Nerve fiber layer astrocytes of the primate retina: morphology, distribution, and density. Thomas E. Ogden

Transcription

1 Nerve fiber layer astrocytes of the primate retina: morphology, distribution, and density Thomas E. Ogden The nerve fiber layer of several species of primate has been studied with Golgi silver impregnations and Nissl stains. The morphology, distribution, and density of fibrous astrocytes of the nerve fiber layer are described. Two classes of astrocytes are distinguished: elongated cells whose processes parallel the nerve fiber bundles and cover a rectangular area more than two times longer than wide ivithout vascular contact and stellate cells whose processes conspicuously cross the nerve fiber bundles to cover a slightly oval area and make unspecialized vascular contacts. The density of glial nuclei in the nerve fiber layer is roughly proportional to the thickness of the layer all across the retina, so that higher densities are seen in specialized regions such as the arcuate nerve fiber bundles and near the optic disc. Key words: primate, retina, nerve fiber layer, astrocyte, neuroglia, morphology This study concerns the astrocytes of the nerve fiber layer of the primate retina, a well-defined central nervous system fiber tract which contains no myelinated fibers or oligodendrocytes and is conveniently displayed in a block-stained retinal whole mount. Astrocytes are stellate-shaped neuroglial cells, occur in large numbers in most parts of the central nervous system, and are classed by light microscopy asfibrousor protoplasmic according to the distribution and character of their processes, which are elongated and thin in the fibrous variety and shorter and membranous in the protoplasmic variety. Astrocytes were observed in the vertebrate retina by the earliest neurohistolo- From The Estelle Doheny Eye Foundation and the Department of Physiology, University of Southern California, Los Angeles. Supported by NIH grant EY Submitted for publication Nov. 23, Reprint requests: Thomas E. Ogden, The Estelle Doheny Eye Foundation, 1355 San Pablo St., Los Angeles, Calif gists (reviewed by Polyak 1 ) but were not studied in detail until the application of the neuroglial stains, particularly the silver carbonate method of Del Rio Hortega and its modification by Scharenberg and Zeman 2 to the retina. A number of studies 1 " 9 of a qualitative nature then showed typical astrocytes in the nerve fiber layer only near the optic disc. Elsewhere the fiber layer contained spindle-shaped glial cells, called "lemmocytes" 4 because they resembled the sheath cells of unmyelinated peripheral nerves. The function of glia remains an enigma despite a growing body of data concerning its physiology. 10 " 11 Most authors seem to accept Virchow's 12 original suggestion that the glia at least provides support for neuronal elements as well as other functions. Recently, considerable evidence has accumulated to suggest that the neuroglia may act as a spatial sump into which potassium pours during neuronal activity, thus preventing its buildup in the extracellular spaces. 13 The latter function requires close apposition of glial and neuronal processes, yet it has been known /78/ $01.20/0 Assoc. for Res. in Vis. and Ophthal., Inc. 499

2 500 Ogden Invest. Ophthalmol. Visual Sci. June 1978 Fig. 1. Photomicrographic montages of nerve fiber layer fibrous astrocytes stained by Golgi impregnation and seen in retinal flat mount. A, Elongated type from a human retina 2 mm temporal to the fovea. B, Elongated type from the far nasal periphery of a rhesus monkey retina. (Calibration 50 fi.)

3 Volume 17 Number & Primate retinal astrocytes 501 Fig. 2. Photomicrographs of nerve fiber layer fibrous astrocytes stained by Golgi impregnation and seen in retinal flat mount. A, Stellate type from the retina of a baboon, about 2 mm superior to the fovea. Note vascular contacts indicated by the arrowheads. (Calibration 35 /LA.) B, Stellate type from the retina of a spider monkey near the optic disc. (Calibration 25 fx.)

