Synaptic Pathology, Altered Gene Expression, and Degeneration in Photoreceptors Impacted by Drusen

Abstract

purpose. Drusen are risk factors for age-related macular degeneration and have been shown to negatively impact cells of the RPE and retina. In this study, the effects of drusen on the synaptic machinery of retinal photoreceptors are investigated.

methods. Human donor eye tissue containing retina, RPE, and choroid was processed for confocal immunofluorescence microscopy, laser capture microdissection, and light and electron microscopy. Tissue sections were immunostained with a panel of antibodies to synapse-associated proteins. Populations of photoreceptors over drusen and normal populations of photoreceptors were microdissected from fresh frozen tissue, RNA was purified, and quantitative PCR was performed to compare relative levels of gene expression.

results. The number of photoreceptor synaptic terminals is reduced in regions of the outer plexiform layer over drusen, synaptic proteins are mislocalized in photoreceptor cells, and synaptic terminals are often observed within the outer nuclear layer. Photoreceptors over drusen also increase expression of the stress response proteins apolipoprotein E and αB-crystallin. Abnormal immunolabeling patterns are not restricted to photoreceptors directly over drusen but are also observed in cells flanking drusen. Gene expression analysis confirms reductions in the expression of genes coding for synapse-associated proteins and signal transduction proteins and increases in the expression of apolipoprotein E and αB-crystallingene transcripts. Ultrastructural analysis of photoreceptor synaptic terminals over drusen reveals significant abnormalities, and cell counts show a reduction in photoreceptor density directly over, and lateral to, drusen of all sizes.

conclusions. Photoreceptors overlying and flanking drusen exhibit morphologic and biochemical signs of degeneration. The expression of synapse-associated proteins decreases in photoreceptor synaptic terminals, whereas the expression of stress-response proteins increases. Reductions in photoreceptor cell densities over, and flanking, drusen suggest that these degenerative effects eventually result in the death of photoreceptors.

Drusen are extracellular deposits that reside between the retinal pigment epithelium (RPE) and Bruch’s membrane and are a significant risk factor for the development of both the “wet” and the “dry” forms of age-related macular degeneration (AMD). ^{1 2 3 4 5 6} As drusen increase in size with age, they encroach on and physically displace the RPE monolayer, eventually compromising RPE function. Studies investigating the composition of these extracellular deposits suggest they contain molecules derived from multiple sources, including the RPE, choroid, and plasma (for a review, see Hageman et al. ⁷ ). Many proteins that mediate inflammatory processes have been identified in drusen, suggesting that they may induce chronic localized complement activation. ^{8 9 10} This hypothesis is supported by recent data linking risk-conferring haplotypes in the gene encoding the complement regulatory molecule, factor H, to persons with AMD. ^{11 12 13 14}  

Even though drusen are a known risk factor for AMD, little is known about how they affect retinal structure and function or their impact on vision. RPE cells overlying and adjacent to drusen often exhibit morphologic signs of cell death, ^{10 15 16} and numerous histopathologic studies have shown that RPE degeneration occurs over drusen in AMD. Often a continuous RPE monolayer is no longer maintained over drusen, resulting in direct contact between photoreceptor (PR) cells and drusen. Because PRs are dependent on the RPE for metabolite exchange, structural support of outer segments, and phagocytosis of shed disc membranes (for a review, see Boulton and Dayhaw-Barker ¹⁷ ), the absence of underlying RPE most certainly produces an unfavorable environment for these cells. Thus, RPE degeneration has been hypothesized to account for the loss of PR cells observed in eyes diagnosed with AMD. ¹⁸ In fact, Dunaief et al. ¹⁹ demonstrated massive retinal degeneration and PR apoptosis over regions of RPE geographic atrophy in advanced AMD. 

Our studies have shown cellular and molecular changes in PRs that directly overlie soft and hard drusen within and outside the macula. ²⁰ One of the changes is a decompartmentalization of the abundant PR protein opsin. Whereas opsin is primarily localized to PR outer segments in normal retina, rod opsin immunolabeling is abundant in inner segments, cell bodies, axons, and axon terminals of PRs that overlie drusen. Opsin mislocalization by PRs is commonly observed after experimental retinal detachment ^{21 22} and in numerous animal models of PR degeneration. ^{23 24 25} Drusen are also associated with the deflection and shortening of PR outer segments and a decrease in cytochrome oxidase. ²⁰ These drusen-associated PR defects are observed over drusen of all sizes, even small sub-clinical drusen (<63 μm). 

