Experimental Eye Research 119 (2014) 61e69

Contents lists available at ScienceDirect

Experimental Eye Research journal homepage: www.elsevier.com/locate/yexer

Sera from patients with seropositive neuromyelitis optica spectral disorders caused the degeneration of rodent optic nerve Yoshiko Matsumoto a, Akiyasu Kanamori a, *, Makoto Nakamura a, Toshiyuki Takahashi b, Ichiro Nakashima b, Akira Negi a a b

Division of Ophthalmology, Department of Surgery, Kobe University Graduate School of Medicine, 7-5-1, Kusunoki-cho, Chuo-ku, Kobe 650-0017, Japan Department of Neurology, Tohoku University School of Medicine, Sendai, Japan

a r t i c l e i n f o

a b s t r a c t

Article history: Received 4 June 2013 Accepted in revised form 12 December 2013 Available online 24 December 2013

Neuromyelitis optica (NMO) is an autoimmune inflammatory, neurodestructive disease primarily targeting the optic nerve and spinal cord. An autoantibody against water channel protein aquaporin-4 (AQP4), which is expressed at endofeet of astrocytes has been implicated in the pathogenesis of NMO. We evaluated the impact of sera of seropositive patients with NMO spectrum disorders (NMOSDs) on the rodent optic nerve and retina. Serum was obtained either from patients with seropositive NMOSD (AQP4þ), seronegative patient with idiopathic optic neuritis (AQP4), and healthy volunteers (control). Anti-AQP4 antibody in a serum was measured by a previously established cell-based assay. The patients’ sera were applied on the optic nerve after de-sheathed. Immunohistochemistry showed that at 7 days after the treatment, the area of the optic nerve exposed to the AQP4þ sera lost expression of both AQP4 and glial fibrillary acidic protein. Also, Human-IgG immunoreactivity and marked invasion of inflammation cells were observed in the optic nerve treated with AQP4þ serum. Immnoreactivity of neurofilament was reduced at 14 days after the treatment, not 7 days. Real-time polymerase chain reaction revealed the reduced gene expression of neurofilament in retina from the eye that was exposed to the AQP4þ sera at 14 days. Retrograde fluorogold-labeling on the retinal flatmount disclosed the significantly reduced number of retinal ganglion cells when the AQP4þ sera were applied. The present model has demonstrated that the sera from patients with seropositive NMOSDs led to the regional astrocytic degeneration and inflammatory cell invasion in the optic nerve, resulting in the ultimate loss of RGCs and their axons at areas beyond the injury site. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.

Keywords: neuromyelitis optica aquaporin-4 optic neuritis serum optic nerve nerve fibers astrocyte

1. Introduction Neuromyelitis optica (NMO) is an autoimmune inflammatory disease of the central nervous system (CNS) associated with the first inflammatory demyelinating lesions mainly in optic nerve and spinal cord, leading to blindness and paralysis (Jarius and Wildemann, 2010; Kitley et al., 2012). The majority of NMO patients are seropositive for autoantibodies (NMO-IgG) against extracellular epitopes on aquaporin-4 (AQP4) (Lennon et al., 2004, 2005; Misu et al., 2006). AQP4, one of plasma membrane water channels, is expressed in astrocytes throughout the CNS (Nielsen et al., 1997). The establishment of animal models of NMO is required for understanding the pathogenesis of NMO and for testing of candidate therapies. Intravenously administered recombinant NMO-IgG

* Corresponding author. Tel.: þ81 78 382 6048; fax: þ81 78 382 6059. E-mail address: [email protected] (A. Kanamori).

could not pass the blood-brain barrier into CNS and was not able to cause neuroinflammation in mice (Ratelade et al., 2011). To induce NMO-like lesion in CNS by peripheral NMO-IgG or recombinant AQP4 antibody administration, pre-existing neuroinflammation should be produced by experimental autoimmune encephalomyelitis caused by myelin basic protein (MBP), MBP peptide or MBP-reactive T cells (Bennett et al., 2009; Kinoshita et al., 2009a). However, such pretreatments by immunization perhaps intercept for the results because the administration of MBP or MBP-reactive T cell caused optic neuritis in rodents by itself (Hu et al., 1998; Milici et al., 1998). Direct administration of NMO-IgG into CNS tissue would be ideal as NMO-model. Recently, Saadoun and the colleagues established an excellent model in which NMOlike lesion was produced in brain of naïve mice (Saadoun et al., 2010). They demonstrated that intra-cerebral administration of IgG from NMO-patients produced lesions with NMO-like characteristics including inflammation, loss of AQP4 and GFAP immunoreactivity and myelin loss (Saadoun et al., 2010).