4 502 Ogden Invest. Ophthalmol. Visual Sci. June p Fig. 3. Camera lucida drawings of nerve fiber layer fibrous astrocytes stained by Golgi impregnation. A, Elongate type seen in a 100 fx thick cross-section of the retina of a cebus monkey. The dotted line indicates the position of the internal limiting membrane (ILM); the arrows indicate small astrocyte excrescences through the ILM. B to E, Retinal flat mounts. B, Elongate type from the owl monkey. C, Intermediate type from a mangabey monkey. D, Stellate type from a human retina; bv, blood vessel. E, Elongate type from a rhesus monkey. F, Stellate type from a rhesus monkey. since the work of Cohen 14 that the axons of the nerve fiber layer of the rhesus monkey have few glial contacts. Cohen sampled only a few retinal areas, yet the nerve fiber layer varies greatly with position across the retina, and it is clear from the work of Ikui et al. 15 that there are regional variations of astrocyte density within the layer. In this paper, the variation of morphology and distribution of astrocytes are described, and it will be shown in seven different primate species that typical fibrous astrocytes are to be found throughout the retina in the nerve fiber layer with a frequency that is roughly proportional to its thickness. Methods This study involved the retina of nine rhesus, two Cebus, one mangabey, one spider and six night monkeys, one baboon (Papio hainadryas) and six humans. Each of the lower primates had been used in electrophysiological studies of the optic nerve at the chiasm. All were tranquilized with Semylan and anesthetized with thiopental and nitrous oxide for 4 to 6 hr. No experimental manipulation involving the retina or orbit was done, and every retina appeared normal by fundoscopy at the time of sacrifice of the animals. The human retinas were obtained 4 to 6 hours postmortem from accident victims in whom the eyes were grossly intact. Retina] whole mounts were prepared as described by Ogden et al. 16 The posterior half of the globe was immersed in Tris-buffered normal saline (ph 7.4); retinas were detached, cut in four places to facilitate flattening, mounted like a Maltese cross under a layer of Telfa gauze, and fixed by immersion in 3% glutaraldehyde. Astrocyte morphology was studied in whole

5 Volume 17 Number 6 Primate retinal astrocytes 503 mounts stained with the Colonnier modification of the Golgi-Kopsch procedure, 17 then dehydrated in alcohol, cleared in xylene, and embedded in Epon. Cells were drawn by camera lucida and photographed. Measures of cell size and process length were made directly from the drawings. Measures of astrocyte density were made from Nissl-stained whole mount retinas of three rhesus monkeys; astrocyte nuclei were easily recognized by virtue of their darkly stained, compact, oval appearance and position in the nerve fiber layer, except in the area centralis where ganglion cell density was very high and the fiber layer very thin. These retinas were also cross-sectioned at 50 JX as described by West, 18 for additional counts through the area centralis. Measures of nerve fiber layer thickness were obtained from the retinal cross-sections. The plane of section was parallel to the horizontal meridian so that the data could be related to the position of the fovea. It is estimated that about 1 mm of retina was lost adjacent to the optic disc in most specimens; thus the position of the disc is not precisely represented in these data. Tissue prepared for electron microscopy was fixed by immersion in cold (4 C) 2% glutaraldehyde and 2% paraformaldehyde in phosphate buffer at ph 7.4 for 2 hr, washed in phosphate buffer for 12 hr, post-fixed in 1% osmium tetroxide for 1 hr, and block-stained with uranyl acetate. Thin sections were further stained with uranyl acetate and lead citrate and viewed with a Zeiss 10 electron microscope. Linear retinal shrinkage due to processing was determined as described by Ogden, 19 by measuring the decrease in separation of a line of carefully spaced retinal holes. Shrinkage varied from 8% to 12% in different specimens as found in previous studies. 19 " 20 The data presented, however, are not corrected for shrinkage. Results The following data were obtained from a study of 328 relatively isolated nerve fiber layer astrocytes viewed in Golgi-stained retinal whole mounts. The processes of these cells are located in the nerve fiber layer. They are long and slender, rarely branched, and of relatively uniform thickness, 1 to 2 /A in Golgi preparations (Figs. 1 to 3). The processes remain within the fiber layer throughout their course, and those adjacent to the internal limiting membrane may bear short excrescences which are insinu O z LU WIDTH (mm).6 Fig. 4. Scattergram showing relative length and width of retina covered by the processes of 328 nerve fiber layer astrocytes of seven primate species. Two populations of cells are evident. Those between the lines have processes covering a roughly circular retinal area and are called stellate; those above the upper line are substantially longer than wide and are referred to as elongate. ated among the Miiller cell end feet at the surface of the retina (Fig. 3, A). The astrocyte perikarya are also located in the nerve fiber layer except in those regions such as the perifovea and far periphery where the fiber layer is less than 5 \x thick. Here the perikarya lie within the ganglion cell layer. Most of these cells were easily classified as elongate (44%) or stellate (49%). The elongate cells are characterized by a process pattern with a ratio of length to width of at least 2:1 and averaging 4.5:1 (±1.87 S.D.). The processes, as shown in Figs. 1, A and B, and 3, A and B, are relatively straight and parallel the nerve fiber bundles, between which most lie. None of these cells was seen to make a specialized vascular contact. The stellate

