Ultrastructural Aging of the RPE–Bruch’s Membrane–Choriocapillaris Complex in the d-Galactose–Treated Mouse

Abstract

purpose. Low-dose d-galactose treatment in mice induces accelerated aging due to advanced glycation endproduct (AGEs) formation. The purpose of this study was to identify ultrastructural aging in the retinal pigment epithelium (RPE)–Bruch’s membrane–choriocapillaris.

methods. Five-month-old C57Bl6 mice were injected daily with d-galactose or control buffer for 8 weeks. Eighteen-month-old mice were also treated with control buffer for 8 weeks. Eyes were prepared for electron microscopy and AGE-specific fluorescence at ex = 370 nm/em = 440 nm and ex = 330 nm/ex = 390 nm.

results. d-Galactose treatment induced AGE-specific fluorescence in lens and RPE/choroid compared to buffer-treated controls. In d-galactose–treated animals, the RPE had dilated and fewer basolateral infoldings. Bruch’s membrane had alterations that included significant thickening, sub-RPE and prominent outer collagenous layer deposits, and choriocapillaris basement membrane duplication/splitting and thickening. The choriocapillaris endothelium displayed fenestration loss.

conclusions. Ultrastructural aging to the RPE–Bruch’s membrane–choriocapillaris developed in mice treated with low-dose d-galactose. These changes could contribute to age-related changes that promote early age-related disease.

Histopathological changes have been identified in aging retinal pigment epithelium (RPE)–Bruch’s membrane–choriocapillaris complex. Basal deposits within Bruch’s membrane are a critical finding. The location and composition of these accumulations separate “normal” aging from age-related macular degeneration (AMD). Initial aging-related deposits appear in the outer collagenous layer. ^{1 2} Basal laminar deposits (BlamD), which occur between the RPE cell and its basement membrane, also develop early during aging, and correlate with AMD only when they thicken and accumulate vesicular structures. ^{3 4} Basal linear deposits (BlinD) are more severe changes that occur in the inner collagenous layer during AMD, and appear chronologically later than BlamD. ^{1 2 3 5 6 7 8 9} The origin, whether from the RPE and/or choriocapillaris, or the inter-relationship of the different deposits, is incompletely understood. A greater understanding of how these deposits form could provide insights into how the RPE–Bruch’s membrane–choriocapillaris complex transitions from aging to age-related disease. 

Advanced glycation endproducts (AGEs) are a heterogeneous group of structures that form between the primary amino group of a protein and a carbohydrate-derived aldehyde group that have been implicated in aging and age-related disease. ¹⁰ In determining how AGEs influence aging and if they contribute to the transition from normal aging to age-related disease, AGEs were found to increase with age in human Bruch’s membrane, to localize to basal deposits and drusen, and to induce an aging phenotype to RPE cells in vitro. ^{11 12 13} Song et al. recently reported that d-galactose treatment, below a dose that induces early diabetic-like retinopathy, induces AGE formation, and accelerates aging in mice. ¹⁴ The purpose of this study was to determine whether d-galactose treatment induced ocular AGE formation and ultrastructural aging to the RPE–Bruch’s membrane–choriocapillaris complex. 

Methods

Mice

All experiments were conducted according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the research was approved by the institutional research boards at UC Davis and Johns Hopkins Medical Institutions. Female C57BL/6 mice (4 months old) were purchased from Charles River (Boston, MA). The mice were fed standard rodent chow and water ad libitum, housed in plastic cages, kept in a 12-hour light–dark cycle, and given a 4-week adaptation period before experiments began. Group 1 consisted of 5-month-old mice (n = 5) that were killed at the start of the project. Mice (5 months old; n = 5) were given daily s.c. injections for 8 weeks of either Phosphate Buffered Saline 0.12 mL (group 2) or d-galactose 50 mg/kg (Sigma, St. Louis, MO) in 0.12 mL (group 3). 18-month-old control mice (n = 5) were given daily s.c. injections for 8 weeks of PBS 0.12 mL (group 4). Mice were killed after the 8-week treatment period ended. The amount of d-galactose that an animal received per day in this model is 1000× less than from the 30–50% d-galactose–rich diets model that simulate diabetic retinopathy. ¹⁵ In the 30% d-galactose model, assuming 170 mg food/gram body weight, a mouse ingests 50 mg/gm body weight (Kern T, personal communication, 2003). 

