Ординатура / Офтальмология / Английские материалы / Ocular Neuroprotection_Levin, Polo _2003

C.Specialized Methods for Regional Photoreceptor Loss

Figure 5 shows a different systematic sampling strategy employed to assess the relative rate of cone and rod degeneration surrounding areas of nearly complete photoreceptor loss vitread to fibrovascular scars and RPE atrophy in eyes with late exudative age-related maculopathy [5]. Here we counted photoreceptors at 0.1 mm intervals along arbitrarily chosen tracks that crossed the boundary between intact and degenerated photoreceptor mosaic. Photoreceptor degeneration was expressed as loss relative to controls as a function of distance from the margin of the intact photoreceptor mosaic. Counts were compared to those in matched locations in control eyes using the same measure described in the previous section for maps. In the figure, the dashed line indicates the variability in cell density among control eyes, calculated as follows. At each retinal location, we computed the differences between log(density) for each pairwise comparison of one randomly chosen control eye and the other controls. This computation, done for both cones and rods, established 95% confidence intervals for differences among the controls. Results were similar regardless of which control was chosen. Then we computed the differences in log(density) for pairwise comparisons of the ARM eye and the controls. Directional differences between ARM eyes and controls that fell below the lower confidence limit for controls were considered significant loss. We then reported the percentage of counting sites with significant loss and the percentage of sites with loss where either rod or cone loss predominated.

VII. COUNTS IN SECTIONS

It is possible to count cells in sections through the full thickness of the retina, (i.e., in the plane orthogonal to a retinal whole mount). Two different approaches have been employed for counting retinal cells in sections, depending the layer of interest. In both cases, one must carefully attend to the location of samples in order to ensure that counts are taken from comparable locations (see Sec. VI A).

In the mammalian retina, photoreceptor nuclei form a thick layer in the outer nuclear layer (ONL), typically with a single row of cones external to multiple rows of rods. Many studies of retinal degeneration and treatments in animal models have quantified photoreceptors as either rows of ONL nuclei or thickness of ONL layer [19]. Use of ONL thickness compares favorably with counts of photoreceptor nuclei and can be done quickly [20]. This approach is suitable for quantifying the more numerous rods but is less suitable for quantifying the less numerous cones, because the sample of cones is very small in single sections [21]. Note that counts of cells along the length of a retinal section (typically expressed as

cells/100 m) assumes that the counted particles (usually nuclei) do not differ in size between treatment and control groups, which may not be true for degenerating photoreceptors. Counts along single slices also assume the overall size of the retina does not differ between groups. It is possible to section the retina serially and determine a total number of cells using a subsample of sections in order to obtain a sample size comparable to that achievable in a whole-mount. Such an approach is feasible for very small eyes, such as embryonic tissues.

The current standard for cell counting in sections is the disector, an unbiased counting method that uses counts obtained in parallel planes separated by a known distance in the tissue (Fig. 6). The rule is to count only those particles that appear within an unbiased counting frame in one plane (the reference plane) but not in the matched plane (the look-up plane). The number of nuclei (conventionally called Q ) is contained within the volume of the disector. The volume of the disector is equal to the area of the counting frame, multiplied by the distance between planes. If the specimen has been sampled systematically by multiple disectors, the number per volume is calculated by summing the area and Q over all disectors. Reference and look-up planes are separated by a distance that is one-quarter to one-third that of the particles being counted. There should be only one counted particle per cell, and the fewest assumptions are required for smooth convex objects, so cell nuclei are the typical counting particle. In order to count nuclei that are 8 m in diameter, planes that are 2–3 m apart are required.

An efficient way to implement a disector is to track through successive focal planes in a thick slice of tissue, a method known as the optical disector. It is possible to use a high numerical aperture objective to identify the reference and look-up planes a relatively thick section ( 25 m). Confocal microscopy can also be used to optically section tissue for counting of fluorescent cells and will likely be used more frequently for this purpose as these instruments become more widely available. The alternative to an optical disector is the physical disector, which involves either semi-thin or ultrathin sections prepared for transmission conventional electron microscopy. In order to identify the same cells in the reference and look-up planes, the sections are aligned by reference to local fiducial landmarks such as blood vessels. A physical disector is difficult to apply to the ONL of the retina. The rod nuclei are morphologically homogeneous and form a paracrystalline array like ball bearings in a box, and there are no nearby blood vessels. Thus, the reference and look-up planes can be aligned spuriously, introducing inaccuracies into the number of cells missing from the look-up plane. However, in the inner nuclear and ganglion cell layers, the presence of capillaries and multiple, distinctive cell populations together facilitate the chore of aligning sections. Disector methods have been successfully used for determining cell densities in these layers [13], and they are the method of choice in contemporary morphometric studies of brain [22].

