Influence of retinal NMDA receptor activity during autoimmune optic neuritis

Autoimmune optic neuritis (AON), a model of multiple sclerosis‐associated optic neuritis, is accompanied by degeneration of retinal ganglion cells (RGCs) and optic nerve demyelination and axonal loss. In order to investigate the role of N‐methyl‐d‐aspartate (NMDA) receptors in mediating RGC degeneration, upstream changes in the optic nerve actin cytoskeleton and associated deterioration in visual function, we induced AON in Brown Norway rats by immunization with myelin oligodendrocyte glycoprotein. Subsequently, visual acuity was assessed by recording visual evoked potentials and electroretinograms prior to extraction of optic nerves for western blot analysis and retinas for quantification of RGCs. As previously reported, in Brown Norway rats RGC degeneration is observed prior to onset of immune cell infiltration and demyelination of the optic nerves. However, within the optic nerve, destabilization of the actin cytoskeleton could be seen as indicated by an increase in the globular to filamentous actin ratio. Interestingly, these changes could be mimicked by intravitreal injection of glutamate, and similarly blocked by application of the NMDA receptor blocker MK‐801, leading us to propose that prior to optic nerve lesion formation, NMDA receptor activation within the retina leads to retinal calcium accumulation, actin destabilization within the optic nerve as well as a deterioration of visual acuity during AON.

Inflammatory demyelination of optic nerves, termed optic neuritis, is a common syndrome associated with MS. This is accompanied by degenerative changes in the retina, consisting of a loss of retinal ganglion cells (RGCs) together with thinning of the retinal nerve fibre and inner nuclear layers in patients (Galetta et al., 2015;Green, McQuaid, Hauser, Allen, & Lyness, 2010;Kupersmith, Garvin, Wang, Durbin, & Kardon, 2016;Syc et al., 2012). Interestingly, retinal nerve fibre layer thinning has also been reported in MS patients in the absence of clinically defined optic neuritis (Bock et al., 2010;Petzold et al., 2010;Talman et al., 2010).
In order to investigate the mechanisms underlying early retinal neurodegeneration in optic neuritis, we have used the model of myelin oligodendrocyte glycoprotein (MOG) immunization of Brown Norway rats. This model of experimental autoimmune encephalomyelitis (EAE) has a high incidence of autoimmune optic neuritis (AON) as well as other clinical and pathological features similar to those in MS patients (Meyer et al., 2001;Storch et al., 1998). In addition, RGC loss has been shown in this model to precede inflammatory demyelination of optic nerves (Fairless et al., 2012), to correlate with the timing of increases in retinal calcium levels and calpain activity (Hoffmann et al., 2013), and to involve NMDA receptor activity (Sühs et al., 2014). We also previously showed that ultrastructural changes in RGC axons of the optic nerve were present during the onset of disease (Fairless et al., 2012) as was elongation of Nodes of Ranvier (Stojic, Bojcevski, Williams, Diem, & Fairless, 2018).
In this study, in order to investigate further the impact of NMDA receptor activity on RGC degeneration in AON, we have correlated the timing of RGC degeneration with changes in the actin cytoskeleton of the optic nerve and visual disturbances. Through intravitreal injection of glutamate and also intravitreal application of an NMDA receptor blocker into rats with AON, we demonstrate that retinal NMDA receptors contribute to upstream changes in the optic nerve, probably by regulating retinal calcium levels.

