): 540/25–565-630/60 and 480/30–505-535/40 for rhodamine and autofluorescence, respectively. Images were captured with a digital camera (Retiga 4000R Fast; Q Imaging, Burnaby, BC, Canada). Drusen immunofluorescence was evaluated from saved images at 270× on a computer monitor by two independent graders (C-ML, CAC). Immunofluorescence and differential interference contrast (DIC) images of the same field, combined as separate layers (in Photoshop, Adobe Systems, Mountain View, CA) were toggled back and forth to facilitate identifying unlabeled tissue. Each druse, numbered on an image hard copy, was evaluated as positive or negative, with reference to a same-exposure-length image of a control (IgG) section also visible on the monitor. Intensity values at least threefold higher in experimental sections than control sections (determined with Photoshop; Adobe Systems) were considered specific. Individual graders’ scores were combined, and each druse was considered positive (graders agreeing), negative (graders agreeing), or uncertain (graders disagreeing). In reporting the percentage of immunoreactive drusen, positive labeling and uncertain labeling are combined. Digitized images were composited (Photoshop CS; Adobe Systems). 

Statistical Analysis

It was apparent that drusen immunoreactivity could be diffusely distributed across the druse or more highly concentrated within a thin rim around the noncleaved druse edge, with or without immunoreactivity in the druse center (see the Results section). To assess the labeling pattern as a function of age, immunoreactivity was assessed as diffuse, rim, both, or other. For analysis, the proportions of diffuse and both were combined, and the proportions of rim and both were combined. Under the null hypothesis that age and ApoC-I and E immunoreactivity patterns are unassociated, the correlation of proportion of drusen with diffuse and rim staining patterns and age were analyzed (JMP Statistical Discovery Software ver. 5.1; SAS Institute Inc., Cary, NC). 

Cell Culture

ARPE-19 and HepG2 cell lines were obtained from the American Type Culture Collection (Manassas, VA) at passage 22 and subjected to RT-PCR after two passages. ARPE-19 cells were plated in T-75 flasks or 6-well plates and grown for 4 weeks in Dulbecco’s minimum essential medium (DMEM)/F12 (1:1) supplemented with 10% fetal calf serum (FCS), as described. ³⁶ Medium was changed twice weekly. HepG2 cells were grown in six-well plates in DMEM containing 10% FCS for 5 days with a medium change every other day. 

Reverse Transcription–Polymerase Chain Reaction

Total RNA was isolated from human retina, RPE, ARPE-19 cells, and HepG2 cells, as described. ²¹ Primers used for RT-PCR are listed in Table 2 . To distinguish between amplified mRNA and genomic DNA, primers were designed to span intron boundaries. Two pairs of primers were designed for each gene. First-strand cDNA was synthesized with reverse transcriptase (Omniscript; Qiagen, Valencia, CA). A commercial PCR system (PCR Core System; Promega, Madison, WI) was used. The cDNA was denatured for 4 minutes at 94°C before cycling. The reaction was amplified through 30 cycles of 45 seconds at 94°C (denaturing), 45 seconds at 55°C to 66°C (annealing), and 1 minute at 72°C (extension), then incubated for 10 minutes at 72°C. 

Results

Gross Appearance of Isolated Drusen

The gross appearance of isolated drusen and their distributions within different eyes are shown in Figure 1and Table 1 , respectively. All extramacular drusen, when cleaved, physically inverted, and viewed from their basal aspects, could be classified into two morphologic types. One, ovoid and largely transparent, had a thin and highly reflective shell around the drusenoid dome. This organization was inferred from the highly reflective rim around the basal aspect en face (Fig. 1C 1E) , a feature not visible from the apical surface (Fig. 1E) . The other was irregular in shape and cloudy (Fig. 1D) , with occasional single corelike structures within (Fig. 1F) . These differences could not be discerned among drusen in situ before their cleavage from BrM and inversion (Fig. 1D) . Although the number of eyes examined is too small for confident generalization, it is noteworthy that five of six eyes from donors aged ≥75 years had predominantly cloudy drusen, whereas three of four eyes from donors aged <75 years had predominantly transparent drusen. 

Apo Immunofluorescence: Prevalence, Remodeling, and Pattern

Apolipoprotein immunoreactivity was detected with a rhodamine-conjugated secondary antibody, and the specificity of labeling was confirmed by examining each section with filter sets for rhodamine and tissue autofluorescence, referring to DIC images for overall tissue texture (Fig. 2) . Figure 2Cshows specific labeling within a druse for ApoC-I (compared with Fig. 2F 2acontrol section incubated with goat IgG), which did not match either druse or RPE autofluorescence (compared to Fig. 2B ). 

