Ординатура / Офтальмология / Английские материалы / Aging and Age Related Ocular Diseases_Lutjen-Drecoll_2000

Table 5. Quantitative measurement (ng/mg protein) of ·B-crystallin in the retrolaminar optic nerve and heart muscle of human donors by dot blotting

Discussion

In the present study, the distribution of ·B-crystallin staining was investigated in the heart muscle, known to show constitutively high amounts of the protein [25], in nonlenticular eye tissues with variations in staining intensity of different cell populations [23, 26] and in a number of glands not studied before. In none of these tissues were significant age-related changes in ·B-crystallin staining found. Using dot blot methods as a semiquantitative evaluation method, no increase in ·B-crystallin was detected in rat heart muscle, but there was more ·B-crystallin in 5- to 6-year-old than in 2-year-old cows. A tendency towards an increase in ·B-crystallin mRNA with age was also seen in rat optic nerve tissues. Age-related increases in ·B-crystallin expression in rat heart muscle and kidney have been described by Bhat and Nagineni [8], Kato et al. [25] and Iwaki et al. [27]. However, in these studies only developmental stages up to 20 days of life were investigated. Katoh-Semba and Kato [21] found a tendency towards increased amounts of ·B-crystallin in mouse brain tissue with increasing age. In that study, the oldest age group of normal animals was 8 months. Unfortunately, we were not able to obtain tissues from various age groups of pigs and cows, but the values obtained from young and middle-aged porcine and bovine optic nerves showed no differences in ·B-crystallin expression. In hu-

man tissues derived from middle-aged and old donors, there were no age-related differences in the protein expression either, but the data are of limited value as the material was derived from donors with different postmortem times.

On the other hand, in all species studied pronounced interindividual differences were found. If a molecule functions as a stress protein, individual differences might well reflect differences in individual stress situations.

Immunohistochemical staining for ·B-crystallin also revealed differences in the distribution and amount of the protein in the five species investigated. These differences may be due to differences in the specificity of the antibody. On the other hand, the DNA sequence of ·B-crys- tallin shows great homologies between all species investigated [28, 29]. The same primer could be used for the detection of ·B-crystallin mRNA in the various species so that the staining differences might in fact be due to species differences in ·B-crystallin expression. In the eye, positive staining of the ciliary muscle and localized staining of the outer part of the trabecular meshwork were only seen in primates [23, 26]. It is well established that in the mammalian species there is a well-developed fovea centralis and accommodation system only in higher monkeys and human eyes [for a review, see 30]. It is therefore tempting to speculate that the characteristic staining pattern in the primate ciliary muscle and outer part of the trabecular meshwork with its connections to the ciliary muscle tendons might be related to mechanical stress during accommodation [31].

Species differences were also seen in semiquantitative evaluation of the protein using the dot blot method. The values in rat, porcine and bovine heart muscles were in the same range as those described in the literature [25, 32]. In all mouse tissues evaluated there was less protein expression than in the other species. In human tissues, the values measured were the same as those found in porcine and bovine tissues even if the material was obtained after longer postmortem times. It is possible that the values in the living tissue are much higher. Differences in the absolute life span of the species might be another factor involved in the observed species differences. However, this hypothesis cannot be proven by our studies due to lack of material from species with longer life spans.

In summary, our studies show that an increase in ·B- crystallin with age may occur but is not a general phenomenon in tissues constitutively expressing this protein.

References

1Bloemendal H: Molecular and Cell Biology of the Eye Lens. New York, Wiley, 1981.

2Ingolia TD, Craig EA: Four small Drosophila heat shock proteins are related to each other and to mammalian ·-crystallin. Proc Natl Acad Sci USA 1982;79:2360±2364.

3Klemenz R, Frohli E, Aoyama A, Hoffmann S, Simpson RJ, Moritz RL, Schafer R: ·B-Crys- tallin accumulation is a specific response to Ha-ras and v-mos oncogene expression in mouse NIH 3T3 fibroblasts. Mol Cell Biol 1991;11:803±812.

4Dasgupta S, Hohman TC, Carper D: Hypertonic stress induces ·B-crystallin expression. Exp Eye Res 1992;54:461±470.

5Horwitz J: Alpha-crystallin can function as a molecular chaperone. Proc Natl Acad Sci USA 1992;89:10449±10453.

6Piatigorsky J: The twelfth Frederick H. Verhoeff lecture: Gene sharing in the visual system. Trans Am Ophthalmol Soc 1993;91:283± 297.