0014-4835/$ e see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.exer.2013.12.010

62

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

However, the usefulness of intra-cerebral administration is limited in investigating the white matter lesion, but not the optic nerve lesion. To date, only ex vivo model was available for evaluating the optic nerve damage relevant to NMO. Damage of the optic nerve, which comprises of the retinal ganglion cell (RGC) axons and astrocyte glia, leads to RGC loss by the Wallerian degeneration. The direct exposure of AQP4 antibody from NMO-patients to the optic nerve would be more easy and reproducible procedure to further understand the pathogenesis of optic neuritis in NMO. Here, we demonstrate that such a procedure in fact generates an NMO-like lesion in the optic nerve and neural cell loss in the retina and could be used to develop optic neuritis relevant to NMO. 2. Methods

procedures, the animals were anesthetized by intraperitoneal injection of 14 mg/kg ketamine and 4 mg/kg xylazine. To expose the optic nerve to human serum, a superior conjunctival approach was chosen. After the optic nerve was exposed by blunt dissection, the posterior ciliary artery was identified on the ventromedial aspect of the nerve sheath. Avoiding damage on the artery, the optic nerve sheath was incised longitudinally about 1 mm at 2.5 mm behind the eye with an 18-gauge fine needle as previously described (Kanamori et al., 2010a). The collected serum (2 ml) was injected underneath the optic sheath using custom-made 32-gauge needle with Hamilton-syringe, and then Medical Quick AbsorberÒ soaked with serum was left beside the optic nerve. The operator was masked to the positivity of AQP4 antibody in each serum. Preservation of the retinal circulation was confirmed by an indirect ophthalmoscopy.

2.1. Correcting patient serum The institutional review board of Kobe University approved the study protocol. We collected serum from Japanese patients with idiopathic or optic neuritis with NMO spectrum disorders (NMOSDs) or from normal subjects. The diagnosis of optic neuritis was based on subacute onset of decreased visual acuity, a visual field defect, a relative afferent pupillary defect and gadolinium enhancement of the optic nerve on the fat-suppressed magnetic resonance imaging (MRI) at the onset of optic neuritis. Table 1 summarized the patient demographics. An AQP4 titre in each patient was measured by a cell-based assay (Takahashi et al., 2007; Kitley et al., 2012). The treated rats were divided into three groups based on the serum samples: AQP4 antibody seropositive (AQP4þ), AQP4 seronegative (AQP4) and normal subjects (control). Each rat was treated with serum from a single sample. In experiments involving four or six rats, we used different serums from patients with NMO-optic neuritis (P1 to P4 or P1 to P6) and different serums from patients with idiopathic optic neuritis (C1 to C4). 2.2. Administration of serum on optic nerve The experiments in this study were approved by the Animal Care Committee of the Kobe University Graduate School of Medicine and followed the Association for Research in Vision and Ophthalmology Resolution on Care and Use of Laboratory Animals. Male Sprague-Dawley rats, ranging in weight from 200 to 300 g (CLEA Japan, Osaka, Japan), were housed at room temperature (24  2  C) in the Kobe University animal facility with ad libitum access to food and water under a 12-h light/12-h dark cycle. For all

