Ultrastructural Changes in Bruch’s Membrane of Apolipoprotein E–Deficient Mice

Abstract

purpose. To examine the histologic and ultrastructural changes in Bruch’s membrane (BM) in apolipoprotein E deficient [ApoE(−)] mice in comparison with age-matched control animals.

methods. Two-month-old (group 1) and 8-month-old (group 2) normal control C57BL/6 mice and 2-month-old (group 3) and 8-month-old (group 4) ApoE(−) mice were studied. All groups of mice were fed a standard rodent diet. The mice were killed, serum lipid levels were determined, and the eyes were ultrastructurally examined using standard techniques to measure the thickness of BM. The area fraction of electron-lucent (EL) particles in BM was quantified using point-counting stereology.

results. The serum cholesterol levels of the ApoE(−) mice were significantly higher than those of the control mice (P = 0.0001). There was a significant thickening and EL particle accumulation in BM associated with age in the control animals. Group 2 had a thicker BM and more EL particle accumulation than group 1 (P = 0.0410 for thickness; P = 0.0042 for particle accumulation). Age-related changes were not seen in ApoE(−) mice; thickness and accumulation were similar in groups 3 and 4 (P = 0.50, thickness; P ≅ 1.0, accumulation). Significant thickening and accumulation were seen in young ApoE(−) mice (group 3) versus young control animals (group 1; P = 0.008, thickening; P < 0.0001, EL particle accumulation). Group 4 ApoE(−) mice did not have a thicker BM or more EL particles than group 2 control animals (P = 0.2910, thickness; P = 0.35, EL particle accumulation).“ Membrane-bounded” material (material between two membranes) was present significantly more frequently in ApoE(−) mice.

conclusions. ApoE(−) mice exhibit accumulation of EL particles at an earlier age and have more membrane-bounded material in BM than control mice. This material has ultrastructural similarities to basal linear deposit, which accumulates in age-related maculopathy.

The cause of age-related maculopathy (ARM), the most common cause of irreversible blindness in Europe and the United States, is largely unknown. Rare macular disorders have been linked to several genomic loci, ^{1 2 3 4 5} and there is evidence that genetic factors are also important in the development of ARM. ^{6 7 8 9 10} A relationship between different apolipoprotein E (ApoE) alleles and the incidence of ARM has recently been reported. ^{11 12 13 14 15} ApoE is a polymorphic protein involved with metabolism of plasma lipids and in central nervous system lipid homeostasis. ¹⁶ Clinical studies have reached contradictory conclusions concerning the relationship between ARM and different ApoE alleles. ^{11 12 13 14} Accumulation of material in the area of the retinal pigment epithelium (RPE) and Bruch’s membrane (BM), including electron-lucent (EL) particles, membranous debris, basal laminar deposit (BlamD), ^{17 18} and basal linear deposit (BlinD), ^{19 20 21} occurs with aging and ARM. Whereas BlamD is not thought to be specific for ARM, ^{17 18} the presence of BlinD correlates with ARM. ^{19 20 21} There is also an age-related accumulation of lipid, presumably including cholesterol in BM. ²² ApoE(−) mice have increased serum cholesterol levels. ^{23 24} If BM debris accumulation arises from serum lipid including cholesterol, then hypercholesterolemic ApoE(−) mice should exhibit particle accumulation in BM when compared with control mice. 

Materials and Methods

Mice

All experiments were conducted according to the Declaration of Helsinki and Guiding Principles in the Care and Use of Animals. Female normal C57BL/6 and C57BL/6-ApoE(−) mice were purchased from Jackson Laboratories (Bar Harbor, ME). ApoE(−) mice are derived from embryonic stem cells in which the murine ApoE gene has been inactivated by gene targeting. The mice have no ApoE and have hypercholesterolemia due to elevated levels of very low- and intermediate-density lipoproteins. These mice have been produced in two laboratories ^{23 24} and are characterized by spontaneous, pronounced hypercholesterolemia and extensive aortic atherosclerosis, even when fed a low-fat diet. ^{23 24} The mice had free access to a standard rodent chow (laboratory rodent diet 5001; PMI Nutrition, Brentwood, MO) and water, were housed in plastic cages, and were kept on a 12-hour light–dark cycle. Four groups of mice were examined: 2-month-old C57BL6 mice (group 1; n = 10), 8-month-old C57BL6 mice (group 2; n = 10), 2-month-old ApoE(−) mice (group 3; n = 10), and 8-month-old ApoE(−) mice (group 4; n = 20). 