7Longoni S, Lattonen S, Bullock G, Chiesi M: Cardiac alpha-crystallin. II. Intracellular localization. Mol Cell Biochem 1990;97:121±128.

8Bhat SP, Nagineni CN: ·B-Subunit of lens-spe- cific protein ·-crystallin is present in other ocular and non-ocular tissues. Biochem Biophys Res Commun 1989;158:319±325.

9Leach IH, Tsang ML, Church RJ, Lowe J: ·B- Crystallin in the normal human myocardium and cardiac conducting system. J Pathol 1994; 173:255±260.

10Atomi Y, Yamada S, Strohman R, Nonomura

Y:·B-Crystallin in skeletal muscle: Purification and localization. J Biochem 1991;110: 812±822.

11Iwaki T, Iwaki A, Goldman JE: ·B-crystallin in oxidative muscle fibers and its accumulation in ragged-red fibers: A comparative immunohistochemical and histochemical study in human skeletal muscle. Acta Neuropathol 1992;85: 475±480.

12Scotting P, McDermott H, Mayer RJ: Ubiqui- tin-protein conjugates and ·B-crystallin are selectively present in cells undergoing major cytomorphological reorganization in early chicken embryos. FEBS Lett 1991;285:75±79.

13Quinlan RA: Assembly of intermediate filament proteins is modulated by the chaperone activity of alpha-crystallins. Invest Ophthalmol Vis Sci 1993;34:989.

14Iwaki T, Kume-Iwaki A, Liem RKH, Goldman JE: ·B-Crystallin is expressed in non-lenticular tissues and accumulates in Alexander's disease brain. Cell 1989;57:71±78.

15Head MW, Corbin E, Goldman JE: Overexpression and abnormal modification of the stress proteins ·B-crystallin and HSP27 in Alexander disease. Am J Pathol 1993;143:1743± 1753.

16 Iwaki T, Tateishi J: Immunohistochemical demonstration of alphaB-crystallin in hamartomas of tuberous sclerosis. Am J Pathol 1991; 139:1303±1308.

17Iwaki A, Iwaki T, Goldman JE, Ogomori K, Tateishi J, Sakaki Y: Accumulation of ·B-crys- tallin in brains of patients with Alexander's disease is not due to an abnormality of the 5)- flanking and coding sequence of the genomic DNA. Neurosci Lett 1992;140:89±92.

18Renkawek K, deJong WW, Merck KB, Frenken CWGM, vanWorkum FPA, Bosman GJCGM: ·B-Crystallin is present in reactive glia in Creutzfeldt-Jakob disease. Acta Neuropathol 1992;83:324±327.

19Renkawek K, Voorter CEM, Bosman GJCGM, vanWorkum FPA, deJong WW: Expression of ·B-crystallin in Alzheimer's disease. Acta Neuropathol 1994;87:155±160.

20Murano S, Thweatt R, Reis RJS, Jones RA, Moerman EJ, Goldstein S: Diverse gene sequences are overexpressed in Werner syndrome fibroblasts undergoing premature replicative senescence. Mol Cell Biol 1991;11: 3905±3914.

21Katoh-Semba R, Kato K: Age-related changes in levels of the ß-subunit of nerve growth factor in selected regions of the brain: Comparison between senescence-accelerated (SAM-P8) and senescence-resistant (SAM-R1) mice. Neurosci Res 1994;20:251±256.

22Duhamel RC, Johnson DA: Use of nonfat dry milk to block nonspecific nuclear and membrane staining by avidin conjugates. J Histochem Cytochem 1985;33:711±714.

23Flügel C, Liebe S, Voorter C, Bloemendal H, Lütjen-Drecoll E: Distribution of ·B-crystallin in the anterior segment of primate and bovine eyes. Curr Eye Res 1993;12:871±876.

24Welge-Lüssen U, Eichhorn M, Bloemendal H, Lütjen-Drecoll E: Classification of human scleral spur cells in monolayer culture. Eur J Cell Biol 1998;75:78±84.

25Kato K, Shinohara H, Kurobe N, Inaguma Y, Shimizu K, Ohshima K: Tissue distribution and developmental profiles of immunoreactive ·B-crystallin in the rat determined with a sensitive immunoassay system. Biochem Biophys Acta 1991;1074:201±208.

26Siegner A, May CA, Welge-Lüssen U, Bloemendal H, Lütjen-Drecoll E: ·B-Crystallin in the primate ciliary muscle and trabecular meshwork. Eur J Cell Biol 1996;71:165±169.