By these methods, all Apos examined was detectable in at least some drusen (Fig. 3and Table 3 ). Figure 3shows each Apo type drusen considered both positive and negative. Immunofluorescence intensity and labeling pattern varied, with some drusen exhibiting diffuse labeling and others with labeling largely confined to the apical rim. RPE fluorescence (e.g., Fig. 3F ) represented autofluorescence detectable through the rhodamine filter set in some cases in which low immunofluorescence necessitated long exposure times. For some drusen, immunoreactivity could be localized to basal laminar deposit by reference to the matching DIC view (e.g., Fig. 3H 3P ). Table 3indicates that a median of 230 drusen was evaluated by two graders for each Apo (range, 163–263). The percentages of immunoreactive drusen were (in descending order) 99.5% for ApoE, 99.5% for J, 93.1% for C-I, 80.4% for B, 61.0% for A-I, 59.2% for A-II, 57.7% for C-II, and 16.6% for C-III. That all Apos colocalized to individual drusen is supported by five drusen in three eyes that were probed with all antibodies. Positive and negative labeling, designations requiring agreement between two graders, can be considered most definitive, with uncertain labeling an index of fluorescence intensity. ApoE- and J-positive labeling was found in virtually all (98%–99%) drusen. Uncertain labeling was negligible for these Apos, because their fluorescence was intense. Similarly, uncertain labeling for the highly prevalent ApoC-I, moderately prevalent ApoC-II, and slightly prevalent ApoC-III was <10%. Uncertain labeling for ApoA-I, A-II, and B was >15%, because their immunofluorescence was not intense. 

To evaluate a possible microscopic correlate to the age-related differences in gross drusen appearance (Table 1) , the patterns of ApoC-I and E immunoreactivity, two Apos with strong signals (Table 3) , were analyzed further in 443 drusen (Table 4) . Apolipoprotein immunoreactivity was either diffusely distributed throughout the druse (56.7%), confined to its rim (16.0%), or both (21.2%). The proportion of diffuse labeling decreased significantly with age for ApoE (P = 0.034) and E/C-I combined (P = 0.027), but not for ApoC-I alone (P = 0.131). A trend toward more drusen with rim labeling with age for ApoE (P = 0.06) did not reach significance at α = 0.05. 

In 6.1% of drusen, intensified immunoreactivity for all Apos was confined to a shell approximately one fourth of a druse diameter in width and excluded from internal cavities (Figs. 4B 4D 4F 4H 4J) . DIC imaging revealed evidence of organized substructure reminiscent of active remodeling (Figs. 4A 4C 4E 4G 4I) . Similar views obtained at higher resolution using 1-μm sections revealed involution of druse content associated with cellular invasion. 

Expression of Apo mRNA Transcripts

We previously reported evidence of ApoB-100 and A-I expression in neurosensory retina, native human RPE, and the ARPE-19 cell line. ^{21 22 24} Figure 5shows evidence that other Apolipoprotein genes are expressed in these tissues, as assessed by RT-PCR and using genuine lipoprotein-secretor HepG2 cells as a positive control. RT-PCR products of the expected size were found for ApoA-II in retina and HepG2 cells, and for ApoC-III, in HepG2 cells only, consistent with and expanding on our previous report. ²¹ RT-PCR products for ApoC-I, C-II, E, and J were found in retina, RPE, and ARPE-19 cells. 

