Chronology of Optic Nerve Head and Retinal Responses to Elevated Intraocular Pressure

Abstract

purpose. To determine the chronology of optic nerve head and retinal responses to elevated intraocular pressure (IOP).

methods. After 1 to 39 days of unilaterally elevated IOP, experimental and fellow rat eyes were examined for morphology and immunohistochemical labeling alterations and for ganglion cell DNA fragmentation.

results. Mean IOP for the experimental eyes was 36 ± 8 mm Hg, an approximately 15-mm Hg elevation above normal values. By 7 days of pressure elevation above 40 mm Hg, endogenous immunostaining for brain-derived neurotrophic factor and neurotrophin 4/5 was absent from the nerve head and superior retina, whereas normal labeling was present in the inferior retina and distal optic nerve of these same eyes. These changes were preceded by a loss of gap junctional connexin43 labeling and astrocytic proliferation in the nerve head and by increased retinal ganglion cell layer apoptosis in the retina. Nerve head depletion of neurotrophins coincided with evidence of axonal degeneration, loss of astrocytic glial fibrillary acidic protein staining, and spread of collagen VI vascular immunolabeling. After longer durations at these same pressures, neurotrophin labeling returned to nerve head glia and scattered retinal ganglion cells.

conclusions. Optic nerve head and retinal responses, including the depletion of endogenous neurotrophins, are readily identified in the rat eye after experimental IOP elevation. However, the apparent chronology of these responses suggests that the withdrawal of neurotrophic support was not the only determinant of retinal ganglion cell apoptosis and axonal degeneration in response to pressure.

Although many glaucoma risk factors have been identified, ^{1 2} elevated intraocular pressure (IOP) is the best defined, ^{3 4} and controlling IOP is the mainstay of glaucoma therapy. Although the mechanism by which increased IOP damages optic nerves is unknown, knowledge of tissue responses to pressure may provide direction in the development of therapeutic strategies to prevent neural injury and protect the optic nerve in human glaucoma. 

A current attractive hypothesis is that pressure-induced axonal transport obstruction at the optic nerve head inhibits the retrograde delivery of neurotrophic substances to retinal ganglion cells (RGCs), thereby triggering programmed cell death (apoptosis). RGC apoptosis rates are increased in both experimental and human glaucoma. ^{5 6 7}  

The neurotrophin hypothesis led us to predict that specific alterations in RGC, axonal, and optic nerve head protein distributions occur sequentially after pressure elevation. Elevated IOP, by obstructing retrograde transport, should diminish the retinal and nerve head distribution of brain-derived neurotrophic factor (BDNF) and neurotrophin 4/5 (NT4/5), neurotrophins known to be trophic factors for adult RGCs. ^{8 9 10} These neurotrophic factors are presumably produced by the lateral geniculate and superior colliculus, complexed with TrkB (a Trk gene family tyrosine kinase receptor glycoprotein) and delivered by retrograde transport to RGC soma. Neurotrophin deficiency should activate RGC early response genes, especially c-jun, whose mRNA and protein levels have been shown to be elevated after various RGC injuries. ^{11 12 13 14} Obstruction of transport should also result in the abnormal accumulations of axonal proteins in affected RGCs, as demonstrated after other optic nerve injuries. ^{15 16 17} These evidences of RGC injury should precede increased apoptotic rates. RGC death should result in detectable axonal degeneration. Because the process of axonal degeneration triggers gliosis and tissue remodeling in the optic nerve, ^{18 19} evidence of these processes should appear relatively late after pressure elevation. 

We used our recently developed rat glaucoma model, which allows analysis of numerous eyes at a range of pressure levels during the early period of elevated IOP exposure, to test this hypothesis. In this model, aqueous outflow pathways are sclerosed by hypertonic saline episcleral vein injection, resulting in chronically elevated IOP and inner retinal atrophy, optic nerve degeneration, and optic nerve head remodeling similar to that seen in human glaucoma. ^{20 21} This model also offers significant practical advantages, including the ability to monitor IOP frequently without anesthesia, ²² as well as the cost-effectiveness of using rodents. These attributes allow us to study, for the first time, the simultaneous dynamic response of the retina and the optic nerve head to sustained pressure elevation. 

In these experiments, we determined whether the tissue responses predicted by the neurotrophin hypothesis could be detected in the rat eye after elevation of IOP and whether the apparent chronology of these responses was consistent with the predicted sequence. By using immunohistochemistry, we were able to simultaneously localize and identify pressure-induced alterations in nine proteins in both the retina and optic nerve head of each experimental eye. Using these same eyes, our observations were then correlated with evidence of RGC apoptosis and axonal degeneration in parallel retinal sections and optic nerve cross sections, respectively. 

Materials and Methods

Animal IOP Manipulation

All experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Unilateral elevation of IOP was produced in 22 male Brown Norway rats by hypertonic saline episcleral injections into aqueous veins, as previously described. ^{20 23} A calibrated TonoPen XL tonometer (Mentor, Norwell, MA) was used for frequent monitoring of IOP in unanesthetized rats, ²⁴ with each daily IOP value determined as the mean of 10 valid readings. The beginning of the period of pressure elevation was defined as the day in which the experimental eye IOP was significantly higher than that of the fellow eye (P < 0.05; Student’s t-test), a difference of at least 3 mm Hg. Generally, IOP was measured daily until 5 days after significant elevation of IOP, and then every other day. Eyes with various degrees and durations of pressure elevation were obtained to estimate the time course and pressure dependence of tissue responses. 