Consistent with these drusen-associated abnormalities in PRs, measurable visual deficits have also been attributed to drusen. Smith et al. ²⁶ reported a significant correlation between age-related macular changes associated with early AMD and a Stiles-Crawford effect. In addition, macular drusen have been correlated with abnormal retinal sensitivity measurements in the central visual field ^{27 28} and with slower rates of dark adaptation ^{29 30 31 32 33} before the clinical diagnosis of AMD. The recently developed shape discrimination sensitivity test attributes measurable visual deficits to drusen in the absence of detectable changes in visual acuity (Wang YZ, et al. IOVS 2005;46:ARVO E-Abstract 3315). These studies clearly support a direct role for drusen in the compromise of visual function. 

In this study, we detail significant attributes of drusen-associated PR abnormalities that include alterations in gene and protein expression, synaptic defects, and cell death. These pathologic effects are not limited to retinal regions directly overlying drusen but extend laterally a considerable distance from sites of drusen deposition. 

Materials and Methods

Tissue

Whole human donor eyes were provided by the Central Oregon Lions Eye Bank (Portland, OR) and the National Disease Research Interchange (NDRI; Philadelphia, PA). Eyes were collected 1 to 8 hours postmortem and shipped in 4% paraformaldehyde in 0.1 M sodium cacodylate buffer on ice or fresh frozen on dry ice. Fixed eyes were processed by removing the anterior chamber and were stored in 0.4% paraformaldehyde at 4°C. The postmortem time interval to freezing for frozen tissue ranged from 1 to 5 hours. Eighteen fixed eyes from 18 donors and 4 frozen eyes from 4 donors were analyzed. Donor ages ranged from 49 to 86; three fixed eyes and one frozen eye had been clinically diagnosed with dry AMD. 

Immunohistochemistry

Retina/RPE/choroid tissue was dissected from eyes in 10- × 10-mm segments, embedded in 5% agarose in 0.1 M phosphate-buffered saline (PBS), and sectioned at 100 μm using a vibratome (Technical Products International, Polysciences, Warrington, PA). Tissue sections were rinsed in PBS, blocked at 4°C for 6 hours in PBTA (PBS with 0.5% bovine serum albumin [Sigma, St. Louis, MO], 0.1% Triton X-100 [Roche, Indianapolis, IN], and 0.05% sodium azide [Sigma]) containing 5% normal donkey serum (Jackson ImmunoResearch Laboratories, West Grove, PA). Sections were then incubated overnight at 4°C in primary antibodies (Table 1) , diluted in PBTA, rinsed, and incubated overnight at 4°C in donkey anti–mouse, anti–rabbit, or anti–goat IgG secondary antibodies conjugated to Cy2, Cy3, or Cy5 fluorochromes (Jackson ImmunoResearch Laboratories). Sections were then rinsed in PBS, mounted in 5% n-propyl gallate in glycerol, and examined on a laser scanning confocal microscope (Olympus Fluoview 500; Olympus America, Melville, NY). Optimal iris and gain functions were determined for each primary antibody and maintained constant during the examination of all sections labeled with that probe. Images were acquired using Olympus software; pseudo double-labeled images were generated by optimizing the Cy3 (red) channel to allow visualization of autofluorescence from lipofuscin pigment in RPE cells and Bruch’s membrane. Drusen were identified using antibodies to apolipoprotein E ³⁴ (ApoE). For each antibody, data were collected over a minimum of 10 drusen in each eye. Documented immunolabeling patterns were consistent in all eyes, those that were normal and those with AMD. 

Laser Capture Microdissection

Fresh frozen eyes were cut into eight wedges, embedded in optimum cutting temperature compound, and cut at 8 μm on a cryostat (Leica, Buffalo, NY). Individual tissue sections were mounted on glass slides, immersed in 70% ethanol at 4°C, dehydrated through a graded series of alcohols, and cleared in xylene. Cells (or groups of cells) were microdissected (7.5-μm laser spot size) using a laser capture microdissection system (PixCell 2e; Arcturus, Mountain View, CA). Normal PRs and PRs overlying drusen were microdissected onto separate microdissection caps (Arcturus). Normal PR populations were collected from retina approximately 1 mm from drusen to control for regional variations in rod/cone density. Caps were then inspected, and any contaminating tissue was removed. 

