Microglial Activation in the Visual Pathway in Experimental Glaucoma: Spatiotemporal Characterization and Correlation with Axonal Injury

Abstract

Purpose.: Glia are the main cellular CNS elements initiating defense mechanisms against destructive influences and promoting regenerative processes. The aim of the current work was to characterize the microglial response within the visual pathway in a rat model of experimental glaucoma and to correlate the microglial response with the severity of axonal degeneration.

Methods.: Experimental glaucoma was induced in each right eye of adult Sprague-Dawley rats by translimbal laser photocoagulation of the trabecular meshwork. Rats were subsequently killed at various times from 3 days to 6 weeks. Tissue sections were obtained from globes, optic nerves, chiasmata, and optic tracts for immunohistochemistry and toluidine blue staining.

Results.: This model of experimental glaucoma led to a marked activation of microglia in the retina, optic nerve, and tract. Indeed, microglial activity remained elevated, even after intraocular pressure returned to basal levels. It is postulated that this process accompanies ongoing axonal degeneration. The degree of activation in the optic nerve correlated with axonal damage. Activation was characterized by increased density and morphologic changes. Both major histocompatibility complex (MHC) class I and MHC class II surface proteins were persistently upregulated in optic nerves and localized to microglial cells; however, this did not correlate with any significant T-cell infiltration. Interestingly, MHC class II expression was not detected in the retina.

Conclusions.: The present data may have implications for the study of the pathology associated with the visual pathway in diseases such as glaucoma.

Glaucomatous optic neuropathy (glaucoma), the second leading cause of blindness in the world, ¹ is a neurodegenerative disease characterized by structural damage to the optic nerve and the slow, progressive death of retinal ganglion cells (RGCs). RGCs represent the third-order neurons in the visual pathway. In the retina, their unmyelinated axons converge at the optic nerve head (ONH), where they exit the globe and become the myelinated optic nerve (ON). In the rat, most fibers decussate at the optic chiasm, forming the contralateral optic tract (OT) and synapsing in the lateral geniculate nucleus. In recent years, several rodent models of experimental glaucoma have been developed ² that exhibit many of the characteristics of the human condition. Shared pathophysiological events between such models include axon transport disruption, selective loss of RGCs and their axons, oxidative stress, and reactive gliosis. 

One cell type whose involvement in the pathogenesis of neurodegenerative diseases is increasingly being recognized is the microglial cell. Microglia are the resident immunocompetent cells of the CNS parenchyma and can be viewed as bridging elements between the neuronal and immune systems. They are part of the mononuclear phagocyte system but, under normal physiological conditions, assume a quiescent, ramified form with highly motile processes that monitor the environment. ³ Microglia respond rapidly to the disruption of tissue homeostasis: they proliferate, assume an ameboid morphology, migrate to the site of injury, express a multitude of receptors, ⁴ produce numerous types of cytokines, ^{5 –7} participate in the complement cascade, ⁸ phagocytose cellular debris, and can function as antigen-presenting cells. ⁹ Recent debate has focused on whether microglial activation is harmful or beneficial in CNS injury. ¹⁰ One theory proposed is that in the early stages of disease, moderate activation of microglia may be beneficial and may contribute to the regeneration of damaged tissue, ¹¹ but in an overactivated and chronically dysregulated state, microglia probably exacerbate preexisting damage and contribute to secondary disease progression ^12,13 ; however, the role of microglia may depend on the type and severity of injury. 

Microglia have been implicated in the pathogenesis of various experimental retinal and ON neuropathies. In the retina, these include autoimmune uveoretinitis, ^14,15 diabetic retinopathy, ¹⁶ ischemia, ^17,18 excitotoxicity, ¹⁹ photoreceptor degeneration, ^{20 –23} AMD, ²⁴ and trauma, ^{25 –27} whereas in the ON, a robust microglial response has been demonstrated in ischemia, ^28–29 and experimental allergic encephalomyelitis (EAE). ³⁰ With regard to glaucoma, there is some limited information regarding microglial activation in the retina/ONH region ^{31 –33} in both experimental models and human specimens ³⁴ ; however, to date, no data are available concerning this process in the optic pathway distal to the ONH. The ONH region is considered by some to be the primary site of injury in glaucoma. ³⁵ Some evidence, however, suggests that the distal portion of the ON may be more severely affected by glaucoma. ³⁶ It is of interest, therefore, to discover where the microglial response occurs earliest and with greatest magnitude. Delineating the spatiotemporal microglial response in the optic pathway would assist in addressing this matter. Further subjects of interest relate to the expression of immunologic cell surface markers by microglia during glaucoma; the infiltration, if any, of macrophages or T lymphocytes, such as occurs in ischemic, traumatic, and autoimmune models of ON injury; and whether microglial activation correlates closely with the degree of ON damage. Such a finding would provide the opportunity for using the detection of microglial activation as a surrogate or adjunct marker for ON injury in studies relating to neuroprotection. The aim of the present study, therefore, was to address these issues using a well-characterized rat model of glaucoma. 

Materials and Methods

Animals and Procedures

This study was approved by the Animal Ethics Committees of the Institute of Medical and Veterinary Science and the University of Adelaide and conformed to the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, 2004, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Adult Sprague-Dawley rats (weight range, 200–250 g) were housed in a temperature- and humidity-controlled room with a 12-hour light/12-hour dark cycle and were provided food and water ad libitum. Rats were anesthetized with intraperitoneal injection of 100 mg/kg ketamine and 10 mg/kg xylazine, and local anesthetic drops were applied to the eye. Ocular hypertension was then induced in the right eye of each animal by laser photocoagulation of the trabecular meshwork using a slightly modified protocol of the method described by Levkovitch-Verbin et al. ³⁷ In brief, 80 to 100 spots of 50-μm diameter, 550 to 600 mW power, and 0.5-second duration were applied to the trabecular meshwork. An additional 10 to 20 spots, 100-μm diameter, 450 mW, and 0.5-second duration were delivered to three of the four episcleral veins. A second laser treatment was often given on day 4 or 7, depending on the IOP. If the difference in IOP between the two eyes was less than 8 mm Hg on day 3, these rats were given a second laser treatment on day 4. In the remaining rats, if the IOP difference was less than 8 mm Hg on day 7, they, too, received a second laser treatment. Some animals had persistently raised IOP and consequently received only one laser treatment. IOPs were measured in both eyes at baseline, day 1, day 3, day 7, and at least weekly thereafter using a rebound tonometer (TonoLab; Icare, Espoo, Finland) factory calibrated for use in rats. Rats were killed at various time points after treatment by cardiac perfusion with physiological saline under terminal anesthesia. The number of rats analyzed at each time point was as follows: 3 days (n = 6), 7 days (n = 13), 2 weeks (n = 24), 6 weeks (n = 7). No animals were excluded from the present study for reasons relating to inadequate IOP elevation. Five animals were excluded as a result of death caused by anesthesia and four from death caused by hyphema. 