Anesthetized rats were transcardially perfused with 4% paraformaldehyde as previously described. ²⁰ Globes with attached optic nerves were removed, postfixed overnight, embedded in paraffin, cut in 7-μm sagittal sections, and mounted two sections per slide yielding approximately 15 slides with attached nerve head per eye, plus additional slides with sections of the adjacent retina. For approximately half the eyes, optic nerves were separated from the nerve head 2 mm behind the globe immediately after dissection, postfixed overnight in 5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2), osmicated, and embedded in plastic for optic nerve lesion evaluations, as previously described. ²⁰ This allowed some retrolaminar nerves to remain intact for immunohistochemical labeling, whereas others were available for examination of the lesion areas in optic nerve cross sections. 

Immunohistochemistry

Deparaffinized globe sections were stained using the avidin-biotin-peroxidase complex method as previously described. ²¹ Antibodies to NT4/5 (1:250) and BDNF (1:250; Santa Cruz Biotech, Santa Cruz, CA) were used to detect altered neurotrophin distribution. ^{25 26 27 28} Results with these antibodies were confirmed by immunostaining on other sections from the same eyes using antibodies to both neurotrophins obtained from the laboratory of David Kaplan (Montreal Neurologic Institute, Quebec, Canada). ²⁹ c-jun antibodies (c-jun/AP-1 ab-2, 1:2000, Oncogene, Cambridge, MA and c-jun/AP-1 (N)-G , 1:1000; Santa Cruz Biotech) were used to detect early response gene activation. ^{11 12 13 14} Antibodies to phosphorylated neurofilaments (PNF; SMI 34, 1:15,000; Sternberger Monoclonals, Baltimore, MD) and growth-associated protein-43 (GAP-43, 1:1000; Boehringer–Mannheim, Indianapolis, IN) were used to detect axonal injury, obstructed axonal transport, and perikaryal protein accumulations. ^{15 30 31 32} Optic nerve head glial cell activation and tissue remodeling were evaluated by immunolabeling for altered gap junctional communication, ^{33 34 35 36 37} glial proliferation, ^{18 38 39 40} cytoskeletal organization, ^{41 42} and extracellular matrix deposition, ^{21 43} using antibodies to connexin43 (Cx43, 1:250; Transduction Laboratories, Lexington, KY), proliferating cell nuclear antigen (PCNA, 1:5000; Santa Cruz Biotech), glial intermediate filament protein (GFAP 1:2000, Dako, Carpinteria, CA), and collagen VI (1:250; Southern Biotech, Birmingham, AL), respectively. Although the first three are typical of glial activation after neural injury, the last appears specific to glaucomatous nerve heads. ⁴⁴ For type VI collagen and c-jun/AP-1 ab-2 antigen retrieval before immunostaining, we used predigestion with 0.5 mg/ml trypsin, 15 minutes at 37°C, and with 0.1% pepsin in 0.01 N HCl, 30 minutes at room temperature, respectively. The use of these antibodies for antigen immunolocalization has been documented and published previously, referenced in organ systems including the brain and eye. ^{11 12 21 29 30 31 38 40 41 42 43 44 45 46 47} In general, sections from each of the 22 experimental and control eyes were immunolabeled for each of the nine proteins. With antibodies such as PCNA and c-jun, which do not normally label in the retina, optic nerve head, or nerve, normal labeling intensity and distribution in the corneal epithelial layer provided an internal control on each section. In addition, each immunoassay included sections from experimental and fellow eyes that were exposed to the complete immunohistochemical labeling process, but with appropriately diluted, normal sera or immunoglobulin substituted for the primary antibody. 

The results for each antibody are conclusions determined after the examination of approximately 60 slides produced from four or five individual immunoassays. The antibodies selected for use in this study displayed reproducible and antibody-specific results in preliminary testing, which included unfixed, frozen and fixed, paraffin-embedded normal, and experimental eyes. Other antigens of potential interest, such as the Trk and integrin receptors, heat shock protein-70, bcl-2, c-Fos, kinesin, dynein, and vimentin were not included in the study, because the preliminary testing of the available antibodies to these proteins did not meet these criteria. 

Apoptotic RGC Detection by In Situ Labeling of Fragmented DNA

Fragmented DNA (TdT-dUTP terminal nick-end labeling–positive [TUNEL+]) ^{5 6 48} was detected in retinal sections from experimental and fellow eyes using terminal deoxynucleotidyl transferase (TDT) to label the exposed 3′-OH DNA ends with biotinylated deoxynucleotides (TUNEL method, FragEL DNA Fragmentation Detection Kit; Oncogene). The protocol provided by the manufacturer for paraffin-embedded tissues was followed. In summary, sections were permeabilized with proteinase K at 37°C, endogenous peroxidases quenched, fragmented DNA biotinylated with TDT, biotinylation detected by exposure to a streptavidin–horseradish peroxidase conjugate and the chromogen diaminobenzidine, to form a dark brown, insoluble precipitate. Exposure to the chromogen was limited to 2 minutes to maximize positive control response and minimize nonspecific labeling. In general, four retinal sections of experimental eyes and two of fellow eyes were examined per animal. All RGC layer nuclei, with the exception of spindle-shaped endothelial nuclei, were counted. An observer, masked to the experimental condition of the section, identified and counted the intensely dark brown, condensed, TUNEL+ nuclei. Each assay included negative control slides produced by replacing the TDT in the reaction mixture with dH₂O as well as positive control slides in which sections were predigested with DNase to produce 3′-OH DNA fragments for biotinylation. Statistical analysis was performed using the χ² test for comparison of proportions. ⁴⁹  

Results

Rat Pressure Histories

For each experimental eye, the IOP history is expressed as the mean ± SD of the daily IOP values obtained from the first day of elevated IOP to the day of death. Table 1 illustrates the experimental animal number, the duration of IOP elevation, and the mean IOP for each of the 22 experimental eyes. The pooled mean IOP for the experimental eyes was 36 ± 8 mm Hg. The mean IOP for untreated, fellow eyes was 19.5 ± 2.8 mm Hg. 