| Animals
Female Brown Norway rats (8-10 weeks old, 140-160 g; Charles River, Sulzfeld, Germany; RRID:RGD_737972) were used in all experiments. All animals were kept under environmentally controlled pathogen-free conditions with free access to food and water, and housed in groups of four in UNO polycarbonate type IV cages under a 12 hr light/dark cycle. Animals were acclimatized for 1 week prior to experiments, which were then performed during the hours of 09:00 and 18:00. Animal experiments were performed in an approved animal facility according to the relevant laws and institutional guidelines of the local ethics committees of Baden-Württemburg and Saarland, Germany (approved protocols C-1.2.4.2.1, 35-9185.81/G-36/12, 35-9185.81/G-172/14 and 35-9185.81/G-36/17). The study was not preregistered. Assignment of animals to study groups was randomly performed by arbitrarily marking tails with a skin marker (Fine Science Tools) before using the Randbetween function in Windows Excel. A total of 160 rats were used in this study, of which 4 were excluded (but not replaced) because of the exclusion criteria of unsatisfactory signal-to-noise ratios during magnetic resonance imaging (MRI).
Experiments were exploratory in nature with animal group sizes kept to a minimum to avoid animal suffering, and were based on previous studies conducted within the laboratory (no sample calculation was performed). The different animal groups and the experimental timelines are shown in Figure 1. Immunization emulsion contained 1:1 ratio of 100 μg MOG in phosphate-buffered saline (Sigma-Aldrich) and complete Freund's adjuvant (Sigma-Aldrich) containing 100 μg of heat-inactivated mycobacterium tuberculosis H37RA (Difco Microbiology). Rats were anaesthetized with 5% isoflurane inhalation (chosen for its rapid on-and offset) and injected intradermally at the tail with 200 μl of immunization emulsion. No analgesics were applied following injection because of the potential to interfere with local inflammation and subsequent EAE. Rats were observed daily for clinical signs of EAE that reflect spinal cord pathology. The grading system used to score neurological symptoms were: grade 0, no signs; grade 0.5, distal paresis of the tail; grade 1, complete tail paralysis; grade 1.5, paresis of the tail and mild hind leg paresis; grade 2.0, unilateral severe hind leg paresis; grade 2.5, bilateral severe hind limb paresis; grade 3.0, complete bilateral hind limb paralysis; grade 3.5, complete bilateral hind limb paralysis and paresis of one front limb; grade 4, complete paralysis (tetraplegia), moribund state, or death. Approved protocols required animals exceeding a clinical score of 2.5 for over 48 hr to be euthanized, a criterion that was not met. The clinical onset of EAE coincides with the pathological hallmark of AON-the appearance of inflammatory demyelination in the optic nerve; the period referred to as clinical AON (cAON). However, in this MOG-EAE model RGC loss precedes cAON (Fairless et al., 2012); the period referred to as induction AON (iAON).

| MOG-EAE induction and animal scoring
At desired time-points during AON (day 10 post-immunization (p.i.) for iAON, or day 1-3 after clinical onset of EAE (approximately day 14 p.i.) for cAON) animals were sacrificed by inhalation of 5% isoflurane followed by decapitation. Retinas were further processed for retinal whole-mount immunofluorescence and optic nerves for Western blotting.

| Retina excitotoxicity model -intravitreal injection of glutamate
In order to induce a primary retinal insult, healthy rats received intravitreal injection of glutamate or saline as control. During injections, animals were kept under anaesthesia by inhalation of 5% isoflurane.
In total 100 nmoles of glutamate (4 μl of 25 mM glutamate (Sigma-Aldrich) in sterile saline) was injected intravitreally by puncturing the eye at the cornea-sclera junction with a 33G needle attached to a 10 μl NanoFil syringe and injecting over the course of 5 min.
Subsequently, eyes were treated with a 0.75% povidone-iodine solution (Braunol; B Braun Melsungen AG). Twenty-four hours (24 hr) or 7 days following intravitreal injection of glutamate, animals were sacrificed and tissue was processed as described in the previous section.