Discussion

RPE expression of mRNA for ApoB and microsomal triglyceride transfer protein ^{21 24 37} implies that the RPE has the capacity to assemble and secrete an ApoB-containing lipoprotein particle (ApoB-lp). Further, particles that behave appropriately in density gradients and display ApoA-I and B in two density fractions have been isolated from human BrM. ²² Our current model of a large, possibly novel, ApoB-lp in human BrM, is shown in Figure 6 , with hepatic very low density lipoprotein (VLDL) and intestinal chylomicrons (CM) for comparison. An isolated peripheral druse, postfixed to reveal neutral lipid, reveals numerous solid 83-nm diameter particles (Fig. 6A 6B 6C) . A large RPE ApoB-lp, on the basis of previous work, probably contained ApoA-I, B (presumed ApoB-100) and ApoE ^{9 21 22 23} (Fig. 6D) . VLDL, at 55 to 75 nm diameter, contains ApoB-100, E, C-I, C-II, and C-III (Fig. 6D) . CMs, at 70 to 400 nm diameter, contain more TG in a larger core than VLDL and exhibit ApoB-48, A-I, E, C-I, C-II, and C-III (Fig. 6D) . Three caveats constrain this model. First, this scheme explicitly omits presumptive smaller, high-density particles of RPE origin (e.g., containing ApoE ⁴² ). Second, unlike VLDL and CM, the RPE ApoB-lp core lipid composition is incompletely understood. The high-TG content implied by particle diameter has not been consistently revealed by direct assays. ^{22 43 44} Third, with the exception of ApoB, the molar masses of other Apos per plasma particle vary with metabolic state. This consideration and differences in antibody efficacy mean that the range of observed immunofluorescence intensities should not be construed as indicating Apo composition of individual RPE-lp particles. Despite limitations of the model, it is nevertheless appropriate to ask how the present data bear on it. 

ApoC-I and C-II are now candidates for inclusion on RPE ApoB-lp, in addition to ApoB-100, A-I, and E (previously demonstrated and herein confirmed, although at a somewhat lower prevalence of ApoB-100 and A-I). An intraocular source cannot be concluded as yet for ApoA-II, expressed in retina but not in RPE and ARPE-19, but appears unlikely. The fact that a significant proportion of drusen contained ApoA-II and C-III immunoreactivity (36.4% and 9.0%, respectively), in the absence of evidence for corresponding RPE mRNA transcripts, underscores the reality that a plasma source for some Apos cannot be excluded. That ApoC-I and C-II, at concentrations of 6 and 4 mg/dL, respectively, ⁴⁵ in normolipemic plasma, were so clearly detected in drusen, whereas ApoC-III, at 12 mg/dL, was negligible (see also Ref. ²¹ ), implies either a highly specific ApoC-binding mechanism within drusen or strong evidence for local biosynthesis of these minority ApoCs. 

Ocular functions of Apos A-I, B, E, and J have been discussed elsewhere, ^{9 21 24 46 47} and we focus herein on the potential functional implications for RPE and choroidal biology of intraocular biosynthesis of ApoC-I and C-II, which, like all Apos, serve as enzyme modulators and receptor ligands in addition to lipid transporters. ^{45 48} The human APOC1 and APOC2 genes are members of a 48-kb cluster on chromosome 19 that includes ApoE and pseudo-ApoC1′. At ∼4.7 kb, APOC1 is expressed primarily in liver and, significantly, also in lung, skin, testis, and spleen, where transcription is driven by a promoter different from the one responsible for high-level hepatic transcription. In vitro studies indicate that ApoC-I activates lecithin cholesterol acyl transferase (LCAT), inhibits, among others, lipoprotein (LPL) and hepatic lipases that hydrolyze TG in particle cores, and hinders binding and uptake of VLDL to the LDL and VLDL receptors by displacing ApoE on the ligand particles. Notably, both LCAT and LPL are expressed in RPE and choroid. ^{22 49} Mice expressing human ApoC-I have elevated plasma cholesterol and TG due to elevated VLDL and LDL fractions, elevated plasma nonesterified fatty acids, scaly skin, and hair loss, and ApoC-I deficient mice have 60% higher plasma triglyceride than wild-type and are sensitive to high-fat diets. Of interest, ApoC-I message is reduced, and ApoC-I protein is increased in the brain in Alzheimer disease. ⁵⁰ Regarding ApoC-II, the 3.4-kb APOC2 gene is expressed in liver and intestine only. Its function, revealed by study of human mutations, is clearly one of an LPL activator, although studies in mice suggest that this effect depends on ApoC-I concentration. Recently, ApoC-II has joined Apos A-I, A-II, and E as those readily forming amyloid in the absence of lipid to stabilize their amphipathic α-helical domains, ⁵¹ of potential interest because drusen contain dispersed supramolecular amyloid assemblies. ^{52 53 54} The ocular-specific functions of ApoC-I and C-II remain to be determined, but their presence, especially that of ApoC-II, lends further credence to the overall concept of a large ApoB-lp. 

