Accumulation of Cholesterol with Age in Human Bruch’s Membrane

Abstract

purpose. To determine the cholesterol composition of normal human Bruch’s membrane and choroid as a function of age and retinal location.

methods. Human eyes with grossly normal maculas were preserved <4 hours after donor death. Cryosections of retina and choroid from the macula and temporal equator were stained with filipin to reveal esterified (EC) or unesterified (UC) cholesterol (n = 20, 17–92 years). Filipin fluorescence in Bruch’s membrane was quantified with digital microscopy. Maculas were prepared for lipid-preserving electron microscopy (n = 18, 16–87 years) and for ultrastructural analysis after lipid extraction (n = 2, 85 and 89 years). Punches of macular Bruch’s membrane, 8 mm in diameter, were assayed for cholesterol content by enzymatic fluorometry (n = 10, >70 years).

results. EC and UC in Bruch’s membrane increased with age in the macula. EC was sevenfold higher in macula than in periphery. Sixty percent of total cholesterol was esterified, and Bruch’s membrane EC was 16- to 40-fold enriched relative to plasma. Solid, 100-nm-diameter particles occupied> 30% of the inner collagenous layer in eyes >60 years. Cholesterol accumulated in choroidal arteries and in small age-related drusen.

conclusions. Human Bruch’s membrane ages like arterial intima and other connective tissues for which plasma lipoproteins are the known source of extracellular cholesterol. Age-related maculopathy and atherosclerotic cardiovascular disease may share common pathogenic mechanisms.

In the body, cholesterol assumes two chemical forms, either unesterified (or free, UC) or esterified to fatty acids (EC). ^{1 2} Further, three physical forms (oily droplets, membranes, and crystals) can be distinguished by their relative proportions of UC, EC, and phospholipid. ^{1 2} Lipids that bind the histochemical stain oil red O increase with age in normal human connective tissues, including the sclera, ³ cornea, ^{4 5} and intima, or inner wall, of large arteries. ^{6 7 8} In these tissues, the oil red O-positive material comprises small (60–200 nm) extracellular droplets that are highly enriched in EC relative to UC (69% EC, 22% UC, and 9% phospholipid). ^{9 10 11 12 13} There is good evidence that the source of EC in sclera, cornea, and arterial intima is low-density lipoproteins (LDL) that are transported from plasma into connective tissues. EC-enriched particles are thought to arise from smaller LDL particles (22 nm) by extracellular matrix–mediated trapping of LDL, degradation of LDL protein or phospholipid components, and fusion of the remaining lipid components (for review, see Ref. ¹⁴ ). 

Late age-related maculopathy (ARM), or age-related macular degeneration, ¹⁵ is the leading cause of untreatable new vision loss in elderly individuals, ^{16 17 18} but its causes are poorly understood. The most prominent clinical and histopathologic lesions of early and late ARM involve Bruch’s membrane, a thin connective tissue between the basal surface of the retinal pigment epithelium (RPE) and the choriocapillaris that is traversed by molecules essential for photoreceptor and RPE function. The only known risk factor for early ARM (i.e., drusen and RPE changes) is advanced age. ^{16 17 18} It is therefore important to understand how age-related changes in Bruch’s membrane predispose some individuals for subsequent disease. ^{19 20} In Bruch’s membrane of the macula there is a progressive accumulation of lipids stainable by oil red O ²¹ , and Bruch’s membrane/choroid extracts contain phospholipids, triglycerides, UC, and EC, in descending order of abundance. ²² It is thought that lipid accumulation renders Bruch’s membrane increasingly hydrophobic with age, impeding diffusion between the RPE and choroidal vessels. ^{21 23 24}  

On the basis of comparison with other tissues, it is logical to hypothesize that the oil red O-positive material that increases with age in human Bruch’s membrane is EC-rich particles. This hypothesis is supported by the presence of numerous small, round electron-lucent droplets in adult Bruch’s membrane ^{21 25 26 27 28} and preliminary polarizing microscopy studies demonstrating birefringence patterns consistent with EC. ²⁹ However, only a small proportion of the cholesterol recovered in Bruch’s membrane/choroid extracts is esterified. ²² Resolving the apparent discrepancy between morphologic and biochemical data is important for understanding whether lipid deposition in Bruch’s membrane is an ocular manifestation of a systemic process or a phenomenon unique to the eye. We therefore examined Bruch’s membrane cholesterol in normal donor eyes, using the fluorescent probe filipin to reveal EC and UC as a function of age and retinal location in tissue sections, ^{9 30} an enzymatic fluorometric assay to determine the relative proportions of EC and UC of isolated Bruch’s membrane, ³¹ and lipid-preserving electron microscopy to identify EC-rich droplets. ¹¹ Our data indicate that Bruch’s membrane is highly enriched in EC. 