27Iwaki T, Iwaki A, Liem RKH, Goldman JE: Expression of ·B-crystallin in the developing rat kidney. Kidney Int 1991;40:52±56.

28 Frederiksen PH, Dubin RA, Piatigorsky J: Structure and alternate tissue-preferred transcription of the murine ·B-crystallin gene. Nucleic Acids Res 1994;22:5686±5694.

29Liao JH, Hung CC, Lee JS, Wu SH, Chiou SH: Characterization, cloning, and expression of porcine ·B-crystallin. Biochem Biophys Res Commun 1998;244:131±137.

30Rohen JW: The evolution of the primate eye in relation to the problem of glaucoma; in LütjenDrecoll E (ed): Basic Aspects of Glaucoma Research. Stuttgart, Schattauer, 1982, pp 3±33.

31Welge-Lüssen U, May CA, Eichhorn M, Bloemendal H, Lütjen-Drecoll E: ·B-Crystallin in the trabecular meshwork is inducible by transforming growth factor-ß. Invest Ophthalmol Vis Sci 1999;40:2235±2241.

32Inaguma Y, Hasegawa K, Goto S, Ito H, Kato K: Induction of the synthesis of hsp27 and ·B- crystallin in tissues of heat-stressed rats and its suppression by ethanol or an ·1-adrenergic antagonist. J Biochem 1995;117:1238±1243.

33Iwaki T, Kume-Iwaki A, Goldman JE: Cellular distribution of ·B-crystallin in non-lenticular tissues. J Histochem Cytochem 1990;38:31± 39.

34Lowe J, McDermott H, Pike I, Spendlove I, Landon M, Mayer RJ: ·B-Crystallin expression in non-lenticular tissues and selective presence in ubiquinated inclusion bodies in human disease. J Pathol 1992;166:61±68.

35May CA, Arnold B, Welge-Lüssen U, Arnold W, Bloemendal H, Lütjen-Drecoll E: ·B-Crys- tallin in the mammalian inner ear. Otorhinolaryngology 1998;60:121±125.

36Dubin RA, Wawrousek EF, Piatigorsky J: Expression of the murine ·B-crystallin gene is not restricted to the lens. Mol Cell Biol 1989;9: 1083±1091.

37Gopal-Srivastava R, Haynes JI, Piatigorsky J: Regulation of the murine ·B-crystallin/small heat shock protein gene in cardiac muscle. Mol Cell Biol 1995;15:7081±7090.

38Bhat SP, Horwitz J, Srinivasan A, Ding L-L: ·B-Crystallin exists as an independent protein in the heart and in the lens. Eur J Biochem 1991;102:775±781.

39Iwaki T, Wisniewski T, Iwaki A, Corbin E, Tomokane N, Tateishi J, Goldman JE: Accumulation of ·B-crystallin in central nervous system glia and neurons in pathologic conditions. Am J Pathol 1992;140:345±356.

40Pinder SE, Balsitis M, Ellis IO, Landon M, Mayer RJ, Lowe J: The expression of ·B-crys- tallin in epithelial tumours: A useful tumour marker? J Pathol 1994;174:209±215.

41Nishikawa S, Ishiguro S, Kato K, Tamai M: A transient expression of ·B-crystallin in the developing rat retinal pigment epithelium. Invest Ophthalmol Vis Sci 1994;35:4159±4164.

Key Words

Optic nerve W Glaucoma W ·B-crystallin W Glial fibrillary acidic protein W Vessels

Abstract

Vascular and glial changes of the retrolaminar optic nerve were studied in monkey eyes with increased intraocular pressure (IOP) from 1 to 4 years and with different stages of optic nerve atrophy. In histological crosssections of retrolaminar optic nerves of 11 rhesus and 6 cynomolgus monkeys the entire area, number of axons and vessels and area of pial septa were quantitated and three different kinds of nerve degeneration classified. Ultrathin sections of these different stages were performed and the number of open and occluded vessels was determined. In addition, in cynomolgus monkey optic nerves immunohistochemical staining for ·B-crys- tallin, glial fibrillary acidic protein (GFAP) and vimentin was performed. Even in animals with the same duration of glaucoma and comparable mean IOP values the axon degeneration varied considerably. Independently of axon loss the number of capillaries in the rhesus mon-

keys remained constant, whereas there was a slight decrease in the cynomolgus monkeys. Some of the vessels, especially in the most severely damaged regions, were occluded. The density of glial cells increased whereas the total number remained nearly constant. In control sections all astrocytes stained for GFAP and ·B- crystallin. In the glaucomatous optic nerves the density of ·B-crystallin- and GFAP-positive cells was significantly increased. The vascular reaction in the retrolaminar glaucomatous optic nerves differs from that described in the prelaminar region. We assume that in the postlaminar region in areas with diminished nutritional needs vessels occlude and finally degenerate.