VIII. CHALLENGES OF THE MACULA

For either whole-mounts or sections, the primate macula presents unique challenges for cell counting, due to its extreme thickness and the presence of the fovea. In the macula of human and monkey, ganglion cells are up to six cells deep. In whole mounts, conventional Nissl stains that involve dehydration of tissue reduce thickness by factors up to two-thirds [2]. We used unstained tissue that was cleared with dimethyl sulfoxide rather than ethanol and xylene and viewed with differential interference contrast optics [2,3,6]. We continuously focused through the tissue while counting nucleoli of ganglion cell layer neurons. Studies using fluorescent markers for inner retinal cells would require multiple image planes. On the photoreceptor side of the fovea, the external fovea is an inward dip, where photoreceptor inner segments do not form a single plane. Therefore, macular photoreceptors, like the ganglion cells, also require continuous adjustment of focus to find the optimal plane for counting.

Another challenge to cell counting in the macula is the sharp changes in cell density near the foveal center. The cone distribution has a very small (100m or less) area of very high density in the foveal center that drops by 90% within 1 mm. Therefore, counts in this area require precisely specified sample locations. Poor control of location can insert noise into cell counts and make effects difficult to detect. In our whole-mount studies, we set the point of highest cone density as the foveal center and used a computer-controlled stepper-stage on the microscope, so that position relative to the foveal center could be specified accurately. In studies requiring sections, we sectioned serially into the fovea so that the center could be identified by the absence of ganglion cell and inner nuclear layers, and the absence of Henle fibers cut in cross-section. Other challenges to studies using the macula include long incubation times required for complete penetration of the retina by immunoreagents, the absence of certain well-estab- lished markers in the foveal region [l0], and the predilection for postmortem swelling in inner retina that can impede the identification of the foveal center.

ACKNOWLEDGMENTS

Supported by NIH grant EY06109 and unrestricted funds from Research to Prevent Blindness, Inc., to the Department of Ophthalmology, University of Alabama School of Medicine.

REFERENCES

1.Curcio CA, Sloan KR, Kalina RE, Hendrickson AE. Human photoreceptor topography. J Comp Neurol 1990; 292:497–523.

2.Curcio CA, Allen KA. Topography of ganglion cells in human retina. J Comp Neurol 1990; 300:5–25.

3.Curcio CA, Drucker DN. Retinal ganglion cells in Alzheimer’s disease and aging. Ann Neurol 1993; 33:248–257.

4.Curcio CA, Millican CL, Allen KA, Kalina RE. Aging of the human photoreceptor mosaic: evidence for selective vulnerability of rods in central retina. Invest Ophthalmol Vis Sci 1993; 34:3278–3296.

5.Curcio CA, Medeiros NE, Millican CL. Photoreceptor loss in age-related macular degeneration. Invest Ophthalmol Vis Sci 1996; 37:1236–1249.

6.Medeiros NE, Curcio CA. Preservation of ganglion cell layer neurons in age-related macular degeneration. Invest Ophthalmol Vis Sci 2001; 42:795–803.

7.Nork TM, McCormick SA, Chao G-M, Odom JV. Distribution of carbonic anhydrase among human photoreceptors. Invest Ophthalmol Vis Sci 1990; 31:1451– 458.

8.Young HM, Vaney DI. Rod-signal interneurons in the rabbit retina: 1. Rod bipolar cells. J Comp Neurol 1991; 310:139–153.

9.Miller WH, Snyder AW. The tiered vertebrate retina. Vision Res 1977; 17:239– 255.

10.Haley TL, Pochet R, Baizer L, Burton MD, Parmentier M, Crabb JW, Polans AS. Calbindin D-28K immunoreactivity of human cone cells varies with retinal position. Vis Neurosci 1995; 12:301–307.