Tissue preparation

After sacrifice, eyes were enucleated, and one eye was placed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.08 M cacodylate buffer, pH 7.35. The lens and RPE/choroid of the contralateral eye of each animal were dissected and cryopreserved by immersion in isopentane (Fisher-Aldrich Chemical Co., Inc., Milwaukee, WI) chilled with dry ice, and stored at −80°C until used. 

Advanced Glycation Endproduct (AGE) Fluorescence

Detection of fluorescence at ex = 370 nm/em = 440 nm is a valid and commonly used technique to measure AGEs. ^{16 17} Therefore, RPE/choroid and lens samples were hydrolyzed in glass ampules with 1 mL of 12 N HCl under nitrogen for 24 hours at 110°C. After hydrolysis, HCl was removed from the samples by evaporation on a heating block under a stream of nitrogen. Hydrolyzates were reconstituted in distilled water (total volume 200 μL) and transferred to 1 mL glass bottles equipped with Teflon-sealed caps. 

For measuring fluorescence at ex = 370 nm/em = 440 nm, pilot studies determined that the minimum sample volume needed to obtain consistent results was 50 μL Thus, 50 μL aliquots of the reconstituted lens and RPE/choroid hydrolysates were placed in 96-well ELISA plates with opaque sides and read with an ELISA microplate reader (Bio-Tek Flx800; Bio-Tek Instruments, Inc., Winooski, VT). Fluorescence in RPE/choroid was cross-checked at ex = 330 nm/em = 390 nm to reduce possible tissue fluorescence from lipofuscin. Since these filter sets were not available in the Bio-Tek plate reader, these assays were performed with a micro quartz flow cell (volume = 12 μL) in a Hitachi fluorometer equipped with a xenon bulb (Hitachi, Model F-1050, Pleasanton, CA). Absorbance readings were recorded on the chromatograph as μV deflection. Distilled water was used as the blank. Sensitivity to fluorescence was higher than the Bio-Tek system, most likely due to the small volume of the quartz flow cell and the use of the xenon bulb for excitation. Fluorescence data from this setup suggested that lipofuscin was an unlikely contributing factor to background noise. Fluorescence data were normalized to hydroxyproline (hpro) content (fluorescence units/μg hpro) as determined from the Woessner Assay. Statistical significance was determined by ANOVA. 

Woessner Assay

The colorimetric Woessner assay was used to determine hydroxyproline content of the sample hydrolysates (50 μL). Briefly, 25 μL chloramine-T was added for 20 minutes at room temperature to oxidize the imidazole ring, and the reaction was stopped with 25 μL of 3.5 M perchloric acid. Samples were incubated in 25 μL of 20% dimethylaminobenzaldehyde solution (in ethylene glycol/monomethyl ether) at 60°C for 20 minutes. After cooling, samples were read on a plate reader (Bio-Tek Instruments, Inc.) at 560 nm. A standard curve was generated from dilutions of a hydroxyproline stock solution (10 μg/mL; Sigma, Rockford, IL). The absorbance of the standards was linear with respect to the hydroxyproline concentration. 

Electron Microscopy

The central 2 × 2 mm tissue temporal to the optic nerve was used for electron microscopic analysis. Tissue was postfixed with 1% osmium tetroxide in 0.1 M cacodylate buffer and standard dehydration of the specimens was performed. The specimens were embedded in Poly/Bed 812 resin (Polysciences, Inc., Warrington, PA). Sections 1.0 micron were cut and stained with 2% toluidine blue in 2% sodium borate. Ultrathin sections were then cut, stained with uranyl acetate and lead citrate, and examined with a CM-120 Biotwin electron microscope (Philips, Inc., The Netherlands). 