2.3. Immunohistochemistry For the preparation of cryosections, anesthetized rats were transcardially perfused with 4% paraformaldehyde at 3 days (3D), 7 days (7D) and 14 days (14D) after the treatment. Eye with optic nerve was dissected and then optic nerve was cut at 1 mm behind the eye globe. Eye ball and optic nerve were fixed in 4% paraformaldehyde overnight at 4  C. The post-fixed optic nerves or retinas were cryoprotected in 15% and then 30% sucrose in PBS, embedded in Tissue-Tek optimal cutting temperature compound (Sakura Finetechnical, Tokyo, Japan), and then stored at 80  C until immunostaining was performed. Cryosections (8 mm in thickness) of retina and optic nerve were collected on APS pre-coated glass slides (Matsunami Glass, Osaka Japan). After 10 min fixation with 4% paraformaldehyde at 4  C, the sections were pre-incubated with PBS containing 0.1% Triton X-100 (PBST) for 10 min at room temperature. The sections were then blocked with 10% normal goat serum in PBST (blocking solution) for 1 h, followed by incubation in blocking solution overnight at 4  C with one of the following antibodies: mouse anti-neuronal nuclei (NeuN) (1:100) (MAB377; Millipore, Billerica, MA), mouse anti-neurofilament (NF) (1:200); (N0142; Sigma, St. Louis, MO) mouse anti-glial fibrillary acidic protein (GFAP) (1:200) (MAB360; Millipore), mouse anti-myelin basic protein (MBP) (1:100) (NE1018; Sigma) or rabbit anti-C5b9 (1:200) (ab55811; Abcam, Cambridge, MA). Rabbit anti-aquaporin-4 (AQP4) (1:200) (SC20812; Santa Cruz) was also used. After three washes (20 min each) with PBST, the sections were incubated with fluorescein isothiocyanate-conjugated F(ab)2 fragment anti-mouse IgG and tetramethyl rhodamine isothiocyanate (TRITC)-conjugated antirabbit IgG (AffiniPure; Jackson ImmunoResearch, West Grove, PA)

Table 1 Patient demographics of the 6 patients with NMO-optic neuritis and the 6 patients with idiopathic optic neuritis. Patient

Sex

Age at onset

AQP4 antibody titres in serum

Disease course

The number of optic neuritis

The side of the affected eye

P1 P2 P3 P4 P5 P6 C1 C2 C3 C4 C5 C6

F F F F F F F F M M M F

11 65 53 47 51 38 38 41 23 58 40 26

16384x 16x 65536x 128x 16384x 512x Negative Negative Negative Negative Negative Negative

First attack First attack Relapsing First attack Relapsing Relapsing First attack First attack First attack Relapsing First attack First attack

1 1 7 2 2 3 1 1 1 2 1 1

Left Left Both Left Both Both Right Left Left Right Left Right

LP; light perception. a Visual acuity was evaluated after the treatment for acute phase.

Visual acuitya Right eye

Left eye

20/20 20/20 No LP 20/20 20/400 20/800 20/20 20/20 20/20 20/20 20/20 20/20

20/200 20/60 20/30 20/40 20/40 20/30 20/20 20/20 20/20 20/20 20/20 20/40

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

63

for 1 h at room temperature and counterstained with bisbenzimide (Hoechst dye 33258, 0.5 mg/ml, SigmaeAldrich Japan, Tokyo, Japan). After extensive washing with PBST, the slides were mounted on cover slips with an aqueous mounting medium (PermaFluor; Laboratory Vision Corporation, Kalamazoo, MI). Some sections were incubated overnight with anti-human-IgG conjugated with TRITC (1:100) (A-11014; Invitrogen) after the blocking. To quantify the sections with fluorescence of NF, GFAP or AQP4, 4 eyes treated with the serum from different patients were analyzed in each group. All sections for a single quantification were subjected to immunohistochemical procedures at the same time. Images under a fluorescence microscope (BZ-8000, KEYENCE, Osaka, Japan) were captured at 3 sections from one eye for each quantification area. Microscope setting was kept constant throughout image requisition. Fluorescence intensity unit per area was calculated using “Analysis” tool (0e255, where 0 ¼ dark and 255 ¼ white) in Adobe Photoshop CS5 Extended (version 10.0.1, Adobe Systems, Tokyo, Japan). The mean grey level in each quantified image was used as an arbitrary unit of immunolabeling intensity.

15 min. After centrifugation at 15,000 rpm for 10 min, the supernatant was collected and protein concentration was determined using a Bio-Rad Protein Assay Kit (Bio-Rad Laboratories, CA, USA). Two optic nerve specimens were pooled for one sample. Fifteen mg of proteins from each sample were subjected to 10% SDS-polyacrylamide gel electrophoresis and were transferred onto nitrocellulose membranes. Following 1 h blocking with 5% non-fat milk, the membranes were incubated with primary antibodies overnight. The primary antibodies used were rabbit anti-AQP4 antibody (1:500) (SC20812; Santa Cruz), or mouse anti-beta-actin (1:500) (ab54724; Abcam). After washing, the membranes were incubated with HRP-conjugated anti-rabbit IgG or HRP-conjugated anti-mouse IgG at room temperature for 1 h. Protein bands were visualized with enhanced chemiluminescence reagents (ECL plus; Amersham, Arlington Heights, IL, USA) and LAS-3000 (Fujifilm, Tokyo, Japan). Relative intensities of bands were quantified by densitometric analysis using NIH Image software. The ratios of AQP4 to beta-actin protein level of were compared between AQP4þ and control optic nerve.