Copyright © 2000 S. Karger AG, Basel

Introduction

It is generally assumed that the axonal insult in glaucomatous optic neuropathy is located at or near the scleral lamina cribrosa, a tissue presumed to be easily affected by elevated intraocular pressure (IOP). Histopathological investigations of primate eyes with acute elevation of IOP to

ABC

Fax + 41 61 306 12 34 E-Mail karger@karger.ch www.karger.com

very high levels have shown abnormal accumulation of cell organelles and localized swellings of optic nerve fiber axons within the scleral lamina cribrosa, reflecting blocking of axoplasmic flow [1±5]. In monkey eyes with laserinduced chronic glaucoma, it has been reported that in the prelaminar region of the optic nerve the amount of axonal tissue decreases and is partly replaced by glial tissue [5, 6]. Morphologically no changes in the number of capillaries in the optic nerve head were found in eyes with elevated IOP for up to 4 months [5]. In later stages of optic nerve atrophy, the proportion of capillaries in the optic disk remained constant [7].

Detailed descriptions of morphological changes in the retrolaminar region of glaucomatous optic nerves are rare. In owl monkey eyes, in which an acute high IOP was induced by injection of ·-chymotrypsin into the posterior chamber and maintained for up to 1 week, axonal degeneration and reactive axonal enlargement as well as degenerative changes in oligodendrocytes, an increase in phagocytosing cells and an activation of astrocytes have been described [1]. In monkey eyes with laser-induced chronic ocular hypertension, histological studies revealed that ± as in glaucomatous human eyes ± axons in the upper and lower temporal quadrant start to atrophy [4, 8]. In a previous study, we found that in monkey eyes with laserinduced chronic glaucoma a significant nerve fiber loss developed in the retrolaminar region of the optic nerve but the nerve fiber counts varied greatly between individual eyes [9]. The reason for the interindividual differences is unclear. As ocular hypertension lasted more than 1 year in all eyes and the mean IOP was almost similar, differences in nerve fiber loss could be due to any of the following: differences in variations of IOP not measured, varying duration and IOP in different eyes, structural differences in the lamina cribrosa, differences in the susceptibility of the vasculature or differences in reactivity of the glial cell system.

Regeneration studies of the optic nerve in the retrolaminar region in fish and rat eyes have revealed that the glial environment which surrounds an axon influences its capacity to regenerate after injury [10]. Factors derived from mature oligodendrocytes [11±13] and extracellular matrix produced by astrocytes have been discussed in this context [14±17]. Whether these factors that favor regeneration might also enhance the ability to survive the initial insults of damage to the nerve fibers has not been studied. Since myelinization of the optic nerve fibers starts only in the retrolaminar region of the optic nerve and the proportion of oligodendrocytes and astrocytes also changes markedly in this region, the mechanism of optic nerve

fiber damage in the retrolaminar region could be different from that seen in the prelaminar portion of the optic nerve.

In the present study we investigated the retrolaminar part of the optic nerve in experimentally induced chronic high-pressure glaucoma in monkey eyes, trying to correlate ultrastructural changes, immunohistochemical changes and vascular changes with the individual degree of glaucomatous optic neuropathy.

Materials and Methods

We performed this study in 24 optic nerves ± 12 from rhesus monkeys (Macaca mulatta) and 12 from cynomolgus monkeys (Macaca fascicularis) as indicated in table 1. Eight rhesus monkey optic nerves and 6 from cynomolgus monkeys were from eyes with experimental chronic high-pressure glaucoma, induced by the method described in a previous paper [9]. The remaining eyes (4 of rhesus monkeys and 6 of cynomolgus monkeys) were normal, healthy fellow eyes of these animals, with normal IOP, and these acted as control. All experiments were performed in accordance with the ARVO Resolution for the Use of Animals in Ophthalmic and Vision Research and the local university rules for experiments on primates.