6 504 Ogden Invest. Ophthalmol. Visual Sci. June 1978 A Fig. 5. Photomicrographs of nerve fiber layer astrocytes stained by Golgi impregnation to show the wide variability in perikaryal morphology and numbers of processes emanating from a cell. A, B, C, E, and F, Stellate cells. D and G, Elongate cells. A, B, and C, Rhesus monkey. D, Owl monkey giant astrocyte. E, F, and G, Human. All material retinal flat mounts. (Calibration 10 (JL except E, which is 20 yx.)

7 Volume 17 Number 6 Primate retinal astrocytes Fig. 6. Photomicrographs of retinal whole mounts stained with Golgi impregnation. The meshwork of glial and nerve fibers is well illustrated in these photographs. Note the tendency of the glial perikarya to occur between nerve fiber bundles. A, Mangabey. B, Human. C, Rhesus. D, Aotes. (Calibrations: A, 15 /JL; B and D, 10 fj.; C, 50 ju..)

8 506 Ogden Invest. Ophthalmol. Visual Sci. June 1978 Fig. 7. Diagram of a nerve fiber bundle of the rhesus monkey, traced from a montage of electron micrographs enlarged 30,000x. Glial tissue of both astrocyte and Muller cells is solid black. Of the 508fibersin this bundle, 188 had no glial contact and 52 contacted glia along 10% or less of their circumference. (Calibration 1 fx.) cells (Figs. 2, A and B, and 3, D and F) are very different in appearance. Their processes are shorter and conspicuously cross the nerve fiber bundles, to enclose a roughly circular area. The ratio of process length to width was 1.55:1 (±0.32 S.D.). Most of these cells have vascular attachments, as shown by the arrows in Fig. 2, A, and labeled bv in Fig. 3, D, but these are not specialized and do not at all resemble the "glial end feet" common in astrocytes of the central nervous system and described by Wolter 9 in stellate retinal astrocytes of the inner plexiform layer of the human retina. Only 7% of the cells studied had features of both the stellate and elongate types (Figs. 3, C, and 5, E). These had mostly short radiating processes with a few longer processes passing along the fiber bundles. Two of these intermediate type cells were seen to make vascular contacts. The extent of spread of processes of 305 of the 328 cells is shown in a scattergram in Fig. 4, where the cells are plotted as a function of length and width. The processes of cells above the upper diagonal line paralleled the nerve fibers. Those of cells below the line prominently crossed the nerve fibers. Two populations are clearly shown; process extent of the elongate cells was about 750 /JL and of the stellate cells about 400 fx. The 23 cells not included had one or more long processes paralleling and many shorter processes crossing the fiber bundles. Since the ratio of length to width for the stellate cells was greater than 1.0, they actually covered an oval area of retina. There was considerable variation in the number of primary processes (three to nine) of the fibrous astrocytes (Fig. 5). Branching of processes was uncommon and occurred, if at all, close to the cell soma (Fig. 5, B, D,, and F). The somata were usually ovoidal, although hour glass shapes (Fig. 5, C and E) were common and many with bizarre shapes were found (Fig. 5, B and G). Cell body di-