2.4. Gene expression

Retinal ganglion cells (RGCs) were retrogradely labeled by bilateral stereotactic injection of fluorogold into the superior colliculus 5days before the treatment with patient’s serum to the optic nerve (Kanamori et al., 2010b). Briefly, rats were placed in a stereotactic apparatus (Narishige Co. Ltd., Tokyo, Japan), and the skin of the skull was incised. The brain surface was exposed by drilling the parietal bone to facilitate dye-injection. Five percent fluorogold (Fluorochrome, Denver, CO), 2.1 ml volume in total, was injected bilaterally at 5.5 mm caudal to bregma and 1.2 mm lateral to the midline to a depth of 4.5 mm from the skull surface. To quantify RGC numbers, retinas were dissected and fixed in 4% paraformaldehyde and then extended on a glass slide with ganglion cell layers facing up. Fluorogold-positive, surviving RGCs were identified under a fluorescence microscope (BZ-8000) using a UV filter set (377/407 nm) (Semrock, Rochester, NY, USA). RGCs were counted in 12 areas of 0.072 mm2 each (three areas per retinal quadrant) at 2/6, 3/6 and 5/6 of the retinal radius by an investigator masked to the treatment condition as previously described (Kanamori et al., 2009).

For real-time reverse transcription quantitative polymerase chain reaction (RT-qPCR), optic nerves and retinas were dissected from the eyes immediately after death and directly subjected to the analyses as described below (Naka et al., 2010). Total RNA was extracted from optic nerves using an RNeasy Plus Mini Kit (Qiagen, Valencia, CA). The amount of total RNA was quantified with a SmartSpec spectrophotometer (Bio-Rad Laboratories, Inc.). Total RNA was reverse transcribed using a QuantiTect Reverse Transcription Kit (Qiagen). Quantitative PCR was performed using TaqMan Universal master mix (Applied Biosystems, Foster City, CA) and a target gene probe on a StepOnePlus real-time PCR system (Applied Biosystems). The target gene were NF (assay ID, Rn0070935_m1; Applied Biosystems), GFAP (assay ID, Rn00566603_m1), and AQP4 (assay ID, Rn00563196_m1). GAPDH probe (assay ID, Rn99999916_s1) was used as an internal control. Each sample was run in triplicate. Serial dilutions of the templates were used in quantitative PCR amplifications to generate a linear standard curve of the logarithm of the DNA concentration versus CT for each gene. Fold change calculation was performed with the comparative threshold cycle (Ct) or the DDCt method, based on the formula FC ¼ 2  DDCt, to calculate normalized FC in gene expression of test samples relative to a calibrator sample. The first step in FC analysis was normalization of the target gene expression signal against GAPDH gene expression (DCt). The second step was to calculate the difference in normalized target gene expression between AQPþ and control samples (DDCt). Fold change calculation, which represents the difference in gene expression level, was carried out for each gene individually using the formula described. The significance of the fold change was tested using an unpaired t-test. The relative ratio compared to the control was shown. 2.5. Immunoblotting analysis Protein preparation and immunoblotting were performed as previously reported with a minor modification (Kanamori et al., 2005). Optic nerves (5-mm lengths) were collected and homogenized using a rotor-stator homogenizer (Mixer Mill MM400, Retsch, Haan,Germany) and moved into the lysis buffer (20 mM TriseHCl (pH 7.5), 150 mM NaCl, 1 mM Na2 EDTA, 1 mM EGTA, 1% Triton, 0.5% NP-40, 2.5 mM sodium pyrophosphate, 1 mM betaglycerophosphate, 1 mM Na3VO4, 1 mg/ml leupeptin, and 1 mM PMSF; Cell Signaling Technology, Danvers, MA) and rocked in 4  C for