After microdissection, PR cells were lysed and RNA was extracted using an RNA isolation kit (PicoPure; Arcturus). Briefly, individual caps were incubated in RNA extraction buffer at 42°C for 30 minutes, and cell lysates were spun through RNA isolation columns (PicoPure; Arcturus). Contaminating genomic DNA was removed using a 15 minute on-column DNAse I incubation (Qiagen, Valencia, CA), and RNA was eluted in sterile water. RNA was then quantified and qualified (18s:28s) using a bioanalyzer (Agilent 2100; Agilent, Palo Alto, CA). 

Quantitative Real-Time PCR

Equal amounts of RNA were reverse transcribed for each sample (normal PRs versus PRs over drusen) using random hexamers, and cDNA was generated using reverse transcriptase (SuperScript II; Invitrogen, Carlsbad, CA) as recommended by the manufacturer. PCR primer sets were designed using commercially available software (Primer Express 1.5; Applied BioSystems, Foster City, CA), and the sequences were subsequently BLASTed against the NCBI gene sequence database to confirm specificity. Quantitative real-time PCR (qPCR) was conducted using thermal cyclers (iCycler IQ; BioRad, Hercules, CA). PCR reactions were carried out in 96-well plates using SYBR-green II (Qiagen) as the double-stranded DNA-specific fluorochrome and Platinum Taq DNA polymerase (Invitrogen) as recommended by the manufacturer. Five-step PCR profiles were used to amplify cDNA: 15 seconds at 95°C, 15 seconds at 60°C, 90 seconds at 72°C, 20 seconds at 80°C, and 20 seconds at 84°C, for 45 cycles. Melting temperatures for individual amplicons were determined by melting PCR products at 95°C for 1 minute and reannealing at 60°C for 1 minute, followed by 70 × 15-second 0.5°C temperature increases. 

Photoreceptors were independently isolated from three quadrants of one eye (including the macula), and a single quadrant in two other eyes (five total experiments). Approximately 20,000 photoreceptors were collected from normal retina and retina overlying drusen in each experiment. Primer sets were run in duplicate, and expression data were averaged for each experiment. Baseline fluorescence levels were established independently for each primer set and subtracted to remove nonspecific fluctuations in background fluorescence. PCR fluorescence threshold values were set at 20 relative fluorescent units for all reactions and individual threshold values were automatically calculated for each reaction. Data analysis was conducted well above detectable background and during the early exponential phase of amplification. Data for each experiment was normalized to three housekeeping genes (glyceraldehyde-3-phosphate dehydrogenase, 18s ribosomal subunit, and glucose phosphate isomerase), as described by Vandesomepele et al., ³⁵ to account for slight differences in starting material and small fluctuations in housekeeping gene expression. In addition to the correct calculated melting temperature, amplification of appropriately sized PCR products was confirmed using agarose gel electrophoresis for all reactions. Finally, fold changes in gene expression were averaged across experiments, and standard error of the mean (SEM) was calculated. 

Electron Microscopy

Eye bank tissue stored in 0.4% paraformaldehyde was transferred to 1% paraformaldehyde/1% glutaraldehyde for 48 hours. Tissue from three human eyes was processed using standard Spurr resin embedding protocols (Polysciences). Briefly, 2 × 2-mm tissue blocks were incubated in 2% osmium tetroxide for 1 hour, dehydrated, and embedded in Spurr’s resin. Thin sections were cut, mounted on formvar-coated grids, and stained with lead citrate and uranyl acetate. Serial resin sections were collected when possible for the reconstruction of individual synaptic terminals. Grids were viewed using a transmission electron microscope (JEM-1230; JEOL, Tokyo, Japan) and photographed using a digital camera. 