Tissue Samples from Iowa City (S.S. Hayreh)

The data and clinical course of 8 rhesus monkey eyes with chronic elevated IOP have been described previously [9]. The duration of the elevated IOP varied between 14 and 43 months, and the IOP values ranged from 22 to 42 mm Hg in the treated eyes compared to 17± 19 mm Hg in the normal fellow eyes (table 1). Before enucleation, the animals were perfusion fixed with 4% paraformaldehyde transcardially as described previously [9]. The optic nerve was cut directly behind the lamina cribrosa and at the level of the entrance of the central retinal artery into the optic nerve. This cylindrically shaped optic nerve specimen was then transversally cut into three to four 1- to 2-mm-wide disks and postfixed in Ito's solution [18] for at least 1 week.

Tissue Samples from Uppsala (A. Alm)

The data and clinical course of 6 of the cynomolgus monkey eyes have been described previously [19]. The duration of glaucoma was 4 years. In these eyes IOP measurements were performed only in the first 9 months after treatment and at the end of the experiment. The mean IOP ranged between 29 and 52 mm Hg after treatment and between 7 and 24 mm Hg at the time of sacrifice.

As in these animals microspheres had been injected for measuring blood flow, the animals were not perfusion fixed before enucleation [19]. After enucleation the retrolaminar optic nerve was divided into 2 pieces. The portion between lamina cribrosa and entrance of the central retinal artery was placed in 4% paraformaldehyde for immunohistochemical investigations, the optic nerve posterior to the entering central retinal artery was immersion fixed in Ito's solution for lightand electron-microscopic examination.

Histological and Electron-Microscopic Investigations

Thin slices of all Ito-fixed specimens were postfixed in OsO4, dehydrated in an ascending series of alcohol and acetone, and

embedded in Epon. Semithin sections were stained with toluidine blue and examined by light microscopy.

The entire area of the optic nerve was evaluated using a Quantimet 500 computer (Leica, Cambridge, UK). To determine the total number of axons, each specimen was divided into 4 sectors and 2 regions. In each of the 8 fields an area 1,000 Ìm2 in size was selected at random, the number of myelinated axons determined and converted into the number of the corresponding area. Three different grades of axonal loss were defined if there was a diffuse nerve fiber loss: mild (more than 800,000 nerve fibers; grade 1), moderate (100,000±800,000 nerve fibers; grade 2) and severe axonal loss (less than 100,000 nerve fibers; grade 3). In optic nerve cross-sections with focal nerve fiber loss, grading was performed for each affected area separately: an area with mild damage histologically appeared nearly unchanged; areas with severe damage were characterized by almost complete loss of nerve fibers which were replaced by glial cells. In areas with moderate damage there were still some nerve fibers present. The classification was confirmed by four of the investigators (N.F., M.F., C.A.M., E.L.D.) independently. The `total amount' of glial cells was calculated by multiplying the glial cell density (number of glial cell nuclei, counted in a defined measuring field of 0.115 mm2) with the entire area of nervous tissue in the cross-section.

The areas occupied by nervous tissue and pial septa were measured with the working plate of the Quantimet computer. The pial septum area was calculated as percentage of the total optic nerve cross-section area minus the complete area occupied by neuronal tissue.

The total number of blood vessels was counted on the optic nerve cross-sections with a magnification of !400 and the location of each vessel was recorded on a schematic drawing. Vessel density was calculated by dividing the total number of vessels through the entire cross-section area of the optic nerve.

From optic nerves with different forms of degeneration (grades 1±3 as defined above) within the cross-section, at least two ultrathin sections of each of the different affected regions were examined by transmission electron microscopy. The nuclei of astrocytes, oligodendrocytes and microglial cells (cell types identified according to the descriptions in Kettenmann and Ransom [20]) were counted in a single measuring field of 0.028 mm2, representing approximately 2±5% of the total neural area. Twenty capillaries within the pial septa were selected at random to determine the capillary occlusion rate. Capillaries with a slit-like lumen or no lumen were rated as being occluded.

Immunohistochemistry

For immunohistochemical staining, antibodies against glial fibrillary acidic protein (GFAP), vimentin and ·B-crystallin were used. GFAP was selected as a general marker for astrocytes in the optic nerve [21, 22]. Vimentin, as a general marker of mesoderm-derived tissue, was selected to study the changes of vimentin-positive glial cells such as activated microglial cells and a subgroup of astrocytes [23, 24]. ·B-Crystallin, a member of the small heat shock proteins with chaperone properties, has been demonstrated in oligodendrocytes of brain tissues [25]. Astrocytes in the brain only occasionally

Table 2. Vessel occlusion rate, mean area of pial septa, mean glial density, number of oligodendrocytes, astrocytes and microglial cells in a measuring field of 0.28 mm2 in relation to different stages of glaucomatous optic neuropathy

Figures in parentheses indicate minimum-maximum. Optic nerve damage; 0 = normal appearance (control sections); 1 = mild; 2 = moderate; 3 = severe degeneration of myelinated nerve fibers; n = number of areas used for quantitative measurements.

show expression of ·B-crystallin [26, 27]; however, an increase in ·B-crystallin in astrocytes has been described under pathological conditions [26, 28, 29]. As an increase in ·B-crystallin was found in the trabecular meshwork of human glaucomatous eyes [30], staining for this protein was included to investigate whether ·B-crystallin is also increased in glial cells of optic nerves with glaucomatous optic neuropathy.