Tissue Processing and Histology

Initially, the brain was removed. Next, each eye with ON, optic chiasm, and the proximal part of the OT attached was carefully dissected. From the dissected tissue, a short piece of ON, 1.5 mm behind the globe, was removed for toluidine blue staining. The brain, globe, and attached short proximal segment of ON, distal ON, chiasm, and proximal segment of OT were fixed in 10% buffered formalin for at least 24 hours. After fixation, the brain was positioned in a rat brain blocker (PA001; Kopf Instruments, Tujunga, CA), and 2-mm coronal slices were taken starting from the rostral and proceeding to the caudal portion of the brain. Brain slices, along with the globe and optic pathway, were then processed for routine paraffin-embedded sections. Globes were embedded sagittally, and ONs and chiasmata were embedded longitudinally. In all cases, 5-μm serial sections were cut. In some animals, the retina and ON were removed for cryosectioning rather than for paraffin sections. These tissues were fixed in 10% formalin for 1 hour and cryopreserved in 30% sucrose overnight, and 7-μm sections were taken using a cryostat. 

The short piece of proximal ON taken for histology was fixed by immersion in 2.5% glutaraldehyde with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for 24 hours at 4°C. It was then placed in 1% osmium tetroxide in saline overnight and was washed with cacodylate buffer at room temperature. Subsequently, the tissue was dehydrated in graded alcohols and embedded in epoxy resin (TAAB Laboratories, Aldermaston Berks, UK) for transverse sectioning. Sections (0.75 μm) were cut on an ultramicrotome, mounted on glass slides, and enhanced with osmium tetroxide-induced myelin staining using 1% toluidine blue. 

Immunohistochemistry

Paraffin-Embedded Sections.

Tissue sections were deparaffinized in xylene and rinsed in 100% ethanol before treatment with 0.5% H₂O₂ for 30 minutes to block endogenous peroxidase activity. Antigen retrieval was achieved by microwaving the sections in 10 mM citrate buffer (pH 6.0). Tissue sections were then blocked in phosphate buffered saline (PBS) containing 3% normal horse serum and incubated overnight at room temperature in primary antibody (containing 3% normal horse serum) followed by consecutive incubations with biotinylated secondary antibody (Vector, Burlingame, CA) and streptavidin-peroxidase conjugate (Pierce, Rockford, IL). Color development was achieved with 3′-,3′diaminobenzidine. Sections were counterstained with hematoxylin, dehydrated, and mounted. 

For immunohistochemical double labeling of iba1 and OX-6, visualization of OX-6 was achieved using a three-step procedure (primary antibody, biotinylated secondary antibody, streptavidin-conjugated AlexaFluor 488), whereas iba1 was labeled by a two-step procedure (primary antibody, secondary antibody conjugated to AlexaFluor 594). In summary, sections were treated as described except for the omission of the endogenous peroxidase block and then were incubated overnight at room temperature with anti-iba1 and anti-OX-6. On the following day, sections were incubated with AlexaFluor donkey anti–rabbit IgG 488 (1:250) together with biotinylated anti–mouse IgG antibody (1:250; Vector) for 30 minutes, followed by streptavidin-conjugated AlexaFluor 594 (1:500; Invitrogen, Carlsbad, CA) for 1 hour before mounting (Fluorescence Mounting Medium; Dako, Carpinteria, CA). 

Cryosections.

Tissue sections were initially rinsed in PBS. For OX18, they were then postfixed for 5 minutes in acetone, which improved the signal-to-noise ratio of staining. Sections were washed in PBS before treatment with 0.5% H₂O₂ for 30 minutes to block endogenous peroxidase activity. Antigen retrieval was not required. Subsequently, sections were treated as for paraffin sections. For immunohistochemical double labeling of cryosections, sections were incubated overnight at room temperature with anti-iba1 and either anti-OX18 or anti-OX42. On the following day, sections were treated as for double labeling in paraffin sections. 

Antibodies.

The following primary antibodies were used in the study: anti–mouse cd11b (1:2000, OX-42; Serotec, Raleigh, NC), anti–rabbit CD3 (1:3000, AO452; Dako), anti–rabbit ionized calcium-binding adapter molecule-1 (iba1, 1:50,000, 019–19741; Wako), anti–mouse ED1 (1:500, MCA341; Serotec), anti–mouse major histocompatibility complex (MHC) class II (1:500, OX-6; Serotec), anti–mouse MHC class I (1: 10,000, OX-18; Serotec), anti–mouse SMI-32 (1:10,000; Sternberger, Baltimore, MD). For double labeling, the antibodies were used at the concentrations listed with the exception that iba1, when used in the two-step procedure, was diluted to 1:5000. 