Paraffin Embedding and Cell Counting

Small (6 × 6-mm square) quadrants of tissue stored in 0.4% paraformaldehyde were transferred to 0.137 M phosphate buffer (pH 7.2). Tissue was then dehydrated through a graded series of alcohols, cleared in toluene, and equilibrated into paraffin before overnight penetration at 60°C. Paraffin was hardened and sections were cut, dried onto slides, deparaffinized in xylene, and stained with cresyl violet. Sections were digitally photographed using a microscope (Olympus BX60; Olympus, Tokyo, Japan) and software package (Microfire; Optronics, Goleta, CA). Drusen were identified and measured in digital printouts, and manual cell counts were conducted directly over individual drusen, 0 to 50 μm lateral, and 50 to 100 μm lateral to individual drusen. To account for variability in outer nuclear layer (ONL) cell density among eyes, cell counts/100 μm were expressed as the ratio of normal cell density to cell density over (or lateral to) drusen. Statistical significance was determined using a standard t test for independent samples, and error bars represent the SEM. 

Results

Synapse-Related Protein Immunolabeling Is Altered in Photoreceptors over Drusen

Antibodies against synaptic vesicle proteins label a thick plexus of rod and cone synaptic terminals within the outer plexiform layer (OPL) of the human retina. Antibodies to the vesicular glutamate transporter (vGLUT; Fig. 1a , low magnification; Fig. 1c , high magnification), SV2 (Fig. 1b , low magnification; Fig. 1d , high magnification), synaptobrevin (Fig. 1e) , synaptotagmin (Fig. 1e) , and synaptophysin ²⁰ specifically label a layer of 2 to 3 rod terminals (spherules) and a single, more vitreal layer of larger, sparsely distributed cone terminals (pedicles) within the OPL. In the normal retina, rod and cone PR terminals in the OPL are intensely labeled by antibodies to these proteins, whereas little labeling is associated with axons or cell bodies. Over drusen, however, fewer synaptic terminals within the OPL are labeled by these antibodies, and diffuse labeling is often observed in PR cell bodies (Figs. 1f 1g) . Over many drusen, PR terminal immunolabeling is absent from regions of the OPL, and only diffuse immunolabeling is observed throughout the ONL. 

Antibodies to other synapse-associated proteins demonstrate similarly altered patterns of immunoreactivity in PRs over drusen. In normal retina, antibodies to the presynaptic terminal proteins syntaxin 3 (Fig. 1h)and SNAP-25 (data not shown), the vesicle recycling protein clathrin (Fig. 1j) , and the synaptic ribbon-associated structural protein ribeye (Fig. 1i)also label rod and cone synaptic terminals within the OPL. Over drusen, however, labeling is conspicuously reduced or absent in the OPL, and synaptic terminal-like labeling is often observed within the ONL. For example, antibodies that label the synaptic ribbon-associated protein ribeye typically reveal an organized accumulation of horseshoe-shaped synaptic ribbons at the distal tips of rod spherules and cone pedicles, corresponding to the location of synaptic contacts with second-order neurons in the OPL. ³⁶ In PRs that are situated over drusen, the organization of synaptic ribbons in the OPL is disrupted, and ribeye labeling is scattered throughout the ONL (Fig. 1i) . Antibodies to vGLUT and clathrin colocalize to PR synaptic terminals in the OPL, but over drusen this labeling is often situated deep within the ONL, at the apical tips of individual PR cell bodies (Fig. 1j) . This synaptic terminal labeling is less intense than in normal terminals and is associated with terminals that might have retracted from the OPL toward their respective cell bodies in the ONL (Fig. 1j) . Ribeye-labeled structures, likely synaptic ribbons, are also detected at the apical tip of PR cell bodies in the ONL (Fig. 1k) . This pattern of altered immunolabeling is observed over drusen of all sizes, including small, subclinical drusen (<63 μm; Fig. 1h ). 

Degenerate rod PRs over drusen retain their rod opsin immunoreactivity, facilitating the visualization of rod terminals retracting from their normal locations in the OPL. ²⁰ Unlike rod opsin, cone opsin immunoreactivity is not associated with cone cell bodies or axons, and immunolabeling is often absent from these cells over drusen, ²⁰ making it difficult to identify cone terminal retraction. However, abnormal cone terminal morphologies are observed over drusen. Antibodies directed against phosphodiesterase-gamma (PDE-γ) reveal distorted cone axons and abnormally shaped cone terminals (Fig. 1l) . Cone axons are often convoluted and demonstrate disorganized synaptic protein labeling, including diffuse vGlut (Fig. 1k)and clathrin (Fig. 1l)immunoreactivity. Over drusen, PDE-γ immunolabeling of cone terminals is relatively sparse compared with normal retinal regions, suggesting a decrease in the number of cones (data not shown). Alternatively, sparse PDE-γ immunolabeling over drusen may represent the downregulation of this protein in cone PRs. 