| Measurement of visual evoked potentials and electroretingrams
Five to seven days prior to the first recording, holes were drilled in the skull for the placement of electrodes. Rats were anaesthetized by intraperitoneal injection (i.p.) of ketamine (60 mg/kg; Atarost GmbH and Co.) and xylazine (12 mg/kg Albrecht) (anaesthetics chosen for longevity during procedure, as well as analgesic properties), skin was incised mediosagitally, and holes were drilled above both visual cortices 1 mm frontal and 3.5 mm lateral to lambda (for recording electrodes) and above the motor cortex (1 mm frontal and lateral from bregma for reference electrode). Following surgery, rats were provided drinking water containing tramadol (100 mg/L) for the next 24 hr to minimize potential pain, and the wound was treated with 7.5% povidone-iodine to minimize the risk of infection.
Measurements of visual evoked potentials (VEPs) were performed using a UTAS Visual Diagnostic System (LKC Technologies).
Prior to recording, animals were anaesthetized with ketamine and F I G U R E 1 Schematic of experimental timelines indicating animal numbers used. (a) Autoimmune optic neuritis (AON) timeline as in Figure 2 (total animals used was 68), (b) intravitreal injection protocol as in Figure 3 (total animals 38), (c) MK-801 intravitreal injections in AON as in Figures 4 and 5 (total animals 24), (d) manganese-enhanced MRI protocol as in Figure 4 (total animals 30, with 2 excluded from the healthy and 2 from the MK-801-treatment groups because of inadequate signal-to-noise ratios, but were not replaced). No animals died during experiments. Optic nerve samples were dedicated to analysis either for lysate fractionation (for G/F-actin assay) or standard Western blotting as indicated in the Results

| 697
BOJCEVSKI Et al. xylazine (as above) and pupils were dilated with 0.5% atropine (Ursapharm) prior to dark adaptation for 10 min. During recording, eye desiccation was avoided using Liquifilm® O.K. eye drops (Allergan), and the animal was maintained at 37°C. For recording of VEPs, needle type electrodes were placed in the visual cortex and in the motor cortex at a depth of 0.5 mm, and a ground electrode was placed subcutaneously in the tail. For recording of flash VEPs (fVEPs), animals were placed in a Ganzfeld dome equipped with an LED whole-field stimulator and flash stimuli were presented with a luminance of 2.5 cd * s/m 2 and a frequency of 2 Hz. A total of 100 sweeps were averaged per recording, with measurements being performed in triplicate.
Recording of pattern-evoked visual responses (pVEPs) was performed on animals placed 20 cm in front of a 15" cathode ray tube monitor with the display centred with the pupil axis. Alternating vertical bar stimulation (66% contrast, temporal frequency reversal rate of 2 Hz) was presented at increasing spatial frequencies Pattern electroretinograms (pERGs) were recorded with the UTAS Visual Diagnostic System using a previously established protocol (Mayer et al., 2018). Briefly, rats were anaesthetized with a ketamine/xylazine mixture. pERGs were recorded using a lens-type electrode with gold contact (LKC Technologies) that was placed on the corneal surface of a focused eye (pupils were not dilated). Liquifilm® O.K. eye drops were used to prevent the desiccation of the eye and also to ensure proper contact with the electrode. Recording of pERGs and calculation of their amplitudes was performed in an identical manner identical to that described for pVEP recordings.

| Retinal whole-mounts and immunofluorescence
At the required time-points following intravitreal injection of glutamate or during AON, rats were sacrificed by anaesthesia with 5% isoflurane inhalation followed by rapid decapitation. Eyes were dissected and the vitreous body and pigmented epithelia were removed. Retinal whole-mounts were mounted on nitrocellulose membranes (Membrane Filter Black, white grid; GE Healthcare Life Sciences Whatman TM) with the RGC layer uppermost and fixed with ice cold 4% paraformaldehyde (Sigma-Aldrich) for 10 min, followed by tissue permeabilization with 0.3% Triton X-100 (Sigma-Aldrich) in phosphate-buffered saline for 10 min at room temperature. Following blocking for 30 min at room temperature (10% normal goat serum (Sigma-Aldrich) in 0.3% Triton X-100 in phosphate-buffered saline), retinas were incubated overnight at 4°C with anti-RNA-binding protein with multiple splicing (Rbpms, 1:500; Abcam PLC; RRID: AB_1861759), an RGC specific antibody (Liddelow et al., 2017;Rodriguez, Sevilla Müller, & Brecha, 2014).
The following day, retinas were washed and incubated with goat-anti To determine RGC cell densities, four different areas, aligned longitudinally from the optic nerve head to the outer retinal rim, were taken from each of four retinal segments. Pictures were taken using a conventional Eclipse 80i microscope (Nikon) at 20× magnification. Eight 0.01 mm 2 squares were used within each of the 16 areas (avoiding blood vessels) and RGCs were counted using an ImageJ cell counting plug-in (https://imagej.nih.gov/ij/). All retinal wholemounts were quantified in a blinded manner by the experimenter, with randomly numbered slides being provided by a co-worker.