Immunohistochemical Analysis

Chronology.

No alterations in normal morphology or protein labeling patterns for the examined proteins were detected in eyes from the three rats killed after 1 day of elevated IOP. The first observed immunohistochemical changes occurred in the glial columns of the optic nerve head before any detectable morphologic alterations. These alterations were followed by changes in the distribution of neurotrophins and other proteins. A summary of the apparent chronology of major optic nerve head and retinal responses to elevated pressure is shown in Table 2 , and the anatomy of the rat optic nerve head is illustrated in Figure 1 . The effects of elevated IOP on protein immunohistochemical localization are described in the following section in the order of their appearance. 

Alterations in Optic Nerve Head Gap Junctions and Onset of Cellular Proliferation.

In normal fellow eyes, Cx43 antibody, which labels gap junctional proteins, produced a distinctive pattern of discrete, punctate label at meningeal surfaces, near blood vessels, and in a diffuse band across the transition region of the optic nerve head (Fig. 2A ). Light vascular and meningeal labeling was also seen in the retrolaminar optic nerve, whereas no labeling was detectable anterior to Bruch’s membrane. Beginning at 3 days of pressure elevation (experimental eyes 126, 119, and 125), a loss of punctate labeling for Cx43 was detectable within the nerve fiber bundles of the optic nerve head transition region (Fig. 2B) . All eyes with 1 week or more of mean IOP higher than 40 mm Hg (eight experimental eyes), showed nearly complete loss of Cx43 labeling in the nerve head, accompanied by the loss of normal columnar arrangement of glia (Fig. 2C) . Most eyes with lower mean IOPs (four of five) demonstrated a reduced amount of nerve head label accompanied by less disruption of glial column organization. The distribution in the remaining eye (eye 70) appeared normal. 

PCNA antibodies, which label dividing cells, produced no nuclear labeling in control fellow eye optic nerve heads, although corneal epithelial nuclei on the same sections showed labeling. Scattered glial column nuclear PCNA labeling first appeared in the neck region at 3 days of elevated IOP (experimental eye 126; Figs. 3A 3B ) and was observed in seven eyes with up to 1 week of elevated IOP or with sustained mean IOPs of less than 30 mm Hg. After 1 week or more of IOP higher than 40 mm Hg, a heavy band of PCNA labeling of glial nuclei appeared progressively more distal in the optic nerve (Figs. 3C 3D) , marking an apparent boundary between the more proximal rearrangement of glial nuclei near the transition region and the normal, columnar arrangement of glial nuclei in the distal optic nerve (six eyes). 

Alterations in Neurotrophin Distribution in Both Optic Nerve Head and Retina.

In all normal fellow eyes, antibodies to neurotrophins NT4/5 and BDNF labeled the neural components of the optic nerve head and inner retina, whereas nonneural tissues were unlabeled (Fig. 4) . In the optic nerve head, NT4/5 label was most intense over nerve fiber bundles (Fig. 4C) , whereas BDNF label frequently was heaviest over glial column nuclei (Fig. 4D) . However, it was not apparent that labeling for either neurotrophin was uniquely localized to either axons or glial processes. In the inner retina, both NT4/5 and BDNF labeled the nerve fiber layer, plexiform layers, and processes associated with RGC layer nuclei (Figs. 4E 4F) . 

Elevated IOP changed the neurotrophin distributions in both the optic nerve head and the retina. In the optic nerve head, alterations were first detected after 7 days of mean IOP above 40 mm (three eyes). At that time and through 16 days at similar pressures (six eyes), labeling for NT4/5 and BDNF was dramatically reduced over both nerve fiber bundles and glial columns in the neck and transition region (Figs. 5A 5B ), whereas normal staining intensity was apparent in the distal optic nerve of the same sections. After an extended period of IOP above 40 mm Hg, glial nuclei in the optic nerve head transition region and surrounding cytoplasm again labeled with NT4/5 and BDNF antibodies (Figs. 5C 5D) . That these nerves had few remaining axons is demonstrated by labeling with neurofilament antibodies (PNF, Fig. 5E ). 

Examination of the retinal portions of the same eye sections revealed that 1 week of IOP above 40 mm Hg also altered relative neurotrophin distributions between the superior and inferior retina (three eyes). In the superior retina, overall staining intensity with antibodies to NT4/5 and BDNF was greatly diminished or lost, whereas in the inferior retinal portion of the same section, the labeling pattern and stain intensity was the same as that observed in the fellow, untreated eyes (Figs. 6A 6B 6C 6D , compare with Figs. 4E 4F ). All retinas examined after longer durations of pressures above 40 mm Hg showed either loss of label in the superior region or in both superior and inferior regions in comparison with fellow eye staining patterns (five eyes). In contrast to this overall general loss of retinal labeling, occasional RGC layer nuclei in eyes exposed to 2 weeks or more of IOP higher than 40 mm Hg exhibited strong somal staining with antibodies to NT4/5 and BDNF (Figs. 6E 6F) . 