Altered Synaptic Protein Labeling Extends beyond the Boundaries of Drusen

We previously reported that altered rod opsin immunolabeling occurs directly overlying individual drusen. ²⁰ Rod opsin immunoreactivity, for example, is found in rod cell bodies, axons, and axon terminals overlying drusen but is rarely observed in regions lateral to drusen. Alterations in PR synaptic labeling, however, are frequently observed lateral to individual drusen and directly over these extracellular deposits (Figs. 1f 1g 1h 1i) . In some cases, sparse synaptic terminal labeling in the OPL and diffuse labeling in the ONL are observed hundreds of micrometers lateral to the edges of individual drusen. Our data suggest that larger drusen are associated with more extensive alteration of the overlying OPL, but smaller drusen (<63 μm) also correlate with significant OPL abnormality. 

Stress-Response Protein Immunoreactivity Increases in Photoreceptors Proximal to Drusen

Apolipoprotein E ³⁷ (ApoE) and αB-crystallin ^{38 39} are upregulated in neurons in response to cellular stress. Over drusen, we observe an upregulation of ApoE immunoreactivity in a subset of PRs. In addition to the expression of ApoE by Müller glial cells, as previously reported, ²⁰ cone PRs over drusen also demonstrate increased immunoreactivity for this protein in their cell bodies and axon terminals (Fig. 1m) . This response may be transient because not all cone PRs over drusen are similarly immunoreactive. Increased ApoE immunoreactivity is also apparent in rods that overlie drusen; however, labeling is not as intense as that in cones. An increase in αB-crystallin immunoreactivity is also present in Müller glial cells and cones (including their terminals) overlying and flanking drusen (Fig. 1g) . This pattern of immunoreactivity is striking compared with normal retina (Fig. 1b) . 

Synaptic Terminal Ultrastructure Is Altered over Drusen

To further examine abnormal PR terminal morphology over drusen, thin sections were cut to visualize synaptic terminals at the ultrastructural level. In normal human retina, rod and cone terminals exhibit a stereotypical ultrastructural morphology that includes large synaptic invaginations and multiple synaptic ribbons adjacent to the dendrites of second-order neurons (Fig. 2a) . Synaptic terminals over drusen, however, exhibit altered morphology. Synaptic invaginations in cone pedicles for example, are flattened, extrude the dendrites of second-order neurons, and have fewer or no identifiable synaptic ribbons (Fig. 2b) . Rod spherules over drusen also lack identifiable synaptic ribbons, and large vacuoles are frequently observed within their synaptic invaginations (Fig. 2c) . Serial sectioning confirmed that these observations were consistent throughout individual synaptic terminals (data not shown). 

Altered Patterns of Gene Expression in Photoreceptors over Drusen

qPCR was conducted to compare relative levels of gene expression in PRs that overlie drusen and “normal” populations of PRs lying >1 mm from drusen. Transcripts for the signal transduction genes rod opsin, red cone opsin, blue cone opsin, green cone opsin, and arrestin were lower in PRs over drusen when compared with normal PRs (Fig. 3) . Typically, expression levels over drusen were 50% to 75% of normal levels. Similarly, synapse-related gene expression was also decreased in PRs over drusen. Expression levels for synaptophysin, synaptotagmin 1, synaptobrevin 2, syntaxin 3, the α1f-subunit of the L-type calcium channel, Rab3a, ribeye, and clathrin were reduced to 50% to 80% of normal. The vesicular glutamate transporter was the only synapse-related transcript not observed to decrease in PRs over drusen. In contrast, the expression of the neuronal stress response genes ApoE and αB-crystallin increased in populations of PRs over drusen (Fig. 3) . The relative transcript abundance for these two stress response genes in PRs over drusen was 150% to 250% of normal. 