Methods

Human Eyes

Eyes were obtained from human donors < 4 hours from death. Use of human tissues was approved by institutional review at the University of Alabama at Birmingham (protocol number X900525013). Twenty eyes preserved by immersion in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) after removal of the anterior segment were prepared for filipin histochemistry (ages 17–22 years, n = 3; 32–53 years, n = 6; and 61–92 years, n = 11). Twenty eyes preserved in 2.5% glutaraldehyde and 1% paraformaldehyde in 0.1 M PB were used for lipid-preserving electron microscopy (ages 16–51 years, n = 8; 61–87 years, n = 10) or for ultrastructural analysis after lipid extraction (ages 85–89 years, n = 2). Ten 4% paraformaldehyde-preserved eyes (>70 years) were assayed for cholesterol content. All eyes were inspected internally and lacked grossly visible chorioretinal pathology in the macula. ³²  

Cryosectioning

Seven millimeter-wide samples containing retina, RPE, and choroid from the macula and periphery were cryosectioned for histochemistry. Macular samples included the fovea and the temporal half of the optic nerve head. Peripheral samples included the temporal equator and ora serrata. Samples were infiltrated in 4:1 and 2:1 30% sucrose and Histo Prep medium (Fisher, Norcross, GA) for 30 minutes each, frozen in −70°C isopentane, sectioned at 10 μm, collected on glass slides, and dried at 40 to 60°C. 

Filipin Histochemistry

We localized EC and UC using the fluorescent polyene antibiotic filipin, which binds specifically to sterols and interacts with the 3β-hydroxy group of cholesterol. ³⁰ For EC detection, ⁹ native UC was extracted from cryosections by two 5-minute rinses in 70% ethanol, native EC was hydrolyzed with cholesterol esterase (EC 3.1.1.13; Boehringer Mannheim, Indianapolis, IN) at a concentration of 1.65 units/ml in 0.1 M potassium PB (pH 7.4) for 3 hours at 37°C, and UC newly released by the hydrolysis of EC was stained with filipin (5 mg filipin, dissolved in 1 ml dimethylformamide and diluted in 100 ml PB saline; Sigma, St. Louis, MO). Control sections were incubated in the enzyme vehicle. For UC detection, ³³ sections were hydrated and incubated for 30 minutes in the above filipin solution without prior extraction and hydrolysis. Control sections were incubated in the filipin vehicle. All sections were counterstained with Mayer’s hematoxylin. 

Photography

Filipin-processed sections were photographed on ASA100 black and white film (TMax100; Eastman Kodak, Rochester, NY) using an Optiphot fluorescence microscope (Nikon, Melville, NY) equipped with a 420-nm excitation filter, 520-nm barrier filter, and a 60× plan apochromat objective. Negative film was scanned to create composite images using Photoshop (Adobe, San Jose, CA). 

Quantification of Filipin Fluorescence

The mean fluorescence intensity of Bruch’s membrane in filipin-stained and control sections was determined by digital microscopy. In each section (one per condition per eye), three 240-μm-long segments of Bruch’s membrane were digitized. Sections were viewed on a Leitz Orthoplan microscope with a 50× Fluotar oil objective (Wetzlar, Germany) and a Chromatechnology 83000 (Brattleboro, VT) fluorescence filter set (excitation, 346 nm; barrier, 460 nm). Images were captured at a resolution of 0.18μ m/pixel using a Photometrics CH250 CCD video camera (Roper Scientific, Tucson, AZ) and IP Lab Spectrum 3.2 software (Scanalytics, Fairfax, VA). All images were exposed for 2 seconds to prevent saturation of the camera. To minimize variability due to fluorescence quenching, digitized sites were separated by 1.5 to 2 mm and were identified and focused using bright-field illumination. To minimize variability due to oblique sectioning planes and haze from RPE autofluorescence, only sites where the RPE was attached to Bruch’s membrane and the inner edge of Bruch’s membrane and RPE pigment granules could be focused simultaneously were used. In digitized images, an observer placed 4.56-μm square sampling windows over Bruch’s membrane. The sampling window spanned the thick macular Bruch’s membrane in older eyes and included choriocapillary endothelium and lumina in addition to thin Bruch’s membrane in other specimens. Five sampling windows were placed across the length of each three digitized images per section, with at least two within and three between intercapillary pillars (total, 15 windows). Sampling windows avoided drusen and basal deposits. Fluorescence intensities in each window were summed and expressed in arbitrary units× 10⁻⁶. The SEM was <3% for unlabeled control sections and 0.8% to 8.6% for filipin-labeled sections. Three sections on the same slide differed by <3%. 