Alterations in Neuronal-Specific Proteins in the Optic Nerve Head.

Neurofilaments are major axonal structural proteins, and RGC-synthesized GAP-43 undergoes fast anterograde transport in optic axons. In control fellow eyes, PNF and GAP-43 antibodies labeled axons from their origin in the retinal nerve fiber layer through the optic nerve head and optic nerve (Figs. 7A 7B ). 

Beginning at 7 days after IOP elevation to 40 mm Hg, alterations in both PNF and GAP-43 immunostaining patterns were observed (Figs. 7C 7D 7E 7F 7G) . Enlarged axons intensely staining for both PNF and GAP-43 were apparent in the optic disc and adjacent nerve fiber layer (PNF, 10 eyes; GAP-43, 7 eyes), suggesting obstructed axonal transport (Figs. 7E 7F) . Immunostaining for PNF also revealed giant axonal swellings in the transition region and the retrobulbar optic nerve from these experimental eyes (10 eyes), indicative of axonal injury and degeneration (Figs. 7C 7G) . 

Alterations in Glial Cytoskeletal Proteins in the Optic Nerve Head.

GFAP, the major astrocyte structural protein, labeled processes oriented perpendicular to the nerve fiber bundles in fellow eye optic nerve heads (Figs. 8A ). Nerve head GFAP labeling intensity was diminished by 1 week and greatly decreased after 2 weeks of IOP higher than 40 mm Hg (Fig. 8B 6 eyes). Only after 33 days at 47 ± 10 mm Hg was heavy nerve head GFAP label restored (Fig. 8C) . 

Alterations in the Optic Nerve Head Extracellular Matrix.

In normal, paraformaldehyde-fixed eyes, collagen VI antibodies heavily labeled the surrounding scleral connective tissue of the optic nerve head, whereas the nerve head neural tissues and vasculature were relatively unstained (Fig. 9A ). An apparent increase in type VI collagen immunostaining of blood vessels in the transition region was first seen in nerve heads after 7 days of pressure elevation (two of four eyes; Fig. 9B ). With extended exposure to these pressures, nerve head labeling became more extensive until labeling extended throughout the optic disc including areas formerly occupied by nerve fiber bundles (six eyes, Fig. 9C ). 

Alterations in Neuron-Specific Proteins in the Retina.

The RGC soma of normal, fellow retinas were virtually unlabeled by antibodies to either PNF or GAP-43 (Fig. 10A 10B ). After more than 2 weeks of IOP higher than 40 mm Hg, only a few heavily labeled PNF- or GAP-43–labeled RGC layer cells scattered among unlabeled RGC layer cells could be identified in some experimental eyes (PNF: three eyes; GAP-43: five eyes, Figs. 10C 10D ). No detectable alterations in c-jun immunolabeling patterns were observed in retinas exposed to elevated IOP. 

TUNEL+ Labeling for Apoptotic RGCs

A summary of TUNEL labeling results is shown in Table 3 . TUNEL-labeled RGC layer nuclei were found in sections from 12 of 18 experimental eyes (Figs. 11A 11B ) including two of three experimental eyes sampled after 1 day of elevated IOP, in contrast to only one of 18 fellow eyes. For the experimental eyes examined, the pooled TUNEL+ rate was 1.42:1000 nuclei, whereas in fellow eyes, the pooled TUNEL+ rate was 0.078:1000 nuclei (P < 0.001; χ²). Because only a small sample of each retina was evaluated, correlation of TUNEL+ rate with degree or duration of pressure increase was not possible. However, for the 7 eyes (eyes 81,119, 123–127) with 7 or fewer days of elevated IOP and no alterations in neurotrophin distribution, the pooled number of TUNEL+ nuclei was significantly greater than in fellow eyes (P < 0.05;χ ²). Negative-control sections were consistently unstained (not shown), whereas positive control sections pretreated with DNase demonstrated uniformly TUNEL-labeled RGC layer nuclei (Fig. 10C)  

Optic Nerve Injury

Ten optic nerves from this study were processed for evaluation of optic nerve morphology (see Table 1 ), and the results are presented in Table 4 . A mean IOP less than 30 mm Hg of any duration resulted in focal nerve lesions, predominantly in the superior and central nerve regions (Fig. 12A ). For the two sampled nerves with mean IOP between 30 and 40 mm Hg, a central, focal lesion was observed after 8 days of elevated IOP, whereas the entire nerve was involved after 27 days. Very large central lesions, usually involving the entire optic nerve area (global lesions), were apparent in all sampled nerves from eyes with mean IOP greater than 40 mm Hg (Fig. 12B) . 

Discussion

This first coordinated determination of the retinal and optic nerve head responses to elevated IOP was made practical by a rat model of pressure-induced optic neuropathy that provided adequate tissue to allow the simultaneous immunohistochemical localization of multiple proteins, TUNEL labeling of DNA fragmentation, and morphologic analysis of optic nerves in each experimental eye. In this experiment, we asked whether the chronology of nerve head and retinal responses to elevated IOP was consistent with a hypothesis of RGC loss based on neurotrophin deprivation. 