Reduced Photoreceptor Cell Density over Drusen

Curcio et al. ¹⁸ reported an overall decrease in the number of macular PRs in retinas with heavy drusen loads. To determine whether PR cell density is significantly lower over drusen, PR cell nuclei were counted in the ONL over drusen and in the ONL proximal to drusen and were compared with cell counts from drusen-free regions distal (>1 mm) to drusen. The average density of PRs in the ONL varied from eye to eye, but cell density counts from drusen-free regions were consistent within individual eyes. A consistent reduction in the density of PRs over drusen was observed in each eye examined (Fig. 4) . On average, there was an approximately 30% reduction in the number of PRs over drusen (eye 1, 31%; eye 2, 31%; eye 3, 25%). PR cell densities varied over individual drusen but were consistently lower than normal regions of the same eye. 

To determine whether there is similar reduction in PR density in the ONL flanking drusen, cell counts were conducted in adjacent regions between 0 to 50 μm and 50 to 100 μm lateral to drusen. A 19% reduction in PR density was observed within 50 μm on both sides of individual drusen, and an 18% reduction was observed in regions 50 to 100 μm lateral to drusen (Fig. 4)when compared with drusen-free retinal regions. 

PR cell densities over drusen also correlated with drusen size. Decreased PR densities are observed over large drusen and small subclinical drusen (<63 μm) but are more pronounced over large drusen (Fig. 5) . 

Discussion

We previously reported that degenerative changes occur in populations of PRs overlying drusen, ²⁰ and significant PR cell death has been documented in retinas with heavy drusen loads. ¹⁸ However, the extent of damage to PR cells that overlie drusen and the focal cell loss associated with individual drusen have not been well characterized. This study demonstrates that PRs proximal to drusen exhibit extensive structural, compositional, and transcriptional modifications that may have direct consequences for proper cellular function. Immunohistochemical analyses revealed altered compartmentalization, loss of synaptic vesicle proteins, deconstruction of synaptic ribbons, absence of proteins required for synaptic function, and retraction of synaptic terminals in PRs over drusen. Gene expression analyses reveal a concomitant decrease in the expression of genes encoding synaptic proteins and signal transduction molecules, accompanied by an increase in the expression of stress response genes. These PR changes are associated with drusen of all sizes, including small subclinical drusen (<63 μm), and are not confined to areas directly overlying individual drusen but extend laterally to drusen. Furthermore, a significant reduction in PR cell density is associated with the presence of drusen. This spectrum of degenerative effects is indicative of cellular dysfunction and impending cell death and likely accounts for the visual deficits associated with drusen. 

Alterations in synaptic gene and protein expression have been documented in animal models of retinal degeneration. Many of these studies correlate synaptic abnormality with cellular dysfunction, cell death, and visual deficits. PR abnormalities in retinas that have been experimentally detached from the RPE are similar to those we observe over drusen. They include the decompartmentalization of opsin and synaptophysin, loss of synaptic proteins, altered gene expression (including clathrin and ribeye transcripts), and retraction of PR terminals from the OPL (Lewis GP, et al. IOVS 2005;46:ARVO E-Abstract 2444). ⁴⁰ In this model, PR defects correlate with detectable visual deficits, ⁴¹ and studies have correlated the activation of mislocalized opsin with PR cell death. ⁴²  

Genetic studies also demonstrate the importance of synapse-associated proteins for proper synaptic function. Members of the SNARE family—SNAP-25, syntaxin, and synaptobrevin—form a core complex that mediates vesicle fusion with the presynaptic membrane (for a review, see Sudhof ⁴³ ). When these genes are independently mutated, resulting in complete or partial loss of function, synaptic transmission is severely compromised (syntaxin ^{44 45} ; SNAP-25 ⁴⁶ ; synaptobrevin ⁴⁷ ; synaptotagmin ^{48 49} ). Thus, the observed reduction in the expression of synaptic proteins in PRs over drusen, coupled with mislocalization of the remaining protein, is likely to have significant functional consequences for these cells. 