The paraformaldehyde-fixed specimens were rinsed in Tris-buf- fered saline (TBS, pH 7.2±7.4) and cryoprotected in TBS containing 20% sucrose. Frozen sections were cut at a thickness of 14 Ìm and placed on poly-L-lysine-coated glass slides. After preincubation with Blotto's dry milk solution [31] the sections were incubated with the primary antibody overnight at room temperature. As primary antibody, we used rabbit anti-·B-crystallin (provided by H. Bloemendal, Nijmegen, the Netherlands; dilution 1:400), rabbit anti-GFAP (Bio Genex Laboratories, San Ramon, Calif., USA; dilution 1:200) and mouse antivimentin antibody (Dako, Glostrup, Denmark; dilution 1:50).

After washing in TBS the sections were covered with a Cy3-fluo- rescein-conjugated secondary antibody (Dianova, Hamburg, Germany; dilution 1:500) for 2 h at room temperature. After washing again, the sections were mounted with Kaiser's glycerin jelly and viewed with a Leitz Aristoplan microscope (Wetzlar, Germany). Negative control experiments were performed using TBS or rabbit preimmune serum, substituted for the primary antibody.

Results

Optic Nerves of Normal Control Eyes

All retrolaminar optic nerve sections of the untreated fellow eyes appeared normal with large bundles of myelinated nerve fibers, surrounded by delicate pial septa. The optic nerve cross-section area, axon count and area covered by pial septa are shown in tables 1 and 2.

Immunohistochemical staining for ·B-crystallin and GFAP revealed an almost identical staining pattern. The periphery of the nerve fiber bundles adjacent to the pial septa stained intensely for both antibodies. A delicate network of stained processes was evenly distributed throughout the nerve, whereas cells surrounding the single axons were unstained (fig. 1a, c). In contrast, vimentin staining was seen only in single star-shaped cells within the nerve fiber bundles and in the fibroblasts and vascular cells of the pial septa (fig. 1e). Stained glial cells were found in all nerve fiber bundles but were located mainly centrally. Several fine processes reached towards the marginal zones of the nerve fiber bundles. The periphery of the bundles and the cells surrounding the nerve fibers were unstained. The number of vimentin-positive glial cells in the normal optic nerve varied in the individual monkeys.

Optic Nerves of Eyes with Chronically Elevated IOP

In the glaucoma eyes, the severity of optic nerve fiber degeneration varied from nearly normal to complete loss of nerve fibers (fig. 2). In general, diffuse damage to axons was seen throughout the optic nerve in all glaucoma cases, while several optic nerves revealed prominent axonal loss only in some quadrants. The entire cross-section area was significantly smaller in the glaucoma eyes, as was the number of axons (table 1).

Based on the severity of axonal loss, glaucomatous neuropathy was graded into mild, moderate and severe stages.

Mild Glaucomatous Optic Neuropathy. This group included cases No. 20 for the entire optic nerve and No. 5 and 6 for parts of the optic nerve cross-section.

On light microscopy, the nerve fiber bundle structure appeared nearly normal, but with the exception of case No. 20, the number of axons was decreased (table 1). On electron microscopy, in some regions of the nerve the myelin sheath was disarranged and partly destroyed. The pial septal architecture was well preserved occasionally showing slight thickening of the connective tissue septa. Glial cells were observed at places of pronounced nerve fiber lesions, showing an increase in rough endoplasmic reticulum. Inclusion bodies of degenerative myelin, indicating phagocytosis, were seen in astrocytes but not microglial cells.

Immunohistochemistry revealed findings almost similar to those in the control optic nerve sections. While no changes occurred in the amount or distribution of ·B- crystallin and GFAP staining, a mild increase in vimen- tin-positive cell processes, located mainly in the center of the nerve, could be observed in the glaucomatous optic nerves compared to their contralateral controls.

