Cone Degeneration in Aging and Age-Related Macular Degeneration

Objective  To examine the morphological features of macular photoreceptors in histologically normal retina from normal donor eyes and eyes with age-related macular degeneration (AMD).

Methods  The macular region was excised from 18 donor eyes (aged 22-96 years) and cryosectioned. Sections were stained with hematoxylin-eosin or double immunolabeled using opsin antibodies or synaptic markers.

Results  Three of 8 retinas studied in detail had AMD lesions; the remainder were histologically normal. Immunoreactivity to cone opsin was abnormal in parts of all retinas (3.5%-95.0% of each sample) and was associated with swelling of and altered immunoreactivity in the cone distal axon. In non-AMD retinas, the anomalies were mainly in nonfoveal macular locations. The nature of the anomalies was identical in non-AMD retinas and in parts of AMD retinas adjacent to overt degeneration.

Conclusion  Redistribution of opsin and anomalies in the distal cone axon are common in the aging human macula and may indicate susceptibility to AMD.

Clinical Relevance  The findings are consistent with tests of cone function in aging and early AMD, which suggests that integrated cone functions—including contrast sensitivity, color matching, and short wavelength–sensitive cone sensitivity—are the most reliable prognostic indicators of progression in AMD.

Evidence that cone dysfunction is predictive of (and a reliable measure of severity in) age-related macular degeneration (AMD) has been widely reported (reviewed by Hogg and Chakravarthy¹). Surface color² and color-matching tests,^3,4 measures of cone sensitivity,^5,6 isoluminance color-contrast sensitivity,^7,8 cone-driven multifocal electroretinography responses,^9,10 and cone-adaptation kinetics¹¹ each suggest a higher degree of pathological manifestation in the cone population than is clinically apparent in early AMD. Despite these data, analyses of photoreceptor topography in donor eyes with AMD indicate that rods in the parafovea show the earliest signs of degeneration in early nonexudative AMD, whereas the foveal cone mosaic superficially appears to be well preserved.^12-14 In patients with early AMD, defects in rod function correlate with the histological evidence,^14,15 including observations of an increase in the dark adaptation time constant¹⁶ and a reduction of scotopic sensitivity.¹⁷ To our knowledge, no reports of morphological and immunohistochemical changes in cones correlate with the many reports of cone dysfunction associated with the early stages of AMD.

In this study, we have identified morphological features of photoreceptors, particularly cones, that correlate with the visual dysfunction reported in early AMD. In the absence of traditional histopathological features of photoreceptor degeneration, we observed a number of features of macular cones indicative of pathological changes in non-AMD retinas and in the normal regions of the retina surrounding AMD lesions.

Human eyes were collected with informed consent through the Lions Sydney Eye Bank, Sydney, Australia, with ethical approval from the ethics committees of the University of Sydney and The Australian National University. Eighteen eyes were fixed overnight in 4% paraformaldehyde and then stored in 2% buffered paraformaldehyde for 1 to 12 months at 4°C. Donor eye cups were rinsed in 0.1M phosphate-buffered saline solution, and the fundus was photographed using a dissecting stereomicroscope (Leica Microsystems GmbH, Wetzlar, Germany) and a digital camera (C10plus; Jenoptick, Jena, Germany). These images were used to initially classify normal retinas vs those with possible AMD. The complex consisting of the retina, retinal pigment epithelium (RPE), and choroid was dissected from the eye cup, rinsed in 0.1M phosphate-buffered saline solution, incubated until saturated in a solution consisting of 30% sucrose and phosphate-buffered saline at 4°C, and flattened with radial cuts. The flattened retina was incubated in a solution consisting of 14% gelatin and 30% saturated sucrose at 37°C for 1 hour, then cooled to room temperature and set at 4°C. A block approximately 8 × 10 mm, including the macula, was excised and cryosectioned at 10 or 14 μm parallel to the horizontal meridian. One in every 60 (10-μm) or 43 (14-μm) sections was stained using hematoxylin-eosin (H&E) for histopathological analysis. These sections were graded for a range of histopathological features by 3 experienced graders (M.C.M., P.L.P., and J.M.P.), using published criteria¹² and classified by consensus as normal or AMD affected. The criteria include total basal deposits, number of drusen, RPE morphological features (eg, nonuniform or atrophied), detachment choroidal neovascularization, and inner/outer segment loss.¹²

