Amyloid-β Increases Capillary Bed Density in the Adult Zebrafish Retina

Abstract

Purpose.: Amyloid-beta (Aβ) is an endogenous peptide that becomes dysregulated in AMD and Alzheimer disease. Both of these disorders are marked by extracellular deposits that contain Aβ, highly branched capillary networks, and neurodegeneration. Although Aβ has been implicated in AMD and Alzheimer pathology for decades, its nonpathological function has remained unclear. We recently showed that high levels of monomeric Aβ induce blood vessel branching in embryonic zebrafish brain, and here we report that a similar mechanism may contribute to aberrant blood vessel branching in the retina of adult zebrafish.

Methods.: Transgenic zebrafish expressing enhanced green fluorescence protein (EGFP) in their endothelial cells were sedated and small intraocular injections of PBS were made into one eye and either Aβ or γ-secretase inhibitor were injected into their opposite eye. A week later, the eyes were enucleated and high resolution maps of the retina vasculature were created using confocal microscopy. Comparisons were made between the treatment groups using the general linear model ANOVA.

Results.: We found that Aβ significantly affects capillary blood vessels in the retina. Small volumes of Aβ injected into the eyes of adult zebrafish induced the formation of significantly more endothelial tip cells and capillary bridges—some with loops—near the circumferential vein. These effects were dose-dependent and increased capillary bed density, though there was no effect on larger arterial vessels.

Conclusions.: This study reveals a previously unknown role for Aβ in regulating capillary bed density, providing new insight into the normal biological function. Aβ will help in the development of therapeutic interventions for AMD and Alzheimer disease.

Introduction

Amyloid-beta (Aβ) is produced by the proteolytic cleavage of amyloid precursor protein (APP) through a two-step process involving β- and γ-secretases. Zebrafish have all of the necessary proteins to produce Aβ, including APP and β- and γ-secretases, and APP knockdown effects in zebrafish can be rescued by human APP mRNA. ^1,2 Aβ production from APP proceeds by the same two-step process in neurons and RPE. Initially, a portion of the extracellular region is shed by a β-secretase metalloproteinase (BACE1/2) to produce a C99 intermediate, which is then cleaved by γ-secretase to release Aβ from the cell's plasma membrane. ³ The γ-secretase complex is also critical to the processing of other substrates including cadherins and Notch. Notably, γ-secretase inhibitors induce angiogenesis through their disruption of Notch signaling. ⁴ We previously proposed that the development of aberrant blood vessels that occurs between Alzheimer disease (AD) plaques involves the disruption of Notch signaling by high levels of Aβ. ⁵ That hypothesis has been supported by recent studies showing that excessive APP leads to cerebral hypervascularization in mice and young zebrafish. ^6,7  

AD is the leading cause of dementia in the elderly, affecting more than 5.4 million people in the United States. ⁸ A definitive diagnosis of AD can only be made after postmortem brain pathology shows the presence of neuritic plaques that contain Aβ and neurofibrillary tangles. ^9,10 Transgenic mice that overexpress human APP form plaques and display cognitive impairment. ¹¹ High cholesterol is a major risk factor for atherosclerosis, which in turn increases AD risk and drusen accumulation. People who carry an ε4 allele of the cholesterol chaperone apolipoprotein-E (ApoE-ε4) have a higher lifetime risk of AD with earlier onset, though ApoE-ε4 carriers have a reduced lifetime risk for AMD. ^12,13  

AMD primarily affects adults over the age of 50, resulting in loss of vision in the center of the visual field. The leading cause of irreversible blindness within geriatric populations in the developed world, ¹⁴ AMD is classified as either geographic atrophy (dry) AMD or exudative (wet) AMD. In dry AMD, numerous large drusen form in Bruch's membrane; these extracellular deposits consist of lipids, apolipoprotein, and small amounts of Aβ. ¹⁵ Drusen normally accumulate with age and most people over the age of 40 have small equatorial drusen deposits that accumulate with age. ¹⁶ However, the appearance of large numerous drusen is a common early sign of dry AMD. Increased secretion of Aβ by the RPE leads to elevated levels in the vitreous and its accumulation within subretinal drusen deposits, especially in AMD patients. ^17–20 In wet AMD, neovascular growth of blood vessels in the choroid leads to leakage of blood and proteins, which results in rapid degeneration of overlying photoreceptors. ²¹ Age-dependent risk, aberrant angiogenesis, and extracellular deposits of Aβ occur in AMD and AD, suggesting these disorders may share common mechanisms. ²²  