Cholesterol Assay

Total cholesterol and UC content was determined with an enzymatic fluorometric assay. ³¹ Paraformaldehyde-preserved eyes were used because of availability and because formalin fixation minimally changes cholesterol content. ³⁴ Punches of macular and peripheral retina, RPE, and choroid were obtained with an 8-mm-diameter trephin. The macular punch was centered on the fovea, and the peripheral punch was centered on the temporal equator. The retina was detached, the RPE was removed with a camel hair brush, and major choroidal vessels were removed by a combination of brush and fine forceps. Grossly visible large drusen in peripheral retina were avoided. Small pieces were removed before and after dissection to assess the completeness of choroid removal in 1-μm histologic sections. Bruch’s membrane, retina, and choroidal vessels were rinsed with water, placed in separate pre-weighed 0.25-ml plastic tubes, lyophilized, weighed, and shipped overnight on dry ice. Lipids were extracted with chloroform/methanol (2:1, by volume), distributed to sample tubes, dried by heating, and redissolved with 95% ethanol. An assay solution was added for 30 minutes at 37°C. The assay for total cholesterol used cholesterol esterase to hydrolyze native EC, cholesterol oxidase to generate H₂O₂, and peroxidase to catalyze the reaction of H₂O₂ with p-hydroxyphenylacetic acid. The resulting fluorescent product excited at 325 nm and emitted at 425 nm. For determination of UC, cholesterol esterase was omitted. EC was the difference of total cholesterol and UC. Cholesterol content was expressed as nmol/g dry weight. 

Electron Microscopy

Glutaraldehyde-preserved maculas were divided horizontally through the fovea. One tissue block was processed conventionally. ²⁸ The other tissue block was processed by the osmium-tannic acid-paraphenylenediamine (OTAP) method to preserve small extracellular lipid particles. ¹¹ Tissues were postfixed with 1% osmium tetroxide in 0.1 M sodium cacodylate (NaCaco) buffer (2.5 hours), 1% tannic acid in 0.05 M NaCaco buffer (30 minutes), and fresh 1% para-phenylenediamine in 70% ethanol (30 minutes). Silver sections were cut from adjoining block faces with a diamond knife, collected on mesh grids, and stained with lead citrate. Electron micrographs spanning the full thickness of Bruch’s membrane were taken at 7500 to 8000× and enlarged three times. Five micrographs per preparation were examined, with two within and three between intercapillary pillars. Other OTAP-processed blocks from 10 eyes were sectioned in a horizontal plane. 

Extraction Studies

Cryosections were extracted with ethanol and chloroform/methanol (2:1, with 1% hydrochloric acid) before filipin histochemistry and digital microscopy. Tissue blocks of the macula of two eyes were subject to the same solvents before osmication and processing for conventional electron microscopy. The number of electron micrographs containing lipid-rich components was determined by an observer unaware of the solvent. 

Stereologic Analysis of OTAP Preparations

The proportion of the inner collagenous layer occupied by lipid-rich particles (area fraction) was measured in 16 eyes of different ages using point-counting stereology. ³⁵ A transparency containing a 4-mm² grid was taped to each of three prints for each eye, and the inner collagenous layer was delimited. Intersections of the grid overlying lipid particles and other inner collagenous layer constituents were scored by an observer. The area fraction was the ratio of intersections overlying particles to the total number of intersections. The median number of points scored per eye was 1112 (range, 621-1682), and the SEM was <10%. Results from younger and older eyes were compared using a t-test. Diameters of lipid-rich particles were measured in six eyes using a digitizing tablet. 

Statistical Analysis

The relationship among filipin fluorescence intensities and age in the macula and periphery was analyzed with multiple linear regression, adjusting for age to examine the independence of associations (e.g., macular EC and peripheral EC) with the effect of age removed. We report the slopes of regression lines as crude parameter estimates and the slopes after adjustment. A P value of 0.05 was considered significant. 

Results

Histochemistry

The identity of neutral lipids in Bruch’s membrane was determined by histochemistry (Fig. 1) . In sections that were extracted with ethanol and hydrolyzed with cholesterol esterase to reveal EC, Bruch’s membrane exhibited an intense, diffusely distributed fluorescence (Fig. 1A) . The RPE and retina were negative for EC, with the exception of occasional retinal arteries (not shown), and the choroid was lightly labeled (see below). The specificity of this reaction was verified by eliminating cholesterol esterase (Fig. 1B) . Because Bruch’s membrane fluorescence intensity was similar in all control sections of each eye (data not shown), we conclude that it represents background autofluorescence. ³⁶ In sections that were exposed to filipin without prior extraction and hydrolysis, cellular membranes of the neural retina, RPE, and choroid were fluorescent, as was Bruch’s membrane (Fig. 1C) . From these results, we conclude that Bruch’s membrane contains both EC and UC. 