9 Volume 17 Number 6 Primate retinal astrocytes 507 Fig. 8. Topographic representation of the density of nonvascular nuclei in the nerve fiber layer of the left eye of a rhesus monkey. Counts were made from a Nissl-stained retinal whole mount at the position of the small circles and at many locations between them. The isodensity lines are drawn by eye and are approximate. Density is greatest at the disc D but is high over the arcuate fibers. (Calibration 2 mm; densities xloo/mm 2.) Cross-sections were made in the regions of the large circles to measure nerve fiber layer thickness which was, from left to right, 17 fx (1806 cells/mm 2 ), 53 fx (3187 cells/mm 2 ), 26 fx (2351 cells/mm 2 ), 12 fx (1432 cells/mm 2 ), and 6 (x (860 cells/mm 2 ). mensions varied between 8 and 16 fx (major axis) and 6 to 8 /u. (minor axis). These general features of the fiber layer fibrous astrocytes were found in each of the retinas studied; species differences were not evident. The only exception was found in the far periphery of the owl monkey. Two giant astrocytes were stained which had sparse processes and were substantially larger than cells of the more central retina of Aotes or of the other species studied (Fig. 5, D). The nerve fiber layer is a dense meshwork of nerve and glial fibers (Fig. 6). These fibers are largely parallel, and the glial somata tend to lie between the nerve fiber bundles (Fig. 6, A, C, and D); their processes join and parallel the bundles in the case of the elongate cells or prominently cross the bundles in the case of the stellate cells. It is remarkable that electron microscopy of the nerve fiber layer shows few glial processes within the nerve fiber bundles. The diagram shown in Fig. 7 was prepared by tracing the outline of the nerve fibers in a typical fiber bundle. Glial processes are heavily shaded. Most are large and irregular in shape and are assumed to be processes of Muller cells; only six profiles are small and round enough to suggest that they may be astrocyte processes. It appears thus that most of the parallel glial fibers must pass along the surface of, or between, the nerve fiber bundles where they are very difficult to distinguish from the processes of Muller cells Of the 508 nerve fibers in

10 508 Ogden Invest. Ophthalmol. Visual Sci. June 1978 this bundle, 188 had no glial contact, and 52 contacted glia along 10% or less of their circumference. This study of Golgi-stained material revealed astrocytes throughout the retina except in the region of the fovea, contrary to the claim usually published that fiber layer astrocytes occur only near the optic disc. 1 ' 4 Therefore it was of interest to determine the distribution and density of these cells across the retina. Golgi stains are not suitable for quantitative studies of cell populations, so that it was necessary to obtain these measures from Nissl-stained material. Nisslstained astrocyte nuclei are dark, compact, and oval and easily distinguished from ganglion cell nuclei except in the area centralis where the latter may have a similar appearance. The primate retina contains large numbers of perivascular glia in intimate relationship to blood vessels. 7 The morphology of this glia is very different from that of the nerve fiber layer astrocytes with which this study is concerned, and care was taken not to include nuclei associated with blood vessels in the nerve fiber layer nuclear counts. For the study shown in Fig. 8, astrocyte nuclear counts were made at 1000X magnification with an oil-immersion lens at 0.25 to 1.0 mm intervals across the retina; only nuclei which were within the fiber layer were included. Astrocyte density was highest (3500/mm 2 ) adjacent to the optic disc. Ridges of high density extended across the temporal retina above and below the area centralis, corresponding to the position of the arcuate nerve fibers. Density fell rapidly toward the area centralis and gradually in other directions from the optic disc. No astrocytes were found in the fovea or perifovea, as Ikui et al. 15 also observed, but it is possible that some were not identified because of the high ganglion cell density in that region. The accuracy of this map is greatest where astrocyte density is above 500 cells/mm 2, since the fiber layer is thicker and the difference in retinal level of the astrocyte and ganglion cell nuclei is appreciable. Studies such as that shown in Fig. 8, done on three retinas of three rhesus monkeys with comparable results, suggest a rough proportionality of astrocyte density and nerve fiber layer thickness. This relationship was verified in the case of the retina shown in Fig. 8, which was re-embedded and crosssectioned in five areas along the horizontal meridian 1 mm above the fovea (large circles in Fig. 8). Measures of fiber layer thickness, obtained from these sections, showed a reasonable correspondence of fiber layer thickness and astrocyte density at each location (see legend to Fig. 8), averaging 104 nuclei/mm 2 IjX. Discussion This study has demonstrated typical fibrous astrocytes in the nerve fiber layer of several primates, including man, and confirms the brief observations of a number of earlier neurohistologists. 1 In modern times, Wolter 4 " 9 has presented a series of studies of retinal glia revealed by the silver carbonate method. This special glial stain did not impregnate any typical astrocytes in the nerve fiber layer, although such cells were found deeper within the retina. In the nerve fiber layer, Wolter 4 found many spindleshaped cells with two elongated thin processes which ran along the nerve fiber bundles. He was struck by the resemblance of these cells to the Schwann cells of sympathetic nerves (lemmocytes) and reasoned that the unmyelinated retinal nerve fibers should be accompanied by a "sheath" cell. Wolter used light rather than electron microscopy and thus did not realize that the retinal nerve fiber bundles contain little glial tissue and are much larger than the glial cells. He assumed that the spindle-shaped cells he observed were combined to form "fibers of Remak" 22 in which the nerve fibers were embedded, as is the case in peripheral nerves, but ultrastructural studies 14 show that this is most unlikely. Lessell and Kuwabara 3 confirmed Wolter's observation of "lemmocyte-appearing" cells in the nerve fiber layer of human retinas stained with the silver carbonate method. These cells were not described by the early neurohistologists and have not appeared in the material of the present study. Thus their staining seems a peculiarity of the silver carbonate method. Silver im-