Evaluation of Histology and Immunohistochemistry

All assessments of ON injury were performed in a randomized, blinded manner. Loss of RGC axons in the ONs of glaucomatous eyes was assessed using a previously developed semiquantitative ON grading scheme based on toluidine blue-stained cross-sections. ³⁸ In summary, zones of equal damage were defined on pictures showing the entire cross-section taken with the 10× microscope objective. The percentage of the nerve cross-section occupied by each zone was determined using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). Representative photographs of each zone were then taken using a 40× objective. The severity of damage (0%, 15%, 30%, 45%, 60%) was then estimated for each zone by two independent graders using the templates shown by Chauhan et al. ³⁸ The overall axonal damage was calculated by summing the products of the mean percentage of damage within each zone and the area of the nerve occupied by the zone. This number was then multiplied by 10 and rounded to obtain an entire number between 0 and 10. Hence, grade 0 corresponds to no damage, grade 5 to 50% axon loss, and grade 10 to 100% axonal loss. Of note, if the calculated damage grade was zero but the nerve contained at least 20 damaged axons within the whole cross-section, the grade was recorded as 1 as a nominal indication that the nerve was damaged. An example of optic nerve grading is shown in Supplementary Figure S1. 

To ascertain whether ON injury correlated with the microglial response, for each animal, transverse sections of the proximal ON were graded for severity of damage by conventional examination of toluidine blue–stained transverse sections (as described), whereas longitudinal sections of the adjacent medial ON were immunolabeled for nonphosphorylated neurofilament heavy (SMI-32) and for the microglial markers iba1 and ED1. Immunostained sections, each expressing a representative level of immunoreactivity, were photographed at 200× magnification. They were then imported into NIH Image J 1.42q software, where they underwent color deconvolution to separate the diaminobenzidine reaction product from hematoxylin counterstain. ³⁹ SMI-32 and ED1 images were subsequently analyzed with regard to the specifically stained area in pixels using the in-built functions of the ImageJ software. For iba1, the number of nuclei with immunoreactive perikarya and processes was counted. Statistical analysis of correlations was performed (Prism 5.0b; GraphPad Software Inc., La Jolla, CA) using nonparametric tests (see Fig. 9 for the results). 

Results

Validation of the Experimental Model of Glaucoma

The present model gave rise to a consistent elevation of IOP (Fig. 1A), with the peak pressure most commonly occurring at day 1 after lasering (35.2 ± 8.0 mm Hg [mean ± SD]). Most eyes needed a second treatment at day 4 or 7 to maintain elevated IOP levels for longer than 2 weeks. By 3 weeks, IOP values had returned to basal levels (Fig. 1A). The amount of ON damage, as assessed by semiquantitative grading of toluidine blue–stained transverse sections, ³⁸ ranged from nominal damage to an axonal injury grade of 6, equating to a loss of axons of approximately 60% (see Figs. 1B–D for representative images). The mean axonal injury grade at 1 week was 1.6 ± 1.3, whereas at 2 weeks, the mean axonal injury grade was 1.8 ± 1.4. These figures equate to a loss of axons of approximately 16% and 18%, respectively, and are consistent with those recorded in other studies ^{40 –42} with similar elevations of IOP. 

Figure 1.

 

(A) IOP profiles of treated (red) and control (gray) eyes after the induction of experimental glaucoma by unilateral laser treatment. Data are expressed as mean area ± SD, where n = 44 (days 0, 1 and 7), n = 31 (days 8 and 14), and n = 7 (days 21, 35, and 42). (B–F) Transverse sections of representative ONs stained with toluidine blue showing different severities of ON injury. Control ON (B), treated ON showing moderate damage (grade 2; C), treated ON showing severe damage (grade 6; D). High-magnification images of control (E) and damaged ONs (F). Scale bars: 25 μm (A–D); 10 μm (E, F).

Spatiotemporal Characterization of Microglial Activation

When considering the microglial response during experimental glaucoma, some general points are worth making regarding the disease model. First, the induction of elevated IOP results in a chronic, relatively mild axonal injury. It is not thought to cause a focal lesion as occurs in other well-characterized paradigms of ON damage, including ON crush or transection, models of ON ischemia, or EAE. Second, unambiguous identification of the site of injury in human and rodent models of glaucoma has proved difficult, but some evidence suggests the ONH region is the primary site. ⁴³ Third, considerable variability exists among animals with regard to the extent of axonal loss and the microglial response. The photomicrographs shown are from representative animals, but other rats killed at the same time points displayed lesser or greater damage. 

Retina.

Previous studies have shown microglial activation in the rat retina after elevation of the IOP induced by cauterization of the episcleral veins. ^32,33,44 Similar results were obtained in the present study after the induction of raised IOP caused by laser photocoagulation of the trabecular meshwork. In control retinas, a relatively sparse population of ramified iba1-positive microglia was present, predominantly within the inner plexiform layer (Fig. 2A). ED1 immunoreactivity was not observed (Fig. 2B). At 3 days after lasering, iba1-positive microglia were marginally more numerous and were labeled more robustly (data not shown). By 7 days, iba1-labeled microglia were abundant within the inner retina and were observed in close proximity to RGC bodies and their axons located in the nerve fiber layer (Fig. 2C). A proportion of these microglia were now ED1 positive (Fig. 2D). This coincided with a decrease in the number of RGCs immunopositive for Brn-3, a transcription factor downregulated before cell death (data not shown). The pattern of iba1 immunoreactivity was similar at 2 weeks after lasering, whereas ED1 immunoreactivity was more prevalent (Figs. 2E, 2F). By 6 weeks, ED1-positive cells were no longer evident, and the density of iba1-labeled cells had decreased markedly (Figs. 2G, 2H). Indeed, in some animals, iba1 immunolabeling was indistinguishable from control rats, although in other animals with severe degeneration an enhanced microglial presence was retained. Unlike in the white matter (see below), most iba1- and ED1-positive microglia in the retina retained a ramified morphology at all time points after the elevation of IOP. 

Figure 2.

 

Microglial activation within the retina after the induction of experimental glaucoma. In the retinas of control rats, a relatively sparse population of ramified iba1-positive microglia was present within the inner retina (A). No ED1-positive cells were found (B). In treated retinas, an increased number of iba1-labeled cells was evident that was notable in the nerve fiber layer (C, E, G). Peak activity was noted at 2 weeks after lasering (E). Subtle signs of activation were still present at 6 weeks (G). Microglia were also immunopositive for ED1 (D, F, H). Maximum activation was noted at 2 weeks (F). ED1 was no longer present 6 weeks after the insult (H). NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer. Scale bar, 50 μm (A–H).