Ultrastructural abnormalities observed in PRs over drusen are similar to those reported in other models of PR degeneration and support the hypothesis that PRs over drusen are reduced in their capacity to function properly. In the P347S rod opsin mutant mouse and the rds mouse, rod opsin mislocalization, vacuolized synapses, and floating synaptic ribbons were observed in PRs before cell death (Peng Y-W, et al. IOVS 2004;45:ARVO E-Abstract 5360; Peng Y-W, et al. IOVS 2005;46:ARVO E-Abstract 5346). In the retinal degenerative RCS rat, which harbors an RPE cell mutation in the MertK gene, synaptic terminals that extrude the dendrites of second-order neurons and form-flattened synaptic contacts are also indicative of degenerating PRs. ⁵⁰ In both these animal models, synaptic defects have been correlated with significant reductions in visual function. 

PR cell loss occurs in eyes with heavy drusen loads, ¹⁸ but the focal loss of cells over individual drusen and a correlation between drusen size and PR cell loss has not been documented. Because PR cell density decreases adjacent to and directly over drusen, it is unlikely that drusen merely displace PRs. Rather, large numbers of PRs are dying. Some PRs, especially cones, appear to be pushed aside by drusen that encroach on the ONL, but these cells are rarely displaced by more that a few microns. The reduction in PR density that is detectable up to 100 μm away from individual drusen suggests that the area of retina negatively impacted by these deposits is extensive. For example, a 100-μm diameter druse that affects an area of retina extending 100 μm in all directions from its borders would be predicted to impact an area of retina almost 10 times greater than its physical area. Although most of the data in this study were collected outside the macula in non-AMD eyes, similar reductions in cell numbers were documented in AMD eyes and from cell counts taken within the maculas of non-AMD eyes. The implications of these observations are significant when one considers multiple large drusen in a small area such as the macula, the region of the retina responsible for a significant proportion of human vision. 

The cellular and molecular abnormalities observed in PRs over drusen are classic indicators of degeneration and impending cell death. However, unlike most animal models of PR degeneration, PRs over drusen are likely to undergo gradual degeneration that occurs over many years rather than an acute degenerative program that lasts days or weeks. Our inability to detect TUNEL-positive cells over drusen (data not shown) suggests that cell death is a rare occurrence. In rapid forms of PR degeneration, and even in advanced AMD associated with geographic atrophy, apoptotic TUNEL-positive cells can be detected. ¹⁹ Since drusen accumulate slowly over many months or years and TUNEL labeling is detectable only during a short window of apoptotic cell death (∼2 hours), it is likely that PRs over drusen undergo a longer, more protracted time course of degeneration. Initially, visual transduction and synaptic functioning is likely to be compromised in these cells while a protective stress response is mounted. Gene expression changes correlate with changes in synaptic protein expression and localization and increases in ApoE and αB-crystallin expression, demonstrating an overall shift in the transcriptional profile of the PR near drusen from that of a functional neuron to that of a compromised cell. During this time, rod opsin mislocalization is likely to occur in rods, with protein present in the cell membrane until the cell reaches its ultimate fate. Finally, the retraction of synaptic terminals and a significant reduction in the number of PR nuclei in areas impacted by drusen suggest that these compromised cells eventually degenerate and die. 

The presence of drusen in the macula is diagnostic for early AMD, yet little is known about the role these extracellular deposits play in vision loss. Recent genetic data support hypotheses that drusen act as local sites of complement activation, suggesting that the immune system may contribute to the compromise of RPE cells overlying and flanking drusen deposits. ^{7 8 9 10} The data presented here clearly link drusen proximity to PR degeneration and cell loss. However, it remains to be determined whether complement-mediated events contribute to retinal degeneration and vision loss associated with drusen in early AMD. Studies addressing the role of complement-related mechanisms in drusen deposition and photoreceptor degeneration are clearly warranted. 

 Supported in part by Cottage Hospital of Santa Barbara, National Eye Institute Grants EY11527 (LVJ) and EY11521 (DHA), and by generous benefactors of the Center for the Study of Macular Degeneration at UCSB.

 Submitted for publication June 16, 2005; revised August 10 and September 6, 2005; accepted October 14, 2005.

 Disclosure: P.T. Johnson, None; M.N. Brown, None; B.C. Pulliam, None; D.H. Anderson, None; L.V. Johnson, 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.

 

Table 1.