Fig. 1a±f. Immunohistochemical staining of cross-sections through the postlaminar optic nerve of cynomolgus monkeys. a, c, e Control nerves. b, d, f Severely damaged glaucomatous nerves. a, b Staining for ·B- crystallin. c, d Staining for GFAP. e, f Staining for vimentin. In the control optic nerve, several cells within and bordering the nerve fiber bundles stain for ·B-crystallin and GFAP, whereas only single cells within the bundle as well as fibroblasts and vascular cells within the pial septa stain for vimentin. In glaucomatous nerves, the density of cells staining for ·B-crystallin and GFAP increases significantly as does the number of vimentin-positive cells. !40.

a,b

c,d

e,f

Moderate Glaucomatous Optic Neuropathy. This group included cases No. 22 and 23 for the entire optic nerve and cases No. 5, 6, 9±12 and 19 for parts of the optic nerve cross-section.

On light microscopy, this stage of optic neuropathy showed extensive degeneration of axons and their myelin sheaths (fig. 2a, c), which was either diffuse throughout the entire cross-section or focal within some areas of the optic nerve. The pial septa were clearly thickened with increased amounts of collagen fibers. Both astrocytes and microglial cells showed activation indicated by an increase in rough endoplasmic reticulum and myelin inclusion bodies. In this stage of neuropathy, however, phagocytosis was more prominent in the microglial cells than in the astrocytes.

Immunohistochemistry showed an increase and thickening of ·B-crystallin- and GFAP-positive glial processes throughout the areas of pronounced optic neuropathy. The periphery of the bundles adjacent to the pial septa was heavily stained, whereas the remaining cells surrounding nerve fibers were unstained. In addition, at some places numerous focal thickenings of processes stained for GFAP but not for ·B-crystallin. The number of vimentin-positive glial cells in these areas was increased and their processes formed a dense network in the remaining neuronal tissue.

Severe Glaucomatous Optic Neuropathy. This group included cases No. 7, 8, 21 and 24 for the entire optic nerve and cases No. 9±12 and 19 for parts of the optic nerve cross-section.

On light microscopy, severe optic neuropathy was characterized by an almost complete loss of nerve fibers, replaced by glial tissue (table 1, fig. 2b, d). On electron microscopy, the density of astrocytes had increased markedly, forming glial scars. The total number of astrocytes was, however, the same as or even smaller than in the control optic nerves. The density of microglial cells also increased in all but one (case No. 8) of the severely damaged optic nerves. Oligodendroglial cells had almost com-

pletely disappeared. A large number of lipid spherules and remnants of myelin were seen throughout these areas. In some cases the pial septa were much thicker than in normal nerves; the connective tissue at places appeared hyalinized. Other sections showed only a little thickening of connective tissue in individual pial septa. The cytoplasm of the astrocytes and microglial cells contained myelin debris and lipid spherules but only a modest amount of rough endoplasmic reticulum.

Immunohistochemically, nearly all cells between the connective tissue septa stained for GFAP and ·B-crystal- lin (fig. 1b, d). In addition, processes were seen extending into the septa. These processes stained for GFAP but not for ·B-crystallin. In contrast to the ubiquitous distribution of GFAP and ·B-crystallin in the gliotic tissue, only single vimentin-positive cells were present within the remaining nerve fiber bundles, but they appeared more numerous than in the control eyes (fig. 1f).

Vascular Reaction

In the contralateral controls of the glaucoma eyes, the localization and morphology of the vessels were the same as described in normal optic nerves. The capillaries in the connective tissue septa showed normal height of endothelial cells and were nearly completely surrounded by pericytes (fig. 3a). In rhesus monkey eyes with increased IOP maintained for up to 4 years, the density of vessels within the retrolaminar region of the optic nerve increased significantly, whereas the absolute number remained nearly constant. In cynomolgus monkey eyes with an increased IOP induced 4 years earlier but with no IOP measurements during the last 3 years prior to the experiment, the

density of vessels also increased slightly, but the absolute number decreased when compared to the contralateral control eyes (table 1). The differences in the two series of animals may be due to the fact that the optic nerve sections in the rhesus monkeys were cut distal from the entrance of the central retinal artery while in cynomolgus monkeys sections proximal from the entrance were investigated. However, these findings were independent of the severity of the nerve fiber damage.