Optic Nerve and Tract.

Throughout the ON and OT of control rats, numerous iba1-positive microglia with small cell bodies and delicate processes were noted (Figs. 3 4 5 6 7–8). These cells were commonly aligned parallel to the axon bundles in the ON and OT, although in the laminar region of the ONH they were sometimes aligned perpendicularly to axons. As in the retina, minimal ED1 immunolabeling was detectable (Figs. 4 5 6 7–8). 

Figure 3.

 

Microglial activation and axonal damage 3 days after the induction of experimental glaucoma. After 3 days of experimental glaucoma, iba1-positive microglia in the ONH (B), ON (F) and OT (J) showed cell body hypertrophy and retraction of processes and were more numerous than in the ONH (A), ON (E), and OT (I) of control rats. In addition, SMI-32 abnormalities were clearly visible in the ONH region (D, showing loss of background staining and subtle neurofilament abnormalities), ON (H), and OT (L), compared with control rats (C, G, K). Scale bars: 100 μm (A–D); 25 μm (E–L).

Figure 4.

 

Microglial activation and axonal damage within the ONH and proximal ON 7 days after the induction of experimental glaucoma. Sections taken from a control (A–C) and an injured (D–M) animal are shown. Microglial activation and increased microglial density were apparent throughout the ONH and proximal ON, as evidenced by robust labeling for iba1 (D) and ED1 (E). In addition, macrophages were present in the vitreous humor (D, E, red arrows). Higher microglial activity and increased axonal cytoskeletal abnormalities (F, black arrows) occurred at the transitional zone between unmyelinated and myelinated axons. This region is shown at higher magnification in (G–I): iba1 (G), ED1 (H), SMI32 (I). Double-labeling immunofluorescence of ED1 (J, K, red) and myelin basic protein (MBP; K, green) and of SMI-32 (L, M, red) with MBP (M, green) was performed to reveal the precise site of myelination. This confirmed that greater microglial activity and increased axonal cytoskeletal abnormalities were present in the myelinated portion of the ON. Asterisk: note the gap in axonal fibers, caused by a penetrating blood vessel also visible in (D–F). Scale bars: 200 μm (A–F, J–M); 50 μm (G–I).

Figure 5.

 

Temporal characterization of microglial activation and axonal cytoskeletal damage in the ON during experimental glaucoma. Sections taken from the distal ON in representative animals are shown. Control (A–C), 7 days (D–F), 2 weeks (G–I), and 6 weeks (J–L) after the induction of elevated IOP. In control rats, iba1-labeled microglia showed a classical ramified morphology (A) and were ED1-negative (B). Axonal fibers are homogenously labeled by SMI-32 (C). At 7 days, a greater number of iba1-positive microglia was noted (D), together with expression of ED1 (E), whereas numerous axons showed SMI-32 abnormalities (F). There was an increased number of iba1- and ED1-positive microglia at 2 weeks (G, H) and again at 6 weeks (J, K). Note the change in morphology of ED1-positive microglia at 6 weeks; they have a foamy appearance indicative of phagocytic activity (K). In contrast, abnormal SMI-32 immunolabeling gradually decreased at later time points (I, L), leaving a reduced number of lightly stained, surviving axons. Scale bar: 50 μm (A–L); 25 μm (inset).

Figure 6.

 

Quantification of iba1 and ED1 immunoreactivities in the distal ON in control rats at 1, 2, and 6 weeks after the induction of experimental glaucoma. There was a gradual increase in iba1 immunoreactivity over time compared with controls. ED1 immunoreactivity was minimal in control ONs, increased steadily during the first 2 weeks, and was dramatically higher by 6 weeks. Data are expressed as mean area ± SEM.

Figure 7.

 

Temporal characterization of microglial activation and axonal cytoskeletal damage in the OT during experimental glaucoma. Sections taken from representative animals killed 1 week (A–C), 2 weeks (D–F), and 6 weeks (G–I) after the induction of elevated IOP are shown. Arrows: boundaries of the OT. In the 7 day animal (grade 1), iba1-labeled microglia displayed retraction in processes indicative of activation (A). There was no evidence of ED1-positive microglia within axonal tissue (B); however, perivascular staining was observed (inset). Axonal damage, denoted by intensely stained SMI-32 abnormalities, was clearly evident (C). In the 2-week rat (grade 2), an increased number of iba1-positive microglia was noted (D), together with the expression of ED1 (E) and numerous SMI-32 abnormalities (F). In the 6-week rat (grade 5), robust labeling for iba1 (G) and ED1 (H) was observed throughout the OT. There were few SMI-32 abnormalities, but the entire right side of the OT showed axonal loss. Note that in the region of heavy axonal loss, there was greater microglial activity. Scale bar: 100 μm (A–I); 10 μm (inset).

Figure 8.

 

Microglial response at the level of the optic chiasm 2 weeks after the induction of experimental glaucoma. In control rats, iba1 labeled ramified, quiescent microglia throughout the distal ON and OT (A), whereas no ED1 immunoreactivity was detectable (C). In treated rats, increased microglial density, morphologic changes, and expression of ED1 were observed within the injured ON and the contralaterally projecting OT (E, iba; G, ED1). Areas within the boxed regions in (A), (C), (E), and (G) are shown at higher magnification in the accompanying images to the right (B, D, F, H). Of interest, there was minimal evidence of any injured, noncrossing fibers in the ipsilaterally projecting OT. Scale bars: 400 μm (A–D); 200 μm (E–H).

By 3 days after the induction of experimental glaucoma, iba1-positive microglia in the ONH (Figs. 3A, 3B), ON (Figs. 3E, 3F), and OT (Figs. 3I, 3J) typically showed some retraction of processes and hypertrophy of the cell body and were marginally greater in number than in controls. There was no evidence of a spatial gradient in microglial activity along the optic pathway. The increased microglial activity was concurrent with early indications of axonal cytoskeleton damage, as evidenced by abnormalities in nonphosphorylated neurofilament heavy (SMI-32), at the ONH (Figs. 3C, 3D), and along the extent of the ON (Figs. 3G, 3H) and the OT (Figs. 3K, 3L). 