| Globular versus filamentous actin ratio assay
Actin network dynamics were evaluated on optic nerve lysates by measuring the ratio of globular versus filamentous actin (G/Factin ratio) using the G-actin/F-actin In vivo assay (Cat. #BK037; Cytoskeleton Inc.). Tissue preparation and protocol was performed according to the manufacturer's instructions. Briefly, optic nerve lysates were collected in F-stabilization buffer and ultra-centrifuged at 100,000 g, 37°C for 1 hr, using an Optima Max-E ultracentrifuge (Beckmann Coulter Inc.) with a TLA-55 fixed-angle rotor (Beckmann Coulter). Next, supernatant was carefully removed (G-actin fraction) while the pellet was re-suspended in F-actin depolymerizing buffer (F-actin fraction). Samples were then subjected to Western Blotting as outlined above. G/F-actin ratio was calculated as the ratio of globular versus filamentous actin band intensity using the ImageJ application for Western blot quantification.

| Manganese-enhanced magnetic resonance imaging
Magnetic resonance imaging (MRI) measurements were performed with a system designed for small animal research (Biospec Avance III 9.4/20; Bruker Biospin GmbH), with a static magnetic field strength of 9.4 Tesla. A linear single channel volume coil (inner diameter 72 mm) was employed for radio frequency transmission and a saddle-shaped surface coil developed for imaging of the rat brain, with a 2 × 2 array setup was used as receiver.
Anaesthesia was induced with 4% isoflurane in oxygen (at a rate of 2 litres per minute) and maintained during the procedure with 2% isoflurane. An aqueous cream was applied to the cornea to prevent desiccation. Animals were placed in prone position on a custom-designed animal cradle, including a water-driven warming system connected to a thermostat. Respiration was monitored via a pres- Intravitreal injections were avoided so as not to disrupt retinal tissue integrity and induce imaging artefacts. Animals were then placed in the magnet, and correct positioning was verified using a tri-directional Fast Low Angle Shot scan. To identify the retina, MRI was performed with a T2-weighted sagittal Rapid Acquisition with Relaxation Enhancement (RARE) sequence (TR/TE, 5,000 ms/57 ms; RARE factor 17) and positioning checked with a T2-weighted coronal RARE sequence (TR/TE, 2,300 ms/39.6 ms; RARE factor 8).
For quantitative image analyses, regions of interest were selected, and signal intensities normalized to the intensity of background brain tissue using Paravision 5.1 software (Bruker). To calculate manganese enhancement, the increase in normalized signal intensities after manganese injection was calculated as a percentage of the signal before manganese injection. The signal increase following manganese injection in iAON rats was then compared to that occurring in healthy controls. False colour images were generated using ImageJ software (National Institutes of Health), by application of rainbow red-Syc-blue colour scaling.