Semiquantitative Grading System

The average BM thickness was determined from the thinnest and thickest measurements of BM, as previously described, by a masked observer. ¹⁸ The Wilcoxon rank-sum test was used to compare the mean average BM thickness in different mouse groups. P < 0.05 was considered statistically significant. ¹⁸ To evaluate changes to Bruch’s membrane, micrographs were evaluated by a masked observer. A minimum of 50 sections was evaluated per tissue sample. A severity score of 0 to 12 points was assessed using the protocol established by Cousins et al. ¹⁹ The categories of Bruch’s membrane abnormalities included (from 0 to 3 points severity scale): thickness, continuity, and content of deposit between RPE cell membrane and basement membrane, deposits external to the RPE basement membrane, and choriocapillaris abnormalities. Groups were compared by determining the mean and SD, and ANOVA was used to determine significant differences. The frequency of BlamD-like deposits was also assessed using the Cousins protocol designation of ‘Any BLD’ as the presence of a discrete focal nodule between the RPE cell membrane and basement membrane in at least one micrograph. ‘Moderate BLD’ was defined as the presence in at least three micrographs of continuous BLD under two or more cells, deposit thickness ≥ 20% of the RPE cross-sectional thickness, or the presence of banded structures within the BLD. Additional frequency assessment used the following criteria: ‘Any outer collagenous layer (OCL) deposit’ was defined as any deposit in the OCL in at least one micrograph with a severity rating of 1, and ‘Moderate OCL deposit’ as any OCL deposit in at least three micrographs of severity scale ≥ 2. ‘Any choriocapillaris (CC) change’ was defined as a severity rating of at least 1 in at least one micrograph, and ‘Moderate CC change’ as a severity rating ≥ 2 in at least three micrographs. Differences in relative frequency were assessed by Fisher’s Exact Test. 

Results

Formation of AGEs and Lipid Peroxidation Products in Ocular Tissue

AGE-specific fluorescence was identified in lens and RPE/choroid samples from d-galactose–treated animals. Figure 1 shows a marked increase in AGE-related fluorescence at ex = 370 nm/em = 440 nm in d-galactose–treated RPE/choroid (P < 0.0001) and lens (P = 0.0034), and for RPE/choroid at ex = 330 nm/em = 390 nm (P = 0.032) compared to PBS controls. 

Bruch’s Membrane Thickening with d-Galactose Treatment

Table 1 shows that Bruch’s membrane thickness increased significantly after d-galactose treatment. The average thickness was 0.44 ± 0.11 μm in 5-month-old C57BL6 mice before treatment, and unchanged at 0.49 ± 0.1 μm in 7-month-old mice that received PBS injections for 8 weeks. The mean Bruch’s membrane thickness of d-galactose–treated mice increased to 0.76 ± 0.19 μm (P = 0.0006, group 2 vs. 3). For comparison, the Bruch’s membrane thickness of 20-month-old mice that received PBS treatment was 0.95 ± 0.28 μm (P = 0.041, group 3 vs. 4). 

Ultrastructural Changes to RPE–Bruch’s Membrane–Choriocapillaris with d-Galactose Treatment

The RPE in all sections of all mice treated with d-galactose had fewer, but enlarged basolateral infoldings compared to PBS treated controls (Figs. 2A 2B) . The RPE basement membrane was indistinct and irregularly thickened adjacent to the infoldings. Focal sub-RPE cell granular deposits were observed in d-galactose–treated mice (Fig. 2C) . No sub-RPE deposits were seen in PBS-treated mice. Focal areas of choriocapillaris basement membrane reduplication or splitting, and thickening were also seen in all d-galactose–treated eyes (Fig. 3) . Focal deposits of proteinaceous-like material were also seen in the outer collagenous layer that were contiguous with the choriocapillaris basement membrane in all eyes treated with d-galactose, and one deposit seen in one eye treated with PBS as shown in Figure 4A . These changes were associated with a loss of choriocapillary endothelial cell fenestrations. Membranous structures were seen within larger outer collagenous layer deposits (Fig. 4B) . 

For comparison, 18-month-old mice treated with PBS injections also showed ultrastructural changes to the RPE–Bruch’s membrane–choriocapillaris complex. The RPE showed a more exaggerated loss of basolateral infoldings with sub-RPE proteinaceous deposits within the spaces than did d-galactose–treated mice. In some of the sub-RPE deposits, 70-nm banded material was seen (Fig. 5) . Prominent outer collagenous layer deposits were observed. 