The actual chronology observed is summarized in Table 2 . Although we found apparently increased rates of TUNEL+ RGC layer nuclei at every duration of pressure elevation, the first detected immunohistochemical alterations were localized to the optic nerve head. At 3 days after IOP elevation, Cx43 immunostaining was reduced in the nerve head transition region, and sporadic PCNA glial column nuclear labeling appeared. At 7 days in eyes with higher IOP levels, reduced optic nerve head and retinal neurotrophin label was observed. Simultaneously, nerve head GFAP label was lost, collagen IV spread near blood vessels, and axonal swellings (identified by PNF antibodies) appeared. Examination of optic nerve cross sections revealed focal and global lesions as early as 4 and 7 days after IOP elevation, respectively. After 2 weeks or more of elevated IOP, both RGC layer soma and optic nerve head glia again labeled with neurotrophin antibodies. At approximately 1 month of elevated IOP, occasional RGC layer somal PNF and GAP-43 labeling appeared, and, glial GFAP label was restored in the optic nerve head. 

This study presents the first immunolocalization of NT4/5 and BDNF to the adult mammalian optic nerve head and retina. Both of these neurotrophins have previously been shown in studies in vitro and in vivo to support RGC survival. ^{9 10 17 25 26 28 50} In addition, it demonstrates loss of immunostaining for the neurotrophins in the nerve head and superior retina of experimental eyes after IOP elevation. Alterations in neurotrophin distribution occurred concurrently with indications of axonal degeneration, as revealed by the appearance of PNF-labeled giant axonal swellings in immunostained sections, and morphologic evidence of large lesions in optic nerve cross sections. The loss of neurotrophin labeling in the neck and transition region of the optic nerve head probably resulted from obstructed retrograde axonal transport of neurotrophins coupled with axonal degeneration, as well as loss of any endogenous production by optic nerve head glial cells or retinal cells. ^{29 51 52}  

The initial restriction of the retinal neurotrophin changes to the superior retina suggests that trophic support for RGCs may be compromised by IOP elevation earlier in this location than in the inferior retina. This observation correlates with our previous finding that axons in the superior temporal quadrant of the rat optic nerve appear more vulnerable to pressure-induced degeneration. ²⁰ Our demonstration of the depletion of immunohistochemically labeled neurotrophins from the inner retina is the first evidence that endogenous neurotrophic support to retinal RGCs is reduced after IOP elevation and is consistent with the hypothesis that this loss contributes to RGC apoptosis. 

However, several aspects of the observed chronology suggest that, instead of a single mechanism of neural injury, elevated IOP results in a more complex response, affecting both retinal and optic nerve head tissues. First, the results of our TUNEL labeling of retinal sections suggest that RGC apoptosis is an early and continuing response to elevated pressure. Despite the relatively limited sample of retinal tissue available in these sections, TUNEL+ RGC layer nuclei occurred in two thirds of experimental eyes compared with only one of the fellow control eyes, and TUNNEL+ nuclei were observed throughout the course of IOP elevation. The same proportion of TUNEL+ retinal samples from experimental eyes has been reported in a recent study of chronic primate glaucoma. ⁵ The pooled rate in experimental eyes was almost 20 times greater than that in control eyes (P < 0.001; χ²) and, for both experimental and control eyes, of the same order of magnitude as respective rates previously determined in rat retinal flatmounts after 1 to 6 weeks of elevated IOP after venous occlusion. ⁷ In addition, when only the experimental eyes with 7 or fewer days of elevated IOP and with normal neurotrophin distribution were examined, TUNEL+ labeling was also significantly increased, suggesting that RGC apoptosis rates may increase before depletion of neurotrophic support. This suggests that factors other than neurotrophin withdrawal may induce RGC apoptosis. 

Secondly, the earliest immunohistochemical or morphologic alterations occurred in the optic nerve head and may reflect direct astrocytic responses to elevated pressure. An initial dramatic loss of labeling for Cx43 gap junctional protein in the optic nerve head transition region was coupled with scattered neck region PCNA nuclear staining. Among neural cells, Cx43 is specific for astrocytes, and its distribution within the transition region in our study is consistent with an astrocytic localization. Loss of Cx43 immunoreactivity after brain injury is associated with a reorganization and functional uncoupling of ultrastructurally detectable astrocytic gap junctions within the lesion area. ^{33 34 35 36} Therefore, we interpret the loss of Cx43 label observed here as a disruption of astrocytic gap junctional communication in response to elevated IOP. ^{37 53}  

In normal neural tissue, astrocytic coupling through gap junctions is thought to form a syncytium, allowing metabolic and electrical communication as well as providing spatial buffering for local ionic and metabolite imbalances. ⁵⁴ The disruption of the nerve head astrocytic syncytium in response to elevated IOP could act to isolate the area of injury. This may protect surrounding neural tissue by restricting the spread of injurious ions or metabolites, but it may also intensify local damage by reducing the ability of the metabolic effects of injury to be buffered by distribution throughout the tissue. ³⁴ The correlation of Cx43 loss with the appearance of astrocytic mitosis indicated by PCNA staining ³⁸ suggests that the disruption of the transition region astrocytic syncytium immediately precedes astrocytic proliferation. In this study, there was no apparent recovery of Cx43 labeling during continued exposure to elevated IOP, suggesting that reduced astrocytic communication and buffering capacity is a continuing effect of pressure elevation. 