A series of sections immediately adjacent to the H&E-stained sections was double immunolabeled with antibodies against long/medium wavelength–sensitive opsin (L/M opsin) and rhodopsin, following antigen retrieval (Revealit Ag; available at http://www.immunosolution.com) using standard procedures. Details of the antibodies used are given in Table 1. Immunofluorescence was viewed using a laser scanning microscope (Carl Zeiss, Inc, Jena), acquired using software from the manufacturer (PASCAL, version 4.0; Carl Zeiss, Inc) and prepared for publication using image-editing software (Adobe Photoshop CS2; Adobe Systems, Inc, San Jose, California).

Two-dimensional maps of the macula were created by first plotting on graph paper the identified histological landmarks in a full set of H&E-stained sections from each specimen (eg, vessel cross-sections and the edge of an AMD lesion), using microscope stage coordinates as a guide for position in the x-axis and the distance between plotted sections as a guide for location in the y-axis. Then, the distributions of defined grades of L/M opsin and rhodopsin labeling (described in the “Results” section) from the adjacent series of sections were plotted on the same axes, with reference to the histological features identified in the H&E-stained sections. In this way, a virtual flat mount was reconstructed representing data sampled from every 0.6-mm strip of each macula.

Adult Sprague-Dawley rats were humanely killed using an intraperitoneal injection of an overdose of pentobarbital sodium (>60 mg/kg [Valabarb; Jurox Pty Ltd, Silverwater, Australia]) with ethical approval from the animal ethics committee of The Australian National University. The left eyes of 4 rats (the same age and same sex) were removed and immediately fixed in 4% paraformaldehyde. Protocols for the right eyes simulated the handling conditions of the human donor tissue. After left eye enucleation, rats were left at room temperature for 1 hour, then stored at 4°C. Right eyes were collected at 6-hour intervals up to 24 hours after death and fixed in 4% paraformaldehyde. All eyes were stored in fixative for 2 weeks before further processing. The retina-RPE-choroid complex was then dissected from the eye cup, embedded in gelatin, and cryosectioned at 14 μm. Sections were stained using H&E, and adjacent sections were double immunolabeled with antibodies to L/M opsin, rhodopsin, and vesicular glutamate transporter 1 (vGluT1) to mimic processing of the human material.

Rat retinas with fixation delays of 6 hours or more showed organizational changes in the photoreceptor inner and outer segments and outer nuclear layer and some modification of immunoreactivity compared with the controls; these changes were identified as resulting from postmortem decay (Figure 1C-H). Three specific features attributable to postmortem decay were identified. First, we noted photoreceptor sloughing, in which the layer of inner and outer segments is thrown into folds and appears detached at the level of the external limiting membrane (Figure 1D, E, G, and H). Second, “beading” of the cone axonal process—suggestive of breakdown of the axonal membrane—was evident in L/M opsin–labeled sections (Figure 1E and F). Third, intense rhodopsin immunoreactivity was seen in the membrane of the rod somata and inner segments (Figure 1E). Immunoreactivity along the whole cone cell membrane without axon beading (seen in some of the human retinas) was not observed in any of the rat retinas. Immunoreactivity for vGluT1 was not altered by fixation being delayed up to 24 hours (Figure 1C, F, and I).

Ten of the 18 human retinas examined had features resembling artifacts identified in the delay-fixed rat retinas, including photoreceptor sloughing (Figure 1D-H) and beading of the cone axonal process (Figure 1E, arrow). Data from these eyes are not included in this analysis. Redistribution of rhodopsin into the membrane of the soma was seen in some of the remaining human samples, but was not considered an artifact because (1) immunoreactivity of the rod soma and axon was very low compared with the artifactual immunoreactivity in delay-fixed rat retinas; (2) this feature was regional and not found throughout the specimen; and (3) low immunoreactivity for rhodopsin in the inner segment and soma was detected in human retinas in the absence of other artifacts (described by Léveillard et al¹⁸).