 

Primary Antibodies

		Manufacturer/Location
Apolipoprotein E	Goat polyclonal	Chemicon/Temecula, CA
αB-crystallin	Rabbit polyclonal	Accurate Scientific/Westbury, NY
Clathrin light chain	Mouse Monoclonal	Synaptic Systems/Gottingen, Germany
CtBP2 (Ribeye)	Mouse monoclonal	BD Biosciences/San Jose, CA
PDE-γ (73–87)	Rabbit polyclonal	Cytosignal/Irvine, CA
Synaptotagmin 1,2	Rabbit polyclonal	Synaptic Systems/Gottingen, Germany
Syntaxin 3	Rabbit polyclonal	Synaptic Systems/Gottingen, Germany
SV2	Mouse monoclonal	D.S.H.B./Iowa City, IA
Synaptobrevin/VAMP2	Mouse monoclonal	Synaptic Systems/Gottingen, Germany
vGLUT1	Rabbit polyclonal	Synaptic Systems/Gottingen, Germany

Figure 1.

 

Synaptic terminal labeling in normal photoreceptors and photoreceptors over drusen. (a-e) Synaptic terminals within the OPL in normal adult retina. Lipofuscin autofluorescence demarcates the RPE in a and b. Antibodies to αB-crystallin (green) do not label normal retina in (b). (c-d) High magnification of the OPL in a and b, respectively. (e) High magnification image of OPL in normal retina. (f) PR synaptic terminal labeling is reduced in the OPL over drusen, and labeling is associated with cell bodies in the ONL (arrows). (g) Synaptic terminal immunoreactivity is reduced in PR terminals over drusen. αB-crystallin immunostaining (green) is present in Müller glial cells over drusen and in cone terminals overlying and adjacent to drusen (arrows). (h) Redistribution of protein from synaptic terminals to PR cell bodies (arrow) over a small subclinical druse. (i) Synaptic ribbons (red) are noticeably absent from the OPL and sparsely distributed throughout the ONL (arrows) over drusen. Autofluorescence (green) allows the faint visualization of the druse and PR inner and outer segments. (j) Antibodies double label PR synaptic terminal-like structures associated with PR cell bodies (arrows) in the ONL over a druse (not shown). (k) Antibodies label PR synaptic terminals in the ONL over drusen (large arrows). A distorted cone axon (small arrow) is also visible. Lipofuscin autofluorescence (red) can be visualized in the RPE. (l) A distorted cone PR over a druse (not shown) contains diffuse clathrin-immunoreactivity (red/yellow, arrow) throughout the axon. Compare clathrin labeling (red) over drusen in j and l with clathrin labeling (red) in the normal retina (a). (m) A small druse (blue) is associated with increased apoE immunoreactivity in cone cell bodies (large arrows) and axon terminals (small arrows). Green autofluorescence demarcates the RPE and photoreceptor inner and outer segments. All unlabeled scale bars, 10 μm.

Figure 2.

 

Electron micrographs of PR synaptic terminal ultrastructure. (a) A cone pedicle (p) and a rod spherule (s) exhibit characteristic morphology and contain many synaptic ribbons (arrows) in normal retina. (b) A flattened cone pedicle (p) over drusen contains no visible synaptic ribbons and appears to have externalized the dendrites of second-order neurons (arrows). (c) A rod spherule synaptic complex (s) over drusen contains a large vacuole (v) and no synaptic ribbons. Magnification: (a) ×6000; (b) ×10,000; (c) ×20,000.

Figure 3.

 

Altered gene expression in photoreceptors over drusen. Laser capture microdissected photoreceptors over drusen exhibit reduced expression of signal transduction-related genes and synapse-related genes and increased expression of two stress-response genes (ApoE and αB-crystallin). Normalized data from five experiments were averaged and expressed as “fold changes” in gene expression in PR populations over drusen compared with normal PR gene expression. Error bars represent SEM.

Figure 4.

 

PR cell density in the normal retina compared with that over and adjacent to drusen. Three donor eyes exhibit similar reductions in the density of PRs directly over drusen and in regions adjacent to drusen (0–50 μm and 50–100 μm).

Figure 5.

 

PR cell density over drusen of different sizes. A significant reduction in PR cell density is observed over drusen. These reductions are observed over large drusen and over small, subclinical drusen (<63 μm). Error bars represent SEM (*P < 0.001).

The authors thank Monte Radeke, Kellen Betts, and Vanessa Gray for technical assistance and Robert Avery and George Primbs for insight into the clinical realm of AMD.