At 7 days after lasering, iba1-positive microglia throughout the proximal (Fig. 4) and distal (Fig. 5) ON and OT (Fig. 7) were more numerous than at 3 days and displayed a partially activated morphology, with a proportion of microglia expressing ED1-positive granules. In the proximal segment of the ON, there was a regional disparity in the microglial response. More iba1 (Figs. 4D, 4G) and ED1 (Figs. 4E, 4H) activities were typically observed in the retrobulbar ON than in the ONH. This was also the case for SMI-32 immunolabeling, which featured more abnormalities in the retrobulbar ON (Figs. 4F, 4I). Double-labeling immunofluorescence of myelin basic protein with ED1 and SMI-32 demonstrated that the increases in microglial activity and axonal cytoskeletal abnormalities occurred at the transition zone, where the axons become myelinated (Figs. 4J–M). 

From 7 days to 6 weeks, throughout the myelinated ON and OT, there was a gradual increase both in the number of microglia and in their activation state. This is illustrated in Figures 5 and 6. Figure 5 shows sections from the distal ON of representative animals killed at 1, 2, and 6 weeks each with moderate ON damage (grade 2) according to toluidine blue grading, whereas Figure 6 is a quantitative representation of the overall results. At 1 week after injury, ongoing neurofilament dephosphorylation was prevalent (Fig. 5F), accompanied by a rise in the number of iba1-microglia (Figs. 5D, 6) but relatively modest ED1 accumulation (Fig. 5E, 6) compared with controls (Figs. 5A–C). At 2 weeks after lasering, neurofilament dephosphorylation was similar to that seen at 1 week (Fig. 5I), and there was a moderate increase in the number of iba1- and ED1-positive microglia (Figs. 5G, 5H, 6). By 6 weeks after lasering, ongoing neurofilament dephosphorylation had decreased markedly in most animals (Fig. 5L); however, the number of iba1-positive cells was higher than at 2 weeks (Figs. 5J, 6), and the abundance of ED1 was dramatically greater (Figs. 5K, 6). ED1-positive microglia displayed an activated morphology, with some cells adopting a macrophage-like, foamy appearance (Fig. 5K). 

Findings similar to those found in the ON were noted in the OT. Figure 7 shows sections through the OT (at bregma −3.10) in representative rats killed at 1, 2, and 6 weeks. The number of iba1-labeled microglia increased with time after lasering and severity of injury (Figs. 7A, 7D, 7G). Very few ED1-positive cells were noted in the 1-week-old rat (Fig. 7B), but increased perivascular ED1 staining (Fig. 7B, inset) was seen. ED1 abundance was greater in the 2-week-old animal (Fig. 7E) and was markedly higher in the 6-week-old rat (Fig. 7H), which, like many animals, featured asymmetric damage (Figs. 7G–I). 

At the optic chiasm, most ON axons from each eye meet, cross the midline, and project to the contralateral OT. In the adult rat, 5% to 10% of RGC axons do not cross the midline of the optic chiasm, projecting instead to the ipsilateral half of the brain. Figure 8 shows iba1 and ED1 immunoreactivities at the level of the optic chiasm in a control rat (Figs. 8A–D) and in a rat that underwent induction of experimental glaucoma 2 weeks previously (Figs. 8E–H). A robust microglial response was seen in the ipsilateral ON and contralateral OT in the treated animal; however, there was negligible evidence for any microglial activation in noncrossing fibers in the ipsilaterally projecting OT or indeed of any axonal cytoskeleton breakdown in the ipsilateral OT (data not shown). This was the case at all time points. 

Correlation between Microglial Activation and Optic Nerve Injury

We were interested in whether ON injury correlated with the microglial response, particularly at early time points featuring mild or moderate pathologic changes. To achieve this aim, rats were killed at 1 and 2 weeks after the induction of experimental glaucoma. Transverse sections of the proximal ON were graded for severity of damage by conventional examination of toluidine blue–stained transverse sections, whereas longitudinal sections of the adjacent medial ON were immunolabeled for SMI-32 and for the microglial markers iba1 and ED1. Iba1 labels both quiescent and activated microglia, thus providing an index of microglial density, whereas abundance of the lysosomal antigen ED1 offers a measure of microglial phagocytic activity. Subsequently, both measures of axonal injury were correlated with each microglial marker. The overall results showed a highly statistically significant correlation between axonal injury and the microglial response (Fig. 9; Supplementary Table S1). In summary, the greater the degree of ON injury (whether assessed by grading of toluidine blue–stained sections or by immunostaining for SMI-32), the greater the number of microglia and the greater the abundance of ED1 granules expressed by microglia. A number of specific observations can also be made: first, ON damage as assessed by grading of toluidine blue–stained transverse sections showed a greater correlation with both microglial markers at the 2-week than at the 1-week time point; second, the correlation was stronger for ED1 than for iba1; third, ON damage as assessed by extent of abnormal SMI-32 immunostaining correlated better with iba1 than with ED1. 

Figure 9.

 

Correlations between axonal injury and iba1 and ED1 immunoreactivities in the ON after the induction of experimental glaucoma. (A, B) Correlations between semiquantitative grading of toluidine blue–stained transverse sections of the proximal ON and number of iba1 microglia or abundance of ED1 immunoreactivity in longitudinal sections of the distal ON at (A) 1 week and (B) 2 weeks. (C) Correlations between the amount of abnormal SMI-32 staining and the number of iba1 microglia or abundance of ED1 immunoreactivity in longitudinal sections of the distal ON at 2 weeks. Each data point represents one animal.