Although Aβ has been implicated in AD and AMD for decades, surprisingly little is known about its physiological/nonpathological function. We recently showed that high levels of Aβ can induce aberrant blood vessel branching in the brains of developing zebrafish, ⁷ and here we have investigated if a similar mechanism may contribute to abnormal blood vessel branching in the retina of adult zebrafish. 

Materials and Methods

Reagents

Human Aβ1-42 was purchased from BioMer Technology (Pleasanton, CA). The product was received as a lyophilized powder that was resuspended by first adding water, and titrating with 1N NaOH until the pH was neutralized and the suspension dissolved completely (i.e., the solution became clear). The solution was then mixed with ×10 PBS, diluted to a final volume of 1 mg/mL (×1 PBS final) and stored as frozen aliquots. Gamma secretase inhibitor (GSI IX/DAPT/N-[N-(3,5-Difluorophenacetyl-L-alanyl)]-S-phenylglycinet-butyl ester in solution) was purchased from EMD Biosciences (San Diego, CA); diluted to 5 mg/mL with DMSO; and stored as aliquots at −20°C. All other reagents were from VWR unless otherwise noted. 

Zebrafish Care and Maintenance

Tg (kdr:EGPF)s843 transgenic zebrafish, expressing EGFP in vascular endothelial cells were obtained from the Zebrafish International Resource Center (ZIRC), housed at 28.5°C, and maintained under a 10-hour dark/14-hour light cycle. ²³ All husbandry and experiments were approved by and conducted in accordance with guidelines established by the Institutional Animal Care and Use Committee at Western University of Health Sciences; and they adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Eye Volume Measurements

Zebrafish eyes were photographed using a camera mounted to a stereo microscope and analyzed using a Java-based image processing program (ImageJ, National Institutes of Health [NIH], Bethesda, MD) to determine vitreal volume, which was used to determine the final molarity of treatments. The depth of the fish eyes were measured after enucleation. Afterward, the vitreal volume was calculated using the formula (4/3 · π · radius_{eye_width} · radius_{eye_height} · depth_eye) − (4/3 · π · radius_lens ³). Eyes were measured from a series of ages ranging from 3 to 20 months. The horizontal eye diameter ranged from 1.93 mm ± 0.07 at 3 months to 2.46 mm ± 0.13 mm at 20 months (R ² = 0.69; P = 3.7e-6); the vertical eye diameter ranged from 1.85 mm ± 0.09 at 3 months to 2.41 mm ± 0.26 mm at 20 months (R ² = 0.56; P = 1.7e-3). 

Injections

Fish were sedated using 0.016% tricaine methanesulfonate (Tricaine; VWR, Radnor, PA) and placed under a dissecting microscope for injections. A tricaine methanesulfonate-containing (Tricaine; VWR) water drip system was used to respire/hydrate the zebrafish during the procedure. Injections were made using a nanoliter injection (Nanoject II; Drummond Scientific Company, Broomall, PA) and pulled glass capillary pipettes. Vitreous volumes were used to calculate injection volumes so that equal concentrations of the drug were administered. Intraocular injections ranged from 70 to 700 nL. An equal volume of PBS was injected into the opposite eye as a control. All fish had PBS injected into the right eye and the left eye was injected with one of the following: 225 nM Aβ, 2.25 μM Aβ, or 1 μM γ-secretase inhibitor (GSI). A 0.5% (v/v) phenol red solution was added so the injections could be seen entering the eye. Each injection was made into the aqueous humor via the dorsalmost aspect of the cornea. Fish were allowed to recuperate prior to being returned to their housing tank. Injection sites healed within 3 days, leaving no observable visual or physiological defects. 