Tissue Preparation

The different groups of mice were killed, blood samples for lipid analysis ²⁵ were taken, and the 12 o’clock position of the eyes was marked. The eyes were then enucleated and the right eyes placed in 2.5% glutaraldehyde. The left eyes were fixed with 4% paraformaldehyde in phosphate-buffered saline. After 24 hours, the central area of the posterior pole was identified according to the 12 o’clock mark, and a rectangular piece of tissue measuring 1.5 × 1 mm including the optic nerve head was removed. Central areas of the right eyes were again placed in 2.5% glutaraldehyde and processed for electron microscopy. Central areas of the left eyes were rinsed with buffer and stored for further processing. 

Electron Microscopy

For electron microscopy, tissue was postfixed with 1% osmium tetroxide in 0.1 M cacodylate buffer. Standard dehydration of the specimen was performed, the specimen was embedded in epoxy resin (LX-112; Ladd Research Industries, Burlington, VT), and semithin sections (1.0 μm) were cut and stained with 2% toluidine blue in 2% sodium borate. Ultrathin (silver) sections were then cut, stained with uranyl acetate and lead citrate, and examined using an electron microscope (mode 100CXII; JEOL, London, UK). A section containing the optic nerve head and central area was scanned at ×1900 magnification with ×10 binoculars. Representative photographs were taken approximately 1 mm from the center of the optic nerve head at ×29,000 and printed at ×72,500 magnification. The thickness of BM was measured in two representative photomicrographs per case. The thinnest and thickest parts of BM were measured and the average thickness was determined. The ultrastructural characteristics were determined using defined criteria. ^{17 18 19}  

Stereological Measurements

Stereology was used to quantify the amount of non–membrane-bounded and membrane-bounded profiles present in each BM. ²⁶ Electron micrographs at ×72,500 magnification were analyzed. A transparent grid with 7-mm spacing, approximately double the average diameter of the non–membrane-bounded particles, was placed over the micrograph. The volume fraction of BM overlying EL particles was determined by counting the grid point boxes filled with EL particles and total grid points in BM. All micrographs used for this procedure had at least 46 grid points across the BM, which allowed for a 5% error. 

Statistical Methods

The Wilks–Shapiro test was used to test whether the distribution of cholesterol level, thickness of BM, and volume fraction of the EL profiles was normal. From these results, the Wilcoxon rank sum test was used to determine whether mean cholesterol level was higher in ApoE(−) mice than in the control animals. The Wilcoxon rank sum test was also used to compare the mean average BM thickness in different mouse groups. With the assumption that the volume fraction of the EL particles was distributed normally in each mouse group, t-tests were performed to compare the amount of lipid particles in different groups. P < 0.05 was considered statistically significant. 

Results

Serum Lipid Levels

C57BL6 mice had an average cholesterol blood level of 68 ± 12 mg/dl and triglyceride level of 44 ± 22 mg/dl. ApoE(−) mice had an average cholesterol level of 651 ± 176 mg/dl and triglyceride level of 56 ± 23 mg/dl (Table 1) . 

