Effects of Cholesterol and Apolipoprotein E on Retinal Abnormalities in ApoE-Deficient Mice

Abstract

purpose. To examine the pathologic changes in the retina of apolipoprotein E (apoE)-deficient mice fed a high-cholesterol diet.

methods. ApoE-deficient mice (ApoE) were maintained on either regular mouse chow (ApoE-R) or a high-cholesterol diet (ApoE-C) for 25 weeks. Age-matched control C57BL/6J mice (C57) were also maintained on either regular mouse chow (C57-R) or a cholesterol-containing diet (C57-C). Retinal function was assessed by dark-adapted electroretinography (ERG). The eyes were embedded, sectioned, and analyzed by histologic and immunohistochemical methods, as well as by light and transmission electron microscopy.

results. After the 25-week feeding period, ERG tracings of ApoE-C mice revealed significant increases of a- and b-wave implicit times when compared with the C57-R group of mice. In addition, there were reductions in oscillatory potential (OP) amplitudes in the ApoE-C group. However, a- and b-wave amplitudes appeared to be unchanged among the four groups of mice. Light microscopic examination of the retinas showed that compared with control C57-R mice, ApoE-C mice had significantly lower cell numbers in the inner and outer nuclear layers (85.1% ± 4.6%, P < 0.05 and 81.4% ± 3.7%, P < 0.01 of C57-R controls, respectively). Transmission electron microscopy of apoE-deficient mice revealed cells of the inner nuclear layer with condensation of nuclear chromatin and perinuclear vacuolization in focal areas. Bruch’s membrane was also found to be thicker, and its elastic lamina appeared disorganized and discontinuous. Immunohistochemistry demonstrated diminished or no immunoreactivity for carbonic anhydrase II and calretinin in the retinal layers of apoE-deficient mice.

conclusions. Overall, there were increasing abnormalities of retinal function and cellular morphology among the four groups of mice in the order of C57-R < C57-C < ApoE-R < ApoE-C. These findings suggest that apoE and/or cholesterol play an important role in retinal function.

Age-related macular degeneration (ARMD) is the leading cause of irreversible blindness among older people. Epidemiologic studies have suggested that nutritional parameters such as dietary fat and cholesterol may play a role in the pathogenesis of ARMD. ¹ Investigators studying a population from the Beaver Dam Eye Study and Nutritional Factors in Eye Disease Study found that high intake of saturated fat and cholesterol is correlated with a higher risk for early age-related maculopathy (ARM). ² Several studies have also shown a relationship of cardiovascular disease with ARMD, ^{3 4 5 6} although others have not been able to verify this finding. ^{7 8} Dietary fat and, in particular, cholesterol are positively linked to increased incidences of coronary heart disease (CHD), ^{9 10} and evidence suggests that abnormal lipid levels may contribute to the development of ARMD, either directly or through the promotion of vascular disease. 

The association of lipoprotein metabolism and neurodegenerative disorders, including age-related neurodegenerative diseases such as Alzheimer’s disease, has been examined in recent years. Given the cholesterol requirement of neuronal cells of the nervous system, it is reasonable to theorize that in neuronal cells, there is an intimate relationship between cholesterol homeostasis and their development, maintenance, and repair. Apolipoprotein E (apoE) plays a central role in serum cholesterol homeostasis through its ability to bind cholesterol and other lipids and to mediate their transport into cells. ^{11 12} In the central nervous system (CNS), apoE is the primary protein component of CNS lipoproteins and is produced by glial cells. ¹³ Although apoE has been implicated in neuronal regeneration, ¹⁴ little is known about lipid delivery and clearance within the CNS and even less of the role apoE plays in the processes. 

Epidemiologic studies have demonstrated genetic association of the ε4 allele of apoE with late-onset familial and sporadic Alzheimer’s disease, ^{15 16} and it has been postulated that apoE is directly involved with cerebral amyloidogenesis. ^{17 18 19 20 21 22 23} Earlier studies investigating the relationship between ARM and the different apoE alleles found no significant differences between control subjects and patients with AMR. However, more recently, the apoE gene polymorphism has been found to be genetically associated with ARMD. Klaver et al. ²⁴ reported that there was a slightly increased risk of ARMD associated with the apoE ε2 allele (odds ratio 1.5; 95% confidence interval [CI ] 0.8–2.82). More dramatically, there was a significant association with a decreased risk of ARMD with apoE ε4 (odds ratio 0.43; 95% CI 0.21–0.88). 