11 Volume 17 Number 6 Primate retinal astrocytes 509 pregnations are, of course, notoriously capricious, and it is not surprising that typical fibrous astrocytes should be revealed by silver chromate and atypical glial cells by silver carbonate stains. In this study, an occasional elongated fibrous astrocyte was seen which was only partially impregnated. Such cells could resemble the cells described by Wolter as lemmocytes, so it is also possible that the cells so designated by him were actually partially impregnated astrocytes. Regardless of their nature, the choice of the term lemmocyte to describe these cells was unfortunate, particularly since it has been accepted in the literature. 3 ' 21 Lemmocytes are Schwann cells associated with unmyelinated nerve fibers in peripheral nerve 23 and have neural crest origin. Retinal astrocytes probably differentiate from glioblasts, as do optic nerve astrocytes and the other retinal glia. They certainly do not represent migrated neural crest material. The present study confirms and extends the preliminary work of Ikui et al., 15 who counted the numbers of glial nuclei in the ganglion cell and nerve fiber layers of one horizontal meridian through a human retina. They counted a total of 369 nuclei in their series of 1 /x thick sections along the meridian. Over half the total were found within 6 mm of the optic disc, and none were found in the fovea, through which their meridian passed. They did not report nerve fiber layer thickness, but their results seem comparable to those reported here. The accuracy with which the identification of astrocyte nuclei can be made varies across the retina. Where the nerve fiber layer is thickest, accuracy is highest. Where the nerve fiber layer is very thin, astrocyte nuclei may lie adjacent to ganglion cell nuclei, even though their processes extend down into the nerve fiber layer (Fig. 4, A). The identity of some of these nuclei was ambiguous, and they were not counted, although they probably should be considered to be nerve fiber layer astrocytes. Difficulty also increased with the thickness of the ganglion cell layer. Thus measures from the parafovea and the far periphery are less accurate than those from other areas of the retina. The data presented here suggest the presence of two classes of fiber layer astrocytes: an elongated type whose processes run for long distances among the nerve fiber bundles but have no obvious vascular contacts and a stellate type whose processes are shorter, conspicuously cross the fiber bundles, and make casual vascular contacts. These cells seem ill-suited for several of the various functions suggested for glia. 10 Ul 13 ' Their axonal contact is probably too limited to present an effective sump for extracellular ions released during axonal activity, a function much more suitably ascribed to the ubiquitous Muller cell. Also, the elongate cells certainly, and the stellate cells probably, have insufficient vascular contacts to function efficiently as a metabolite pathway between vessels and neurites. This function seems more appropriate for the typical astrocytes with massive vascular end feet found in the central nervous system and within the retina in the ganglion cell and inner plexiform layers, but not in the nerve fiber layer. Certainly the network of astrocyte fibers paralleling and crossing the nerve fiber layer suggests a structural supportive function for these cells in the normal retina. In a pathological retina, of course, all types of astrocytes are probably mobilized and participate in the formation of retinal gliosis and preretinal proliferative responses to injury. 24 With the exception of the giant astrocytes found in the far periphery of the retina of Aotes, the morphology and dimensions of the retinal astrocytes were similar in all the species studied. This suggests that any of these subhuman primates may be used as a suitable animal model for studies of glial function that are to be applied to the human. The suitability of Aotes, however, as an animal model for glial studies may be questioned, since it may have developed glial specialization not present in other primates or man. REFERENCES 1. Polyak, S. L.: The Retina, Chicago, 1941, Chicago University Press. 2. Scharenberg, K., and Zenian, VV.: Zur Leistungsfahigkeit und zur technik der hortegaschen silber-