That drusen have a life cycle of nucleation, expansion, coalescence, and alternative endpoints of calcification or involution, is most strongly supported by studies employing longitudinal angiography and transmission electron microscopy, ^{55 56} with corroborative results emanating from long-term fundus observation in populations. ⁵⁷ Information on the molecular and cellular basis of these phenomena is sparse. Nevertheless, detailed immunohistochemical analysis of extramacular drusen ^{8 10 58 59} shows immunoglobulin G and vitronectin immunoreactivity homogenously distributed throughout small drusen and concentrated along irregular surfaces and internal cavities in larger drusen. Our Figure 4demonstrates similar effects with respect to Apo localization that we, like other investigators, interpret as stigmata of druse degeneration and in some cases, phagocytosis by invading cells. Reportedly, 40% of extramacular drusen contain cores associated with CD1a-CD83-CD86-immunoreactive dendritic cells, ⁵⁸ whereas we found <6% of peripheral drusen with morphologic signs of active remodeling, suggesting cells. Despite these quantitative differences, the concept that newly formed drusen are degraded by invading phagocytes remains distinctly possible. ¹⁰ Furthermore, the fact that EC, a major lipoprotein core component, is prominent in both normal BrM and in drusen of all sizes, ^{19 60} and the Apos in drusen are plausibly associated with EC (Fig. 6) , we argue that the postulated cellular invasion and remodeling are secondary to a primary process of constitutive lipoprotein secretion. 

Within peripheral drusen, Apos could be more highly concentrated within a thin exterior shell, a pattern potentially relevant to both a previously reported EC-rich exterior shell ^{19 21 54} and the optical properties of drusen. Of note, in patients with early-adult-onset grouped drusen, drusen visualized in color fundus photographs are larger than they are when visualized in fluorescein angiograms. ⁶¹ This finding implies that a hydrophilic domain in a druse center is smaller than the actual druse, which, by necessity, contains a hydrophobic shell that is conceivably neutral lipid rich. Those investigators proposed that drusen cores (discussed earlier) ⁶² could account for a hydrophilic center. We propose that an EC-rich, Apo-immunoreactive rim could account for a hydrophobic shell. A shell immunoreactive for complement protein C5 ⁸ resembles the thin rims we observed, raising the possibility that Apos and proteins involved in inflammatory response colocalize. Note that the druse shell defined by Apo immunoreactivity (Fig. 3)differs from the outer zone present in remodeled drusen (Fig. 4)in that it is much thinner. Finally, we found that drusen acquire more diffuse Apo immunoreactivity with age, a process resulting in reduced contrast between the rim and the interior. It is unclear whether this process is attributable to either more particles, or more likely, to larger particles. It is well known that large TG-rich lipoprotein particles (VLDL and CM) at the top of density gradients form a characteristically cloudy layer. In either case, it may be possible to detect such changes in the living fundus using the clinical imaging methods described elsewhere. ⁶¹  

Strengths of our study include use of an enriched druse preparation, at least two primers for all genes, and the relative consistency of findings across eyes. Limitations include the relatively small number of eyes and the long interval between fixation and experiments that may account for reduced Apo prevalence relative to previous work. ²¹ Finally, we emphasize that several important questions remain unanswered. First, finding Apos colocalized within individual drusen does not necessarily mean that they are colocalized on the same lipoprotein particle. Second, the Apo composition of macular drusen has not been addressed. Finally, the cellular localization and function of Apos expressed in neurosensory retina (ApoA-II, C-I, and C-II) are unknown. These gene products may be expressed in Müller cells, believed to secrete a lipoprotein particle into the vitreous ^{63 64} ; in retinal ganglion cells, which contain ApoB and MTP immunoreactivity ²⁴ ; or both. Answering these questions will be the focus of future investigations. 

 Supported by National Eye Institute Grants EY06109, the International Retinal Research Foundation, unrestricted departmental funds and a Lew R. Wasserman Merit Award from Research to Prevent Blindness, Inc., the EyeSight Foundation of Alabama, and a Roger Johnson Prize in Macular Degeneration Research.

 Submitted for publication November 9, 2005; revised December 9, 2005; accepted April 26, 2006.

 Disclosure: C.-M. Li, None; M.E. Clark, None; M.F. Chimento, None; C.A. Curcio, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Starting Sample Size and Predominant Druse Type

Age	Gender	Race*	Starting Sample Size	Predominant Druse Type, †
58	F	W	160	TrOv
66	M	AA	131	Cl
69	M	W	375	TrOv
73	M	W	82	M
75	M	W	880	TrOv
75	M	AA	202	Cl
80	M	W	217	Cl
88	M	W	245	Cl
88	F	AA	100	Cl
93	F	W	160	Cl

*  White, African-American.