Details of the 8 eyes further analyzed are shown in Table 2, and examples of their histological appearances are shown in Figure 2. Three retinas had histopathological features consistent with advanced AMD¹² (mean age, 85 years), 4 were aged non-AMD eyes (mean age, 84 years), and 1 was from a healthy young adult (aged 22 years). The histopathological features of all eyes are shown in Table 2.

Varying patterns of opsin immunoreactivity were observed at different locations in each retina, including the young adult retina (specimen 0703). In normal retinas, L/M opsin was confined to the outer segment, as expected (Figure 3A [green]), or distributed throughout the photoreceptor membrane (Figure 3B and C [green]). Similarly, aberrant rhodopsin labeling was identified throughout the cell membrane at levels of immunoreactivity lower than in the outer segment (Figure 3C [red, arrows]). In AMD retinas, L/M opsin or rhodopsin immunoreactivity of the entire photoreceptor membrane (as shown in Figure 3B and C) was commonly detected contiguous with areas of degeneration (Figure 3E and F), and photoreceptors lacking inner and outer segments had rhodopsin or L/M opsin–immunoreactive cell membranes (Figure 3E-G).

We defined 3 grades of opsin immunoreactivity (0-2), in which grade 0 is the expected/normal pattern (Figure 3A) and grade 2 characterizes degenerating cells that lack inner/outer segments and are immunoreactive throughout their membranes (Figure 3D-G). The intermediate grade (grade 1) describes cells with moderate to intense (L/M opsin) or low to moderate (rhodopsin) immunoreactivity of the inner segment and soma, sometimes involving the axon and synaptic terminal. The differences in intensity of labeling of the cell membranes of rods and cones in the intermediate grade reflects our observation that subtle changes in cone immunoreactivity, but not in the rods, were widespread in histologically normal retina.

The graded immunolabeling for L/M opsin and rhodopsin are shown in 3 representative virtual flat mounts in Figure 4; a quantitative assessment of the proportions of the different grades of labeling in 8 retinas is shown in Figure 5. Retina specimen 0605 had a large disciform scar (Figure 4A) covering about 50% of the sample area (Figure 5B). In H&E-stained sections, photoreceptors in the central portion of the scar were absent or had lost their photoreceptor morphological features (hatched area), whereas those on the margin were obviously degenerating (gray zone). Grade 2 opsin immunoreactivity of photoreceptors (Figures 4B and C) is consistent with the degenerative condition observable in H&E-stained sections. However, photoreceptors in the remaining histologically normal retina (white area, Figure 4A) were grade 1 for L/M opsin (Figure 4B) and rhodopsin (Figure 4C). Although 50% of the retina was histologically normal, less than 10% of the sample area had photoreceptors with normal (grade 0) opsin immunoreactivity (Figure 5B). Specimen 0609 was histologically normal throughout the sample area (Figure 4D). However, approximately 37% of the sample area was grade 1 with respect to opsin immunoreactivity, with sparing of the foveal region (Figure 4E). A much smaller area (approximately 2%) showed grade 1 rhodopsin labeling (Figure 4F). The young adult retina was histologically normal, with normal rod labeling throughout (Figure 4G), but a narrow region of anomalous cone opsin immunolabeling (approximately 3.5% of the sample area) was detected in the vicinity of the optic disc (Figure 4H).

All 5 non-AMD retinas included regions of anomalous cone labeling (grade 1, 3.5%-75.0% of the sample areas). In contrast, only 3 of the 5 non-AMD retinas had anomalous rhodopsin immunolabeling (<10% of the sample area; Figure 5A). In each of the 3 AMD retinas, the proportion of the samples with anomalous cone immunolabeling (90%-100%) is much larger than that of the histological lesions (30%-55%) (Figure 5B).

Where L/M opsin immunoreactivity was noted in the axon and pedicle, abnormal enlargement or swelling of the distal axon was also detected (Figure 6A-C) and was associated with aberrant distribution of immunoreactivity for vGluT1 (Figure 6B and C). This transporter is normally confined to the presynaptic terminal¹⁹ (Figure 6E and F). However, in regions of grade 1 L/M opsin labeling, moderate to intense immunoreactive vGluT1 was present in the pedicle and appeared to accumulate in axonal swellings (Figure 6B).