Using sections of macula and periphery that were stained with filipin for EC and UC, we investigated the effects of age and retinal location on cholesterol deposition in Bruch’s membrane (Fig. 2 and Table 1 ). The range of mean fluorescence intensities (×10⁻⁶ units) for each eye was 1.2 to 18.0 for EC, 4.0 to 14.2 for UC, and 1.1 to 2.6 for controls in the macula and 1.1 to 5.3, 0.3 to 4.6, and 1.0 to 4.0, respectively, in the periphery. EC increased significantly with age in both macula and periphery (Figs. 2A 2B) . In the macula, EC was absent from donors < 20 years old, increased sharply with age to reach high and variable levels (fourfold range) in donors > 60 years old. Macular EC was much higher than peripheral EC at all ages > 30 years (median macula/peripheral ratio for eyes, 6.9), and the difference increased significantly with age (P = 0.02). Macular and peripheral EC in individual eyes were significantly associated when the effect of age was removed using multiple linear regression analysis (Table 1) . Regarding UC, the macula exhibited a significant age-related increase (Fig. 2C) , but the peripheral retina did not (Fig. 2D) . Unlike EC, UC was present in the youngest specimens (likely because choriocapillary endothelial membranes ³⁷ were included in the measurements) and did not increase as markedly with age in the macula as EC. Macular and peripheral UC in individual eyes were not significantly related when the effect of age was removed (Table 1) . 

In addition to the intense filipin fluorescence due to EC in Bruch’s membrane, there was also a faint but specific filipin fluorescence due to EC in the walls of choroidal arteries of the same sections (Figs. 3A , 3B ). Veins were unlabeled (Figs. 3A 3B) . Table 2 shows that EC was present in choroidal arteries of the macula and periphery only in eyes > 60 years old. Thus, EC accumulates with age in choroidal arteries, as it does in Bruch’s membrane. In contrast to Bruch’s membrane, however, EC deposition in choroidal arteries appears significantly later in life, reaches qualitatively lower levels, and is unrelated to retinal location. 

Cholesterol Assay

The relative proportions of EC and UC were directly assayed in Bruch’s membrane, retina, and choroid of 10 eyes from donors > 70 years of age with high expected cholesterol content. Table 3 shows that 39.1 ± 18.8 nmol EC/g dry weight were recovered from 8-mm-diameter punches of macular Bruch’s membrane. There was an almost threefold range in macular EC content (15.5–68.4 nmol/g). The proportion of total cholesterol that was esterified was less variable at 59.6% ± 6.7%. In contrast, both the content and proportion of EC was low in choroidal arteries (mean, 7.4 nmol/g; 15.0%) and very low (mean, 2.0 nmol/g; 3.6%) in macular retina of the same eyes. Peripheral Bruch’s membrane in three eyes contained 2.7- to 11-fold less EC than macular Bruch’s membrane. Together, these biochemical data validate the results obtained using filipin histochemistry. 

Histologic analysis revealed that the RPE and major choroidal arteries were completely removed from Bruch’s membrane, and the choriocapillaries and some veins were still attached. Because EC bound to plasma lipoproteins in the remaining vessels would affect estimates of Bruch’s membrane cholesterol content, we computed the magnitude of this potential contamination in the worst case, that is, if the choriocapillaries and veins were intact and completely filled with blood. Using standard clinical measures and published estimates of choroidal morphology, ^{38 39 40 41} we estimated that the choriocapillaries underlying an 8-mm-diameter macular punch contain 0.39 nmol EC in 0.09 μl plasma. The choriocapillaries and veins together contain 0.94 nmol EC in 0.13 μl plasma. Thus, blood remaining in the choriocapillaries could account for only 2.5%, and choriocapillaries and veins together only 6%, of the EC determined in our samples. Accordingly, Bruch’s membrane EC content is enriched at least 16- to 40-fold relative to plasma. 

Ultrastructure

The intensity and diffuse distribution of filipin fluorescence in Bruch’s membrane (see Fig. 1 ) suggests that cholesterol-containing particles are densely packed and smaller than the resolution limit of light microscopy. To identify these particles, we used conventional and lipid-preserving electron microscopy. Conventional electron microscopy revealed numerous small round particles with electron-lucent interiors (“droplets”) in Bruch’s membrane of older eyes (Fig. 4A ). Droplets were scattered throughout the collagenous layers. They were also grouped within coated membrane-bounded bodies (CMBBs), ²⁷ 91% of which occurred in the outer collagenous layer, and were associated with coiled membranes, the presumed remnants of CMBBs (Fig. 4A) . Droplets have been previously described as vesicles ^{25 42} (i.e., membrane-bounded structures with aqueous contents). However, in preparations in which lipid-rich particles are preserved by the OTAP method (Fig. 4B) , droplets were solid, electron-dense particles that formed a single size distribution (minimum/median/maximum diameters; 54/98/156 nm for inner collagenous layer; 56/112/225 nm for outer collagenous layer). These results indicate that electron-lucent droplets are not vesicles but rather solid particles whose lipid-rich contents were extracted by conventional tissue processing. 