Loss of gap junctions and cellular proliferation was followed by decreased nerve head GFAP immunoreactivity, suggesting that astrocytic cytoskeletal reorganization is associated with these changes. ⁵⁵ Because GFAP immunoreactivity is known to be reduced initially at the site of optic nerve crush injury, ^{42 56} our study offers further support to previous evidence that the optic nerve head is the site of nerve injury in glaucoma. ⁵⁷  

Thirdly, extracellular matrix alterations, represented by increased type VI collagen labeling, occur simultaneously with neurotrophin depletion. Collagen VI deposition is characteristic in both human and rat glaucomatous optic nerve head. ^{21 43} The results of the present study indicate that the deposition begins after nerve head glial cell proliferation and occurs simultaneously with the initial evidence of active axonal degeneration and neurotrophin depletion. The early timing of this response may have important implications for the progression of glaucomatous axonal injury, because collagen VI has been shown to inhibit retinal neurite outgrowth in vitro and may also have inhibitory effects in vivo on recovery of injured axons or axonal regenerative efforts. ⁵⁸ In addition, deposition of this and other extracellular matrix materials probably underlies the extensive remodeling and structural changes seen in advanced glaucomatous optic disc cupping and may induce changes in the physical properties of the lamina cribrosa that could contribute to the susceptibility of remaining nerve fibers to degeneration in end-stage glaucoma. 

Finally, expected early retinal responses were either late occurring and sporadic or absent. In the retina, abnormal perikaryal accumulations of PNF, GAP-43, and Jun proteins have been identified as early markers of RGC response after optic nerve transection. ^{11 15 59} In contrast, after IOP elevation, RGC labeling with phosphorylated neurofilaments and GAP-43 was only a rare and late occurrence, and increased labeling was not observed using c-jun/AP-1 antibodies from the same sources used by others. ^{11 45} These apparent differences in response to elevated IOP compared with axotomy may reflect methodologic differences, such as using retinal wholemounts rather than the globe cross sections that were used in this study. Furthermore, gradual, and persistent injury caused by elevated IOP, in contrast to the highly traumatic and global injury of axotomy, may result in actual differences in the rate, sequence, or intensity of the injury responses. 

It is striking, however, that the same experimental eyes with RGC labeling for the PNF and GAP-43, proteins associated with the axonal cytoskeleton and neurite extension, ^{15 16 30 31 32 46} also showed RGC somal labeling for NT4/5 and BDNF. This suggests that an endogenous expression of neurotrophins by RGCs may be a delayed response to pressure-induced injury and that this expression may be followed by regenerative processes involving cytoskeletal protein synthesis. Similar evidence of enhanced endogenous BDNF expression by RGCs after optic nerve crush has been reported. ⁶⁰ Alternatively, these specific RGCs could be those whose axons remain intact, because even in the most affected optic nerves, intact axons could still be identified. 

Similarly, the localization of BDNF and NT4/5 neurotrophin labeling to the glial cells that occupy the nerve head after extensive axonal degeneration and continued exposure to elevated IOP indicates that these astrocytes can produce neurotrophins. Schwann cell neurotrophin production has been observed after peripheral nerve axotomy. ^{61 62} Whether the neurotrophins identified in these optic nerve head glial cells act as autocrine regulators of endogenous glial function or in a paracrine manner toward injured and potentially regenerating RGCs is unknown. However, astrocytes genetically modified to secrete BDNF have been shown to enhance RGC survival in vitro. ²⁷ The localization of neurotrophins to nerve head astrocytes and RGCs in these experimental eyes with extensive axonal degeneration demonstrates that, in the adult eye, potential neurotrophin sources exist in addition to those derived by retrograde transport. 

In summary, using our rat model of experimentally elevated IOP, we have demonstrated that depletion of the endogenous neurotrophins BDNF and NT4/5 occurs in experimental optic nerve heads and retinas during the process of apoptotic RGC death. However, that neurotrophin losses were not detected until after evidences of increased RGC apoptosis and altered optic nerve head astrocytic function were apparent suggests that loss of neurotrophic support may be only one of several processes by which elevated IOP results in axonal degeneration and RGC loss. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1997.

 Supported by National Institutes of Health Grant RO1EY10145; Alcon Laboratories, Fort Worth, Texas; and unrestricted funds from Research to Prevent Blindness.

 Submitted for publication February 9, 1998; revised July 14, 1999; accepted August 16, 1999.

 Commercial relationships policy: N.

 

Table 1.

 

Experimental Eye IOP History

Rat Number	Elevated IOP (Days)	IOP*
124	1	32
123	1	37
127	1	36
126	3	27 ± 5
119	5, †	28 ± 6
125	6	27 ± 10
81	7, †	24 ± 15
115	7, †	42 ± 7
116	7, †	44 ± 7
122	7, †	45 ± 3
121	8, †	37 ± 5
70	11	26 ± 9
117	13	41 ± 4
118	14	38 ± 4
120	14, †	43 ± 6
57	16	54 ± 1
58	16	47 ± 13
82	18, †	27 ± 3
86	27, †	33 ± 11
51	29	36 ± 10
66	33	47 ± 10
85	39, †	27 ± 9

*  Mean mm Hg ± SD. The mean IOP for untreated fellow eyes was 19.5 ± 2.8 mm Hg.

†  Optic nerve lesion evaluated, see Table 4 .

Table 2.