Expression of Immunologic Cell Surface Markers

The morphology and distribution of complement receptor type 3 (cd11b, OX42) immunoreactive microglia in control and injured tissues closely matched those of iba1. In control animals, OX42 was expressed in ramified microglia with small cell bodies and delicate processes, which, in the ON, were typically aligned parallel to the nerve bundles (Fig. 10A). At 1 week (data not shown) and 2 weeks (Fig. 10B) after the induction of experimental glaucoma, OX42-positive microglia throughout the optic pathway were more numerous and tended to have larger cell bodies and shorter, thicker processes. Double-labeling studies showed that OX42 was exclusively localized to iba1-positive microglia; however, a small percentage of iba1-positive microglia were not immunolabeled by OX42 (Figs. 10C, 10D). 

Figure 10.

 

Microglial expression of complement type 3 receptor (OX-42, A–D), MHC class I (OX18, E–H), and MHC class II (OX-6, I–L) during experimental glaucoma. (A, E, I) Control ONs. The remaining images are of sections from moderately damaged ONs 2 weeks after the induction of elevated IOP. OX-42 is constitutively expressed by quiescent microglia in the normal ON (A) and is persistently upregulated on activation (B–D). Interestingly, a few iba1-positive microglia did not express cd11b (D). OX-18 is not exclusively, although it is predominantly, expressed by microglia in control ONs (E). Marked upregulation occurs in activated microglial cells (F–H). OX-6 is absent from control ONs (I) but is robustly and exclusively expressed by activated microglia (J–L). Scale bars: colorimetric images, 50 μm; immunofluorescence images, 25 μm.

In control animals, MHC class I (OX18) lightly stained a population of cells that had the morphology of OX42- and iba1-positive microglia, featuring small cell bodies and fine processes (Fig. 10E). OX18 also faintly labeled a few cells with lateral processes that bore greater resemblance to astrocytes. After 1 week (data not shown) and 2 weeks (Fig. 10F) of experimental glaucoma, there was a marked upregulation in the number of OX18-labeled cells that showed an altered morphology analogous to that observed for OX42. Double-labeling studies revealed that most OX18-positive cells colocalized with iba1-positive microglia (Figs. 10G, 10H). 

MHC class II (OX6) was absent from the optic pathway in control animals (Fig. 10I) except for the presence of occasional perivascular microglia, which constitutively express MHC class II. After the induction of experimental glaucoma, numerous OX6-positive cells with the morphology of OX42- and iba1-positive cells were found throughout the ON. Cells were typically lightly stained at 1 week but were strongly labeled by 2 weeks (Fig. 10J). Double labeling demonstrated that the induction of OX6 expression occurred exclusively in iba1-positive microglia (Figs. 10K, 10L). Interestingly, and unlike OX42 or OX18, there was a spatial difference in the expression of OX6 within the optic pathway. No expression of OX6 was found in the retina at any time point after the induction of experimental glaucoma, even in animals that featured numerous ED1-positive retinal microglia, indicating ongoing neuronal damage and phagocytic activity. This finding is illustrated in Figure 11, which shows adjacent sections from the retina (Figs. 11A, 11B), ONH (Figs. 11C, 11D), and proximal ON (Figs. 11E, 11F) from one representative animal killed after 2 weeks and stained for ED1 and OX6. ED1 immunoreactivity was seen in all three locations. In contrast, robust labeling of OX6 was apparent in the ON, but only very limited expression of OX6 was evident in the ONH, and no staining was detectable in the retina. Of note, OX6 labeling in the neck of the ONH varied markedly between animals, with most rats showing few OX6-positive cells while others featured several OX6-labeled microglia. The presence or absence of OX6 immunoreactivity seemed unrelated to the severity of ON injury or grade of microglial activation within the ON. 

Figure 11.

 

Spatial pattern of OX6 expression during experimental glaucoma. Adjacent sections from one representative animal killed 1 week after the induction of elevated IOP and stained for ED1 (A, C, E) and OX6 (B, D, F) are shown. Three locations are shown: midperipheral retina (A, B), ONH (C, D), and proximal ON (E, F). Numerous ED1-positive ramified microglia were present within the retina (A). ED1-labeled cells were somewhat reduced in density in the prelaminar and laminar portions of the ONH (C, arrows) but were numerous within the myelinated portion of the proximal ON (E). Interestingly, the pattern for OX6 was strikingly different from that for ED1. OX6 was absent from the retina (B). Note, however, the presence of many OX6-positive cells in the choroid (B, arrows). Expression of OX6-positive microglia in the ONH varied between animals, but typically only sparse labeling was observed, as in this rat (D, arrow). In the proximal ON, OX6 was robustly expressed on cells with the distinct features of activated microglia (F). Scale bar, 50 μm (A–F).

Infiltration of Leukocytes

Very limited T-cell infiltration was found in the damaged ON (Figs. 12A, 12B) or retinas (data not shown) using an antibody that recognizes the pan T-cell marker CD3. The number of infiltrating cells tended to increase both with the severity of injury and with the passage of time after insult (Fig. 12C). The increased density of T cells observed in perivascular areas (Fig. 12D) was consistent with invasion from blood vessels. No infiltration of neutrophils, as defined by immunoreactivity to myeloperoxidase, was found in injured ONs during experimental glaucoma (data not shown). 

Figure 12.

 

(A–D) Infiltration of T cells in the distal ON during experimental glaucoma. Occasional T cells, as identified by the pan T-cell marker CD3, are present in control ONs (A). Very limited infiltration occurred in moderately damaged ONs after 2 weeks (B). Severely damaged (grade 5) nerves at 6 weeks show greater T-cell infiltration (C). This is particularly evident around large blood vessels (D). (E–H) Presence of macrophages during experimental glaucoma. Immunolabeling for iba1 (E, G) and ED1 (F, H) in a moderately (grade 2; E, F) and a severely (grade 5; G, H) injured ON 6 weeks after the induction of elevated IOP showed occasional cells with the morphologic features of macrophages (large, round cells with foamy cytoplasm but no processes; E–H, arrows). However, most cells resembled phagocytic resident microglia, even in the severely damage ON. Of note: clear distinction between the two populations is not possible based on iba1 and ED1 immunolabeling. Scale bars: 100 μm (A–C); 25 μm (D–H).