To determine whether droplet cholesterol composition resembles that of EC-rich particles isolated from other tissues (i.e., 69% EC and 22% UC), ^{5 13} we examined the effects of solvents on filipin histochemistry and ultrastructure in conventional preparations. We expected that droplet ultrastructure would be affected by solvents that reduce both EC and UC fluorescence. We further expected that the ultrastructure of UC-containing membranes in the RPE, choriocapillary endothelium, and CMMBs would be affected only by solvents that reduce UC fluorescence. Table 4 shows that treatment with ethanol eliminated Bruch’s membrane UC fluorescence, produced somewhat distorted droplets, and reduced the visibility of cellular and CMMB membranes. Treatment with chloroform/methanol eliminated UC and EC fluorescence, converted droplets into distorted empty spaces embedded within an electron-dense matrix, and completely removed all membranes. These results are consistent with our expectations. 

The distribution of droplets changed with age in a manner consistent with their identification as EC-rich particles. In eyes < 60 years, droplets were scarce and scattered through the inner collagenous layer or associated with CMMB in the outer collagenous layer (Figs. 5A , 5B ). In contrast, droplets in eyes > 60 years old were abundant in the inner collagenous layer and formed a broken or continuous layer external to the RPE basal lamina (Figs. 5C and 6 , asterisk). Tangential sections of Bruch’s membrane revealed rows of droplets aligned along fibers in the inner collagenous layer (Fig. 6) . The proportion of the inner collagenous layer occupied by droplets in eyes ≥ 67 years (36% ± 7.0%, n = 10) was significantly higher than in eyes ≤ 51 year (7% ± 2.7%, n = 6, t = −11.6, P < 0.0001). 

Cholesterol and Age-Related Pathology

Older eyes exhibit many mild pathologic changes in Bruch’s membrane, including calcification and the accumulation of basal deposits and drusen. ^{28 43 44} In eyes > 60 years used for filipin histochemistry, maculas contained only small drusen and patchy basal deposits, and the periphery contained numerous small drusen and basal deposits, as previously characterized by light microscopy. ⁴⁵ Below we describe patterns of filipin fluorescence and lipid particle distribution that are consistent with cholesterol involvement in age-related pathology. In the macula, Bruch’s membrane from eyes > 60 years was thick and intensely labeled (Fig. 7A ). Most eyes (9/11) from older donors exhibited intermittent bands of less intense filipin labeling that corresponded to patches of calcification ⁴⁴ (Fig. 7B) . Ultrastructural and histochemical observations (Figs. 7C 7D 7E) suggested a progression between the thickened band of EC-rich particles external to the RPE basal lamina (see Figs. 4 and 5 ) and some small drusen. Figure 7C shows an EC-rich band overlying a small accumulation of a homogeneous material. Figure 7D shows EC-rich deposits on the inner surface of Bruch’s membrane that are too small to be considered drusen. Figure 7E shows a small druse with an EC-rich rim. Electron micrographs comparable to the histochemical observations in Figures 7D and 7E have been published previously (Ref. ⁴⁶ , Figs. 2 and 5D ; Ref. ²⁸ , Fig. 4B ). Finally, the small drusen (Figs. 7F 7G) and basal deposits (Figs. 7H 7I) present in the periphery of these eyes were virtually all positive for both EC (Figs. 7F 7H) and UC (Figs. 7G 7I) . 

Discussion

Throughout adulthood, normal human connective tissues accumulate extracellular EC-rich droplets (perifibrous lipid) that appear to originate from infused plasma lipoproteins. ⁸ In this article we provide evidence that Bruch’s membrane shares with sclera, cornea, and arterial intima a linear, age-related increase in EC and UC, a high degree (60%) of enrichment with EC, and an abundance of 100-nm-diameter droplets. Our use of filipin histochemistry and digital microscopy permitted specific labeling and precise localization of samples to Bruch’s membrane, and histochemical results were verified by direct chemical assay. 

Using ultrastructural techniques that preserve lipid particle morphology and extraction experiments, we demonstrated that the electron-lucent droplets recognized in human Bruch’s membrane for more than 30 years ^{21 25 26 27 42} are the EC-rich particles suggested by filipin histochemistry. Droplets resemble the solid, EC-rich particles seen in other tissues with regard to size and alignment along matrix fibers. ^{5 47} A membranous shell of UC and phospholipid ¹² could account for the vesicular appearance of Bruch’s membrane droplets in tissue examined by conventional electron microscopy. Droplets also occur in coated membrane-bounded bodies, ^{27 46} implicating these enigmatic structures in lipid trafficking. In the maculas of older adults, EC-rich particles occupy one third of the inner collagenous layer and nearly 100% of a narrow sublayer external to the RPE basal lamina. These results provide a basis for the finding that hydraulic conductivity across Bruch’s membrane and choroid isolated from older donors improves markedly when the inner collagenous layer is removed. ²⁴  