Retina Flatmount

At 7 days postinjection (dpi), fish were euthanized with a lethal dose of tricaine methanesulfonate (Tricaine 250 mg/L; VWR). Their eyes were enucleated and fixed in 4% paraformaldehyde at 4°C overnight. The eyes were then rehydrated in PBS, and the lenses removed to leave an eyecup that was flatmounted on a slide using a mounting medium (VECTASHIELD; Vector Laboratories, Inc., Burlingame, CA) with propidium iodide. 

Confocal Analysis

Retina imaging was done with a 3-color confocal microscope (Nikon A1; Nikon Instruments, Inc., Melville, NY) using a ×20 objective. All vascular structures within the retina could be captured within a 5-μm image stack. Image stacks were projected into a single image for each region of the retina. Imaging of the entire retina required confocal projections from a 7 × 7 grid that were pieced together using a graphics editing program (Adobe Photoshop; Adobe Systems, Inc., Mountain View, CA) and a commercial presentation program (Microsoft Office PowerPoint; Microsoft Corp., Redmond, WA) to generate whole retina maps of the vasculature. 

Retinal Anatomy and Statistical Analysis

In the zebrafish eye, the blood supply enters the retina via the central retinal artery at the optic nerve head and splits into six or seven radiating blood vessels that project outward, and each of these arterial vessels subdivide a farther three to five more times to form the capillary plexuses. ²⁴ The last of these vessel branches are circumferential vein capillaries (CVC) of approximately 12 to 20 μm in diameter. All of the capillaries drain into the circumferential vein (CV), which surrounds the lens. On the ventral surface of the eye, there is a region with abundant branching that appears only once in each eye and serves as a reliable landmark. The branching/radiating vessels immediately adjacent to this area were analyzed. These regions were traced from the optic nerve to the CV and the number of blood vessel tips, branches, bridges, and loops were counted. Tips sometimes joined with another emerging tip, forming a bridge or loop between CVCs. Total tip and bridge counts were made by three independent, blinded observers and averaged for each retina. Total CVCs connecting to each CV were also tabulated. The number of tips and small bridges/loops were then compared between the treated eyes and the PBS control eyes from the same fish using ANOVA, applying the general linear model with Bonferroni correction in statistical analysis software (SPSS, version 13; IBM Corp., Armonk, NY). The tips/bridges per CVC were similarly compared. 

Results

Eye Volume Measurement and Injection

Zebrafish continue to grow throughout life, so we measured the vitreal volume for adult zebrafish from 3 to 20 months of age. Variances for vitreal volume were as low as 4.3% among age groups and did not exceed 8.9%. Regression analysis using statistical analysis software (IBM Corp.) showed that as fish aged, the vitreous volume increased significantly (R ² = 0.87; P = 2.5e-14). The average vitreous volume at 3 months was 2.72 ± 0.21 μL, whereas by 20 months, the vitreous had increased to 5.14 ± 0.27 μL (Fig. 1). Vitreous volumes were used to calculate injection volumes so that equal concentrations of the drug were administered. Intraocular injections ranged from 70 to 700 nL. An equal volume of PBS was injected into the opposite eye as a control. Visual acuity was assessed using the optokinetic responses before and after a series of injections to verify that the injection procedure did not adversely affect the visual abilities of the fish. The fish did not have any significant decrease in visual acuity as a result of the intraocular injection. 

Figure 1

 

Eye volume and injection procedure. (A) Eyes were measured for their vertical and horizontal diameters (indicated with red arrows) using a Java-based image processing program (NIH). Inset: Eyes were enucleated to obtain eye depth (red arrows). (B) Injections were made through the dorsal cornea (yellow arrow). Note the needle passing into the corneal space in the top right of the image. (C) Vitreal volume increase with age (R ² = 0.87; P = 2.47e-14; n = 31).

Blood Vessel Characterization in the Zebrafish Retina

To explore the effects of Aβ, we employed a transgenic zebrafish model that expresses GFP in vascular endothelial cells, so that we could analyze complete maps of the retinal vasculature in adult zebrafish eyes. Six or seven arteries radiate outward from the optic artery near the optic nerve (at the center of each image), toward the CV (Fig. 2). Each of these major branches divides an additional three to five more times before connecting with the CV as CVCs. ^24,25 Regions nearest the CV (the capillary area and collecting vein region ²⁶ ) showed the most angiogenic activity in the form of endothelial tip cells and bridge formation on CVCs (Fig. 3). We observed a heavily branched area in adult zebrafish retina that corresponds to the ventral region, which was used as a consistent landmark (Fig. 2, yellow boxes) as microvascular branching decreased with distance from this ventral branched region. Tip cells projected laterally between major vessel/CVC branches and did not appear to project up or down from the z-axis of the retina (see Supplementary Material and Supplementary Video 1). 

Figure 2

 

Flatmounted 6-month-old Tg (kdr:EGPF)s843 retinas showing the retinal vasculature. Each image is comprised of ∼49 confocal image projections that were stitched together. Each retina is representative of the four injections: PBS; 225 nM Aβ; 2.25 μM Aβ; or 1 μM γ-secretase inhibitor (GSI). Red boxes indicate areas of interest—that is, the areas adjacent to the landmark area of high blood vessel density (yellow boxes). White boxes are the regions shown in Figure 3. Scale bar = 500 μm.

Figure 3

 

Microvascular branching near the CV junction. Images are select detailed sections shown as the white boxed regions in the retinas from Figure 2. Red arrows indicate bridges where endothelial tips (yellow arrows) fused. The creation of a bridge usually creates a loop structure or even loops within loops. Yellow arrows point out the endothelial tips extending from CVCs. Note the increased vasculature in the form of tips and bridges/loops in the Aβ and GSI-treated retinas. Scale bar = 25 μm.

Monomeric Aβ Induces Angiogenesis in the Adult Zebrafish Retina

Aβ and γ-secretase effects on capillary bed density were investigated by microinjecting the eyes of adult zebrafish with 225 nM Aβ; 2.25 μM Aβ; 1 μM γ-secretase inhibitor (GSI); or PBS as a contralateral control. Injection sites healed within 3 days, leaving no observable visual or physiological defects, and the fish were allowed to recover for 7 days. At 7 dpi, fish were sacrificed and the eyes were enucleated, fixed, and imaged using confocal microscopy. Overlapping confocal image stacks were pieced together to generate maps of the entire retinal vasculature in each eye (Fig. 2). Eyes injected with 2.25 μM Aβ or 1 μM GSI showed significantly more endothelial tip cells and capillary branches than the contralateral (control) eye of the same fish (Figs. 3, 4). The number of endothelial tip cells, along with the number of bridges (tips extending from one radiating blood vessel to an adjacent vessel) were counted in all eyes and compared using ANOVA and the general linear model. Aβ increased the number of branches in a dose-dependent manner (Fig. 4). The highest dose, 2.25 μM Aβ, nearly doubled the number of branches per retina: 71 branches compared with 38 (P = 4.5e-3). A similar increase was also observed when we assessed the number of branches and tip cells per CVC using the same statistical analysis: 2.2 branches or tips per CVC compared with 1.3 in PBS-treated retinas (P = 0.017). GSI also increased the number of tip cells and branches relative to CVC number with an average value of 4.3 branches or tips per CVC (P = 2.4e-5). These findings demonstrate that high physiological concentrations of Aβ induce angiogenic remodeling in the adult zebrafish retina. 

Figure 4

 

Aβ-induced retinal neovascularization. (A) The estimated marginal means of tips and bridges observed in 2.25 μM Aβ and 1 μM GSI-treated retinas were significantly increased compared with PBS controls using univariate analysis with Bonferroni correction. **2.25 μM Aβ, P = 4.5e-3; 1 μM GSI, P = 6.5e-4. (B) The estimated marginal means of tips and bridges per CVC ratio is significantly higher in both 2.25 μM Aβ and 1 μM GSI-treated retinas compared to their PBS controls. A dashed line indicates the average value of the PBS injection for comparison across the other treatments. Error bars represent the standard error of the mean (n = 16 for PBS; n = 6 for 225 nM Aβ; and n = 5 for 2.25 μM Aβ and 1 μM GSI). *P = 0.017; **P = 2.4e-5.

Discussion

Aβ and GSI had their strongest effects on the CV capillaries, with little or no effect on larger arterial branches. Aβ increased the number of endothelial tip cells (Figs. 3, 4), and increased the number of vessel branches starting approximately two-thirds distal to the optic nerve region, with most growth occurring among vessels draining into the CV. Our initial experiments showed some branching beginning at 3 days postinjection, but we were able to reliably find changes 7 days postinjection. These findings are consistent with previous work showing that high levels of Aβ induce blood vessel branching in the brains of young zebrafish. ⁷ Results presented here extend that work and demonstrate an angiogenic effect of Aβ that impacts capillary bed density in adult animals, after development is complete. It should be noted that in contrast to mammals, including humans, the zebrafish have a proliferating ciliary marginal zone that occurs at the boundary between the ciliary epithelium and neural retina. The ciliary marginal zone region is a circumferential, continuous annulus around the retina periphery where progenitor cells are populated supporting a growing and regenerating retina. ²⁷ The angiogenic effect seen after Aβ treatment may be partially facilitated by the ciliary marginal zone. 

While the Aβ effects were not as robust as GSI or hypoxia-induced neovascularization, high doses of Aβ did induce tip cell formation and branching in capillary beds nearest the CV. By contrast, hypoxia has been shown to induce branching of arterial blood vessels in the retina of adult zebrafish. ²⁵ Therefore, capillary bed density may be especially sensitive to high levels of Aβ, which is consistent with AD brain pathology wherein highly dense, aberrant capillary beds form between the plaques in affected brain regions. ²⁸  

Aβ is produced throughout life and tissues that do not clear it efficiently have higher levels of the peptide, which may act through a negative feedback loop to limit further processing of the C99 intermediate (of APP) by γ-secretase. This feedback inhibition could reduce γ-secretase processing of other substrates, including Notch—a major regulator of blood vessel stability. ⁵ As a physiological γ-secretase inhibitor, Aβ may promote VEGF signaling and disrupt neovascular perfusion. ²⁹ These effects could be differentially impacted by ApoE alleles as this cholesterol chaperone is produced by pericytes that surround capillary endothelial cells and may affect the integrity of blood capillaries in the brain. ³⁰  

Anti-Aβ therapy has been shown to protect RPE and prevent vision loss in an AMD mouse model ^22,31 and to improve cognition in mouse models for AD. ⁹ However, recent clinical trials with these biologicals, as possible treatments for AD, have been disappointing, ^32,33 though anti-Aβ therapies have not been tried in AMD patients. While dry AMD has yet to find a viable treatment option, wet AMD patients have been successfully treated with anti-VEGF therapeutics, which has become standard care for those patients. ³⁴ A similar strategy focused on vascular factors might be applied to dry AMD and AD patients, though cost and mode of entry may be problematic, especially for AD patients. 

In conclusion, we have shown that high concentrations of Aβ have physiological effects on capillary bed density in vivo. This finding provides a novel avenue to explore the physiological function of APP including its cleavage product Aβ. Furthermore, by understanding these new mechanisms, we may gain unique insights into therapeutic targets for two costly and devastating diseases: AMD and AD. 

Supplementary Materials

pdf S1 

mov S2 

Footnotes

 Supported by the Western University of Health Sciences College of Optometry to DJC, College of Biomedical Sciences to DWE, and a grant from the California Institute for Regenerative Medicine to DWE (RN1-00538).

Footnotes

 The authors alone are responsible for the content and writing of the paper.

Footnotes

 Disclosure: K. Cunvong, None; D. Huffmire, None; D.W. Ethell, None; D.J. Cameron, None