| Statistics
All data are presented as their mean values ± standard error of the mean (SEM). Statistical analyses were made using SigmaPlot 13.0 software (Systat Software Inc.). Where two experimental groups were compared, if data were normally distributed (as assessed by the Shapiro-Wilk test), statistical significance was assessed by two-tailed Student's t test. If data failed the Shapiro-Wilk test, statistical significance was assessed by Mann-Whitney rank sum test. Where more than two groups were compared in normally distributed data (as assessed by the Shapiro-Wilk test), ANOVA combined with post hoc Tukey's method was used. If data failed the Shapiro-Wilk test, Kruskal-Wallis one-way analysis of variance on ranks followed by Dunn's method was used.
No test for outliers was performed. Exact p values are given in the results section with three levels of significance defined: *p ≤ .05 was considered significant, **p ≤ .01 was considered strongly significant; ***p ≤ .001 was considered highly significant. All data were used for analysis unless a clear reason for exclusion was apparent -criteria for exclusion included damage to the retina during dissection (e.g. if the area around the optic disc was damaged or missing), or bleeding from the eye during intravitreal injection. Experiments were performed at least twice, with all data pooled for analysis.

| Progressive RGC loss during AON is detectable with anti-Rbpms labelling
Following MOG-immunization, there are two distinct stages of AON.
The first is the induction or pre-clinical stage (which we term iAON; in this study d10 p.i. has been used) during which there are no signs of spinal cord injury (i.e. no clinical deficit reflected in the classical EAE scoring system). During this period, no changes in the optic nerve in terms of demyelination or axonal loss are detectable (Fairless et al., 2012), as confirmed in this study (Figure 2a). However, it has previously been reported that progressive retinal deterioration, indicated by a loss of RGCs, already begins during this stage (Fairless et al., 2012;Hobom et al., 2004). This was confirmed in the current study through immunolabelling of RGCs within retinal whole-mounts with an antibody against the relatively novel RGC marker RNA-binding protein with multiple splicing (Rbpms, Rodriguez et al., 2014). A significant decrease in RGCs was observed during iAON of about 7% compared to healthy, unimmunized controls (iAON, 2,875 ± 46 Rbpms + cells/mm 2 , n = 7; healthy, 3,076 ± 53 Rbpms + cells/mm 2 , n = 7; p = .047; Figure 2b).
The second stage of AON is the clinical phase (which we term cAON; in this study days 1 to 3 after EAE onset has been used, approximately day 14 p.i.), which is characterized by optic nerve demyelination, immune cell infiltration and axonal loss (Figure 2a), and typically occurs in parallel to the onset of clinical EAE symptoms reflecting spinal cord lesions (Fairless et al., 2012;Meyer et al., 2001).

| Intravitreal injection of glutamate mimics retinal and optic nerves changes during iAON
In the retina, glutamate excitotoxicity can be induced following intravitreal injection of glutamate (Schori et al., 2001;Sisk & Kuwabara, 1985;Zhou et al., 2007). In order to explore whether the changes that we observe during AON (Figure 2) reflect over-activation of glutamate receptors, glutamate was injected into the eye of healthy rats and retina and optic nerves were subsequently analysed.

| Intravitreal injection of glutamate leads to a deterioration in visual acuity
Next, we investigated the effect of intravitreal injection of glutamate on visual acuity. To address this we first performed recordings of fVEPs. In this study, we observed no change in fVEP amplitudes following intravitreal injection of glutamate compared to saline-injected controls 24 hr after the injection (Saline, 160 ± 30 µV, n = 3, Glutamate, 147 ± 17 µV, n = 8, p = .694; Figure 3i). The same was true for the fVEP latencies (Saline, 38 ± 1 ms, n = 3; Glutamate, 35 ± 1 ms, n = 8, p = .233; Figure 3j), and was also observed at 7 days post-injection ( Figure S2b,c).
However, since fVEP responses reflect the activity of all cells in the pathway from photoreceptors to the visual cortex, we next performed recordings of pVEPs which specifically give information about the pathway from RGCs to the visual cortex, and are considered a more sensitive parameter, for example for assessment of MS pathology (Halliday & Mushin, 1980). Following intravitreal injection of glutamate we observed a significant decrease in pVEP amplitudes compared to saline-control 24 hr post-injection (Table 1; Figure 3l), and a similar decrease was observed at 7 days post-injection ( Figure S2e). pERGs were then recorded to give specific information regarding RGC activity within the retina without involvement of the optic nerve (Porciatti, 2015). Similar to the pVEP measurements, pERG amplitudes following intravitreal injection of glutamate were significantly decreased compared to saline-control group 24 hr post-injection period (Table 1; Figure 3n), and a similar decrease was observed at 7 days post-injection ( Figure S2g). TA B L E 1 pVEP and pERG amplitudes following pattern stimulation at the indicated spatial frequencies, measured 24 hr following intravitreal injection of either glutamate or saline