Our data contrast sharply with those of Holz et al., ²² who indicated that the proportion of EC in macular Bruch’s membrane was only 14% and that EC content was unrelated to age. The previous study, ²² which used thin-layer chromatography to assay lipids in unfixed tissues and did not remove the choroid from Bruch’s membrane, recovered one tenth as much total cholesterol per unit area as we did. Our higher yield may be due to using tissues that were preserved in paraformaldehyde quickly after death, presumably cross-linking EC-rich particles to a surrounding proteinaceous matrix. Our higher proportion of EC may be due to our removing the choroid. Because the choroid contains many cells and is 50- to 100-fold thicker than Bruch’s membrane, its presence would inflate Bruch’s membrane UC content relative to EC. In humans, two thirds of total plasma cholesterol is transported by LDL, ³⁸ and 64% of the cholesterol in LDL is esterified. ⁴⁸ The low proportion of Bruch’s membrane EC found by Holz et al., ²² relative to total lipids, was the basis of that study’s conclusion that plasma was an unlikely source for Bruch’s membrane lipids. Although we did not examine other lipid classes (e.g., triglycerides or phospholipids), our demonstration that EC accounts for a high proportion of total cholesterol suggests that this conclusion should be re-evaluated, at least for cholesterol.⁷  

Although it is thought that the RPE produces many Bruch’s membrane constituents, ^{49 50} there is currently little reason to suspect that the RPE is a source of EC-rich droplets. It is possible that the large-diameter (1–2 μm), oil red O-positive droplets occasionally seen in RPE cells (lipoidal degeneration ^{51 52 53} ) could be released into Bruch’s membrane upon cell death. However, we did not detect EC within RPE cells using filipin, and the uniformly small size of Bruch’s membrane droplets is inconsistent with their originating from the breakdown of larger droplets. ¹² Another potential source of Bruch’s membrane EC is the photoreceptor outer segment membranes that are regularly ingested by RPE. However, outer segment membranes are poor in UC compared with typical plasma membranes, ⁵⁴ and it is not yet known if the RPE can esterify cholesterol. Although the RPE cannot be conclusively excluded as a source without further data, it is more parsimonious to postulate that cholesterol deposition in Bruch’s membrane reflects a well-described systemic process than to invoke a new mechanism involving the RPE. 

We found a highly variable but marked (sevenfold) difference in the degree of cholesterol accumulation of macular and peripheral Bruch’s membrane. This difference is far greater than can be explained by differences in the number of photoreceptors in the macula and at the temporal equator (macula/periphery ratio = 2.7 for rods, 1.9 for cones; calculated from Ref. ⁵⁵ ). Regional differences in age-related EC accumulation also occur in the cornea, which has higher EC near the limbus (arcus lipoides) than in the center. ⁵ In the cornea this effect is attributed to greater vascular perfusion near the limbus and a tightly packed matrix that retards LDL diffusion toward the center. ⁵ A similar combination of a lipoprotein-retaining milieu and a high blood flow that saturates retentive components could underlie the predilection of macular Bruch’s membrane for extracellular cholesterol. Notably, the choroid has the highest blood flow in the body, and blood flow in the macula is eightfold higher than that in the periphery. ⁵⁶ High blood flow alone is insufficient to account for regional differences in Bruch’s membrane cholesterol, however, because the EC content of the choroidal arteries was low and we could not detect a difference in age-related EC accumulation between macular and peripheral arteries using histochemistry. Understanding the relative roles of blood flow and matrix in EC accumulation will benefit from additional information about regional differences in Bruch’s membrane composition. ^{26 57}  

Although Bruch’s membrane is the wall of a capillary bed, it is intriguingly similar to the inner wall of an artery. The arterial intima is located between two diffusion barriers: an endothelial cell layer and a dense elastic layer. ⁵⁸ Throughout life intima thickens adaptively to the mechanical stresses of blood flow and wall tension. ⁵⁹ Collagen, elastin, and proteoglycans at sites of intimal thickening specifically interact and bind with plasma LDL. ^{60 61} Similarly, Bruch’s membrane is located between the choriocapillaris endothelium and the RPE component of the blood–retina barrier, it thickens threefold throughout adulthood, ^{27 40} and it contains extracellular matrix molecules that could potentially interact with lipoproteins. ^{49 50 62 63 64} Consistent with this model is our previous demonstration of immunoreactivity for apolipoprotein B (apo B), the principal protein of LDL, in peripheral Bruch’s membrane associated with sub-RPE deposits. ⁴⁵ Little immunoreactivity was detected in normal macular Bruch’s membrane, despite high neutral lipid content, ⁴⁵ suggesting that apo B, if present, is normally degraded and cleared efficiently. EC accumulates in Bruch’s membrane of rabbits with extremely high levels of plasma lipoproteins ⁶⁵ but not in mice with moderately elevated levels. ^{66 67 68} The role played by plasma lipoproteins in normal retinal homeostasis is poorly understood. The RPE has LDL receptor activity, ⁶⁹ and plasma LDL transports docosahexanoic acid destined for photoreceptors. ⁷⁰ However, the choriocapillaris endothelium reportedly excludes molecules as large as LDL, ^{71 72} which is inconsistent with our findings. Studies to clarify these issues are warranted. 