The severity and frequency of ultrastructural changes to Bruch’s membrane were assessed by the semiquantitative grading scale established by Cousins et al., ¹⁹ as summarized in Table 2 . The pretreatment and PBS-treated controls showed little ultrastructural changes using this criteria while d-galactose–treated eyes showed a significantly higher severity score (P < 0.001). d-Galactose–treated eyes had a similar score as old controls (P = 0.27). The frequency of ultrastructural changes to Bruch’s membrane was significantly different between PBS controls and d-galactose–treated eyes. d-Galactose–treated eyes had deposits throughout Bruch’s membrane, but moderate changes were confined for the most part, to the outer collagenous and choriocapillaris region. 

Discussion

Song et al. reported that d-galactose s.c. injected at a dose below that which simulates diabetes mellitus, induces premature aging with increased serum AGE content, memory latency time and error rate, skin hydroxyproline content, and decreases motor activity, lymphocyte mitogenesis, interleukin-2 production, and superoxide dismutase enzyme activity. ¹⁴ AGE-related fluorescence was measured in both lens and RPE/choroid and premature age-related ultrastructural changes to the RPE–Bruch’s–choriocapillaris complex after 8 weeks of d-galactose treatment were identified. This model was distinct from the 30–50% d-galactose–rich diet model which simulates diabetic retinopathy after 21 to 26 months, in that animals received 1000× less d-galactose per day. ¹⁵  

The eye is susceptible to AGE formation due to its high oxidative/photo-oxidative stress environment and high ascorbic acid concentration. The conditions for AGE formation may be enhanced by d-galactose treatment. While d-galactose could serve as a substrate for AGE formation, it is also known that d-galactose induces redox disequilibrium by promoting oxidation of ascorbate to dehydroascorbate, which along with its fragmentation products, become precursors for AGEs. ^{20 21} AGE-fluorescence data suggested that less AGE forms in the RPE/choroid than the lens. The relatively reduced AGE formation between RPE/choroid and lens could be explained by differences in oxidative and photo-oxidative exposure, oxidative defense systems, and ascorbate metabolism. AGEs have an established role in matrix expansion, such as in diabetic nephropathy and atherosclerosis, by increased expression of growth factors (TGF-b, IGF-1, PDGF-b, VEGF, CTGF), matrix proteins (collagen I, IV, laminin, fibronectin), and reduced susceptibility to matrix proteases. ^{22 23 24 25 26 27} Some of these AGE-related alterations may participate in the development of accelerated aging in the RPE–Bruch’s membrane–choriocapillaris complex. 

Mice that received d-galactose had RPE cells with fewer, but enlarged basolateral infoldings, irregular RPE basement membrane thickening, and focal sub-RPE deposits which was similar to human RPE aging reported by Feeney-Burns et al. ²⁸ These changes suggest RPE cell injury just as, for example, renal tubular epithelial cells develop dilated basolateral infoldings during acute tubular necrosis and acute interstitial nephritis. ^{29 30} The RPE basement membrane thickening and focal granular sub-RPE deposits were localized between widely spaced basolateral infoldings, which suggested that the injured RPE cell participates in the deposit formation. 

Changes to outer Bruch’s membrane were a common finding in d-galactose–treated animals. The choriocapillaris basement membrane thickening and reduplication or splitting was similar to the changes that Marshall et al. identified in aging human eyes. ³¹ The most prominent deposits in d-galactose–treated mice, however, developed in the outer collagenous layer, accounted for much of Bruch’s membrane thickening, and appeared to be a normal aging development that was accelerated by d-galactose treatment. The location of these deposits away from intercapillary regions argued against a sectioning artifact. Killingsworth reported that outer collagenous layer deposits accounted for the age-related increase in human Bruch’s membrane thickness ¹ whereas van der Schaft et al. found that outer collagenous layer deposits form as early as 19 years of age, and that 74% of eyes had outer collagenous layer deposits compared to 17% with inner collagenous layer deposits in a study of elderly non-AMD maculas. ² These studies indicate that outer collagenous layer deposition is an aging process that is distinct from inner collagenous layer deposition, or basal linear deposits (BlinD), which are specific for AMD. Further study is necessary to determine whether these aging-related deposits contribute to pathologic BlinD formation. 