12 510 Ogden Invest. Ophthalmol. Visual Sci. June 1978 karbonat Methoden, Arch. Psychiatr. 188:430, Lessell, S., and Kuwabara, T.: Retinal neuroglia, Arch Ophthalmol. 70:671, Wolter, J. R.: The cells of Remak and the astroglia of the normal human retina, Arch. Ophthalmol. 53: 832, Wolter, J. R.: The astroglia of the human retina and other glial elements of the retina under normal and pathological conditions, Am. J. Ophthalmol. 40:88, Wolter, J. R.: Uber besondere Astroglia an der Innenflasche der Retina, Klin. Monatsbl. Augenheilkd. 129:224, Wolter, J. R.: Perivascular glia of the blood vessels of the human retina, Am. J. Ophthalmol. 44:766, Wolter, J. R.: Glia of the human retina, Am. J. Ophthalmol. 48:370, Wolter, J. R.: Silver carbonate techniques for the demonstration of ocular histology. In Smelser, G., editor: The Structure of the Eye. New York, 1961, Academic Press, Inc. 10. Somjen, G. G.: Electrophysiology of neuroglia, Annu. Rev. Physiol. 37:163, Watson, W. E.: Physiology of neuroglia, Physiol. Rev. 54:245, Virchow, R.: Cellularpathologie in irhe Begrundung auf physiologische und pathologische Gevvebelehe, ed. 4, Berlin, 1846, August Hirschwald. 13. Kuffler, S. W., and Nicholls, J. G.: From Neuron to Brain, Sunderland, Mass., 1976, Sinauer Associates, Publisher. 14. Cohen, A. I.: Electron microscopic observations of the internal limiting membrane and optic fiber layer of the retina of the rhesus monkey (M. mulatto), Am. J. Anat. 108:179, Ikui, H., Uga, S., and Kohno, T.: Electron microscope study of astrocytes in the human retina using ruthenium red, Ophthalmol. Res. 8:100, Ogden, T. E., Green, J. D., and Peterson, R. G.: Graded impregnation of nervous tissue stained by the Golgi procedure, Stain Technol. 49:81, Colonnier, M.: The tangential organization of the visual cortex, J. Anat. 98:327, West, R. W.: Superficial warming of epoxy blocks for cutting of ^m sections to be resectioned in the nm range, Stain Technol. 47:201, Ogden, T. E.: The receptor mosaic of Aotes trivirgatus: distribution of rods and cones, J. Comp. Neurol. 163:193, Steinberg, R. H., Reid, M., and Lacey, P. L.: The distribution of rods and cones in the retina of the cat (Felis domesticus), J. Comp. Neurol. 148:229, Hogan, M. J., and Feeney, L.: The ultrastructure of retinal vessels. III. Vascular-glial relationships, J. Ultrastruct. Res. 9:47, Remak, R.: Ueber die Wiedererzeugung von nerven Fasern, Virchows Arch. Pathol. Anat. 23:441, Nageotte, J.: Sheaths of the peripheral nerves. Nerve degeneration and regeneration. In Penfield, W., editor: Cytology and Cellular Pathology of the Nervous System, New York, 1932, Hoeber, Inc. 24. Laqua, H., and Machemer, R.: Glial cell proliferation in retinal detachment (massive periretinal proliferation), Am. J. Ophthalmol. 80:602, Copyright information The appearance of a code at the bottom of the first page of an original article in this journal indicates the copyright owner's consent that copies of the article may be made for personal or internal use, or for the personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc., P.O. Box 765, Schenectady, N.Y , /518/ , for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale.