Oligodendrocyte Dysfunction after Induction of Experimental Anterior Optic Nerve Ischemia

Abstract

purpose. The early response and survival of oligodendrocytes after axonal stroke and their potential contribution to neuronal survival in vivo have not been adequately addressed. The purpose of this study was to investigate the changes occurring in the retina and optic nerve (ON) in anterior ischemic optic neuropathy (AION), using a c-fos transgenic mouse model.

methods. A new mouse model of AION (rodent AION) was developed to evaluate the in vivo stress response of oligodendrocytes and retinal ganglion cells (RGCs) in a transgenic mouse strain, using the immediate early stress-response gene c-fos, RT-QPCR technology, immunohistochemistry, and electron microscopy. Confocal microscopy was used with cell-specific antibodies to characterize the timing of cells responding to rAION. The TUNEL assay detected cells undergoing apoptosis. Ultrastructural changes were analyzed by electron microscopy.

results. In rAION, oligodendrocytes rapidly respond in vivo to ischemic ON damage, with c-fos activation as an early detectable event. Early evidence of progressive oligodendrocyte stress, is followed by demyelination, wallerian degeneration of the ON, and oligodendrocyte and RGC death far from the primary lesion.

conclusions. After rAION induction oligodendrocytes, as well as RGCs, undergo progressive stress, with dysfunction and apoptosis. The findings lead to a proposal that progressive retrograde oligodendrocyte stress, away from the primary lesion, is an important factor after ischemic optic neuropathy. Postinduction demyelination must be addressed for effective neuroprotection of ischemic and hypoxic white matter.

Anterior ischemic optic neuropathy (AION) is an optic nerve (ON) stroke and is the leading cause of sudden ON-related vision loss in the United States, affecting 2.3 to 10.2 per 100,000 people over 50 years of age. ¹ This disorder results from interruption of the blood supply to the anterior part of the ON, ^{2 3 4} leading to death of retinal ganglion cells (RGCs). Because RGCs are central nervous system (CNS) neurons, AION may be considered to be a form of isolated CNS infarct. Thus, understanding the underlying cellular responses in AION may result in a better understanding of other ischemic events involving CNS white matter. 

Patients with AION typically present with sudden, painless loss of vision associated with swelling of the ON head in the affected eye, followed by disruption of the normal nerve architecture, ⁵ RGC death, and permanent vision loss. ^{6 7} Unfortunately, the sequence of events and mechanisms responsible for neuronal death after ischemic axonal injury are still poorly understood, ⁸ despite several animal models to generate various kinds of optic neuropathies, including nerve trauma, ^{7 9 10} glaucoma, ¹¹ and primate combined retinal and ON occlusion similar to arteritic ischemic neuropathy ¹² and several in vivo studies seeking the mechanisms of ischemic damage to central white matter axons using an isolated ON model. ¹³ However, until recently, there has been no appropriate in vivo animal model for evaluating pure axonal ischemia. 

Recently, an AION rat model (rodent AION), closely resembling human AION was described by one of us. ¹⁴ This model makes it possible for the first time to study in vivo, the sequence of events that occur during isolated ON ischemia. In the present report we describe the development of a new mouse model of ON ischemia. Although mice are smaller and more difficult to manipulate in this system than are rats, the ability to use mice enables the use of transgenic technology to help evaluate specific gene contributions and to dissect involvement of specific cellular responses after isolated in vivo CNS axonal ischemia. To analyze the early responses of CNS neurons and their supporting cells, we used a transgenic mouse strain constructed with the c-fos gene promoter linked to a β-galactosidase (β-gal) reporter. ¹⁵ c-Fos is an immediate-early stress-response gene that is rapidly expressed after various types of cellular stress. ^{15 16 17} Expression of c-fos is strongly implicated in the activation program of cell death in various types of neuronal cells. ^{15 18 19} This strain can be used to identify early stress-related changes in the ON and retina that are directly affected by ON ischemia. 

Methods

Mice

All animal protocols were approved by the UMB Institutional Animal Care and Utilization Committee (IACUC) and were handled in accordance with ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. C57B6 (NHSD) wild-type mice were obtained from Jackson Laboratories (Bar Harbor, ME). We obtained a transgenic c-fos/LacZ mouse strain on a C57B6/6J background, containing a transgene constructed by fusion of 500 bp of the c-fos promoter to the β-gal gene. ²⁰ c-Fos induction results in blue-stained cells after incubation with the appropriate agent. The transgenic mouse strain was originally generated in mice with an FVB background, which has an rd (photoreceptor degeneration) phenotype. ²¹ Transgenic animals also possessed the rd allele. Thus, c-fos/LacZ homozygotes express the rd phenotype. Heterozygotes, however, have a normal number of retinal ganglion cells and normal ON morphology. Only data from heterozygotes (CD1XC57B6LacZ) were used for analysis of the results. Mice were housed under standard conditions and fed food and water ad libitum. 

Induction of Mouse AION

AION was induced in ketamine/xylazine (80 and 4 mg/kg, respectively; IP) anesthetized mice age 3 to 4 months. A custom-designed, plastic fundus corneal contact lens was use that enabled direct in vivo visualization of the retina and ON head. After intravenous administration of 0.05 mL of 2.5 mM rose-bengal (RB) in phosphate-buffered saline (PBS), the ON head was illuminated with a frequency doubled yttrium-aluminum-garnet (YAG) diode laser (535 nm), 300-μm spot size, 50-mW power, for a constant duration. One eye of each animal was left untreated as an internal control. Other negative controls included anesthesia alone, RB without laser treatment, and laser illumination without RB injection. As an additional positive injury control, we surgically dissected the orbit, directly identified the ON, and crushed the ON with a jeweler’s forceps. After induction, animals were allowed to recover, and the retina and ON of the living animals were photographed with a digital camera (D1X ver. 5.3 Mpixel; Nikon Corp., Tokyo, Japan). Animals were subsequently euthanatized by CO₂ inhalation followed by cervical dislocation at appropriate intervals, after which tissues were harvested and prepared for appropriate assays. As a positive control, we crushed the ON of c-fos/LacZ mice and evaluated the temporal expression of ON/β-gal. 

India Ink Retinal Vasculature Tracing

To evaluate whether rAION results directly in capillary closure, we perfused deeply anesthetized mice by intracardiac puncture with 3 mL India ink, after sectioning the inferior vena cava. After decapitation, eyes were isolated, fixed in 4% paraformaldehyde-PBS, cleared with dehydration through xylene, and flatmounted for direct microscopy. 

Retinal Fluorescein Angiography

To measure blood–retinal barrier breakdown, we used fluorescein angiography. The degree of fluorescein in the vitreous is reflective of the degree of retinal vascular leakage and manifests as indistinct vascular borders progressing to diffuse hazy fluorescence. Five days after induction, the mouse pupils were dilated with one drop of 1% atropine sulfate, and 0.1 mL 25% fluorescein (AK-Fluor 25% AMP; Akorn, Decatur, IL) was injected intraperitoneally. Successive photographs of the right and left retina were taken with a small animal fundus camera (Genesis; Kowa, Tokyo, Japan, with Elite Chrome 400 film; Kodak, Rochester, NY). First photographs were obtained when the fluorescence was evident in the eye, typically within 20 seconds of fluorescein injection. The time lapse between alternating right and left eye retinal photographs averaged 10 seconds. 

Detection of c-Fos Expression

After treatment, animals were euthanatized at 1, 3, 6, 9, 14, 21, 30, 45, and 60 days by CO₂ inhalation, followed by cervical dislocation. After euthanasia, the globes, ONs, and brain were dissected and fixed for 24 hours in 2% paraformaldehyde-2% PIPES (piperazine-N-N′-bis(2-ethanesulfonic acid)) buffer. LacZ activity was detected as previously reported. ²² Retinas were either flatmounted on slides or paraffin fixed and 6-μm-thick sections used for analysis and staining. LacZ-positive RGCs and ON cell bodies were counted at magnifications of 40× (flatmount) and 100× (section). 

Histologic Analysis and RGC and ON Cell Quantitation

After rAION induction, animals were euthanatized at 1, 3, 6, 9, 14, 21, 30, 45, and 60 days by CO₂ inhalation followed by cervical dislocation. Globes and ONs were dissected and fixed for 24 hours in 4% paraformaldehyde-PBS. Tissues were paraffin embedded, sectioned, and stained with H&E. Retinal step-cut (300-μm interval) cross sections were collected from five animals at each time period. The number of RGCs in 18 sections (three sections/step × six steps) were collected and averaged from each retina. Statistical analysis was performed using the t-test for two independent samples, to evaluate the significance of the difference between RGC counts in treated and control eyes. Longitudinal sections were cut from the ONs and analyzed for c-fos expression (blue cell count) and cellular architecture changes. 

cDNA Preparation

We used a two-stage reverse-transcriptase–based quantitative polymerase chain reaction (RT-QPCR) to evaluate c-fos and myelin basic protein (MBP) gene expression changes after rAION induction. ONs were dissected from groups of three mice at each time point and snap frozen in dry ice. Total RNA was isolated with a mini kit (RNeasy; Qiagen, Hilden, Germany), with DNase I (RNase free DNase; Ambion, Austin, TX) treatment to remove genomic DNA contamination. Approximately 3 μg of total RNA per time point was used for generating complementary DNA (Retroscript kit; Ambion). 

Quantitative Real-Time Polymerase Chain Reaction

The Sequence Detection System (Prism 7900; Applied Biosystems, Inc. [ABI], Foster City, CA) was used to perform RT-QPCR analysis of mouse MBP and c-fos. The primers used in the study are shown in Table 1 . 

Mouse glyceraldehyde phosphate dehydrogenase (GAPDH) was used to normalize the cDNA input levels. Reactions were performed in a 20-μL volume containing 3 μL cDNA, 0.5 μM each of forward and reverse primers, and buffer included in the master mix (SYBR Green I; Qiagen). Duplicate RT-QPCR reactions were performed for each gene, to minimize individual tube variability. Standard curves were obtained using untreated mouse cDNA for each gene PCR assay. 

PCR cycling conditions were an initial denaturation step of 95°C for 10 minutes followed by 50 cycles of 1 minute denaturation at 95°C and 1 minute of annealing and extension at 60°. Sample runs were performed in duplicate and an average taken for each time point. 

The cycle threshold (Ct) difference was used to calculate the amount (x-fold) of change in gene expression as x = 2 − ΔΔCt, where ΔΔCt = ΔCt (c-fos ON after rAION induction) − ΔCt (c-fos control ON); ΔCt (c-fos after rAION induction) = Ct(c-fos ON after rAION induction) − Ct(GAPDH); ΔCt(c-fos control ON) = Ct(c-fos control ON) − Ct(GAPDH). Statistical analysis of ΔΔCt was performed with a t-test for two independent samples, with significance set at P < 0.05, compared between treated and control groups. ²³  

In Situ Apoptosis Analysis

Mouse retinas and ONs were collected at 2, 6, 9, 14, and 21 days after rAION induction and fixed in 4% paraformaldehyde. Longitudinal cross sections were cut 10 μm thick and used for in situ TUNEL staining with a the fluorescein-tagged apoptosis detection system (FITC, catalog No. 1684795l; Roche Diagnostics, Indianapolis, IN). Results were analyzed with a confocal microscope (Fluoview X; Olympus, Tokyo, Japan) at 488-nm wavelength. The number of TUNEL-positive cells in the ON and retina sections were determined and compared with the contralateral control (untreated) retina and ON of each animal and also with other appropriate positive and negative controls. 

Confocal Immunohistochemistry

Fixed paraffin sections were rehydrated and preincubated with normal serum of the same species used to produce the secondary antibody. Sections were incubated overnight at 4°C with the appropriate primary antibodies. Postincubation, sections were reacted with the appropriate Cy3 and Cy5-fluorescent dye-tagged secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA), at 1:400 dilution. Specificity controls for each antibody included incubation of sections without primary antibodies. Primary antibodies were rabbit polyclonal anti-β-gal (Novus Biologicals NB 600-305); mouse monoclonal anti-β-gal (Z378A; Promega, Madison, WI), rabbit polyclonal anti-cleaved activated caspase-3 (2305-PC-020; Trevigen, Gaithersburg, MD), mouse monoclonal anti-galactocerebroside (Gal C, MAB342; Chemicon, Temecula, CA), and rabbit polyclonal anti-CNPase (cyclic nucleotide phosphohydrolase; Sigma-Aldrich, St. Louis, MO). Sections used for confocal microscopy were mounted with glycerin-DAPI (4′,6′-diamino-2-phenylindole) gelatin mounting medium, coverslipped, and examined with a confocal microscope (FluoviewX; Olympus). Immunohistological light microscopy was performed with the biotinylated universal antibody kit and 3-amino-9-ethylcarbazole (AEC) reagent (Vector Laboratories, Inc., Burlingame, CA). 

Electron Microscopy

ONs were dissected and fixed at 4°C in paraformaldehyde-glutaraldehyde buffer, and embedded in Araldite-Epon. Tissues were infiltrated with 5% uranyl acetate and lead citrate. Ultrathin sections (50 nm) were floated onto copper grids and examined by transmission electron microscopy (TEM; Philips, Eindhoven, The Netherlands). Axonal quantitation at 21 days was performed by counting nine grids in five widely spaced ON regions, for both control and rAION-induced eyes. The number of axons ± SD was averaged for each region. 

Results

Gross and Histologic Appearance of the Mouse ON after Anterior ON Ischemia

One day after induction, examination revealed that the untreated ON head was normal in appearance (Fig. 1A) . The ON border (arrow) was well demarcated from the retina (ret), whereas the treated ON head was moderately swollen (Fig. 1B ; double arrows), with peripapillary nerve fiber layer (NFL)edema and obscuration of disc vessels (Fig. 1B ; single arrow). ON head swelling typically resolved within 6 days (data not shown). Histologic comparison of laser-treated control (Fig. 1C)and induced (Fig. 1D)eyes showed significant differences. The control eyes had a central depression in the ON, at the inner retinal border (Ret) near the nerve fiber layer (Fig. 1C ; NFL). However, in induced eyes, there was anterior ON head swelling (NFL; arrows), with vacuolization of the nerve substance (Fig. 1D ; arrowhead). Three days after treatment, laser control eyes had intact ONs, with columns of dark oligodendrocyte nuclei (Fig. 1E) , surrounded by eosinophilic axonal layers. However, 3 days after treatment, the rAION-induced eyes had a central necrotic lesion of the nerve just behind the ON–retina junction (Fig. 1F) , with normal-appearing peripheral oligodendrocyte nuclear and axonal columns. 

Effect of Ischemic Microvascular Damage to the ON Head on Intraretinal Vessels

Fluorescein angiography (FA) was used in vivo after induction to validate the hypothesis that ischemic damage produced by this model selectively affects ON head microvessels, while preserving larger, intraretinal vessels. These results are seen in Figure 2 . A photograph of the laser-treated, noninduced (no dye) retina and ON is seen in Figure 2A , and that of an induced eye is shown in Figure 2B . The normal mouse ON had a distinct border, and vessels were visible above the pigmented retina (Fig. 2A) . In contrast, 1 day after induction the eye showed the ON with swelling (ON edema), indistinct borders and pallor, and obscuration of the smaller disc vessels due to edema (Fig. 2B ; arrow). Some changes in retinal RPE pigmentation were apparent around the optic disc. The changes in rAION were distinguishable from those in central retinal vein occlusion (CRVO; Fig. 2C ) or central retinal artery occlusion (CRAO; Fig. 2D ), which can be induced with longer (>15 seconds) induction times. We have seen venous abnormalities at lower induction times, but rarely (<10%). However, because of intrinsic variability of animals, strain differences, and sexes, and because each investigator may have slightly different induction parameters, it was essential that all animals be evaluated before and 1 to 2 days after treatment, to ensure consistent AION induction, rather than other retinopathies such as CRVO or CRAO. Laser treatment of the ON head, without RB injection, resulted in a normal-appearing FA, without disc leakage at any phase (Figs. 2E 2G) . This treatment without RB did not induce subsequent optic neuropathy or RGC death. Five days after rAION induction, FA demonstrated microvascular occlusion of the ON, with intact intraretinal vessels (Fig. 2F) . There was leakage from the disc noted as a faint fluorescent “cloud” around the ON large vessels in late angiogram phases (Fig. 2H ; arrow). Disc leakage from ischemic nerves began in the early-intermediate phase (1.5 minutes after injection) and increased with time to a maximum leakage 3 minutes after injection. 

Analysis of control, laser-treated and rAION eye vasculature at the level of the ON immediately after treatment was determined using India ink-perfusion (Fig. 2I 2J 2K) . The untreated control showed filling of the capillaries (Fig. 2I ; black arrow) originating from the central retinal artery (CRA), which directly supply the ON at the level of the choriocapillaris (CC). After laser treatment without rAION induction, the ON capillaries remained patent (Fig. 2J ; black arrow). After rAION induction, there was a lack of microvascular perfusion of the ON (Fig. 2K) . Thus, rAION induction in mice appears to result from direct closure of the capillaries supplying the axons of the ON. 

Assessment of the Postischemia Stress Response

We assessed the immediate ischemic stress response in the ON and retina by identifying cells that express c-fos after rAION induction. After treatment, c-fos/LacZ transgenic animals were euthanatized at 1, 2, 3, 6, 9, 14, and 21 days. c-Fos gene activation in transgenic animals is detectable by assaying for β-gal, which is visible as a blue product in cells expressing this protein. These results are shown in Figure 3 . 

Eyes treated with a laser, but without dye (sham induction) showed only a few blue cells (Fig. 3A) . In contrast, retinal c-fos gene activation was apparent 1 day after rAION induction (Fig. 3B) . Quantitation of blue cells in rAION and laser-treated (no dye) control cells revealed significantly more LacZ-positive cells in rAION-induced eyes (Fig. 3C ; first three animals; gray bars), than in the contralateral (control) eyes (Fig. 3C ; black bars). In contrast, there were nearly equivalent numbers of blue cells in laser treated eyes, compared with contralateral controls (Fig. 3C ; animals 4–6). These blue cells were almost exclusively RGCs (Fig. 3D ; arrow, and 3E; arrows). Retinal cells in other layers in induced eyes showed no evidence of c-fos activation. 

Long-term (>30 days), laser treated, uninduced eyes showed no change in RGC numbers (Fig. 3F) , while induced eyes showed a loss of RGCs (Figure 3G ; visible as increased spacing between nuclei in the RGC layer compared with Fig. 3F ). RGC loss was detectable in rAION-affected eyes 14 days after induction, reaching a maximum loss of 49% by 21 days after induction (Fig. 3H) . The reduction in RGC numbers in 10 treated eyes, compared with the control eyes, was statistically significant (P < 0.05; Fig. 3H ). 

There were no changes in the number of cells in other retinal layers (compare INL and ONL in Figs. 3F 3G ). 

ONs from either control or sham-treated (control/no laser Fig. 4A ; laser only; Fig. 4D ) eyes had no blue cells, indicating a lack of c-fos gene activation (Fig. 4A) ; however, blue cells were observed soon after induction of ischemia in the retina and the anterior portion of treated ONs (Fig. 4B ; arrows). The c-fos-expressing cells in the ON had elongated oval nuclei and were located in a linear pattern (Fig. 4B ; arrowheads). The nuclear appearance and distribution of the ON cells suggests that these were oligodendrocytes. c-Fos expression became progressively apparent along the ON, in retrograde fashion. By 2 days after rAION induction, β-gal expression was apparent throughout the treated prechiasmal ON (data not shown). By day 3, β-gal activity was detectable in cells at the prechiasmal ON of the induced eye (Fig. 4C ; rAION), although there was little if any detectable activity in uninduced ONs (Fig. 4C ; Con). There were β-gal-positive cells on both sides posterior to the optic chiasm (Fig. 4C ; arrowheads), suggesting that c-fos activation occurred sequentially in these structures. β-Gal activity was subsequently noted in the contralateral superior colliculus (which receives approximately 95% of the terminating crossing RGC axons in rodents ²⁴ ) and in the contralateral lateral geniculate nucleus. 

c-Fos activation in the anterior visual pathway of c-fos/LacZ mice with crushed ONs (positive control) was similar to that in the rAION mouse model, with expression of LacZ-positive cells up to the chiasm at 2 days after treatment (Fig. 4E ; arrow) and crossing the chiasm at 3 days (4F; arrowhead). The crush-induced response was more intense (more blue cells) than that induced by the rAION method. 

Identification of Cells Expressing c-Fos in the ON

In immunohistochemical staining for oligodendrocytes, β-gal-positive cells were present 1 day after rAION induction, but only near the ON–retina junction (Fig. 5A ; arrow). There were no β-gal-positive cells in animals treated with laser alone (Fig. 5B) . There was a loss of CNPase staining in rAION-induced ONs by 3 days (data not shown). ON staining with Luxol fast blue revealed central demyelination in longitudinal sections of the treated ONs 6 days after induction (Fig. 5C ; arrow). 

TUNEL staining of the control ONs showed no TUNEL-positive cells at any time point (Fig. 5D) . In contrast, TUNEL staining of ONs after rAION induction showed a progressive increase in TUNEL-positive cells beginning at 6 days (Fig. 5E) , with more TUNEL-positive cells at 9 days (Fig. 5F ; arrow). TUNEL positive cells were still apparent at 14 days after induction (Fig 5G ; arrow), with a columnar pattern suggestive of oligodendrocytes. Both light and confocal microscopic techniques suggest that these rAION-induced changes were occurring in oligodendrocytes. Double immunostaining with oligodendrocyte markers (GAL-C and CNPase) and c-fos staining showed clearly that oligodendrocytes were the primary cell type affected. 

Detection of Ultrastructural Changes and Demyelination in the ON after rAION

TEM of control ONs revealed compact bundles of axon fascicles containing myelinated axons, with normal-appearing septae (Fig. 6A ; arrow). One day after rAION induction, axons in treated nerves were swollen, with distortion and flattening of the normal appearance of many axons and reduction in the apparent thickness of the septae (Fig. 6B ; arrow). Three days after rAION induction, vacuolization and swelling were clearly apparent (Fig. 6C ; arrowhead), and early demyelination was apparent centrally (Fig 6C ; arrows). Demyelination progressed over time, with intralaminar splitting of the myelin sheaths (Fig. 6D ; arrow) and axonal collapse (Fig. 6D ; arrowhead). Intralaminar myelin splitting was more intense 9 days after induction (Fig. 6E ; arrow), with persistent axonal swelling (Fig. 6E ; double asterisks) and axonal collapse (Fig. 6E) . 

Axonal loss was evident in ONs by 21 days after induction in rAION eyes (Fig. 7) . Control ONs showed myelinated axons of various diameters (Fig. 7A ; single arrow). Different ON regions showed an overall loss of axon numbers, that varied from region to region (Fig. 7D) . There was an average 60% loss of normal-appearing axons in rAION-treated ONs compared with control ONs. The peripheral ON regions (Fig. 7B)showed fewer changes in the overall number of axons (Fig. 7D ; area 4; compare rAION (filled bar) with control (open bar). Demyelinating or degenerating axons were present as multilayered, collapsed structures (Fig. 7B ; double arrows). Other (central) regions showed more extensive axonal loss (Fig. 7C) . 

Quantitative Analysis of c-Fos Expression in the Anterior Ischemic Induced and Control Nerve, Using RT-QPCR

There was no change in GAPDH expression between treated and control ONs between 6 and 9 days after induction (Fig. 8A ; GAPDH). MBP mRNA expression was lower at 6 days after induction, than at earlier times (data not shown). However, MBP mRNA levels were higher at 9 days, than at 6 days (Fig. 8A) . By 28 days after induction, MBP levels were still reduced in rAION eyes (Fig. 8B ; OD), compared with contralateral control eyes (Fig. 8B ; OS). 

c-Fos mRNA expression was elevated in equivalent lengths of rAION-induced nerves at days 1, 6, and 9 after induction. c-Fos mRNA levels were also always higher in AION-induced versus control nerves (Fig. 8C ; compare induced [OD]with uninduced contralateral eye [OS]). c-Fos levels increased over time, with levels at 9 days higher than those at 6 days (Fig. 8D ; compare 6 and 9 days). However, total c-fos mRNA levels were low compared with GAPDH (Fig. 8D ; compare GAPDH and c-fos cycle numbers). 

Discussion

We have developed and characterized a mouse model of ischemic optic neuropathy that resembles human AION, both grossly and histologically, making it a potentially useful tool in investigations of potential treatments of this common vision-threatening condition. In clinical AION, ophthalmoscopic findings include early optic disc swelling followed by progressive pallor of the optic disc. Histologic findings include progressive loss of the retinal NFL and RGCs, the presence of lipid-laden macrophages, and loss of central ON substance and gliosis. ²⁵ Many of the changes in the human disease also are present in our model, but with an accelerated progression. 

The use of the photosensitive dye (RB) coupled to 540-nm wavelength (green) light to generate superoxide radicals has been reported previously. ²⁶ The superoxide radicals generated by this photochemical method produce a direct endothelial injury limited to the treated vessels, because of the nature of the focused laser light aimed at the ON. RB-induced microvascular thrombosis also has been extensively studied, and it is clear that thrombosis occurs by superoxide radical-induced vessel damage to the intima of the affected vessels. ^{26 27 28} RB-induced thrombosis has been used in a number of systems to study the effects of ischemia, ²⁹ with ischemic damage confirmed by triphenyltetrazolium chloride (TTC) staining. ²⁶  

In our model, FA showed leakage only around the ON, thus demonstrating that there was no generalized breakdown of the blood–retinal barrier. These findings are similar to those described in human AION ³ and suggest that the technique we use to create our model produces occlusion only in ON head microvessels. Other evidence that the ON is selectively damaged in our model includes the observations that isolated laser irradiation of the ON head, without RB injection, did not induce optic neuropathy or produce subsequent RGC death; that there was a lack of severe vessel engorgement and retinal edema visible in CRVO and CRAO; and that intravenous perfusion of rAION mouse retinas with India ink 10 to 30 minutes after induction demonstrated a lack of microvascular perfusion of the capillaries at the treated ON head, at the level of the choriocapillaris, with filling of the inner retinal vessels of both eyes. 

Furthermore, as in the case of our rat model, ¹⁴ India ink filling of the ON capillaries in our mouse model demonstrated microvascular capillary nonperfusion at the ON head, with sparing of the larger retinal vessels and inner retinal circulation. The demonstration of acute loss of vascular supply to the anterior ON but not the retina suggest that, similar to previous studies, ^{26 30} the chronic changes were probably due largely to isolated ON ischemia. 

One of the objectives of this work was to evaluate directly the early response of the cells primarily affected by AION (e.g., the retinal ganglion cells). Mice have the advantage of a well-defined genome, and many transgenic strains can be used to analyze the expression of specific genes. We used transgenic c-fos/LacZ reporter mice ³¹ to characterize the early sequence of events after rAION, because c-fos is an immediate-early stress response gene that is implicated in RGC apoptosis, survival, and regeneration. ¹⁶ This transgenic mouse strain shows strong LacZ expression in neurons that ultimately die after ischemia-induced stress. ^{17 32} RGC LacZ expression is present early, 1 day after rAION. Coupled with postneuropathy cell counts, the early expression of the c-fos/LacZ transgene predicts a significant RGC loss in affected eyes. 

We also detected early c-fos expression in ON oligodendrocytes in treated eyes, as shown by both morphologic characteristics and immunostaining. These findings are consistent with those of other studies showing that oligodendrocytes are vulnerable to ischemia ^{33 34 35 36 37} ; however, the rapid involvement of ON oligodendrocytes in vivo, shown by c-fos expression and confirmed by quantitative analysis of c-fos and MBP mRNA levels, has not been described previously, to our knowledge. Our findings indicate that oligodendrocytes rapidly become dysfunctional after ischemia of the anterior portion of the rodent ON. This report also demonstrates for the first time that oligodendrocyte involvement rapidly progresses along CNS axons, involving the retrochiasmal CNS by 3 days after rAION induction. Because oligodendrocytes are essential for neuronal cell body and axon survival ³⁸ as well as for myelin assembly, ³⁹ the rapid and progressive loss of oligodendrocyte function may explain the progressive demyelination that occurs, both in our model and in many types of optic neuropathy, including human AION. ^{40 41}  

Because one oligodendrocyte myelinates many axons, demyelination may lead to additional loss of axonal function and, ultimately, to additional RGC death in the ischemic surround. However, RGC loss can lead to axonal degeneration with associated oligodendrocyte dysfunction and demyelination. 

In our mouse model, oligodendrocytes in the anterior ON undergo apoptosis within 2 days after induction of rAION. Although this early death may represent a direct toxic effect of the induction procedure, the region of oligodendrocyte apoptosis is identical with the region that demonstrates necrosis in human AION ^{25 42} and correlates with the portion of the mouse ON supplied by retrograde retinal capillaries directly affected by the rAION procedure. ⁴³ TUNEL-positive cells were detectable in the ONs of treated eyes several days before such cells were identified in the retina of such eyes (2 days vs. 6 days). These findings, along with the finding of necrosis and loss of central ON material at day 3 after induction, suggest that local dysfunction and death of oligodendrocytes precedes, rather than follows, RGC death and could contribute to it. Indeed, initial oligodendrocyte dysfunction may produce a domino effect, resulting in progressive axonal and oligodendrocyte damage throughout the length of the ON. Wallerian degeneration has been shown in other models of optic neuropathy, including ON transection and crush. ^{44 45 46}  

Despite the enormous potential value of this model in the study of the pathophysiology and potential treatments of human AION, some issues must be considered. The vasculature of the anterior ON of mice differs somewhat from that of rats and humans. Mouse ON lacks a choroidal blood supply, relying instead on recurrent peripapillary retinal branches from the intraretinal portion of the central retinal artery. Thus, in mice, the ON may be damaged by thrombosis of vessels supplying the NFL overlying the ON, vessels within the ON head itself, or recurrent branches of the central retinal artery that supply the retro-ocular ON. ⁴³ Mice also lack a lamina cribrosa. These differences may explain differences in the extent of damage to the ON in our mouse model of AION, compared with human AION. Another issue that must be considered is that there are only a few histologic studies of early clinical AION in humans to compare with our model and none in which immunostaining and electron microscopy have been performed. Thus, it is difficult to compare the findings in our model with those of human AION. Nevertheless, the histologic findings in our model are similar to those that have been described in the few human cases for which tissue comparisons are available. We noted ON hypercellularity in late stages, which probably represents leukocyte infiltration. Some thickening of the ON pial sheath was also observed. These possible inflammatory reactions should be investigated in future studies. 

A potential limitation of our model relates to the rd mutation found in many mouse strains. The c-fos/LacZ strain of mouse we used has the rd photoreceptor mutation on a C57B6 background. This results in outer retinal (photoreceptor) loss in the early postnatal period in homozygotes. ²¹ Thus, several the animals we used in early studies showed the rd phenotype in addition to transgene expression. This could be a limitation in our study, because the rd mutation can cause alterations in the vasculature in the outer layers of the retina and possibly also in RGCs. ⁴⁷ We repeated many of our experiments in heterozygotes that do not exhibit the rd phenotype and also in wild-type animals. We did not detect any differences in the RGC or ON oligodendrocyte responses in heterozygote or wild-type animals compared with the rd homozygotes. This suggests that our results are unlikely to have been influenced by the presence of the rd mutation in our model. 

In addition to demonstrating that oligodendrocytes may be primarily damaged in rAION and the importance of this finding in the selection and assessment of potential treatments for human AION, our AION model may be useful in evaluating neuroprotective treatments in vivo in the CNS, including those aimed at preserving poststroke oligodendrocyte loss and reducing demyelination-induced secondary damage. Indeed, the model, coupled with transgenic technology, can be used as a platform for further analysis of the genetic sequence of events after AION and other isolated ischemic axonopathies. Recently, several neuroprotective agents designed to protect oligodendrocytes have been suggested as potential poststroke treatments. ⁴⁸ Our findings provide a further basis for this form of therapy. 

 A portion of this report was presented orally at the 2003 annual North American Neuro-Ophthalmology Society (NANOS) meeting, Snowbird, Utah.

 Supported by the Zanvyl and Isabel Krieger Fund (NG-C), Baltimore, Maryland; an unrestricted grant from Research to Prevent Blindness (SLB), the V. Kann Rasmussen Foundation, Denmark (SLB); and National Eye Institute Grant R01-EY015304-01 (SLB). NG-C is a recipient of a Wilmer research award.

 Submitted for publication May 17, 2004; revised October 30 and December 7, 2004; accepted December 20, 2004.

 Disclosure: N. Goldenberg-Cohen, None; Y. Guo, None; F. Margolis, None; Y. Cohen, None; N.R. Miller, None; S.L. Bernstein, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Primer Sequences

Name	Accession	Sequence
GAPDH	M32599	5′ AAC GAC CCC TTC ATT GAC 3′ (sense)
		5′ TCC ACG ACA TAC TCA GCA C 3′ (antisense)
MBP	BC004704	5′ TGA TGG CAT CAC AGA AGA GAC 3′ (sense)
		5′ GCC CAG GAC GGC TGC GGG CAT 3′ (antisense)
C-fos	GenBank V00727	Ambion, Inc., catalog no. 5402

Figure 1.

 

Gross and histologic appearance of control and rAION mouse ocular tissues. (A) Ophthalmoscopic appearance of normal ON head and surrounding retina (RET). The ON was present as a well-defined optic disc with sharp margins (arrow). The large intraretinal vessels emerged from the ON in a radial pattern. (B) Appearance of retina and ON head 1-day after rAION induction. ON head edema was present with blurring of the disc margin (double arrows), as well as blurring of the central retinal vessels (single arrow), indicating NFL edema. (C) Histologic appearance of normal ON head and surrounding retina (RET). Arrows: compact, thin NFL. Note the excavation in the nerve center and patent intraretinal vessel (asterisk). (D) Higher magnification shows histologic appearance of ON head and surrounding retina 1 day after rAION induction. H&E staining. The NFL was edematous (arrow), as indicated by vacuolization (arrowhead), elevation (double arrows), and loss of the central excavation. (E) Histologic appearance of normal ON. An H&E-stained longitudinal section through the ON of the untreated eye shows the linear pattern of intraneural nuclei, typical of oligodendrocytes. (F) Histologic appearance of H&E-stained longitudinal ON section 3 days after rAION induction. There was a central lesion (asterisk) with loss of staining and ON swelling (note increased distance between nuclear columns) and vacuolization in the anterior portion of the nerve, just behind the globe–ON junction. Note the hypercellularity and thickening in the pial sheath.

Figure 2.

 

Gross appearance and fluorescein angiography of control and rAION eyes. (A) Ophthalmoscopic appearance of normal pigmented mouse (C57Bl6) ON head, before FA. The ON and retina (Ret) are indicated. The ON–retina border was distinct (arrowhead). The central retinal vessels (Crv) were located in the ON head center. (B) Appearance of mouse retina and ON head 1 day after rAION induction. ON head edema was apparent, with whitening and blurring of the ON–retina junction and retinal vessel borders, due to NFL edema (arrow). (C) Appearance of mouse retina and ON head with CRVO 1 day after induction. All retinal veins were swollen (arrows), and light retinal edema masked the underlying choriocapillaris. (D) Appearance of mouse retina and ON head with CRAO 1 day after induction. There was dense retinal edema, with masking of the choriocapillaris, retinal hemorrhage, and gross dilation of retinal vessels. (E) FA, early phase (1.5 minutes), control eye. No dye leakage from the disc. (F) FA, late phase (3 minutes), control eye. No dye leakage from the disc. (G) FA, early phase, 5 days after rAION induction. There was complete filling of the large intraretinal vessels. Vessel dye leakage at the optic disc center was minimal (1.5 minutes). (H) FA, late phase (3 minutes), same retina as in (E). Dye leakage from the optic disc was visible as a faint halo and blurring of the central disc (arrow). (I) Normal ON vasculature (India ink visualization). Vessels surrounding the ON at the level of the choriocapillaris (CC) formed a ring around the ON (white arrow). The central retinal artery (CRA) passing through the ON was visible. Capillaries emerged from the CRA to supply the ON (seen as thin black vessels; black arrow). (J) India ink visualization of mouse ON vasculature 15 minutes after laser treatment (no dye induction). CC vessels at the level of the ON formed an incomplete ring around the nerve (white arrows, normal variant). Capillaries supplying the ON were patent and were visible as a black meshwork (black arrow). (K) India ink characterization of ON vasculature 15 minutes after rAION induction (dye+laser). The CC vessels around the ON and central retinal artery (CRA) were patent and fill with ink. Few if any patent capillaries were present within the ON. Magnification: (I–J) ×250.

Figure 3.

 

Histology and LacZ expression after rAION in c-fos/LacZ transgenic retinas. (A) Absence of LacZ expression in an unstained retinal flatmount of a transgenic mouse control eye after development in blue II solution. The ON and retina (Ret) are indicated. (B) LacZ expression in a transgenic mouse 1 day after rAION induction (flatmount). Gray cells in the retina (arrows) indicate LacZ expression through the c-fos promotor activation. LacZ expressing cells were in the RGC layer, as seen at higher magnification in (E) and in a sagittal section of the retinal layer in (D). (C) Graph of LacZ-expressing cells in rAION-induced and control retinas. rAION (animals 1–3) and laser control (animals 4–6). ( ): treated eyes. (▪): contralateral eyes (untreated controls). There was an increased number of LacZ-positive cells in rAION-induced eyes, compared with contralateral control eyes (animals 1–3). Nearly equal LacZ-positive cells were seen in laser control and contralateral untreated retinas (compare date for animals 4–6. (D) Retinal LacZ expression 3 days after rAION induction. Blue cells in the RGC layer (arrow), indicated c-fos activation only in this cell layer. (E) Retinal flatmount 1 day after rAION induction. Blue cells were only in the RGC cell layer (arrows). (F) Control retina section at 21 days; H&E stain. RGCs were present in a continuous layer. The inner nuclear (INL) and outer nuclear (ONL) layers were intact, with multiple nuclear layers in each. (G) rAION retina section 21 days after induction; H&E stain. RGC loss was indicated by increased RGC nuclei spacing (arrow). INL and ONL were intact and similar to the control. (H) Histogram of RGC loss, at days 1, 2, 9, 14, 21, and 28. Note the significant loss of RGCs at days 14 and 21 after rAION induction. Magnification: (D) ×100; (E) ×400; (F, G) ×250.