 

Chronology of Pressure-Induced Responses^*

Optic Nerve Head	Retina
3 Days’ elevated IOP
Cx43 immunolabeling diminishes in optic nerve head nerve fiber bundles	No detected alterations
Scattered PCNA optic nerve head glial column nuclear labeling
No morphologic alterations
7 Days’ IOP higher than 40 mm Hg
Loss of nerve head immunolabeling with NT4/5 and BDNF antibodies	Loss of NT4/5 and BDNF label in the superior retinal segment
Apparent nerve head nuclear proliferation
Near complete loss of Cx43 label
Increased vascular labeling with collagen VI antibodies
Reduced GFAP label in the optic nerve head
Enlarged axons, intensely labeled with PNF and GAP-43 at peripheral optic disk and adjacent inner retina
PNF-labeled, versicle-filled, giant axonal swellings first appear
Focal and global nerve lesions apparent in optic nerve cross sections
14 Days’ IOP higher than 40 mm Hg
Glia labels with antibodies to NT4/5 and BDNF	Loss of NT4/5 and BDNF label in the superior or both retinal segments
Collagen VI deposition extends in nerve head
GFAP label decreases dramatically
Axonal labeling with PNF greatly reduced
Global optic nerve lesions
More than 14 Days’ IOP higher than 40 mm Hg
Glia labels with GFAP	Heavy label of NT4/5 and BDNF in isolated RGC soma
	Occasional RGC somal accumulations of GAP-43 and PNF

*  Apoptotic (TUNEL-labeled) RGCs were detected in experimental eyes throughout the course of elevated IOP.

Figure 1.

 

Anatomy of the rat optic nerve head. The nerve head (trichrome stain) consists of the unmyelinated neck (N) and transition (T) region at the level of the posterior sclera (S). In the transition region, the thickened walls of the vasculature (arrow, fuchsia stain) form the equivalent of the primate lamina cribrosa. The retina (R) appears at the top and the fully myelinated optic nerve (ON) at the bottom of the figure. Small, pale glial nuclei (pale lavender hematoxylin stain) are oriented in columns that parallel the course of the nerve fiber bundles. Original magnification, ×170.

Figure 2.

 

Gap junctional communication. In fellow eyes, discrete, punctate labeling of Cx43 (arrowheads), indicating gap junctions, was present in a diffuse band across the transition region of the optic nerve head (A). After 3 days with an IOP of 27 ± 5 mm Hg, labeling was dramatically diminished, especially within transition region nerve fiber bundles (B), indicating a disruption of the astrocytic syncytium. After longer durations of elevated IOP, Cx43 labeling almost completely disappeared, illustrated by the same region from a nerve head after 7 days at 45 mm Hg (C). Illustrated sections from a normal fellow eye (A), and experimental eyes 126 (B), and 122 (C); brown, specific label; blue, hematoxylin nuclear counterstain. Original magnification, ×550.

Figure 3.

 

Cellular proliferation. At 3 days of elevated IOP, PCNA labeling, indicating cell division, was first detected in the optic nerve head among glial nuclei in normally arranged columns (A, B illustrates mitotic figure at arrowhead). After longer durations of elevated IOP (C, 13 days at 41 ± 4 mm Hg), the optic nerve displayed obvious cellular proliferation, with loss of glial nuclei columnar arrangement apparent (detail, D) in the transition region (T), whereas a band of PCNA-labeled columnar glial nuclei was located more distally in the optic nerve (ON detail, E). Illustrated sections from experimental eyes 126 (A, B) and 117 (C, D, and E); brown, specific label; blue, hematoxylin counterstain. Original magnification, (A) ×180; (B) ×450; (C) ×120; and (D, E) ×250.

Figure 4.

 

Neurotrophin distribution in the normal optic nerve head and retina. In fellow optic nerve heads, NT4/5 labeling was densest over nerve fiber bundles, apparently associated with membranes or vesicles (A, detail in C). BDNF label appeared as a dense, finely particulate deposition throughout the optic nerve and nerve head, particularly over the nerve head glial column nuclei (B, detail in D). The inner retina, including the nerve fiber layer (arrowhead) labeled with both NT4/5 (E) and BDNF (F) antibodies. All illustrated sections from normal fellow eyes; brown, specific label; blue, hematoxylin counterstain. Original magnification, (A, B) ×70; (C, D, and E) ×360.

Figure 5.

 

Neurotrophin alterations in the optic nerve head after elevated IOP. After 7 days of IOP more than 40 mm Hg, alterations in optic nerve head (ONH) neurotrophin staining patterns were apparent, accompanied by obvious cellular proliferation and glial column disruption. NT4/5 (A) and BDNF (B) immunolabeling was diminished in the optic nerve head, both in comparison with the corresponding optic nerve and with nerve head labeling intensity in fellow eyes. After extended periods of pressure elevation (33 days at 47 mm Hg), nerve head glia strongly label with antibodies to NT4/5 (C) and BDNF (D). The glial association of this labeling was demonstrated by PNF-labeling (E), which revealed only a few residual axons (arrowhead) remaining in these nerves. Experimental eyes 116 (A, B) and 66 (C, D, and E); brown, specific label; blue, hematoxylin counterstain. Original magnification, ×90.

Figure 6.

 

Neurotrophin distribution in the retina after IOP elevation. After 1 week of IOP higher than 40 mm Hg, NT4/5 (A) and BDNF (B) staining intensity in the superior retina was diminished, compared with the inferior retina from the same experimental eye, which demonstrates a normal staining pattern for NT4/5 (C) and BDNF (D). Illustrated retina was from experimental eye 115 after 1 week at 44 ± 7 mm Hg IOP. After 2 weeks or more at this level of pressure, prominent RGC layer cytoplasmic staining of some isolated neurons with NT4/5 (E) and BDNF (F) antibodies appeared in experimental eyes (example from eye 57, 16 days at 54 ± 1 mm Hg). Brown, specific label; blue, hematoxylin counterstain. Original magnification, ×420.

Figure 7.

 

Axonal proteins in the optic nerve head. Normal fellow eye staining patterns are illustrated for the axonal proteins PNF (A) and GAP-43 (B). After pressure elevation, dramatic alterations in PNF (C, eye 117, 13 days at 41 ± 4 mm Hg) and GAP-43 (D, eye 115, 7 days at 42 ± 7 mm Hg) are observed. In the anterior, neck portions of these optic nerve heads, intensified axonal staining (arrowheads) for PNF (E) and GAP-43 (F) suggests obstructed anterograde axonal transport. Throughout the transition region of the optic nerve head and retrolaminar optic nerve, numerous giant axonal swellings are present (arrows), labeled by PNF antibodies (C, detail in G), indicative of axonal injury and degeneration. Brown, specific label; blue, hematoxylin counterstain. Original magnification, (A, B, and D) ×90; (C)× 45; (E, F, and G) ×460.

Figure 8.

 

Glial structural proteins in the optic nerve head. In normal fellow eyes, GFAP labeling antibodies labeled transversely oriented processes within the optic nerve head (A). A loss of GFAP labeling intensity in the nerve head and proximal optic nerve was observed after elevated IOP (B, eye 117, 13 days at 41 ± 4 mm Hg). The recovery of nerve head GFAP label was seen in experimental eye 66 after 33 days at 47 ± 10 mm Hg (C). Brown, specific label; blue, hematoxylin counterstain. Original magnification, ×115.

Figure 9.

 

Optic nerve head extracellular matrix deposition. After 7 days of mean IOP higher than 40 mm Hg, type VI collagen staining appeared to be intensified around transition region blood vessels (compare experimental eye 116 in B with fellow eye in A). Nerve heads experiencing 2 or more weeks of similarly elevated IOP (C, eye 66, 32 days at 47 ± 10 mm Hg) showed extensive deposition of this collagen. Hematoxylin counterstain. Original magnification, ×115.

Figure 10.

 

Retinal RGC layer perikaryal labeling. In normal fellow retinas (top), RGCs perikarya were virtually unlabeled by the axon-specific proteins PNF (A) and GAP-43 (B) antibodies. After extended periods of elevated IOP, immunostaining revealed only a few RGC layer somata heavily labeled with antibodies to PNF (C) and GAP-43 (D, bottom, eye 86, 27 days at 33 ± 11 mm Hg). Brown, specific label; blue, hematoxylin counterstain. Original magnification ×480.

Table 3.

 

TUNEL+ Labeling for Apoptotic Retinal Ganglion Cells

	n	Experimental Eyes	Corresponding Fellow Eyes
Selected experimental eyes*	7	22/16,921, †	0/5,799
All experimental eyes	18	47/33,119, ‡	1/12,760

 Data are the ratio of TUNEL+ labeled RGCs to total number of RGCs examined.

*  Eyes with fewer than 7 days of elevated IOP and normal neurotrophin distribution.

†   P < 0.05 by χ² analysis.

‡   P < 0.001 by χ² analysis.

Figure 11.

 

TUNEL labeling of RGC layer nuclei. TUNEL labeling detects DNA fragments indicative of cellular apoptosis. (A) Intensely dark brown labeling of condensed perinuclear chromatin indicative of early apoptosis (arrow) in a TUNEL+ nuclei stood out among aqua-counterstained, TUNEL-negative RGC layer nuclei in the retina of an experimental eye exposed to 1 day of 36 mm Hg IOP. (B) Heavy, uniform nuclear labeling indicated a more advanced stage of apoptosis in another TUNEL+ nuclei (eye 116). (C) TUNEL+ control samples, pretreated with DNase to form DNA fragments, yielded uniformly labeled RGC layer nuclei (B). Original magnification, ×575.

Table 4.

 

Optic Nerve Lesion Evaluation

Rat Number	Elevated IOP (Days)	IOP*	Lesion Description
119	5	28 ± 6	Small focal
81	7	24 ± 15	Small focal
115	7	42 ± 7	Large focal
116	7	44 ± 7	Global
122	7	45 ± 3	Global
121	8	37 ± 5	Large focal
120	14	43 ± 6	Global
82	18	27 ± 3	Small focal
86	27	33 ± 11	Global
85	39	27 ± 9	Focal

*  Mean mm Hg ± SD.

Figure 12.

 

Optic nerve lesions. Moderate levels or short durations of pressure elevation resulted in localized nerve lesions, characterized by axonal swellings and myelin debris. The appearance of such lesions is shown (A, 39 days at 27 ± 9 mm Hg). Nerves with IOP higher than 40 mm Hg for 7 days or more had lesions that usually encompassed the entire nerve cross section (B, 7 days at 42 ± 7 mm Hg). Within these lesions, many morphologically normal axons were apparent. Toluidine blue stain. Original magnification,× 575.