Electron-microscopic investigation revealed that in the control optic nerves the lumina of all capillaries were wide open, whereas in glaucoma optic nerves some of the lumina were partly or completely occluded (fig. 3b). Most occluded vessels were seen in thickened pial septa (fig. 2c) and were therefore more common in severely damaged optic nerves. On the other hand, there were still regions in the optic nerves with severely damaged nerve fibers, where the vessels were not occluded (fig. 2d). In those capillaries the endothelium appeared flatter than in the controls. None of these vessels showed fenestrations, and pericytes were still present.

The central retinal artery within the optic nerve showed focal thickening of the intima in all but 1 (case

No. 9) glaucomatous optic nerves, and 2 of the control optic nerves (cases No. 1 and 3) also showed mild intimal thickening. There was no correlation between the intimal changes in the central retinal artery and age of the animals, duration of glaucoma, IOP or severity of optic neuropathy.

Pial Arteries

In the control eyes the arterioles in the periphery of the optic nerve cross-sections showed a complete internal elastic membrane and an intima of regular thickness. In most glaucoma eyes (8 of 14), some of the arterioles showed ultrastructural changes. The intima of single vessels was thickened due to an increase in connective tissue in the intimal layer, and some smooth muscle cells had invaded the inner elastic membrane.

Correlation of Morphological Changes with IOP

In the rhesus monkeys, there was no correlation between the severity of optic nerve damage and either last IOP measurement, IOP peaks or mean IOP. In cynomolgus monkeys with increased IOP for 4 years, loss of nerve fibers accompanied by loss of oligodendrocytes and loss of capillaries was more pronounced than in the rhesus monkey eyes with glaucoma for 14±43 months. This may be due to higher levels of sustained IOP in the latter group than in the former ± in the former group a sustained relatively lower level of IOP was maintained by antiglaucoma medication.

Discussion

The majority of our findings agree with what one would expect to occur in atrophy of nervous tissue [21]. Axons are destroyed and the space filled by astrocytes and collapsed connective tissue. The cellular reactions occurring in glaucomatous optic nerve atrophy were similar to those described for optic nerves undergoing Wallerian degeneration: staining for GFAP [22] and vimentin increases [23, 24], while the total number of astrocytes remains constant [32]. Accumulation of GFAP and vimentin also occurs in white matter astrocytes after axonal injury [33, 34]. In human open-angle glaucoma, changes in the extracellular material and in immunohistochemical staining of astrocytes for GFAP and neural cell adhesion molecule have been described [35, 36]. So far, ·B-crystal- lin and vimentin have not been investigated in glaucomatous optic nerves. In contrast to what is seen in the brain, in the normal optic nerve all glial cells which stained for

GFAP also stained for ·B-crystallin, indicating that optic nerve astrocytes express this small stress protein. We do not know whether ·B-crystallin is increased in these cells in glaucomatous eyes, because the staining in the normal eyes was already so bright that differences could not be detected morphologically. Due to fixation the material could not be used for further biochemical investigations.

We found that in glaucomatous optic nerves some vessels were patent while others were partly or completely occluded. Most occluded vessels were seen in thickened pial septa and were therefore more numerous in severely damaged optic nerves. On the other hand, there were still regions in the optic nerves with severely damaged nerve fibers, whose vessels were not occluded; however, in those capillaries the endothelium appeared flatter than in the controls. Quigley and co-workers reported that there were no morphological changes in the number of capillaries in the optic nerve head in eyes with elevated IOP for up to 4 months [5], and in later stages of optic nerve atrophy the proportion of capillaries in the optic disk remained constant [7]. They did not specify whether, in the latter case, the capillaries seen by them were patent or occluded. Morphological demonstration of presence of capillaries gives no information whatsoever about the state of the circulation in them, which is of prime importance in proper nutrition of the optic nerve. We assume, however, that the occluded vessels were the branches of the few abnormal pial vessels in the periphery of the optic nerve. When nutritional needs are diminished in a region of the retrolaminar optic nerve, it is possible that some vessels become closed and others remain open.

From the available information we cannot know whether the differences in morphology between preand retrolaminar glaucomatous optic nerves reflect a difference between white and gray matter in general. We can conclude, at least, in glaucomatous optic neuropathy that in the retrolaminar region of the optic nerve there are not only regional differences in loss of nerve fibers but also in vascular and glial changes.

Acknowledgement

The authors wish to thank Marco Gösswein for the excellent preparation of the photographs. This research was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 539 to E.L.D.), Akademie der Wissenschaften, Mainz (to E.L.D.), European Community (Biomed to E.L.D.), US National Institute of Health (EY 01576 to S.S.H.) and an unrestricted grant for research by Research to Prevent Blindness (to S.S.H.).