In a second study, Souied et al. ²⁵ found the same association of the apoE ε4 allele with a protective factor of ARMD. It was shown that the frequency of carriers of the apoE ε4 allele in the ARMD group was significantly less when compared with age- and sex-matched control subjects (12.1% vs. 28.6%, respectively; P < 0.0009). Their data also showed a lower frequency of the apoE ε4 allele in the ARMD group than in the control group (0.073% vs. 0.149%, respectively; P < 0.006). Furthermore, it was revealed that the decreased frequency of the ε4 allele was mainly due to the subgroup of patients with ARMD with only soft drusen when compared with control subjects (0.045% vs. 0.149%, respectively; P < 0.0009). 

Most recently, Schmidt et al. ²⁶ have reported findings that further support a protective effect of the apoE ε4 allele against ARMD. The investigators found that the odds ratio for apoE ε4 allele carriers among individuals younger than 70 years of age with familial ARMD was 0.24 (95% CI 0.08–0.72). However, the protective effect of the apoE ε4 allele was not observed in patients with familial ARMD older than 70 years of age or in patients with sporadic ARMD. ²⁶ These studies, along with the fact that there is a considerable neuronal cell makeup of the retina, strongly suggest that overall cellular lipid and cholesterol balance is important for normal retinal function. 

Previously, our laboratory observed abnormal changes of the retinal cell layers in cholesterol-fed apoE-deficient mice, including retinal neuronal cell drop-out and cell layer thinning. ²⁷ Recently, Miceli et al. ²⁸ described retinal changes in the retinal pigment epithelium (RPE) and Bruch’s membrane (BM) of C57BL/6 mice fed an atherogenic diet. These pathologic changes included increases of the number and size of autophagocytic and empty cytoplasmic vacuoles in the RPE, thickening and fragmentation of the elastic lamina in BM, and lipidlike droplet accumulation in the RPE. More recently, Dithmar et al. ²⁹ demonstrated that 8-month-old apoE-deficient mice consuming low-fat chow exhibit ultrastructural changes in BM with similarities associated with ARM. The apoE-deficient mouse is an animal model used to study abnormal lipoprotein metabolism and hypercholesterolemia and has been used extensively in cardiovascular and neurologic research. In this study, we used the apoE-deficient mouse model in conjunction with dietary fat and cholesterol intake. We report both functional and morphologic changes in the retina in this animal model system. The findings indicate that this mouse model could provide valuable insight into the role of apoE and cholesterol in retinal function. 

Materials and Methods

Animal Maintenance

Aged-matched control C57BL/6J and apoE-deficient mice were purchased at 4 to 5 weeks of age from Jackson Laboratory (Bar Harbor, ME). Control mice were C57BL/6J mice fed regular mouse chow (C57-R). Three groups of experimental mice consisted of C57BL/6J mice fed a high-cholesterol diet (C57-C), apoE-deficient mice fed regular mouse chow (ApoE-R), and apoE-deficient mice fed a high-cholesterol diet (ApoE-C). Each group contained five to eight mice. Animals were fed ad libitum for 25 weeks with either regular mouse chow composed of 11% fat (animal fat), 0.03% cholesterol, and 17.5% protein (mouse diet 5015; Purina Mills, Inc., Richmond, IN) or a high-cholesterol diet composed of 21% fat (milk fat), 0.15% cholesterol, and 19.5% protein (Teklad Adjusted Calories Western-type diet TD88137; Harlan Teklad, Madison, WI). Both diets contained comparable mineral and vitamin mixes. The regular mouse chow is the normal diet provided to mice at the vivarium at Cedars-Sinai Medical Center and therefore was used as the control diet. The high-cholesterol diet is an atherogenic diet that has been used in previous studies in apoE-deficient mice. ^{30 31} The use of experimental animals was in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Serum Cholesterol Analysis

Blood samples were collected after mice were killed and sera obtained after separating the red blood cells by centrifuge. Only samples that had little or no sign of hemolyzed blood were assayed. Levels of serum cholesterol were measured using a cholesterol diagnostic kit and cholesterol calibrator standards (Sigma Chemical Co., St. Louis, MO). Levels of cholesterol were calculated from the linear range of the cholesterol standards. 

Electroretinography

The electroretinographic (ERG) procedure was performed on animals that had been dark adapted for at least 15 hours. Each animal was anesthetized with ketamine and xylazine (intraperitoneally, 88 and 14 mg/kg body weight, respectively) and pupils dilated with 1 drop each 1% tropicamide and 2.5% phenylephrine hydrochloride. After the animal was allowed to stabilize on a 37°C warming pad for 10 minutes, 1 drop 2.5% hydroxypropyl methylcellulose (Goniosol; Ciba Vision Ophthalmics, Duluth, GA) was put in the eye to act as an electrode conductor. The three electrodes attached to the animal were an electrode consisting of a wire-attached 21-G needle placed just underneath the skin of the tail (ground electrode), a silver wire electrode placed just inside the mouth (reference electrode), and a supported silver wire electrode placed in contact with the surface of the eye (test electrode). The mouse was placed in a photopic stimulator chamber where the animal was exposed to flashes of blue light once every 5 to 7 seconds. The a-wave amplitude was measured from baseline to the a-wave trough, and the b-wave amplitude was measured from the a-wave trough to the b-wave peak. 

Tissue Preparation

Animals were anesthetized with carbon dioxide and killed by decapitation. After enucleation, the eyes were fixed in 10% neutral buffered formalin for 4 hours at room temperature and either embedded in optimal cutting temperature (OCT) compound (Tissue Tek II; Laboratory Tek, Naperville, IL) or fixed in 4% paraformaldehyde and embedded in paraffin (TissuePrep 2; Fisher Scientific, Fair Lawn, NJ). Alternately, some eyes were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate-0.2 M sodium phosphate buffer for 12 hours at 4°C and embedded in resin (Spurr; Ted Pella, Inc., Redding, CA). 

Morphometric Assessment

Retinal morphometric analysis was performed on tissue cross sections that bisected the optic nerve in OCT-embedded eyes. Color photomicrographs were taken and subjected to measurement and cell counting by two separate individuals who were blind to the identities of the tissue sections. For all measurements and cell counting, the area of the retina adjacent to the optic nerve head was analyzed. Distant measurements and cell counts were normalized to the overall thickness of the retina to adjust for slight irregularities of the bisectional cut that might exist. Several sections from each eye (three to five sections) were measured and counted and the results were averaged. 

Statistical Analysis

The data from the serum cholesterol, ERG, and morphometric measurements were subjected to statistical analysis by ANOVA on computer (Prism, ver. 3.00; GraphPad Software, Inc., San Diego, CA). The number of mice per group ranged from five to eight animals. All values were normalized to control C57BL/6J mice fed regular mouse chow for the morphometric analysis. P ≤ 0.05 was considered statistically significant. 

Transmission Electron Microscopy

Whole enucleated mouse eyes were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate-0.2 M sodium phosphate buffer, postfixed in 1% osmium tetroxide, stained with 1% uranyl acetate, and embedded in low-viscosity embedding medium (Spurr; Ted Pella). The eyes were sectioned at 1 μm on an ultramicrotome (Ultracut R; Leica, Deerfield, IL). The tissue sections were stained with 2% toluidine blue O and examined under a light microscope to determine areas of interest. Thin sections of approximately 50 to 90 nm were cut and collected on copper grids. These sections were stained with 4% uranyl acetate and Reynold’s lead citrate. Subsequently, the sections were evaluated by transmission electron microscopy (TEM; EM10; Carl Zeiss, Thornwood, NY) and photographed. 

Immunocytochemical Analysis

Paraffin-embedded tissue sections were subjected to protease treatment (Pronase; Calibiochem-Novabiochem Corp., San Diego, CA) for 10 minutes at room temperature and then blocked (Ultra V Block; Laboratory Vision Corp., Fremont, CA) for 10 minutes at room temperature. The sections were incubated with carbonic anhydrase II (CAII) antiserum (catalog number 100-401-136; Rockland Immunochemicals, Gilbertsville, PA; 1:1000) or calretinin antiserum (catalog number A149; Chemicon International Inc., Temecula, CA; 1:100) overnight at 4°C. After washing, secondary biotinylated anti-rabbit antibody (Vectastain ABC Kit PK-4001; Vector Laboratories, Inc., Burlingame, CA) was added to the sections and subsequently visualized by developing with diaminobenzidine (DAB) substrate (Sigma Chemical Co.). 

Results

Serum Cholesterol

After a 25-week feeding period, the levels of serum cholesterol were measured. Whereas control C57BL/6J mice fed regular mouse chow (C57-R) had normal levels of approximately 115 ± 9 mg/dl, the other three experimental groups of mice were found to have elevated levels of serum cholesterol. C57BL/6J mice fed the cholesterol-containing diet (C57-C) and apoE-deficient mice fed regular mouse chow (ApoE-R) had average cholesterol levels of 359 ± 75 mg/dl and 409 ± 53 mg/dl, respectively, whereas apoE-deficient mice fed the cholesterol-containing diet (ApoE-C) were found to have the highest levels of serum cholesterol (1451 ± 126 mg/dl). These values are comparable to those reported by others. ^{30 32}  

Electrophysiological Analysis

ERGs were recorded for the four groups of mice, C57-R, C57-C, ApoE-R, and ApoE-C. Figure 1 shows the dark-adapted responses of representative mice from the groups. From these ERGs, it can be seen that, compared with the control C57-R mouse, implicit times of a- and b-waves were increased for the other three types of mice. Average a- and b-wave implicit times for each group of mice are shown in Figures 2A and 2B . Both a- and b-wave implicit times were found to be prolonged with increasing times in the order of C57-R < C57-C < ApoE-R < ApoE-C. Comparing the ApoE-C group of mice with the control C57-R group, there were statistically significant increases of a-wave (40.4 ± 3.8 msec vs. 33.3 ± 2.8 msec, P < 0.01) and b-wave (83.0 ± 5.1 msec vs. 63.5 ± 4.7 msec, P < 0.001) implicit times. Another observed change in the electrophysiology of the retina occurred with the oscillatory potentials (OPs). The OPs of the four representative mice in the ERG recordings of Figure 1 showed marked reduction of amplitudes in concordance with the delayed a- and b-wave implicit times. This attenuation of OP amplitude was most severe in mice from the ApoE-C group. Despite the changes in implicit times and OP amplitudes, there were no significant differences in either a- or b-wave amplitudes among the groups of mice (data not shown). 

Histologic Analysis

Figure 3 illustrates photomicrographs of retinal cross sections from representative mice of each of the four different groups. Higher magnification of retinal sections from a control C57BL/6J mouse fed regular mouse chow (C57-R) and apoE-deficient mouse fed the high-cholesterol diet (ApoE-C) is shown in Figure 4 . At this magnification, light microscopic evaluation clearly revealed condensation of nuclear chromatin and perinuclear vacuolization in focal areas in the inner nuclear layer (INL) in the ApoE-C mouse, whereas the retinal cell layers appeared normal in the C57-R mouse (Fig. 4) . The location and number of cells suggest that they are amacrine cells. Whereas these structural abnormalities were frequent in the retinas of ApoE-C group of mice, retinas from the other two groups of experimental mice, C57BL/6J fed the high-cholesterol diet (C57-C) and apoE-deficient mice fed regular mouse chow (ApoE-R), showed little or no structural abnormality (data not shown). 

The morphometric analyses of retinal cell density and layer thickness for the groups of mice are summarized in Figure 5 . Comparison of the C57-R control group of mice with the other three groups (Fig. 5A) showed decreases in INL cell numbers (expressed as a percentage of C57-R control mice values) with the order of highest to lowest number of INL cells being C57-R (100% ± 15%) > C57-C (94% ± 15%) > ApoE-R (88% ± 18%) > ApoE-C (84% ± 10%). The decrease in INL cell numbers for the ApoE-C group of mice was statistically significant (P < 0.05) when compared with the control C57-R group of mice. Decreased cell numbers were also found in the retinal outer nuclear layer (ONL) and showed the same order of the groups of mice with C57-R (100% ± 13%) > C57-C (90% ± 10%) > ApoE-R (83% ± 12%) > ApoE-C (80% ± 9%) as seen in Figure 5C . The decreases for mice of the ApoE-R and ApoE-C groups were statistically significant (P < 0.05 and P < 0.01, respectively) when compared with the control C57-R group of mice. The decreases in cell numbers in the INL and ONL are reflected in reduced layer thickness of the experimental groups of mice with ApoE-C mice displaying the greatest degrees of both INL and ONL thinning (86% ± 16% and 84% ± 7% of C57-R controls, respectively). The reductions of both INL and ONL thickness for the ApoE-C group of animals were statistically significant (P < 0.05 for both) compared with the C57-R mice. 

Immunohistochemical Analysis for CAII and Calretinin

Retinal sections were analyzed by staining for expression of CAII and the neuron-specific calcium-binding protein, calretinin, by immunocytochemistry. Immunoreactivity for CAII was observed in the middle of the INL of control C57-R mice (Fig. 6A) , most likely in Müller cell nuclei. ^{33 34 35} The level of CAII immunoreactivity was diminished in the other three groups of mice, with a near absence in the ApoE-C group (Figs. 6B 6C 6D) . Similar to other measured parameters in this study, the overall order of CAII staining in the groups of mice from the most CAII immunoreactivity to the least was C57-R > C57-C > ApoE-R > ApoE-C. 

Figures 7A 7B 7C 7D show calretinin immunoreactivity on retinal sections from representative animals. In control C57-R mice, immunohistochemical analysis demonstrated the presence of calretinin-positive nuclei in the INL (most likely bipolar cells) and ganglion cell layer, as well as three calretinin-positive layers in the inner plexiform layer indicative of synaptic connections of bipolar and amacrine cells. This pattern of retinal staining with calretinin antibody has been shown in amacrine cells (most likely AII amacrine cells), ^{36 37 38} cells in the ganglion cell layer, ³⁹ and the ribbon synapses of gap junctions between amacrine cells and cone bipolar cell terminals. ⁴⁰ The C57-C and ApoE-R experimental groups of mice had decreased levels of calretinin immunoreactivity, whereas the ApoE-C group of animals had little or no positive calretinin immunoreactivity. 

Transmission Electron Microscopy

TEM evaluation was performed on mouse retinas from three groups of mice, C57-R, ApoE-R, and ApoE-C (retinas from the C57-C group were not available for TEM). As shown in the vitreal portion of the INL (Fig. 8) , neuroretinal cells (most likely amacrine cells) from ApoE-C mice had condensation of nuclear chromatin and perinuclear vacuolization in focal areas. In contrast, C57BL/6J control mice fed regular mouse chow displayed very little of these aberrant nuclear morphologic changes. ApoE-deficient mice fed regular mouse chow (ApoE-R) also displayed these morphologic changes in the INL cells, albeit less extensively than in the ApoE-C group (Fig. 8) . In addition, the BM of apoE-deficient mice fed regular mouse chow or the cholesterol diet (ApoE-R and ApoE-C) appeared to be increased in thickness and to be less organized with discontinuation of the elastic lamina along the membrane (Fig. 8 , arrows). 

Discussion

In this study we examined the effects of high cholesterol and the apolipoprotein E gene in mouse retinas. To accomplish this, we used the apoE-deficient transgenic mouse model system and compared a control group with three experimental groups of mice: C57-R, control C57BL/6J mice fed regular mouse chow; C57-C, C57BL/6J mice fed a fat- and cholesterol-containing diet; ApoE-R, apoE-deficient mice fed regular mouse chow; and ApoE-C, apoE-deficient mice fed a fat- and cholesterol-containing diet. The diet used in this present investigation (Teklad Adjusted Calories Western-type; Harlan Teklad) contains 21% fat by weight and 0.15% cholesterol by weight. In the most affected group of mice, this intake of saturated fat and cholesterol resulted in elevated levels of serum cholesterol to over 1400 mg/dl in cholesterol diet–fed apoE-deficient mice (ApoE-C). The serum cholesterol levels of C57BL/6J fed a high-cholesterol diet (C57-C) and apoE-deficient mice fed regular mouse chow (ApoE-R) were also elevated (359 mg/dl and 409 mg/dl, respectively) compared with the C57-R mice (115 mg/dl). Differences in cholesterol levels between C57-R mice and C57-C mice (P < 0.05) and ApoE-R mice and ApoE-C mice (P < 0.05) were statistically significant, demonstrating that the cholesterol diet was effective in elevating serum cholesterol in these mice. 

Other investigators have used the apoE-deficient transgenic mouse model extensively to investigate the pathogenesis of atherosclerosis and, more recently, neurodegenerative disorders. The structural, cellular and functional analyses of this current investigation revealed a variety of alterations in the retinas of the experimental groups of mice after a 25-week feeding period. There were decreases in cell numbers and cell layer thickness (Fig. 4) in the retinas of apoE-deficient mice, indicating degeneration of retinal cell layers. The changes in electrophysiologic response were diminished OPs and delayed a- and b-wave implicit times (Figs. 1 2) demonstrating retinal neuronal dysfunction. It is believed that the generating source of OPs is cells of the inner retina (e.g., amacrine cells and inner plexiform cells). ^{41 42 43} Therefore, it was not unexpected that the retinas of the experimental animals showed a decrease or loss of CAII (Müller) and calretinin (amacrine) immunoreactivity (Figs. 6 7) . This strongly suggests that, in cholesterol-fed apoE-deficient mice (ApoE-C), cells of postsynaptic retinal neuronal circuits are impaired. In addition, in the apoE-deficient mice, the BM appeared thickened and contained breaks indicative of membrane integrity being compromised. In all the analyses, mice in the ApoE-C group were found to exhibit the most adverse changes in the retina. Therefore, these findings strongly suggest that the combination of a hypercholesterolemic state and apoE deficiency is additive in contributing to the pathologic changes observed in the retinas of this mouse model system. 

Previously, various factors have been shown to be associated with ARMD, including hypercholesterolemia, cardiovascular disease, and apoE gene polymorphism. The Eye Disease Case–Control Study (EDCCS) found a positive association of serum cholesterol levels with the exudative form of ARMD (risk ratio [RR] 4.1, 95% CI 2.3–7.3) in the highest cholesterol group when compared with the lowest cholesterol group. ⁷ More recently, Hyman et al. ⁴⁴ reported that an elevated cholesterol level correlates positively with neovascular ARMD (odds ratio 2.2), whereas there is no association with the non-neovascular form of ARMD. However, other investigators have not found a positive association with cholesterol levels and ARMD. ^{8 45} In a study of the relationship between dietary cholesterol with ARM, subjects from the Beaver Dam Eye Study and Nutritional Factors in Eye Disease Study were examined. It was found that human subjects with intake of cholesterol and saturated fat in the highest quintile had increased odds of 80% and 60%, respectively, for early ARM, compared with the lowest quintile. ²  

In our study, the hypercholesterolemic state in the experimental mice seemed to play a key role in bringing about pathophysiological changes in the retina. Whereas the apoE group of mice with the highest level of serum cholesterol (1451 mg/dl) exhibited the most dramatic changes to retinal structure and function, the other two groups of experimental mice, C57-C and ApoE-R, with intermediate levels of serum cholesterol, displayed retinal changes that were between those seen in control C57-R mice and ApoE-C mice. When the correlation of levels of serum cholesterol to the cell numbers and thickness of the INL and ONL and the implicit times of a- and b-waves was determined, there were statistically significant correlations in the serum cholesterol levels and the appearance of the retinal changes. Whereas the mechanism of hypercholesterolemia-induced retinal degeneration remains to be elucidated, our findings with this mouse model system suggest that dietary cholesterol plays a key role in retinal cellular maintenance and function and support the conclusions of human studies relating cholesterol and saturated fat intake with ARM. ²  

Other laboratories have also used the apoE-deficient mouse to investigate the role of risk factors in retinal disorders. ^{29 46} Fliesler et al. ⁴⁶ found that standard rodent chow–fed, hypercholesterolemic apoE-deficient mice did not have alterations in retinal structure or function compared with age-matched control C67BL/6J mice. The findings in the aforementioned study share similarities with the findings of the present study. Overall, the group of apoE-deficient mice fed regular mouse chow in our study did not have dramatic changes of cellular morphology, with significant decreases of cell numbers observed only in the ONL. The findings of Fliesler et al. also agree with those of the current report showing apoE-deficient mice fed regular mouse chow did not exhibit significant changes in their ERG amplitude responses compared with control C57BL/6J mice. However, we observed some changes in the retina of apoE-deficient mice fed regular mouse chow, including reduced OP amplitudes and decreased CAII and calretinin immunoreactivities. 

A possible explanation for the different observations may be the use of different “standard” diets. Typically, various standard rodent diets contain very low fat content (4.5%–5% fat and 0%–0.02% cholesterol), and most likely one of these types of standard diets was used in the study of Fliesler et al. ⁴⁶ The regular mouse chow used on the apoE-deficient mice in the present study was slightly higher in fat content (11% fat and 0.03% cholesterol). Therefore, the few alterations in retinal morphology and function in the regular mouse chow–fed apoE-deficient mice may be due to the slightly higher fat content (11% vs. 4.5%–5%). This would support the notion that increased dietary fat intake plays a role in retinal impairment. 

Whereas the greatest retinal changes were found in the group of ApoE-C mice that were subjected to both a high-cholesterol diet and apoE-deficiency, it is not clear from the data whether the two factors share the same pathway (i.e., hypercholesterolemia) in eliciting the observed retinal alterations or whether there is an additional contribution by apoE to these adverse changes. The group of ApoE-R mice (not subjected to the cholesterol diet) displayed retinal abnormalities second only to that of the ApoE-C mice. When these ApoE-R mice were compared with control C57-R mice, there were several significant differences between the two groups of animals (Fig. 2B 2b -wave implicit time; Fig. 5C , ONL cell number; Fig. 6 , CAII immunoreactivity; and Fig. 7 , calretinin immunoreactivity), which suggests that apolipoprotein E deficiency itself could have an independent role in predisposing the ApoE-R mice to retinal alterations. However, despite being fed regular mouse chow, ApoE-R mice had elevated levels of serum cholesterol as expected in this type of mouse and therefore, the hypercholesterolemic state of ApoE-R mice may have contributed significantly to the retinal abnormalities. 

It is worth noting that C57-C mice (C57BL/6J mice subjected to a cholesterol diet), although their serum cholesterol levels were nearly comparable to those in ApoE-R mice (359 ± 75 and 409 ± 53 mg/dl, respectively), had retinal changes that were consistently (albeit not significantly) greater than those of ApoE-R mice (Figs. 2 5 6 7) . This observation hints at the possibility that apoE is an additional causative factor for the pathologic changes in the retina. A possible explanation is that although the circulating serum levels of cholesterol were the same in these two types of mice, local retinal tissue levels of cholesterol were significantly different between C57-C and ApoE-R mice due to the absence of apoE in the latter group of animals. Another possibility is that, in a mechanism separate from hypercholesterolemia, the absence of apoE itself, particularly in the retina, results in the adverse retinal changes in the ApoE-R mice. Lipid metabolism in the retina probably involves the local production of heterogeneous lipoprotein particles containing apoE and apoJ. ^{47 48 49} After synthesis and secretion of apoE and apoJ by Müller cells, newly assembled lipoprotein particles are secreted into the vitreous and are rapidly transported into the optic nerve and its terminals in the lateral geniculate and superior colliculus. ⁴⁷ With the ligand activity of apoE for cellular LDL receptors and lipid-binding action of apoJ, it appears that this local mechanism of cholesterol transport and delivery plays an important role in providing retinal neurons with lipids needed for cellular membrane maintenance and remodeling. 

We found that compared with control C57-R mice, the three experimental groups of mice exhibited significant decreases in CAII and calretinin immunoreactivities (Figs. 6 7) , particularly in apoE-deficient mice. This could indicate either that Müller and amacrine cell functions are impaired or these cell types are being lost. Therefore, the resultant aberration in retinal circuitry in these hypercholesterolemic mice could be the consequence of disruption in local cholesterol homeostasis in the retinal cell layers. Determination of local retinal tissue cholesterol levels would help to resolve this question. 

The source of cholesterol for photoreceptors and RPE is lipid metabolism from an extracellular source (i.e., serum lipids) instead of de novo synthesis. This uptake of cholesterol is most likely receptor mediated. RPE has been determined to have significant expression of receptors for native LDL ⁵⁰ and may even involve the local production of apoE by RPE cells. ⁵¹ Again, abnormal lipoprotein metabolism stemming from high dietary cholesterol intake and/or apoE deficiency could lead to a disturbance in the cholesterol balance in these cellular layers of the retina. This could help further explain the changes to the BM of the RPE in the ApoE-R and ApoE-C groups of mice (Fig. 8) . The discontinuous and thickened nature of the basement membranes of these mice share similarities with altered RPE basement membranes associated with choroidal neovascularization in human ARMD. ^{52 53} Several other laboratories have also reported basement membranes with appearance of thickening and breaks in C57BL/6J mice fed a diet containing 1.25% cholesterol, ²⁸ apoE-deficient mice fed a low-fat diet, ²⁹ and cholesterol-fed apoE-deficient mice (SJ Fliesler, personal communication, May 2000). 

It is widely believed that a defective RPE is the underlying cause of human retinal and macular diseases and dystrophies, because the first observed clinical changes in ARMD and ARM seem to occur in the RPE. ¹ Therefore, it is possible that the RPE is also altered in this mouse model, and that, as a consequence of abnormal changes in the RPE, photoreceptors degenerate, leading to impairment of the inner retina circuitry. Although the current investigation has not determined whether there are aberrations in the RPE of this apoE-deficient mouse, future studies may demonstrate a role of the RPE in the retinal changes of this animal model. 

The pathophysiological basis of eye diseases involving the dysfunction and degeneration of the retina such as ARMD is most likely to be multifactorial, with both environmental and genetic risk factors. The apoE-deficient mouse model described in the present study addressed two risk factors associated with ARMD: an environmental factor (dietary fat and cholesterol intake) and a genetic factor (apoE). Based on functional and structural analyses, the apoE-deficient mouse should be a valuable tool in elucidating the underlying mechanism of retinal degeneration. 

 Supported by the Guenther Foundation and the Skirball Program for Molecular Biology.

 Submitted for publication August 15, 2000; revised February 15, 2001; accepted March 19, 2001.

 Commercial relationships policy: E (REW, RWL); N (all others).

 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.

 

Figure 1.

 

Electroretinographic tracings from representative mice from the four groups of animals. Scotopic ERG potentials were recorded for the four groups of mice consisting of control C57BL-6J mice fed regular mouse chow (C57-R), C57BL-6J mice fed a cholesterol-containing diet (C57-C), apoE-deficient mice fed regular mouse chow (ApoE-R), and apoE-deficient mice fed a cholesterol-containing diet (ApoE-C). The dark-adapted ERG responses shown are from representative mice in each group.

Figure 2.

 

Implicit times of a- and b-waves. Average a- and b-wave implicit times are shown for the control (C57-R) and experimental (C57-C, ApoE-R, ApoE-C) groups of mice. Comparison of the ApoE-C group with the control C57-R group showed statistically significant increases of a-wave (40.4 ± 3.8 msec vs. 33.3 ± 2.8 msec, P < 0.01) and b-wave (83.0 ± 5.1 msec vs. 63.5 ± 4.7 msec, P < 0.001) implicit times. Despite the changes of the implicit times and OP amplitudes, there were no significant differences in either a- or b-wave amplitudes between any of the groups of mice (data not shown).

Figure 3.

 

Light microscopy of hematoxylin and eosin–stained paraffin sections of mouse retinas. Photomicrographs of retinal sections of representative mice are shown from each of the four groups (defined in Fig. 1 ). Scale bar, 20 μm. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer.

Figure 4.

 

Light microscopy of plastic-embedded sections of mouse retinas. Photomicrographs of retinal sections of representative mice from C57-R and ApoE-C groups of mice (defined in Fig. 1 ). Resin-embedded blocks of mouse eyes were sectioned at a 1-μm thickness and stained toluidine blue O. Scale bar, 20 μm. Abbreviations defined in Figure 3 .

Figure 5.

 

Quantitation of retinal cell density and layer thickness. (A) Comparison of the C57-R control group with the other three groups showed decreases in INL cell numbers (expressed as a percentage of counts in the C57-R control group), with the order of most to least INL cells being C57-R (100 ± 15%) > C57-C (94 ± 15%) > ApoE-R (88 ± 18%) > ApoE-C (84 ± 10%). (C) Decreases in cell numbers were also found in the retinal ONL and showed the same order: C57-R (100 ± 13%) > C57-C (90 ± 10%) > ApoE-R (83 ± 12%) > ApoE-C (80 ± 9%). Decreases in cell numbers in the INL and ONL are reflected in the reduced layer thicknesses in the groups of mice, with the ApoE-C group displaying the greatest degrees of both (B) INL and (D) ONL thinning (86% ± 16% and 84% ± 7% of the C57-R control group, respectively).

Figure 6.

 

Immunocytochemistry analysis for CAII. Retinas from the C57-R (A), C57-C (B), ApoE-R (C), and ApoE-C (D) groups were reacted with antibody to CAII. The presence of positive immunoreactivity for CAII was clearly evident in the central portion of the INL of the control C57-R mice (probably Müller cell nuclei). The level of CAII immunoreactivity diminished in the other three groups of mice, with a nearly complete absence in the ApoE-C group. Scale bar, 20 μm.

Figure 7.

 

Immunocytochemical analysis for calretinin of the neuron-specific calcium-binding protein calretinin. Retina from C57-R (A), C57-C (B), ApoE-R (C), and ApoE-C (D) were reacted with antibody to calretinin. Most notable was the rows of staining in the IPL, probably indicating the presence of calretinin at the gap junctions between the cells of the INL and the ganglion cells, as well as staining of cell nuclei of ganglion INL cells. Retinas from C57-C, ApoE-R, and ApoE-C mice had decreased levels of calretinin immunoreactivity. Scale bar, 30 μm.

Figure 8.

 

TEM evaluation of eyes of mice showed condensation of nuclear chromatin (arrows) and perinuclear vacuolization in focal areas in the INL, with the ApoE-C group being the most affected (top). In addition, BM of apoE-deficient mice fed regular mouse chow and those fed a cholesterol-containing diet, appeared to be thicker and disorganized, with breaks observed (bottom, arrows). Scale bar, 2 μm.

The authors thank Steven Nusinowitz for helpful comments on the manuscript regarding the ERG analysis and Alex Ljubimov and Don Brown for assistance with the statistical analysis.