If Bruch’s membrane can be conceptualized as an arterial intima, it is appropriate to examine diseases of the intima for insight into ARM. Atherosclerotic cardiovascular disease (CVD), a major cause of morbidity and mortality, is a complex disease that proceeds from the normal infusion of plasma LDL and accumulation of EC in thickened intima through advanced lesions containing increased levels of cholesterol and lipoproteins, culminating in neovascularization, inflammation, and thrombosis. ⁷³ Although the cholesterol and lipoprotein content of ARM-specific lesions ²⁸ remain to be determined, our study confirmed that small age-related drusen contain EC and UC, extending our previous observation that these drusen contain neutral lipids and apo B ⁴⁵ (but see Ref. ⁷⁴ ). A plasma origin for neutral lipids in small drusen is therefore likely, consistent with studies demonstrating other plasma proteins in drusen. ^{74 75} EC-rich droplets occur in other age-related and disease-related sub-RPE deposits. ^{76 77 78} In dominant late-onset retinal degeneration, an inherited disorder, a cholesterol-enriched basal laminar deposit also displays intense apo B immunoreactivity. ⁷⁷ Thus, diverse retinal degenerations with sub-RPE debris may involve plasma lipoproteins interacting with a locally retentive matrix material. 

Significantly, our study demonstrates that with regard to the deposition of extracellular cholesterol, Bruch’s membrane ages like the intima of large, atherosclerosis-prone arteries. In fact, the proportion of Bruch’s membrane occupied by EC-rich droplets in older adults is much higher than in normal arterial intima at the same age (1%–3%). ¹² Therefore, it is plausible that the progression from aging to ARM shares pathogenic mechanisms with atherosclerotic CVD. Seeking this link through epidemiologic studies ⁷⁹ is difficult, because the earliest lesions in CVD occur decades earlier than those in ARM. ^{16 44 80} Our data now strengthen the rationale for seeking links between these two diseases at the tissue, cellular, and molecular level. We caution that atherosclerotic CVD and ARM are complex multifactorial diseases, and the geometry, spatial scale, and cellular constituents of arterial intima and Bruch’s membrane differ in important ways. Nevertheless, this analogy can provide a useful conceptual framework for generating testable new hypotheses about the pathogenesis of ARM. 

 Supported in part by National Institutes of Health Grants EY06109 and P30 EY03039 (Vision Science Research Center, UAB); unrestricted funds to the Department of Ophthalmology from Research to Prevent Blindness, Inc.; and the Alabama Eye Institute.

 Submitted for publication July 21, 2000; revised September 22, 2000; accepted September 29, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Histochemical detection of cholesterol in human Bruch’s membrane, using filipin (epifluorescence, A through C). Arrows, Bruch’s membrane; arrowheads, RPE. The filipin fluorescence in (A) through (C) can be easily distinguished from RPE autofluorescence by color (green versus orange). Bar, 20 μm. (A) EC in Bruch’s membrane is labeled by filipin after extraction and hydrolysis. The RPE and retina are unlabeled. (B) Control section, which was extracted but not hydrolyzed, is devoid of filipin fluorescence. Bruch’s membrane is autofluorescent. (C) Filipin labels UC in Bruch’s membrane and in cellular membranes of retina, RPE, and choroid (Ch) in an unextracted section. ONL, outer nuclear layer.

Figure 2.

 

Filipin fluorescence due to EC and UC increases with age in Bruch’s membrane. For EC in the macula (A) and periphery (B) and UC in the macula (C) and periphery (D), fluorescence intensity (×10⁻⁶ arbitrary units) for each eye is expressed as the mean sum of intensities within 15 4.86-μm square windows placed on Bruch’s membrane in the experimental section minus the mean sum of intensities in control sections of the same eye. Negative differences between experimental and control means were set to zero.

Table 1.

 

Histochemically Detected Cholesterol: Age and Retinal Location

Region	Fluorescence	Versus	Slope Parameter, Multiple Linear Regression*
						Crude	Adjusted, †
Macula	;l>EC	Age	0.14, ‡	—, §
		Peri EC	1.31	−1.26
Macula	UC	Age	0.09, ‡	—§ >
		Peri UC	0.86	−0.04
Periphery	EC	Age	0.02, ‡	—§ >
		Mac EC	0.06	−0.81, ‡
Periphery	UC	Age	0.01	—§ >
		Mac UC	0.07	−0.01

 EC, esterified cholesterol; UC, unesterified cholesterol; Mac, macula; Peri, periphery.

*  Slope parameter b, where y = a + bx

†  Adjusted for age.

‡  Significant (P < 0.05).

§  —, not in model.

Figure 3.