Thickness of BM

The average thickness of BM was 0.35 ± 0.13 μm in 2-month-old C57BL6 mice, 0.45 ± 0.11 μm in 8-month-old C57BL6 mice, 0.47 ± 0.09 μm in 2-month-old ApoE(−) mice, and 0.52 ± 0.23 μm in 8-month-old ApoE(−) mice (Table 2) . Eight-month-old C57BL6 mice (group 2) had a significantly thicker BM than 2-month-old C57BL6 mice (group 1; P = 0.0410). Eight-month-old ApoE(−) mice (group 4) did not have a thicker BM than 2-month-old ApoE(−) mice (group 3; P = 0.5000). Two-month-old ApoE(−) mice (group 3) had a thicker BM than 2-month-old C57BL6 mice (group 1; P = 0.0080). Eight-month-old ApoE(−) mice (group 4) did not have a thicker BM than 8-month-old C57BL6 mice (group 2; P = 0.2910). 

Particles in BM

The ultrastructural features of all groups are shown in Figures 1 2 3 4 5 6 7 8 . Ultrastructural analysis revealed two distinctive vacuolar changes in BM. These included non–membrane-bounded EL vacuoles, which were round, occasionally confluent, and scattered throughout both collagenous layers (Fig. 4) . In addition, we observed larger,“ membrane-bounded” vacuoles in both collagenous layers (Figs. 6 8) that measured 40 to 320 nm in diameter. These vacuoles were bounded by bilaminar or multilaminar membranes. We refer to both of these types of vacuoles as membrane-bounded for convenience. The distribution of these changes differed significantly among groups of mice. The non–membrane-bounded material was present in 2 of 10 2-month-old C57BL6 mice and in all mice of the other groups. The membrane-bounded material was absent in mice in group 1, present in only three mice of group 2, and in all ApoE(−) mice (groups 3 and 4, Fig. 8 ). 

Volume Fraction of Material in BM

The volume fraction of membrane-bounded and non–membrane-bounded particles quantified by stereological measurements was 0.29 ± 0.10, 0.41 ± 0.08, 0.50 ± 0.06. and 0.42 ± 0.08 in groups 1 through 4, respectively (Table 3) . Eight-month-old C57BL6 mice (group 2) had a significantly higher volume fraction than did 2-month-old C57BL6 mice (group 1; P = 0.0042). The volume fraction in 8-month-old ApoE(−) mice (group 4) was similar to that in 2-month-old ApoE(−) mice (group 3; P ≅ 1.0). Two-month-old ApoE(−) mice (group 3) had a higher volume fraction than did 2-month-old C57BL6 mice (group 1; P < 0.0001). Eight-month-old ApoE(−) mice (group 4) did not differ from 8-month-old C57BL6 mice (group 2; P = 0.35). The ratios of the average volume fraction and average thickness of groups 2, 3 and 4 to group 1 were similar (1.5 and 1.4 respectively). Assuming that the area of BM is constant, this indicates that the change in volume was due to a change in thickness. 

Discussion

ApoE, first identified in 1973 by Shore and Shore, ²⁷ is a surface constituent of lipoprotein particles (chylomicron remnants, very low-density lipoprotein remnants, and intermediate-density lipoproteins) and is a ligand for lipoprotein recognition and clearance by hepatic lipoprotein receptors. ^{28 29 30} It is a ligand for the chylomicron remnant receptor ¹⁶ as well as for the low-density lipoprotein (LDL) receptor. ApoE is produced primarily in the liver, accounting for two thirds to three fourths of plasma ApoE, although ApoE is produced in most other organs. ¹⁶  

Three different codominant alleles (ε2, ε3, ε4) at a single gene locus are responsible for six phenotypes, three homozygous (E2/2, E3/3, E4/4) and three heterozygous (E3/2, E4/2, E4/3). The most common allele is ε3, and the most common phenotype is Apo-E3/3. ¹⁶ Clinical studies have yielded contradictory results regarding the relationship between ARM and different ApoE alleles. In one study, Souied et al. ¹¹ found a lower frequency of the ε4 allele in patients with exudative ARM, suggesting that the ε4 allele is a potential protective factor for the disease. In another clinical study, Klaver et al. ¹² also found a lower frequency of the ε4 allele in patients with ARM and did not find a significantly higher frequency of the ε2 allele in patients with ARM compared with control animals. Other studies have failed to demonstrate any significant difference in allele frequency between cases and control animals. ^{13 14}  