Rat models of traumatic, ischemic, and autoimmune optic neuropathies all share a common pathologic feature: the presence of a focal lesion in the proximal ON with accompanying infiltration of hematogenous macrophages. We addressed the question of whether laser-induced experimental glaucoma also results in the presence of a focal lesion within the ON with accompanying infiltration of macrophages. After 1 and 2 weeks, there was an absence of ED1- and iba1-positive cells with the physical characteristics of macrophages in any part of the optic pathway, suggesting that the blood brain barrier remains intact in this disease paradigm. At 6 weeks, there was evidence of some cells with a macrophage-like morphology within the degenerating ON (Figs. 12E–H), particularly in rats with more severe damage, which corresponded with the disappearance of myelin basic protein. These cells could be resident microglia that had fully transformed into phagocytic macrophages or infiltrating macrophages. 

Discussion

There is increasing evidence that microglia play a central role in chronic degenerative conditions of the CNS, including Alzheimer's and Parkinson's diseases, multiple sclerosis, amyotrophic lateral sclerosis, and many others. ¹² In the present study we have, for the first time, described the microglial response in the optic pathway of rats with experimental glaucoma. Microglial activation occurred along the entire optic pathway and correlated closely with axonal damage. It was, however, not accompanied by any overt infiltration of neutrophils. Microglia upregulated the expression of immunologic cell surface markers, including complement receptor type 3, MHC class I, and MHC class II, but the expression of MHC class II was limited to cells within the white matter. Despite the increased expression of molecules associated with antigen presentation, only nominal T lymphocyte infiltration was observed. Further studies are warranted to elucidate the role of microglial activation in glaucoma. 

The primary site of injury in glaucoma is yet to be unequivocally identified, but some research points toward the ONH, the site at which RGC axons pass through the connective tissue of the lamina cribrosa. ^43,45,46 Previous studies in rats ³¹ and humans ⁴⁷ have described increased microglial activation in the ONH during glaucoma, but neither study examined other regions of the optic pathway. In the present study, we observed marked axon transport failure at the ONH as early as 3 days after the induction of raised pressure. This was accompanied in the ONH by an alteration in the morphology of iba1-positive microglia from a quiescent to an activated phenotype; yet, there was no indication of a specific or preferential microglial activation at this location at this time point. Throughout the retina, ON, and OT, iba1-positive microglia also showed some retraction of processes and cell body hypertrophy. Similarly, at 7 and 14 days, microglia in the ONH were more numerous and upregulated expression of the lysosomal antigen ED1. Again, these events were ostensibly concurrent in the other regions of the optic pathway. Indeed, somewhat contrary to expectation, there was a clear trend of greater microglial activity beyond the point of myelination in the retrobulbar ON that occurs immediately distal to the ONH. This corresponded with the finding of more numerous SMI-32 abnormalities in the myelinated region of the ON. The combined results are consistent with the results of a study by Schlamp et al., ³⁶ who analyzed axonal degeneration in the DBA/2J mouse model of glaucoma and documented greater structural preservation in the ONH compared with more distal segments of the pathway. 

Five percent to 10% of RGC axons in the adult rat do not cross the midline of the optic chiasm, projecting instead to the ipsilateral half of the brain. These fibers originate from the inferior-temporal crescent of the retina. ^48,49 In the present study, analysis of the optic chiasm revealed minimal evidence of any microglial activation in noncrossing fibers in the ipsilaterally projecting OT at any of the time points analyzed. The obvious conclusion to draw is that there is negligible death of RGCs in the region of the inferior-temporal retina from which these fibers originate. The results confirm and extend earlier observations of preferential damage in the superior segment of the retina and ON in the hypertonic saline and laser models of glaucoma. ^50,51  

One important goal of the present study was to shed light on the temporal relationship between microglial activation and axonal loss in the ON. Two related aspects were of interest: first, to ascertain whether alterations in microglial markers preceded, were concomitant with, developed soon after, or were significantly delayed after axonal cytoskeletal damage; second, to evaluate whether microglial activation can be used in neuroprotection studies as a surrogate or an adjunct marker for ON injury, particularly at early time points featuring mild or moderate pathologic changes (1 and 2 weeks) when axon counting is less reliable. Microglia are known to be very sensitive to disturbances in milieu homeostasis, neuronal function, and disruption, ³ which would seem to put them in an ideal position to quantify the severity of early axonal damage. 

With regard to the onset of microglial activation in the ON, at 3 days after the induction of glaucoma, the earliest time point analyzed, axonal loss as determined by conventional analysis of toluidine blue–stained ONs was marginal, but SMI-32 immunolabeling revealed subtle abnormalities in some axons. Expression of ED1, the rodent equivalent of CD68 whose presence is considered indicative of phagocytosis, was negligible. Conversely, iba1 immunostaining showed clear evidence of microglial activation, as discussed. Iba1 is increasingly used as a marker of microglial activation. It has been shown to be upregulated in a time-dependent manner after injuries such as axotomy ⁵² and focal cerebral ischemia ⁵³ and is thought to contribute to microglial cell migration. ⁵⁴ The present results show iba1 to be a highly sensitive marker of damage in the ON, which allows identification of early pathologic axonal changes soon after injury, before any overt axonal loss and phagocytosis have occurred. 