 

Filipin fluorescence due to EC in choroid. (A) Artery (a) and vein (v), bright field. (B) EC is in artery wall only.

Table 2.

 

Esterified Cholesterol in Choroidal Arteries

	Macula	Periphery
<60 years	1/7	2/7
>60 years	9/11	9/11

 Values are number of eyes with filipin-labeled arteries per total number of eyes.

Table 3.

 

Cholesterol Content of Bruch’s Membrane, Retina, and Choroid

Age (y)	Macula											Periphery
		Bruch’s Membrane			Retina			Choroid			Bruch’s Membrane
		EC	EC/TC (%)	EC	EC/TC (%)	EC	EC/TC (%)	EC	EC/TC (%)
83	21.3	62.3	—	—	—	—	—	—
85	15.5	56.0	—	—	—	—	—	—
76	35.7	62.5	—	—	—	—	—	—
85	43.3	58.8	0.0	0.0	4.5	13.5	—	—
76	20.2	64.8	7.1	8.4	4.7	12.8	—	—
71	39.2	47.6	3.2	4.4	3.9	5.9	9.4	15.4
72	26.8	50.2	0.0	0.0	10.5	15.6	9.8	20.9
80	58.6	68.2	4.0	12.0	10.3	28.7	—	—
87	62.3	59.4	0.0	0.0	5.8	11.7	5.7	44.3
80	68.4	66.0	0.1	0.1	12.4	16.7	—	—
Mean 79.5	39.1	59.6	2.0	3.6	7.4	15.0	8.3	26.9
SD 5.6	18.8	6.7	2.8	4.9	3.5	7.0	2.3	15.4

 EC, esterified cholesterol, in nmol/gm dry weight; TC, total cholesterol.

Figure 4.

 

Lipid-rich particles in intercapillary pillar of foveal Bruch’s membrane, conventional EM (A) and OTAP (B), 85-year-old donor. Basal RPE is at the top. ic, inner collagenous layer; oc, outer collagenous layer; c, collagen fibrils (electron dense in A, negative image in B); arrowheads, RPE basal lamina; asterisks, electron lucent droplets in (A) and solid particles in (B); large double arrows, coated membrane-bounded bodies; large single arrow, remnants of coated membrane-bounded bodies; small double arrows, row of particles external to the RPE basal lamina.

Table 4.

 

Effects of Solvents on Bruch’s Membrane Cholesterol

	Untreated	Ethanol	Chloro/Meth
Filipin fluorescence*
Unesterified cholesterol	Intense	At background, †	At background, †
Esterified cholesterol	NA, ‡	Intense, ‡	At background, †
Ultrastructure, §
Droplets/vesicles	++	++	Distorted
CMMB	+	+	−
Coiled membranes	+++	++	+
RPE membranes	+++	+	−
ChC membranes	+++	+	−

 NA, not applicable; ChC, choriocapillaris endothelium; RPE, retinal pigment epithelium; CMMB, coated membrane bounded bodies.

*  Fluorescence was quantified in two eyes per condition (see Methods).

†  Significant reduction in intensity by t-test.

‡  By definition, the EC condition involved ethanol treatment (see Methods).

§  Frequency of findings in two eyes: −, not present; +, present in<1/2 of the micrographs; ++, present in >1/2 of micrographs; +++, present in all micrographs.

Figure 5.

 

Lipid-rich particles in eyes of different ages, OTAP. Arrowheads, RPE basal lamina; arrows, electron dense lipid particles. c, negative image of collagen fiber in cross-section. (A) 16 years; (B) 51 years; (C) 76 years. X, basal laminar deposit.

Figure 6.

 

Oblique section of Bruch’s membrane (OTAP, 85-year-old donor). RPE is at top. Arrowheads, RPE basal lamina; asterisk, the band of lipid-rich particles. Other particles are disbursed among collagen fibrils (negative image) in the inner collagenous layer.

Figure 7.

 

Cholesterol and age-related changes in Bruch’s membrane. (A through E) Macula; (F through I) periphery. (A, B) Arrowhead, an EC-free gap (A, filipin fluorescence) that corresponds to a calcified patch (B, bright field, hematoxylin). (C) Band of lipid-rich particles (arrows) overlying a druse-like deposit on the inner surface of Bruch’s membrane, OTAP. Arrowheads, RPE basal lamina; X, basal laminar deposit. (D) EC in small, drusen-like deposits; (E) small druse with EC-rich shell; (F) EC in druse; (G) UC in druse; (H) EC in basal deposits; (I) UC in basal deposits.

The authors thank the Alabama Eye Bank for timely retrieval of donor eyes; Shawn Williams of the High Resolution Imaging Facility at UAB for assistance with digital microscopy; Gerald McGwin, Jr., Ph.D., for statistical consultation; David Fisher for graphics assistance; and Steven J. Fliesler, Ph.D., for helpful discussions.