Regional changes in L/M opsin immunoreactivity and the intracellular distribution of vGluT1 are shown in Figure 6E-H. In that retina, L/M opsin immunoreactivity at the fovea and foveal rim (Figure 6E) was normal (green) and vGluT1 (red) was confined to a narrow band corresponding to the cone pedicles (Figure 6E). In the adjacent parafovea (1-2 mm from the fovea), L/M opsin immunoreactivity extended into the membrane of cone somata and proximal axons (arrows), whereas the distal axon and synaptic layer (red, arrowheads) appeared to be normal (Figure 6F). In the perifovea (Figure 6G), cones were L/M opsin immunoreactive throughout the length of the axon (grade 1), and the normally distinct band of vGluT1 labeling of the cone pedicles (Figure 6E and F) is dispersed into a broad band in the perifovea (Figure 6G [red]). Beyond 3 mm from the fovea (Figure 6H), axonal enlargements are more marked (arrows). Abnormal axons (green) and dispersal of vGluT1 (red) is shown at higher magnification in Figure 6B, whereas normal cones at a comparable location are shown in Figure 6C. Double labeling for vGluT1 and synaptic vesicle 2 showed that, in regions where distal cone axons are enlarged, vGluT1 (red), but not synaptic vesicle 2, is redistributed into the axonal swellings (Figure 6D).

We also identified prolapse, or extrusion, of cone nuclei through the external limiting membrane into the subretinal space in most of the aged non-AMD retinas. Prolapsed cones, many immunoreactive for L/M opsin (not shown), were more frequently observed in nonfoveal locations (Figure 7A) and in many cases were preferentially associated with the profiles of large vessels (Figure 7B).

Our analyses indicate that (1) broad areas of histologically normal retina include regions of anomalous photoreceptor immunoreactivity (Figures 4 and 5) and (2) regions of anomalous L/M opsin immunoreactivity are more extensive than regions of anomalous rhodopsin immunoreactivity (Figure 5). Redistribution of opsin from the outer segment to the soma and cell membrane has been described in several studies as part of the retinal response to injury and in disease.^20-23 We found a nonuniform redistribution of photoreceptor opsins, observed in broad regions of retina often involving the optic nerve head, in regions surrounding AMD lesions and in association with morphological changes in cones detected using other immunohistochemical markers.

Our findings are consistent with reports of cone dysfunction in the literature.¹ We observed that the cone anomalies identified in the histologically normal retina adjacent to AMD lesions are identical to those identified in the maculae of aged retinas, commonly in the context of histopathological correlates of early AMD. We suggest that cone dysfunction in aging and early AMD is associated with the cellular changes described herein.

We detected cone anomalies in every specimen and conclude that such anomalies are common, although no assessment of prevalence can be made because of the small number of specimens in the final cohort. These anomalies consistently involved the optic disc region, including a specimen from a 22-year-old patient, in which grade 1 cone anomalies surrounded the optic disc, suggesting retinal changes associated with peripapillary chorioretinal atrophy. Our findings are consistent with a previous study that identified peripapillary chorioretinal atrophy in 100% of eyes from donors aged 39 to 93 years and that found the histopathological features to closely resemble AMD.²⁴ Similar abnormalities have been reported in cone dystrophies^20,25 and in aged and AMD retinas,²⁶ although in those studies the macular region was not specifically investigated. Studies of cone dystrophies^20,25 emphasize cone pedicles rather than the distal axon as the location of the swelling. However, comparison of the enlarged structures described in cone dystrophies with those described herein in the axon indicates strong morphological correspondence. Comparison of our findings with those of Pow and Sullivan²⁶ suggests that enlargements of the distal axon correspond with regions identified as the origin of sprouting neurites. We did not specifically investigate neurite sprouting, although evidence of axonal branching can be detected in some of our sections. Accumulation of vGluT1 in the distal axon may indicate redirection of vesicular glutamate into neurites or more general dysfunctional trafficking of intracellular proteins, including opsin. Abnormalities in the immunoreactivities of histologically normal cones has also been described in retinas from donors who had retinitis pigmentosa.²¹