| NMDA receptor is involved in RGC neurodegeneration during iAON
In order to determine the potential role of NMDA receptors in mediating RGC degeneration and upstream optic nerve actin cytoskeletal changes in iAON, the potent use-dependent blocker of NMDA receptors MK-801 was chosen (Huettner & Bean, 1988

| MK-801 reduces retinal manganese-enhanced MRI signal intensities during iAON
In order to further explore the role of NMDA receptor signalling during AON, we used manganese-enhanced MRI to visualize retinal calcium entry pathways, as has been widely used to study the optic tract (Gadjanski et al., 2009;Yang et al., 2016). We have previously reported that retinal manganese-enhanced MRI signals are increased in iAON animals compared to healthy controls, indicating increased tissue calcium levels during the preclinical stage of AON (Hoffmann et al., 2013). In this study, whereas saline-treated iAON animals had increased manganese-enhanced MRI signal intensities compared to healthy (4.25 ± 1.00%, n = 10), this was reduced in iAON animals receiving MK-801 (−0.61 ± 2.08%, n = 8; p = .044, Figure 4j). Thus, it appears that increased NMDA receptor activation is responsible for the increased retinal calcium levels in iAON.
As for the pVEPs, there was a significant difference in pVEP amplitudes between the three groups; with healthy values being the highest, followed by iAON -MK-801 and lastly by iAON -Saline group (Table 2; Figure 5e). The same was true for the pERG F I G U R E 4 Retinal delivery of MK-801 reduces iAON-associated RGC loss and disturbances in the optic nerve actin cytoskeleton. (a) Immunostaining with anti-Rbpms to label RGCs in whole-mounted retinas from rats during iAON, receiving either saline or MK-801 by intravitreal injection. (b) Quantification of RGC density revealed greater retinal ganglion cell (RGC) numbers in retinas following intravitreal injections of MK-801 compared to saline (both during iAON disease phase; n = 16 retinas per treatment group). In both groups RGC numbers are significantly lower compared to healthy eyes (n = 7). (c) Western blotting of optic nerve lysates with an antibody against actin (42 kDa) following fractionation into globular and filamentous forms (G/F-actin). (d) Intravitreal injection of MK-801 lowers the optic nerve G/F-actin ratio in iAON animals compared to saline treatment (n = 6 animals per group), restoring it to healthy levels (n = 10 animals). (e) Western blotting of optic nerve lysates with antibodies against gelsolin (98 kDa) and the actin-cleavage product, fractin (32 kDa), with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (38 kDa) as a loading control. (f) MK-801 treatment increases the level of gelsolin compared to saline treatment (saline, n = 5; MK-801, n = 4 animals) and restores gelsolin to healthy levels (n = 9 animals). (g) Fractin levels are decreased in optic nerves of iAON rats receiving MK-801 compared to those receiving saline (saline, n = 4; MK-801, n = 5 animals), and are comparable to healthy (n = 8 animals). (h) T1-weighted scan showing region of interest containing retina (white box), and brain area containing olfactory bulb (white ellipse) taken as a control for normalization. (i) Representative image of T1-weighted scans taken before and after (1 hr) MnCl 2 injection of iAON rats which had received systemic saline or MK-801 prior to MRI imaging. (j) Quantification of manganese enhancement of T1-weighted MRI signals in treated animals as % signal increase after MnCl 2 injection normalized to the manganese enhancement of healthy retina. iAON retinas from rats receiving MK-801 had reduced manganese enhancement compared to those receiving saline (saline n = 10, MK-801 n = 8 animals). Scale bars, (a) = 100 µm, (i) = 1 mm. *p < .05; **p < .01; ***p < .001; n.s. -non-significant (ANOVA) amplitudes (Table 2; Figure 5g), suggesting that MK-801 treatment reduced the decrease in visual acuity observed during AON.