Choriocapillaris endothelial cell fenestration loss was observed adjacent to the largest outer collagenous layer deposits. Endothelial cell fenestration loss is a sign of cytotoxic injury, and can be induced by oxidative free radicals in liver sinusoidal endothelium. ^{32 33 34} Gottsch et al. reported choriocapillaris endothelial fenestration loss due to oxidative stress after griseofulvin treatment. ³⁵ Since AGEs induce age-related changes, in part from reactive oxidative species damage, ^{36 37} it is possible that an oxidative stress-related mechanism induces injury to the choriocapillaris endothelium. Fenestration loss was seen adjacent to outer collagenous layer deposits and not with isolated choriocapillaris basement membrane alterations. This observation suggests that basement membrane changes are early events, while the largest outer collagenous layer deposits are later changes that occur after ultrastructural evidence of endothelial cell injury. Observation by Fratto et al. of choriocapillaris endothelial cell fenestration loss in 17 eyes with AMD, but not age-matched controls, supports this theory (Fratto M, et al. IOVS 2002;43:ARVO E-Abstract 1304). 

Aging to the RPE–Bruch’s membrane–choriocapillaris has been identified in several mouse models. Most investigators have reported changes only to the RPE and inner Bruch’s membrane. For example, the RPE cell and basement membrane, but not the choriocapillaris, developed ultrastructural abnormalities in several hyperlipidemia models. ^{18 38 39 40} Rakoczy et al. showed RPE alterations, basal laminar and intraBruch’s membrane deposits in a transgenic mouse overexpressing mutated cathepsin D. ⁴¹  

Other models show changes to both the RPE and choriocapillaris. Majji et al. found age-related disruption of RPE basolateral infoldings, basal laminar deposits, choriocapillaris atrophy, and intraBruch’s membrane choroidal neovascularization in the senescence-accelerated mouse, a strain in which the genetic factors responsible for aging are unknown. ⁴² Cousins et al. found BlamD and ultrastructural damage to both the RPE and choriocapillaris endothelium with estrogen loss or high fat diet and blue light exposure in elderly mice. ^{19 43} The ultrastructural changes to the RPE and choriocapillaris after d-galactose treatment indicate that both cell types are altered. However, the predilection for changes to the outer Bruch’s membrane by severity and frequency analysis suggested that d-galactose alters the choriocapillaris endothelium more than the RPE. Recently, Ambati et al. showed that elderly Ccl-2– and Ccr-2–deficient mice developed AMD-like changes such as drusen deposits, photoreceptor atrophy, and spontaneous choroidal neovascularization. ⁴⁴ In this model, AGEs were present in the RPE/choroid of knockout, but not wild-type controls. It is possible that AGEs could participate in disease development since AGEs trigger inflammatory cascades that are involved in a number of age-related diseases, including atherosclerosis and Alzheimer’s disease. ⁴⁵  

Aging is a complex multifactorial process that produces ultrastructural alterations to the RPE–Bruch’s membrane–choriocapillaris complex. Since the d-galactose aging model is created by s.c. injections over a short duration, a potential benefit is its applicability with other models, which would allow a systematic study of potential environmental and genetic variables that could initiate and propagate aging of the RPE–Bruch’s membrane–choriocapillaris complex. 

 Supported by Grants RO1 EY14005 (JTH), RPB Lew Wasserman Award (JTH), and RO1 EY06473 (LMH), Michael Panitch Macular Degeneration Research Fund, and a gift from Aleda Wright.

 Submitted for publication December 11, 2003; revised January 28, 2004; accepted February 6, 2004.

 Disclosure: H. Ida, None; K. Ishibashi, None; K. Reiser, None; L.M. Hjelmeland, None; J.T. Handa, None

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

 

Figure 1.

 

AGE-specific fluorescence in ocular samples. Fluorescence (fluorescence units/μg hydroxyproline) at ex = 370/em = 440 nm of acid hydrolysates of controls and d-galactose–treated animals. (A) RPE/choroid. P < 0.0001 for pretreatment vs. d-galactose, and PBS vs. d-galactose; (B) Lens. P values for *pretreatment vs d-galactose, **PBS vs d-galactose; (C) RPE/choroid fluorescence at ex = 330 nm/em = 390 nm. P = 0.032 for pretreatment vs. d-galactose and PBS vs. d-galactose.

Table 1.