Our data suggest that, in the absence of AMD, foveal cones may be protected from the damaging changes described herein. In 3 of 5 non-AMD retinas (specimens 0606, 0608, and 0609), the foveal region appeared to be inured from the cone anomalies (eg, Figure 4E), including specimen 0606, in which 75% of the sample area had grade 1 cones. This observation is consistent with physiological evidence because several studies have concluded that visual acuity is not the best predictor or measure of progression in AMD.^2,27,28 Despite this, acuity continues to be used as a principal outcome measure for major clinical trials. Visual acuity is mediated in the central fovea by intraretinal circuits that originate from individual cones and is conveyed to the brain through nonconverging circuits via midget ganglion cells.²⁹ Other measures of visual function, including contrast sensitivity and sensitivity to blue light, rely on integrated cone functions and convey information from many cones that converges on other ganglion cell types (parasol and bistratified ganglion cells, respectively) for relay to the brain. These measures are sensitive to changes in macular function and are good predictors of progression in AMD.^4-6,30-34 Significantly, both originate predominantly from cones in nonfoveal locations.^29,35,36

In our non-AMD retinas, a higher proportion of the sample areas show anomalies in the cone population compared with the rod population (Figure 5A). This finding appears to contradict reports that rods are preferentially vulnerable to aging and degeneration because cone density at the fovea does not change significantly with age, whereas rod density decreases by 30% by the ninth decade of life.³⁷ Furthermore, in AMD retinas, the photoreceptors surviving in and around disciform scars tend to be cones rather than rods, and cones also appear more likely to survive in nonexudative AMD.⁸ We also observed in this study that very few rods survive in the retina bordering an AMD lesion, whereas cones are numerous in the surrounding retina and islands of remnant cones persist even within old scars (Figure 3E-G), confirming that rod death precedes cone death.

However, our data also suggest that, in non-AMD retinas, cones show signs of damage ahead of rods but, as the conditions generating the response progress, damaged cones linger in the retina longer than do rods. It appears that, although signs of incipient photoreceptor degeneration have onset in cones earlier than in rods, rods succumb to cell death signals more readily than do cones. The abundance of prolapsed cones detected, but not rods, also supports this contention. Displacement of photoreceptor nuclei into the outer plexiform layer and subretinal space has been described previously.³⁸ From the morphological features and immunoreactivity to L/M opsin, we identified most of these displaced or prolapsed photoreceptors as cones. We did not detect prolapsed rhodopsin-immunoreactive photoreceptors in the subretinal space. Our observations suggest that cones can be displaced, presumably losing their synaptic contacts³⁸ without immediately succumbing to cell death; on the other hand, if prolapsed rods occur in equal proportions, they appear to die more rapidly.

In conclusion, cone death subsequent to rod loss has been described in a number of retinal disorders, including AMD and retinal dystrophy,^39-42 and several authors have suggested targeting rod survival factors to facilitate cone survival.^18,41,43 Our findings show that anomalies in the distal cone axon are common in the aging human macula and may indicate susceptibility to AMD. If cell death is considered the main indicator of morbidity in the retina and if indicators of incipient degeneration in cones are not recognized, attempts to rescue truly functional cones by using survival factors are likely to be unsuccessful. This and other recent studies^26,44,45 indicate a need for further investigation into the processes of normal aging in the human retina and the role of aging in the emergence and progression of AMD.

Correspondence: Jan M. Provis, PhD, Research School of Biological Sciences, The Australian National University, GPO Box 475, Canberra, Australian Capital Territory 2601, Australia (Jan.Provis@anu.edu.au).

Submitted for Publication: December 21, 2007; final revision received July 16, 2008; accepted July 17, 2008.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Australian Research Council Centres of Excellence program. The antibody to SV2 was obtained from the Developmental Studies Hybridoma Bank under the auspices of the National Institute of Child Health and Development and maintained by the Department of Biological Sciences at The University of Iowa.

Additional Contributions: The tissue coordinators at the Lions Sydney Eye Bank provided assistance. Vicki Chrysostomou, BSc(Hons), assisted with the rat experiments. Jonathan Stone, PhD, and David Pow, PhD, provided feedback on the manuscript.