| D ISCUSS I ON
In this study we demonstrate that, during the induction phase of AON, a decrease in visual acuity and changes in the optic nerve cytoskeleton has also been suggested to interfere with monoamine uptake (Callado, Hopwood, Hancock, & Stamford, 2000) and to block other receptors such as the nicotinic acetylcholine receptor (Amador & Dani, 1991), since the AON changes could also be mimicked by glutamate injection, this would suggest that the NMDA receptor is the likely candidate.
The role of glutamate receptors in mediating neurodegeneration has been well described in many different pathophysiological conditions such as brain trauma, cerebral ischaemia and Alzheimer's disease (Lipton, 2006). This is also true of different pathological conditions of the retina, such as diabetic retinopathy (Gu et al., 2014;Santiago et al., 2009;Santiago, Hughes, Kamphuis, Schlingemann, & Ambrósio, 2008), retinal vein thrombosis (Mosinger et al., 1991;Nivison-Smith, Khoo, Acosta, & Kalloniatis, 2018) and glaucoma (Fu & Sretavan, 2012;Ju et al., 2015). NMDA receptor stimulation might be involved. For example, brain endothelial cells are known to express NMDA receptors whose activation leads to an increase in blood-brain barrier permeability (Sharp et al., 2003;Vazana et al., 2016), and thus it is possible that retinal endothelial cells similarly express NMDA receptors that contribute to pathological mechanisms in AON. Intravitreal MK-801 treatment might therefore reverse blood-retinal barrier-induced damage by decreasing the influx of proinflammatory compounds such as fibrin (Davalos et al., 2012)  Abbreviations: iAON, induction autoimmune optic neuritis; pERG, pattern electroretinogram; pVEP, pattern visual evoked potential. (Furukawa, Smith-Swintosky, & Mattson, 1995;Neely & Gesemann, 1994) cause F-actin disassembly, leading to a destabilization of the actin network. Significant increases in intracellular calcium also activate calcium-dependent proteases such as calpains and caspases, which further destabilize the actin cytoskeleton. One such protease activated by calcium is the actin-severing protein gelsolin (Yin & Stossel, 1979). At the same time, gelsolin is cleaved by calpain/ caspase (Kothakota et al., 1997) which, as an anti-apoptotic factor (Harms et al., 2004), may leave the cell more vulnerable to NMDA receptor activity. In addition, fractin, a calpain/caspase-cleaved actin monomer product which accumulates following induction of apoptosis (Brown, Bailey, & Savill, 1997), also plays a functional role in apoptotic signalling (Schulz, Vogel, Mashima, Tsuruo, & Krieglstein, 2009). Both our findings of gelsolin and fractin changes are in agreement with our previous report of increased calpain activity during iAON (Hoffmann et al., 2013). Moreover, changes in actin network dynamics might be involved in the restructuring of Nodes of Ranvier following both intravitreal injections of glutamate and during early AON, as has been recently reported (Stojic et al., 2018). However, no changes in gelsolin were observed following injection of glutamate despite the protective effects of MK-801 in iAON, suggesting that a decrease in gelsolin levels might be dependent upon NMDA receptor activation in addition to other, currently unidentified, disease-related factors. Conversely, since gelsolin is protective against apoptosis (Harms et al., 2004), it is conceivable that following glutamate-mediated stress its expression might be up-regulated, but for reasons that are unclear (but might reflect the complexity of the disease scenario) this appears to have failed in AON.