11.Kolb H. Amacrine cells of the mammalian retina: neurocircuitry and functional roles. Eye 1997; 11:904–923.

12.Xiang M, Zhou L, Macke JP, Yoshioka T, Hendry SH, Eddy RL, Shows TB. The Brn-3 family of POU-domain factors: primary structure, binding specificity, and expression in subsets of retinal ganglion cells and somatosensory neurons. J Neurosci 1995; 15:4762–4785.

13.Jeon C-J, Strettoi E, Masland RH. The major cell populations of the mouse retina. J Neurosci 1998; 18:8936–8946.

14.Williams RW, Strom RC, Rice DS, Goldowitz D. Genetic and environmental control of variation in retinal ganglion cell number in mice. J Neurosci 1996; 16:7193– 7205.

15.Gundersen HJG, Bendtsen TF, Korbo L, Marcussen N, Moller A, Nielsen K, Nyengaard JR. Some new, simple and efficient stereological methods and their use in pathological research and diagnosis. Acta Pathol Microbiol Immunol Scand 1988; 96:857–881.

16.Mohand-Said S, Deudon-Combe A, Hicks D, Simonutti M, Forster V, Fintz AC, Leveillard T. Normal retina releases a diffusible factor stimulating cone survival in the retinal degeneration mouse. Proc Natl Acad Sci USA 1998; 95:8357–8362.

17.Curcio CA, Sloan KR, Meyer D. Computer methods for sampling, reconstruction display and analysis. Vision Res 1989; 19:529–540.

18.Curcio CA, Sloan KR, Jr., Packer O, Hendrickson AE, Kalina RE. Distribution of cones in human and monkey retina: individual variability and radial asymmetry. Science 1987; 236:579–582.

19.LaVail MM, Yasumura D, Matthes MT, Drenser KA, Flannery JG, Lewin AS, Hauswirth WW. Ribozyme rescue of photoreceptor cells in P23H transgenic rats:

long-term survival and late-stage therapy. Proc Natl Acad Sci USA 2000; 97: 11488–11493.

20.Michon JJ, Li ZL, Shioura N, Anderson RJ, Tso MM. A comparative study of methods of photoreceptor morphometry. Invest Ophthalmol Vis Sci 1991; 32:280– 284.

21.LaVail MM, Matthes MT, Yasamura D, Steinberg RH. Variability in rate of cone degeneration in the retinal degeneration (rd/rd) mouse. Exp Eye Res 1997; 65:45– 50.

22.West MJ, Slomianka L, Gunderson HJG. Unbiased stereological estimation of the total number of neurons in the subdivisions of the rat hippocampus using the optical fractionator. Anat Rec 1991; 231:482–497.

23.Mayhew TM, Gundersen HJG. ‘If you assume, you can make an ass out of u and me’: a decade of the disector for stereological counting of particles in 3D space. J Anat 1996; 188:1–15.

nerve [6–13] or visual cortex [14–19]. Because it is a relatively isolated organ, and because the anatomy, physiology, and pharmacology of many of its neurons are well defined, the retina provides a good model for studying degeneration and neuroprotection within the central nervous system. In addition, the retina can be studied in vivo, or removed from the eye and examined for an extended period of time as an isolated tissue preparation [20–25]. Because removal from the eye does not disrupt its cellular organization, neurons within the isolated retina retain their normal spatial and connective relations (synaptic and gap junctional), and their responses to light stimulation. This is in contrast to other isolated brain preparations where tissue acquisition commonly results in a significant disruption of afferent and efferent connections, neurons at the tissue margins are damaged during isolation, and the functional integrity of the isolated preparation can be assessed only by means of electrical stimulation. The maintenance of normal cellular relations within the isolated retina also is important with respect to neuroprotection, where survival mechanisms need not involve only direct interactions between the drug and the injured neurons.