Kliffen et al. ¹⁵ found BlamD-like material in ApoE3 transgenic mice, although the nature of the inactivated ApoE was not specified. In that study, electron microscopic examination was performed on two eyes and BlinD-like material was not described. ¹⁵ Whereas BlamD is not thought to be specific for ARM, ^{17 18} the presence of BlinD is associated with early and late ARM. ^{19 20 21} Choroidal neovascularization (CNV) in eyes with ARM invades and ramifies in the plane of the BlinD and drusen and not the plane of BlamD. ^{31 32 33 34} In our study, we did not find ultrastructural evidence of BlamD-like material in C57BL6 control animals nor in ApoE(−) mice. Mice had ultrastructural changes in BM that resemble those seen in eyes of aged human donors and donors with ARM. ^{20 35} EL droplets are scattered throughout BM of adult human eyes and form a discrete layer external to the RPE basal lamina in elderly eyes. ¹⁹ All mice except the youngest C57BL6 had similar droplets. Membrane-bounded particles are the principal component of basal linear deposits and large drusen, which are lesions specific for ARM. ¹⁹ Alterations resembling membranous debris increased with age in both the C57BL6 mice and ApoE(−) mice. In both mice and humans, these ultrastructural profiles resemble lipid-rich droplets and vesicles that are extracted by tissue processing for conventional electron microscopy. ³⁶ Other methods are required to establish the biochemical identity of these putative lipid-rich structures. 

There is confusion in terminology regarding BlamD and BlinD. ¹⁷ These terms are evolving. Currently, BlamD refers to electron-dense material with associated fibrous widely spaced collagen located between the plasma membrane and basement membrane of the RPE. ¹⁹ BlinD refers to membranous debris and non–membrane-bounded EL droplets, often but not necessarily located between the basement membrane of the RPE and inner collagenous zone of BM. ¹⁹ BlinD, and not BlamD, appears to be a specific marker for ARM. ¹⁹ The origin and biochemical composition of BlinD is unclear, although the EL droplets and membranous debris in BlinD have ultrastructural similarities to extracellular material found in atheromatous plaques. ^{37 38} This lipid-rich material in atheromatous plaques is thought to be derived from the serum. The membrane-bounded vacuoles that accumulate in BM in ApoE(−) mice have a similar ultrastructural appearance to membranous debris found in BlinD. It is beyond the scope of this study to determine the histochemical properties of these membrane-bounded vacuoles. EL droplets were associated with age or ApoE deficiency, although accumulation of membrane-bounded vacuoles was only associated with ApoE deficiency (and hypercholesterolemia) in our study. Therefore, accumulation of EL droplets and membranous debris in BlinD may be due to separate mechanisms. This may be relevant to discrepancies in the association between the absence of the Apo-ε4 allele and ARM in various studies. 

A difference in the presence of membranous debris was found between the 8-month-old C57BL6 control animals and ApoE(−) mice. These results suggest that ApoE deficiency predisposes to ultrastructural changes in BM. CNV was not present, and other factors, such as exposure to peroxidative injury, ³⁹ may be related to the development of CNV. ApoE(−) mice had average plasma cholesterol levels of 651 ± 176 mg/dl, which is approximately 9.5 times more than that in C57BL6 mice consuming a normal diet. This is much higher than plasma cholesterol levels of C57BL6 mice consuming a high-fat diet, which ranges between 230 and 270 mg/dl. ⁴⁰ The findings in BM in our study may be directly caused by ApoE deficiency; increased plasma lipid levels, inasmuch as other serum components are present in drusen ⁴¹ ; or other factors, including effects at the level of the RPE, because ApoE mRNA and LDL receptor have been demonstrated in the RPE. ^{42 43} Our study shows that ApoE deficiency in mice increases the incidence and amount of age-dependent BlinD-like debris in BM. 