Figure 4.

 

Appearance and progression of LacZ expression in c-fos/LacZ transgenic control and after rAION ON. (A) A histologic section of control (c-fos/LacZ) transgenic retina developed for LacZ expression. Note the colorless appearance of retina and ON, indicating the absence of c-fos promotor activity. (B) Histologic section through transgenic retina 1 day after induction. Blue cells (LacZ positive) at the ON–retina junction (arrows) represented c-fos promotor activation at the site of the primary lesion. There were also blue nuclei columns in the ON (arrowheads), suggestive of early oligodendrocyte stress, that extended retro-orbitally toward the CNS. (C) Appearance of the ON and chiasm 3 days after rAION induction. There was an absence of blue-stained cells in the control nerve (Con). Cells with c-fos promotor activation (blue cells) were present in the ON of the treated eye (AION; arrows). Blue cells were also present in the chiasm on both sides of the CNS (arrowheads). (D) Absence of LacZ expression in ON and tract in a laser control (c-fos/LacZ) transgenic animal 1 day after treatment. (E) LacZ expression in ON 1-day after crush induction. Blue cells reached to the chiasm (junction of ON and tract) but did not cross (arrow). (F) Activation of LacZ activity in ON 3 days after crush induction. Progression of LacZ-expressing cells have reached and crossed the chiasm.

Figure 5.

 

Immunofluorescence labeling of β-gal (LacZ) expression and apoptosis in the ON. (A) β-Gal expression in intrinsic ON cells 1 day after rAION induction. β-Gal-positive cells (arrow) were visible near the retina–ON junction. (B) β-Gal expression in laser-treated (laser control) nerve 1 day after treatment. Few if any positive cells were visible. (C) Luxol-fast blue stain of a longitudinal section through an rAION ON 6 days after induction. Central demyelination was apparent by the loss of specific staining and by the loss of central staining (arrow), with vacuolization (arrowhead) more peripherally in the ON. (D) TUNEL staining of control ON. No TUNEL-positive cells were visible. (E–G) TUNEL staining of rAION-induced ON 6 (E), 9 (F), and 14 (G) days after induction. An increase in TUNEL-positive cells was seen from 6 to 9 days (compare E and F). TUNEL-positive cells in a columnar pattern suggestive of oligodendrocytes were also visible at 14 days (arrow). Magnification: (A) ×400; (B–G) ×250.

Figure 6.

 

Electron microscopy of control and rAION ON sections. (A) TEM of ultrastructural morphology of the cross-section of a control untreated ON, 1 day after rAION induction in the fellow eye. Myelinated ON axons were visible as circular structures, lining individual septae. (B) ON cross section 1 day after rAION induction. Early edema was present, visible as compression and flattening of the axons with reduction in the apparent septal thickness (arrow). (C) ON cross section 3 days after rAION induction. Axonal swelling was present (arrowhead), as well as early splitting of the axonal myelin sheaths (arrows) and further distortion of normal axonal appearance. (D) ON cross section 6 days after rAION induction. Axonal collapse (arrowhead) was present, along with increase in interlaminar splitting of the myelin sheaths (arrow). There is distortion of remaining axons, and vacuolization (double asterisk). (E) ON cross section 9 days after rAION induction. Axonal distortion and increased levels of demyelination were present (intralaminar splitting; arrow), along with axonal collapse. Myelin vacuolization (asterisk) was present.

Figure 7.

 

TEM analysis of ON 21 days after rAION. (A) Control ON. Myelinated axons of various diameters were present in the section (arrow). The myelin sheath was thin. (B) rAION ON (peripheral region). Many myelinated, normal appearing axons were present (arrow). A few degenerating axons were also present, seen as multilaminated structures with collapsed centers (double arrows). (C) rAION ON (central region). Only a few normal-appearing axons remained (arrow), and the rest were in various stages of degeneration. (D) Axon quantitation in different ON regions after rAION. Five similar ON locations for each nerve were analyzed in control (□) and rAION-induced (▪) ONs. There was a variable loss of axons in all regions of rAION-induced ON, compared with control ONs, and an overall 60% loss of intact axons. Bars ± SD. Magnification: (A–C) ×2500.

Figure 8.

 

RT-QPCR graphs. (A) MBP levels (bottom) at 6 versus 9 days after rAION induction ONs. The 9-day curve (9 d) showed higher MBP levels than did the 6-day curve (6 d). There were nearly equivalent levels of GAPDH mRNA expression (top) in both ON sections. (B) MBP levels in ONs in control (OS) and rAION (OD) 28 days after rAION induction. GAPDH mRNA levels were approximately equivalent in both nerves of the same animal (top). There was less MBP mRNA in the rAION-induced ON (bottom). (C) c-Fos levels in ONs at 6 and 9 days after rAION induction. Right curve shows lower c-fos mRNA levels at day 6, compared with the left curve at day 9. GAPDH mRNA levels (GAPDH curves) for both sections are shown. (D) c-Fos mRNA levels in rAION-induced nerves at 6 and 9 days. There was more c-fos mRNA expression in the rAION-induced ON at the later time point, with similar GAPDH mRNA levels (superimposed GAPDH curves for both times).

The authors thank David Knox (Johns Hopkins University) for suggestions concerning histologic analysis, Patrick Tong (Johns Hopkins University) for performing fluorescence angiography in the mice, and Harry Quigley (Johns Hopkins University) for critical reading of the manuscript. SLB thanks Alex M. Bernstein (Chevy Chase, MD) for performing the LacZ retinal flatmount cell counts.