†  Cl, cloudy; M, mixed; TrOv, transparent ovoid.

Figure 1.

 

Isolated drusen. (A) A group of drusen, cleaved from BrM, lay on RPE before harvesting. (B) Harvested drusen suspended in PBS in a Beem capsule. (C) Transparent ovoid druse (black arrow), inverted and viewed from its basal aspect. External surface of the druse dome has a thin reflective shell. White arrowheads: indicate other drusen in situ, before cleavage from BrM, also viewed from their apical aspect. (D) Cloudy drusen (white arrowhead) interspersed with small transparent ovoid drusen (black arrowheads), mostly viewed from the basal aspect after inversion (250 μm). (E) Two transparent ovoid drusen, with thin reflective shells, viewed from the basal aspect. A similar druse, viewed from the apical aspect before inversion (black arrows), is barely visible (125 μm). (F) A single cloudy druse, viewed from its basal aspect, exhibits a core substructure (black arrow; 125 μm). Scale bars: (A, B) 500 μm; C, E, F) 125 μm; (D) 250 μm.

Table 2.

 

Sequences of PCR Primers and Expected Fragment Sizes^*

Gene	Genomic Location, †	Protein	Primers	Expected Size (bp)
APOA2	Chromosome 1 at 158, 005, 156-158, 006, 491	ApoA-II	F: 5′-GCA GCA ACT GTG CTA CTC CTC AC-3′	304
			R: 5′-GCA AAG AGT GGG TAG GGA CAG G-3′
APOC1	Chromosome 19 at 50, 109, 419-50, 126, 154	ApoC-I	F: 5′-TGG TGG TTC TGT CGA TCG TCT-3′	235
			R: 5′-ACC CTT CAG GTC CTC ATG AGT-3′
APOC2	Chromosome 19 at 50, 137, 335-50, 144, 660	ApoC-II	F: 5′-GAA TCT CTC TCC AGT TAC TGG G-3′	306
			R: 5′-GAA TTC AGG CTA GAG TTG GGA G-3′
APOC3	Chromosome 11 at 116, 205, 834-116, 208, 997	ApoC-III(1)	F: 5′-TGC TCC AGG AAC AGA GGT GC-3′	439
			R: 5′-GTA GGA GAG CAC TGA GAA TAC T-3′
		ApoC-III(2)	F: 5′-TCA GTT CAT CCC TAG AGG CAG-3′	515
			R: 5′-CCA GCT TTA TTG GGA GGC CAG-3′
		ApoC-III(3)	F: 5′-CGG GTA CTC CTT GTT GTT GCC-3′	457
			R: 5′-TTA TTG GGA GGC CAG CAT GCC-3′
APOE	Chromosome 19 at 50, 100, 904-50, 104, 489	ApoE	F: 5′-ACT GGC ACT GGG TCG CTT T-3′	163
			R: 5′-GTT GTT CCT CCA GTT CCG ATT-3′
CLU	Chromosome 8 at 27, 510, 369-27, 528, 244	ApoJ, ‡	F: 5′-CTT CCA CGC CAT GTT CCA G-3′	498
			R: 5′-ACC TCA GTG ACA CCG GAA G-3′

*  All primers for human genes were designed based on GenBank gene sequences. F, forward primer; R, reverse primer.

†  Data from www.ensembl.org.

‡  Apolipoprotein J is also known as clusterin.

Figure 2.

 

Specificity of Apo immunoreactivity. Each column shows the same drusen viewed with DIC optics (top), a fluorescein filter set for autofluorescence (middle), and a rhodamine filter set for immunofluorescence (bottom). (A–C) ApoC-I immunoreactivity. The same RPE (arrow) and druse (*) are indicated. (D–F) Control experiment using goat IgG. Scale bar, 40 μm.

Figure 3.

 

Apolipoprotein immunoreactivity in isolated drusen. Each row shows the same drusen viewed with DIC optics (left column) and indirect immunofluorescence, with a rhodamine filter set (right column). Prominent RPE fluorescence in (F) and (L) is autofluorescence detectable through the rhodamine filter in cases in which low specific immunofluorescence necessitated long exposure times. (H, P, arrows) indicate basal deposits. Shown are drusen immunoreactive for Apos (A, B) A-I, (C, D) A-II, (E, F) B, (G, H) C-I, (I, J) C-II, (K, L) C-III, (M, N) E, and (O, P) J. Scale bar, 40 μm.