 

Thickness of Bruch’s Membrane

Group	Age (mos)	Minimum Thickness (μm ± SD)	Maximum Thickness (μm ± SD)	Average Thickness (μm ± SD)
1 Pretreatment	5	0.34 ± 0.094	0.54 ± 0.13	0.44 ± 0.11
2 PBS control	7	0.36 ± 0.093	0.62 ± 0.15	0.49 ± 0.10
3 D-Galactose	7	0.50 ± 0.11	1.02 ± 0.34	0.76 ± 0.19
4 Old PBS control	20	0.62 ± 0.24	1.28 ± 0.37	0.95 ± 0.28

  P-values; Group 1 vs. 2: P = 0.309; Group 2 vs. 3: P = 0.0006*; Group 2 vs. 4: P = 0.0001*; Group 3 vs. 4: P = 0.041*.

 * Statistically significant.

Figure 2.

 

(A) Electron micrograph of the RPE–Bruch’s membrane–choriocapillaris complex after 8-week PBS control treatment. Note the well-developed basolateral infoldings, and nonthickened, organized pentalaminar Bruch’s membrane. RPE-BM, RPE basement membrane; ICL, inner collagenous layer; MEL, middle elastic layer; OCL, outer collagenous layer; CC-BM, choriocapillaris basement membrane. The choriocapillaris (mL) endothelium has fenestrations. The slight thickening at the left side of the image is an intercapillary pillar. (B) RPE–Bruch’s membrane after 8-week d-galactose treatment. The RPE basolateral infoldings are enlarged and less densely spaced. Bruch’s membrane shows disorganization. The asterisk shows an early OCL granular deposit. (C) Inset shows a higher power magnification of granular sub-RPE deposits that appear continuous with the RPE basement membrane.

Figure 3.

 

Choriocapillaris basement membrane changes after d-galactose treatment. Arrows point to areas of choriocapillaris basement membrane reduplication/splitting.

Figure 4.

 

(A) Granular outer collagenous layer deposit after d-galactose treatment. Inset shows a higher magnification view illustrating the choriocapillaris fenestrations (arrows) overlying nonthickened Bruch’s membrane, but fenestration loss adjacent to the outer collagenous layer deposit (asterisks). (B) Outer collagenous layer deposits with membranous debris and loss of choriocapillaris fenestrations after d-galactose treatment. Arrows point to vesicular debris.

Figure 5.

 

RPE–Bruch’s membrane–choriocapillaris complex in a 20-month-old mouse after 8-week PBS treatment. The RPE has dilated and fewer basolateral infoldings compared to young control eyes. The asterisk highlights a sub-RPE deposit that contains a 70-nm banded structure. A granular outer collagenous layer deposit is seen on the left side of the image.

Table 2.

 

Severity and Frequency of Ultrastructural Bruch’s Membrane Changes

Group	Severity Score Mean (± SD)	‘Any’ Sub-RPE Deposit (Eyes)	‘Mod’ Sub-RPE Deposit (Eyes)	‘Any’ OCL Deposit (Eyes)	‘Mod’ OCL Deposit (Eyes)	‘Any’ CC Change (Eyes)	‘Mod’ CC Change (Eyes)
1 Pretreatment	0 (± 0)	0/5	0/5	0/5	0/5	0/5	0/5
2 PBS control	0.1 (± 0.31)	0/5	0/5	0/5	0/5	0/5	0/5
3 D-Galactose	6.5 (± 1.65)	1/5	0/5	4/5	3/5	5/5	5/5
4 Old PBS control	7.1 (± 1.73)	4/5	2/5	4/5	4/5	5/5	5/5
P-value	ANOVA	Fisher’s	Fisher’s	Fisher’s	Fisher’s	Fisher’s	Fisher’s
	Group 1 vs. 2: P = 0.854	Exact test 0.02	Exact test 0.21	Exact test 0.0021	Exact test 0.013	Exact test 3.25 × 10−5	Exact test 3.25 × 10−5
	Group 2 vs. 3: P < 0.0001*
	Group 2 vs. 4: P < 0.0001*
	Group 3 vs. 4: P = 0.273

*  Statistically significant.

The authors thank Shigeru Honda and Noritake Miyamura for their technical help, Li Ming Dong and Michele Melia for their statistical analysis, and Timothy Kern for helpful insights.