To ascertain whether iba1 or ED1 might prove useful in neuroprotection studies as quantitative markers for ON injury, we correlated each antigen with two complementary measures of axonal injury, grading of toluidine blue–stained transverse sections, which is the most frequently used method for estimating axonal loss ² and abundance of SMI-32 abnormalities in longitudinal sections of the ON. SMI-32, which labels nonphosphorylated NF-H, has been shown by others to be an excellent marker of axonal injury ^55,56 and, in our hands, was more sensitive than other axonal cytoskeletal proteins for early detection of pathologic changes in the ON (unpublished observations). The overall results showed a highly statistically significant correlation among all four combinations; nevertheless, minor differences were evident. The first disparity was that ON damage as assessed by toluidine blue staining correlated better with both microglial markers at the 2-week than at the 1-week time point. This may simply be due to the higher number (n) at the later time point. Alternatively, it may relate to the greater reliability of both the grading system and the microglial response in situations of increased injury. Two further observations were that the correlation with ON grading was stronger for ED1 than for iba1, whereas injury as assessed by SMI-32 immunostaining correlated better with iba1 than with ED1. These findings are understandable when viewed from a physiological perspective. Early axonal injury, comprising axonal transport disruption and dephosphorylation of neurofilaments but little measurable axonal loss, would result in increases in SMI-32 abnormalities and iba1 expression, but negligible ED1 expression since the phagocytosis of axonal debris is yet to commence. Later time points would comprise morphologically visible axonal breakdown, myelin disruption, and activation of the phagocytic system. Thus, it makes sense that ED1 and ON grading are closely matched. 

The overall results advocate microglial activation as a useful adjunct quantitative tool for assessment of the status of ON damage in neuroprotection studies, although a cautionary note must be added that we do not yet have enough information to discern the precise relationship between microglial activation and neurodegeneration. Nevertheless, previous studies, both in the ON and the brain, have reached a similar conclusion that the extent of microglial activation reflects the severity of injury; for example, Zhang et al. ⁵⁷ showed that the time course of microglial morphologic change after transient middle cerebral artery occlusion paralleled neuronal damage, whereas Kato et al. ⁵⁸ highlighted that microglia were activated in a graded fashion in response to the severity of neuronal injury after temporal forebrain ischemia. The usefulness of microglia as a means of quantifying damage and neuroprotection has also been exploited in chronic models of injury, such as chronic cerebral hypoperfusion, which features slow, progressive damage to CNS white matter tracts. ^{59 –61}  

Rat models of traumatic (ON transection, ⁶² ON crush ⁶³ ), ischemic (anterior ischemic optic neuropathy, permanent occlusion of the common carotid arteries), ²⁹ and autoimmune (EAE) ³⁰ optic neuropathies all share a common pathologic feature: the presence of a focal lesion in the retrobulbar ON with accompanying infiltration of hematogenous macrophages. 

It is known that blood vessels in the retrobulbar ON are permeable to intravascular tracers such as horseradish peroxidase, prompting the question whether there is a true blood-brain barrier at the optic nerve head. ⁶⁴ This inherent weakness of the blood-brain barrier has been suggested to be responsible for the susceptibility of this region to lesion formation in both EAE and multiple sclerosis. ³⁰ We examined whether there is a similar focal lesion with an accompanying presence of macrophages during experimental glaucomatous optic neuropathy. We found no evidence of a focal lesion or macrophage presence in the ON after the elevation of IOP, suggesting that the blood-brain barrier remains intact in this disease paradigm. 

During glaucoma, the appearance of ED1-positive microglia with an ameboid, foamy, phagocytic appearance in the ON and OT was delayed, occurring between 2 and 6 weeks. This paralleled an increase in myelin basic protein disorganization. It is well known that the clearance of myelin debris after axonal cytoskeletal degeneration is a lengthy process in CNS neurodegenerative conditions because of the failure of microglia to develop into fully functional phagocytic cells, ⁶⁵ coupled with the minimal role played by oligodendrocytes. ⁶⁶ Phagocytic removal of tissue debris helps create a pro-regenerative environment. The inefficiency of white matter phagocytosis in the CNS compared with peripheral nerves is thought to account for the lack of neuronal regeneration under physiological conditions. This is clearly the case in glaucoma. 

Microglia are the major cell population in the CNS with the potential to act as antigen-presenting cells. Under normal physiological conditions, the expression of MHC molecules, which are critical for T-cell interactions, is at low or undetectable levels. After injury, microglia increase the expression of MHC class II. In some, but not all, models of acute and chronic injury, this is associated with a limited T-cell response. After the induction of experimental glaucoma, there was an upregulation in MHC class I and II in the ON and OT, which was most clearly evident at 1 week and which slowly declined thereafter. This did not correlate with any significant T-cell infiltration, suggesting either an insufficient presence of costimulatory molecules or that MHC antigens perform functions unrelated to the induction of an immune response. However, in more severely injured animals, we could observe limited infiltration, which is in accordance with the infiltration of only small numbers in nearly all neurodegenerative diseases. ¹⁰ Interestingly, there was a fundamental difference between the gray matter of the retina and the white matter of the ON and the OT: MHC class II was markedly upregulated on white matter microglia, but no expression was observed in retinal microglia. The same observation was made by Rao et al. ⁶⁷ and Zhang and Tso ⁶⁸ using intracranial and intraorbital models of ON transection, namely that MHC class II-positive cells were localized to degenerating myelinated fibers but not to the retina. The explanation for the regional difference is not simply attributed to the presence of injured myelin in the white, but not gray, matter, because ischemia-reperfusion ¹⁸ and kainic acid-induced excitotoxicity ¹⁹ both induced OX6 expression by resident retinal microglia. The explanation may be related to the fact that these latter models cause injury to more than one neuronal class in the retina, which is not the case with ON transection and glaucoma, or that they cause a more pronounced glial reactivity in general. The degree of retinal stress caused by the present model may simply not be sufficient to provoke the expression of MHC class II molecules on resident microglia or to cause invasion by bone marrow-derived macrophages, which in mice have been shown to be more prone to express MHC class II molecules. ⁶⁹  

Supplementary Materials

Supplementary Figure S1 - (PDF) 

Supplementary Table S1 - (PDF) 

Footnotes

 Supported by National Health and Medical Research Council Grant 508123 and by the OPOS Stiftung zugunsten von Wahrnehmungsbehinderten, St. Gallen, Switzerland (AE).

Footnotes

 Disclosure: A. Ebneter, None; R.J. Casson, None; J.P.M. Wood, None; G. Chidlow, None

The authors thank Mark Daymon, Zhao Cai, and Kathy Cash for expert technical assistance.