 Supported in part by Grant Di 98/99 from the Gertrud Kusen Foundation, Hamburg, Germany; Departmental Core Grant EY-06030 from the National Eye Institute; and an unrestricted grant from Research to Prevent Blindness.

 Submitted for publication December 10, 1999; revised January 19, 2000; accepted February 7, 2000.

 Commercial relationships policy: N.

 

Table 1.

 

Plasma Lipid Levels in C57BL6 and ApoE-Deficient Mice

Group	Total Plasma Cholesterol	Total Plasma Triglycerides	VLDL Cholesterol	IDL Cholesterol	LDL Cholesterol	HDL Cholesterol
C57BL6	68 ± 12	44 ± 22	5.5 ± 1	0	18 ± 3	44 ± 8
ApoE(−) deficient	651 ± 176	56 ± 23	318 ± 86	49 ± 13	270 ± 73	14 ± 4

 Data are expressed in milligrams per decaliter. HDL, high-density lipoprotein; IDL, intermediate-density lipoprotein; LDL, low-density lipoprotein; VLDL, very low-density lipoprotein.

Table 2.

 

Thickness of Bruch’s Membrane

Group	Minimum Thickness	Maximum Thickness	Average Thickness
Two-month-old C57BL6 mice	0.32 ± 0.12	0.38 ± 0.14	0.35 ± 0.13
Eight-month-old C57BL6 mice	0.31 ± 0.06	0.58 ± 0.18	0.45 ± 0.11
Two-month-old ApoE(−) deficient mice	0.30 ± 0.07	0.65 ± 0.13	0.47 ± 0.09
Eight-month-old ApoE(−) deficient mice	0.35 ± 0.08	0.70 ± 0.39	0.52 ± 0.23

 Data are expressed in micrometers.

Figure 1.

 

Group 1. BM (between arrowheads) in 2-month-old C57BL6 mouse was present between the RPE and endothelium of the choriocapillaris (END). Magnification, ×8160.

Figure 2.

 

Group 1. BM (between arrowheads) was free of deposits. The RPE and choriocapillaris endothelium (END) were unremarkable. Magnification, ×49,300.

Figure 3.

 

Group 2. BM (between arrowheads) in 8-month-old C57BL6 mouse. The RPE and choriocapillaris endothelium (END) were unremarkable. Magnification, ×8160.

Figure 4.

 

Group 2. There were numerous non–membrane-bounded vacuolations (∗) in BM (between arrowheads), which lies between the RPE and choriocapillaris endothelium (END). Magnification, ×49,300.

Figure 5.

 

Group 3. BM (between arrowheads) appeared vacuolated in this 2-month-old ApoE(−) mouse. The RPE and choriocapillaris endothelium (END) appeared to be normal. Magnification, ×8160.

Figure 6.

 

Group 3. There were both non–membrane-bounded (∗) and membrane-bounded (arrow) vacuolations in BM (between arrowheads). The RPE was normal, and the choriocapillaris endothelium (END) contained occasional vacuolations. Magnification, ×49,300.

Figure 7.

 

Group 4. This 8-month-old ApoE(−) mouse exhibited vacuolations in BM (between arrowheads). The RPE was normal, and the choriocapillaris endothelium (END) exhibited reactive cytologic changes. Magnification, ×8160.

Figure 8.

 

Group 4. There were both non–membrane-bounded (∗) and membrane-bounded (arrow) vacuolations in BM (between arrowheads). The RPE was normal, and there were scattered vacuolations in the choriocapillaris endothelium (END). Magnification, ×49,300.

Table 3.

 

Average Volume Fraction of Lipid-Rich Particle in Bruch’s Membrane

Group	Average Volume Fraction
Two-month-old C57BL6 mice	0.29 ± 0.10
Eight-month-old C57BL6 mice	0.41 ± 0.08
Two-month-old ApoE(−) deficient mice	0.50 ± 0.06
Eight-month-old ApoE(−) deficient mice	0.42 ± 0.08