Table 3.

 

Apolipoprotein Prevalence in Isolated Drusen Checked with Immunofluorescence^*

Eyes	ApoA-I							ApoA-II							ApoB							ApoC-I
		No. of Drusen	%					No. of Drusen	%					No. of Drusen	%					No. of Drusen	%
				+	+/−	−	+		+/−	−	+	+/−	−		+	+/−	−
2004020L	24	33.3	45.8	20.8	22	63.6	13.6	22.7	25	48.0	28.0	24.0	17	94.1	0.0	5.9
2002061L	7	85.7	14.3	0.0	6	83.3	16.7	0.0	23	65.2	21.7	13.0	6	100.0	0.0	0.0
2003073L	39	17.9	33.3	48.7	34	20.6	23.5	55.9	37	21.6	35.1	43.2	37	78.4	8.1	13.5
2004019L	18	50.0	11.1	38.9	17	58.8	17.6	23.5	14	85.7	0.0	14.3	12	91.7	0.0	8.3
2002018R	75	37.3	8.0	54.7	80	50.0	20.0	30.0	73	72.6	11.0	16.4	101	75.2	15.8	8.9
2002083L	19	84.2	10.5	5.3	22	40.9	54.5	4.5	20	95.0	5.0	0.0	16	81.3	12.5	6.3
2004048L	31	25.8	25.8	48.4	35	2.9	31.4	65.7	32	96.9	3.1	0.0	38	94.7	2.6	2.6
2003131L	3	33.3	0.0	66.7	4	50.0	0.0	50.0	6	83.3	0.0	16.7	9	77.8	22.2	0.0
2004003L	10	100.0	0.0	0.0	22	0.0	0.0	100.0	17	47.1	23.5	29.4	17	100.0	0.0	0.0
2003003L	10	40.0	40.0	20.0	8	37.5	37.5	25.0	10	100.0	0.0	0.0	10	100.0	0.0	0.0
Total	236	41.1	19.9	39.0	250	36.4	22.8	40.8	194	62.9	17.5	19.6	263	84.0	9.1	6.8

Eyes	ApoC-II							ApoC-III							ApoE							ApoJ
		No. of Drusen	%					No. of Drusen	%					No. of Drusen	%					No. of Drusen	%
			+	+/−	−	+	+/−		−	+	+/−	−	+		+/−	−
2004020L	8	25.0	12.5	62.5	15	40.0	26.7	33.3	16	87.5	6.3	6.3	11	90.9	9.1	0.0
2002061L	9	100.0	0.0	0.0	11	0.0	9.1	90.9	8	100.0	0.0	0.0	11	100.0	0.0	0.0
2003073L	20	10.0	25.0	65.0	25	8.0	12.0	80.0	26	100.0	0.0	0.0	24	100.0	0.0	0.0
2004019L	13	69.2	7.7	23.1	9	0.0	0.0	100.0	7	100.0	0.0	0.0	15	100.0	0.0	0.0
2002018R	62	22.6	12.9	64.5	77	3.9	3.9	92.2	92	100.0	0.0	0.0	78	97.4	2.6	0.0
2002083L	14	100.0	0.0	0.0	16	18.8	18.8	62.5	20	100.0	0.0	0.0	21	100.0	0.0	0.0
2004048L	4	100.0	0.0	0.0	36	0.0	0.0	100.0	34	100.0	0.0	0.0	31	96.8	0.0	3.2
2003131L	10	20.0	0.0	80.0	6	16.7	50.0	33.3	9	88.9	11.1	0.0	8	100.0	0.0	0.0
2003003L	8	87.5	12.5	0.0	13	38.5	0.0	61.5	13	100.0	0.0	0.0	12	100.0	0.0	0.0
2004003L	15	100.0	0.0	0.0	15	0.0	0.0	100.0	14	100.0	0.0	0.0	5	100.0	0.0	0.0
Total	163	47.9	9.8	42.3	223	9.0	7.6	83.4	239	98.7	0.8	0.4	216	98.1	1.4	0.5

 * Evaluated by two graders. +, positive, both graders agree; +/−, graders disagree; −, negative, both graders agree.

 † Eyes are sorted by age.

Table 4.