Different visual tests can be used to provide insight into the function of different visual pathway components. In general, fVEPs are considered to reflect the activity of all cells in the pathway from photoreceptors to the visual cortex, with alterations closely correlating with optic nerve damage (Halliday, McDonald, & Mushin, 1972. As such, it is a key parameter for optic neuritis with changes in latency and amplitude reflecting the extent of demyelination and axonal injury respectively (You, Klistorner, Thie, & Graham, 2011). In our study, no changes in fVEPs were detected, which fits with the absence of inflammatory-driven demyelination or axonal loss during the induction phase of AON (Fairless et al., 2012). In contrast, the more sensitive measure of pVEPs is believed to reflect the signalling from RGCs to the visual cortex. The most specific technique, however, for assessing RGC function is through measurement of pERGs (Porciatti, 2015), and has been shown to correlate with the degree of RGC degeneration in AON (Hobom et al., 2004;Mayer et al., 2018;Meyer et al., 2001). Furthermore, this technique is highly sensitive being able to detect RGC dysfunction prior to the onset of death, as demonstrated in models of glaucoma (Porciatti, 2015).
The early degeneration that is seen in this model prior to the demyelination and inflammatory infiltration that characterize optic neuritis is in contrast to the classical concept that secondary degeneration of RGCs occurs as a result of axonal damage in the demyelinated optic nerve (Shindler, Ventura, Dutt, & Rostami, 2008). In this manner, the Brown Norway rat model also differs from the mouse model where significant RGC loss is not seen until after onset of optic nerve demyelination. An explanation for this is probably the early and robust antibody response observed in Brown Norway rats (Stefferl et al., 1999), which has been demonstrated to accumulate in the optic nerve head during the induction phase prior to immune cell infiltration and demyelination (Stojic et al., 2019), a time when subtle changes in the axo-glial junctions are also observed (Stojic et al., 2018). It is also interesting that in MS patients, RGC degeneration in the absence of optic nerve demyelination may also be occurring as evidenced by the observations of retinal nerve fibre layer thinning even in the absence of optic neuritis (Bock et al., 2010;Petzold et al., 2010;Talman et al., 2010). Thus, it is conceivable that subtle, subclinical optic nerve changes may initiate retinal degenerative processes that in turn affect visual performance in an NMDA receptor-mediated manner.
In conclusion, data presented in this study support the hypothesis that in early AON retinal events can lead to anterograde optic nerve changes in actin cytoskeletal dynamics, probably mediated by calcium accumulation and activation of actin-regulatory proteases. In addition, we provide further evidence that NMDA receptor modulation may prove a therapeutically relevant strategy for achieving retinal neuroprotection under autoimmune neuro-inflammatory conditions. Most importantly, MK-801, chosen in this study because of its high specificity and persistent inhibitory kinetics (Halliwell, Peters, & Lambert, 1989;Huettner & Bean, 1988;McKay et al., 2013), was able to protect against visual disturbances in iAON affected eyes. Even though MK-801 is known to have side effects precluding its therapeutic use such as interfering with long-term potentiation (Frankiewicz, Potier, Bashir, Collingridge, & Parsons, 1996) and induction of psychosis (Andine et al., 1999), if applied locally, as in this study was done by intravitreal injection, this blocker might be reconsidered in future therapeutic studies. All experiments were conducted in compliance with the ARRIVE guidelines.

CO N FLI C T S O F I NTE R E S T
The authors have no competing financial interests.

O PE N S CI E N CE BA D G E S
This article has received a badge for *Open Materials* because it provided all relevant information to reproduce the study in the manuscript. More information about the Open Science badges can be found at https://cos.io/our-servi ces/open-scien ce-badge s/.