The cellular organization of the vertebrate retina, which is conserved across most species, also makes it an ideal tissue for use in degeneration and neuroprotection studies. Although the retina contains several million neurons, these are arranged into three distinct layers of nerve cells, separated by two layers of synaptic connections. And while many subclasses of retinal neurons have been described, it is generally agreed that most retinas contain only five major classes of neurons. From outside to inside, these include the photoreceptors, horizontal, bipolar, and amacrine cells, and the ganglion cells [1]. Of these different cell types, perhaps the best studied are the ganglion cells. These neurons represent the final stage of visual processing within the retina and their axons form the optic nerve, a primary site of injury in many optic neuropathies. In addition, ganglion cells and their axons give rise to the different functional streams that compose the central visual pathway [2,5] Finally, because they are relatively large and form a distinct layer near the inner retinal surface, ganglion cells are more accessible, both in vivo and in the isolated retina than most other neurons [1].

III.GANGLION CELL CLASSIFICATIONS AND METHODS OF STUDYING

To date, at least thirty morphologically distinct types of ganglion cells have been reported in the vertebrate retina [26–28]. Despite this relatively large number, however, our current understanding of ganglion cell structure and function is based almost exclusively on a select group of neurons from the cat and primate retinae. In the cat, the three major anatomical classes of ganglion cells are the

alpha, beta, and gamma cells, and there is direct intracellular evidence that these correspond with the Y-, X- and W-cells described physiologically [2,29–31]. Of these different classes of ganglion cells, the alpha and beta cells have been described most completely. In brief, alpha cells represent about 5% of the ganglion cells in the cat retina. They have the largest cell bodies (25–35 m diameters), and large, radially oriented, dendritic arbors that commonly originate from five or six primary dendrites and display a regular pattern of branching. Beta-cells comprise about 55% of the ganglion cells in the cat retina. They have mediumsized somata (18–20 m diameters), and their small to medium-sized dendritic trees often originate from a single primary dendrite that then gives rise to a compact, bushy dendritic arbor. Functionally, Y-cells have high contrast gain, large receptive fields, fast conducting axons, and are considered to be involved primarily with object detection. By contrast, X-cells have small receptive fields, are of highest density in central retina, and are considered to be involved primarily with the analysis of fine detail.

The primate retina also contains three major classes of ganglion cells. Anatomically, these are the parasol, midget, and small-field bistratified cells [4,5,20,22,25,32–36]. Parasol cells represent approximately 10% of the ganglion cells in the primate retina. At all retinal eccentricities, the somata and dendritic fields of these neurons are among the largest in the ganglion cell layer. Similar to the alpha cells of the cat retina, the dendritic arbors of primate parasol cells commonly originate from three or four large primary processes that branch regularly and form a radially symmetric arbor. Functionally, parasol cells have large receptive fields with rapidly conducting axons, and they respond best to achromatic stimuli of high temporal and low spatial frequency [3,5,37–40]. Combined with their relatively uniform distribution acrosss the retina, these neurons also are considered to be involved primarily with object detection. Midget ganglion cells represent about 80% of the ganglion cells in the primate retina. They have medium-sized somata and their small to medium-sized dendritic trees often originate from a single dendrite that then gives rise to a compact, bushy dendritic arbor. Midget ganglion cells have small receptive fields and respond best to chromatic stimuli (primarily red–green) of high spatial and low temporal frequency [3,5,37-40]. Based on these characteristics, and their high density and one-to- one synaptic arrangements with midget bipolar cells within the fovea, these neurons are considered to subserve fine spatial discrimination. Small-field bistratified cells represent about 5–8% of the ganglion cells in the primate retina. These neurons have somata and dendritic fields that are similar in size to those of surrounding parasol cells, and they respond best to short-wavelength (blue) stimuli [4,5,36].

For more than 100 years, anatomists have been interested in the structure and classification of neurons in different regions of the brain, including the retina. Some of the earliest anatomical work in the retina was conducted by Tartuferi

and Ramo´n y Cajal using modifications of the osmium-dichromate-silver staining method of Golgi, and by Dogiel, using the methylene blue staining routine of Ehrlich [41]. More recent studies aimed at defining the different morphological classes of ganglion cells in the vertebrate retina also have relied on these early techniques. In addition, neurofibrillar stains have been used to study the morphologies of ganglion cells in the normal retina [42], and following damage to the optic nerve [8]. A number of different retrograde tracing methods also have been used to examine ganglion cell morphology. Some of these have involved direct application of various dyes and neuronal tracers to either the intact retina or the severed optic nerve [43–46]; others have used micro-injections of tracers into specific retinal target structures in order to define the central projection patterns of specific ganglion cell types [47–49]. The major advantage of these staining techniques over the more traditional Nissl-staining methods [50] is that they provide detailed information about not only the cell soma but also the dendritic arbor. This enhances the accuracy of ganglion cell classification and provides a finer level of detail for determining injuryand/or treatment-related changes in ganglion cell morphology.

Surprisingly, to date few of the techniques described above have been applied to studies of retinal ganglion cell degeneration or neuroprotection. Most likely this is because of their capriciousness. Although silver staining methods reveal exquisite morphological detail, they often stain only select subsets of neurons. In addition, the pattern of staining can be highly variable across retinas, or even within the same retina. Various approaches using retrograde tracers also have been hampered by the fact that they often label only small populations of ganglion cells, the labeling is inconsistent and nonuniform across and within retinas, and the tracer often fails to label fine distal dendrites. Furthermore, in those cases where the insult occurs at the level of the optic nerve, the potential for labeling ganglion cells from a central target are greatly diminished or eliminated. One approach that has been used with relatively good success to study retinal degeneration and neuroprotection in the rat involves placing a piece of gelatin sponge soaked with either Fluoro-Gold (Fluorochrome, Englewood, CO) or the carbocyanine dye DiI (Molecular Probes, Eugene, OR) over the superior colliculus and visual thalamus in vivo [11,51]. Because both of these retrograde tracers are capable of labeling ganglion cells long term, it is possible to prelabel ganglion cells, induce the retinal injury, and still provide an adequate survival period for assessing the treatment strategy. One limitation to this approach, however, is that it is most practical only in small vertebrates, such as the rat, where the central target nuclei can be easily accessed and covered with the tracer-soaked sponge. In addition, the tracers label primarily the cell soma, and the fate of tracer released from degenerating ganglion cells is always a concern. Finally, recent evidence indicates that, following long-term survival periods ( 6 wk), Fluoro-

Gold migrates not only from ganglion cells but from the retina in general (Bal Chauhan, personal communication).

IV. THE ISOLATED RETINA PREPARATION

In our analyses of retinal ganglion cells from cats with neonatal visual cortex damage [19] and monkeys with experimental glaucoma [25], we sought to avoid the limitations of the silver impregnation and retrograde labeling techniques by combining an isolated retina preparation with intracellular staining techniques. This approach provided several advantages. First, it allowed us to visualize and target single neurons. This was especially important in the cats with neonatal visual cortex damage, where there was a significant reduction in the number of surviving ganglion cells. Second, using the isolated retina preparation we were able to examine ganglion cells from matched areas of each retina, thereby reducing sample variability. And finally, like the silver staining methods, the intracellular approach allowed us to label and compare qualitatively and quantitatively morphological features associated not only with the cell soma but also the dendritic tree and intraretinal segment of the axon.

Similar procedures are used to isolate and maintain our cat and primate retinas. These are outlined diagrammatically in Figure 1 and described in more detail below. Following the desired survival period, each animal receives an overdose of pentobarbital sodium. The eyes then are removed and placed into the same solution that will be used to sustain them during the course of the experiment. We have used both Ames media [20–22,46] and the artificial cerebral spinal fluid (aCSF) described by Saito [29] with equal success. If the eyes must be transported, we have found it best to either leave them intact, or design a small chamber for providing oxygenation during transport. The anterior segment of each eye (from the ora serrata forward) is removed using a scalpel blade and pair of small scissors, and the resulting posterior eyecup is placed into a beaker of aCSF (pH 7.4). The beaker contains a stainless steel basket that holds the tissue off the bottom, and the aCSF is oxygenated with 95% O2 and 5% CO2 at room temperature using a glass gas dispersion tube with a fritted ending. Separation of the anterior and posterior segments is performed in a petri dish over gauze that has been moistened lightly with aCSF to prevent the tissue from sticking. Good separation, with complete removal of the vitreous, seems to work best if the initial incision is made 2–3 mm posterior to the ora serrata, and the two segments then are folded back away from each other using two pair of forceps with serrated tips. The goal is to have the vitreous remain firmly attached to the anterior segment and removed from the posterior segment by peeling it away from the retinal surface; pulling the vitreous perpendicular to the surface can