 

Pattern of ApoC-I and ApoE Immunoreactivity in Drusen

Eye	Age (y)	ApoC-I					ApoE					ApoC-I + ApoE
				Drusen* (n)	Diffuse (Prop)	Rim (Prop)	Drusen* (n)	Diffuse (Prop)	Rim (Prop)	Drusen* (n)	Diffuse (Prop)	Rim (Prop)
2004020L	58	16	0.438	0.688	14	0.500	0.786	30	0.466	0.733
2002061L	66	6	1.000	0.167	9	1.000	0.111	15	0.867	0.000
2003073L	69	27	0.814	0.444	24	0.583	0.333	51	0.706	0.392
2004019L	73	11	0.455	0.728	—	—	—	11	0.455	0.728
2002018L	75	102	0.500	0.559	83	0.903	0.337	185	0.681	0.459
2002083L	75	15	1.000	0.200	20	1.000	0.000	35	1.000	0.086
2004048L	80	33	1.000	0.212	31	0.967	0.322	64	0.984	0.266
2003131L	88	7	1.000	0.143	6	1.000	0.000	13	1.000	0.077
2004003L	88	7	1.000	0.429	14	1.000	0.000	21	1.000	0.143
2003003L	93	8	0.875	0.250	10	1.000	0.200	18	0.945	0.223
Mean correlation with age			0.808	0.382		0.884	0.232		0.810	0.311
P			0.131	0.175		0.034	0.060		0.027	0.148
r			0.512	−0.466		0.704	−0.647		0.692	−0.492

 Prop, proportion.

*  Median/minimum/maximum number of drusen: ApoC-I (13/6/102); ApoE (14/6/83); ApoC-I + ApoE (25.5/11/185).

Figure 4.

 

Drusen with evidence of remodeling. (A, C, E, G, I) DIC images; (B, D, F, H, J) epifluorescence images, obtained with a rhodamine filter set. Arrows: remodeling in drusen centers. Shown are drusen with Apos (A, B) A-I, (C, D) B, (E, F) C-I, (G, H) E, (I, J) and J, and (K) 1 μm-thick toluidine-O-blue-stained section showing remodeling druse. (*) Druse; (* *) basal laminar deposits. Scale bar, 40 μm.

Figure 5.

 

Apolipoprotein gene expression. Total RNA was isolated from human retina, RPE, ARPE-19, and HepG2, and RT-PCR was performed. ApoA-II was expressed in retina and HepG2 cells; ApoC-III was expressed in HepG2 cells only; ApoC-I, ApoC-II, ApoE, and ApoJ were expressed in retina (R), RPE (P) and ARPE-19 (19) and HepG2 (H) cells. Expected sizes of RT-PCR products are 304 bp for ApoA-II, 235 bp for ApoC-I, 306 bp for ApoC-II, 515 bp for ApoC-III, 163 bp for ApoE, and 498 bp for ApoJ. M 100 bp DNA ladder. Arrowhead: 500 bp DNA marker.

Figure 6.

 

RPE lipoprotein: a model. (A) One-micrometer-thick section of an isolated peripheral druse, postfixed with osmium tannic acid paraphenylenediamine (OTAP), exhibits a darker brown stain in areas of neutral lipid accumulation around its rim (double arrowhead) and in a pocket at the base near the site of cleavage from BrM (arrowhead). Box: representative area illustrated at higher magnification in (B) and (C). (B) Electron micrograph showing RPE, nucleus (N), basal laminar deposit (BlamD) overlying a druse exhibiting OTAP enhancement (black arrowhead), and a high concentration of OTAP-negative, presumed-neutral, lipid-rich particles ^{19 38} at the druse periphery (white arrowhead). (C) Higher magnification view of OTAP-positive (black arrowhead) and OTAP-negative (white arrowhead) particles. Diameter of OTAP-positive particles, 83.3 ± 9.8 nm. Scale bars: (A) 10 μm; (B) 500 nm; (C) 200 nm. (D) Schematic of hepatic VLDL (60 nm in diameter), intestinal dietary CM (100 nm), and a hypothetical RPE large lipoprotein (RPE-LP, 100 nm), showing neutral lipid cores and surface Apos. VLDL and CM Apo identities are taken from References ^{39 40 41} .

The authors thank the Alabama Eye Bank for retrieving donor eyes; Ramon F. Dacheux II, PhD, for guidance in designing the drusen-harvesting pipette; and Nassrin Dashti, PhD, for helpful discussions.