2.6. Counting of retinal ganglion cells

2.7. Statistics Statistical analyses were performed with Excel software (version 2007; Microsoft) and MedCalc (version 11.6.1.0, Mariakerke, Belgium). All statistical values were judged significant if P < 0.05. Data were shown in mean  standard deviation. 3. Results 3.1. The localized and precedent reduction of AQP4 and GFAP expression in the optic nerve after exposure of seropositive serum We examined AQP4 and GFAP expression in the optic nerve by immunohistochemisty at 3D, 7D, and 14D after exposure of patients’ sera to the optic nerve. AQP4 was co-localized with GFAP immnoreactivity (Fig. 1B). No apparent change was observed at 3D (data not shown). The fluorescence intensities of AQP4 and GFAP in the optic nerve at 7D and 14D were evaluated in the area exposed to the serum (EA), the area near to the eye (eye side; ES) and the area for brain side (BS). In EA, the fluorescence intensity of AQP4 (P ¼ 0.016) and GFAP (P ¼ 0.026) were decreased in AQP4þ group at 7D compared with the control group (N ¼ 4). In contrast, both in ES and BS which were remote from the area exposed to the serum, both AQP4 (ES;

64

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

Fig. 1. GFAP and AQP4 expression in the optic nerve. A: Areas of rat optic nerve for immunohistochemical analyses of GFAP (green). The area near to the eye (eye side; ES), the area exposed to the serum (EA), and the area for brain side (BS) were analyzed. Scale bar ¼ 100 mm. B, C: Representative photos in EA of immunohistochemistry with antibodies against GFAP (green) and AQP4 (red) at 7D (B) and 14D (C). Scale bar ¼ 100 mm. Nuclei are marked with Hoechst (blue). Scale bar ¼ 100 mm. The immunoreactivities of AQP4 and GFAP treated with AQP4 seropositive serum from NMO-patient (AQP4þ) at 7D was decreased compared with control group. D: Quantitative analyzes of fluorescence intensity (FI) of GFAP and AQP4 in EA at 7D (arbitrary unit, N ¼ 4, mean  SD). AQP4þ group significantly lost GFAP and AQP4 expression. *, P < 0.05. E: Immunoblot analysis of AQP4 protein in the optic nerve at 7D. Data are normalized to beta-actin levels in each sample (N ¼ 4, mean  SD).

P ¼ 0.26, BS; P ¼ 0.22) and GFAP (ES; P ¼ 0.98, BS; P ¼ 0.35) did not significantly changed between AQP4þ and the control group at 7D (Fig. 1D). At 14D, the fluorescence intensity of GFAP in the entire regions of the optic nerve in the AQP4þ group was not changed compared with the control group (N ¼ 6). Likewise, the AQP4 fluorescence intensity in the AQP4þ group in the entire regions of the optic nerve was not significantly different from the control group (N ¼ 6). Furthermore, the gene expression of GFAP and AQP4 in the optic nerve was evaluated by real-time RT-PCR. However, theses expressions in the AQP4þ group did not significantly changed compared with control group (GFAP; 7D, P > 0.5, 14D, P ¼ 0.09) (AQP4; 7D, P ¼ 0.18; 14D, P > 0.5). Additionally, we performed immunoblotting for the AQP4 protein in the optic nerve. The AQP4 protein tended to decrease in the AQP4 group compared to the control group, but the decrease was not statistically significant (N ¼ 4, P ¼ 0.19). This may be due to the fact that the AQP4 positive serum only led to localized changes (approximately 1 mm) in AQP4 and GFAP expression in the areas exposed to the sera. Collectively, the exposure of anti-AQP4 antibody positive sera to the optic nerve induced the reduced expression of AQP4 and GFAP in the restricted area of the optic nerve. 3.2. Detection of human-IgG in astrocytes of the optic nerve after exposure of AQP4 seropositive serum Cyrosections of the optic nerve at 7D treated with AQP4þ serum was immunostained with anti-human-IgG and various cell-specific markers. As shown in Fig. 2, human-IgG was co-localized with the astrocyte marker GFAP, but not the oligodendrocyte marker MBP. It was also co-localized with the nerve fiber marker NF in some portion of the optic nerves. Axonal flow might up-take antibody from the damaged axons, but the exact mechanism was not cleared

by our study. Interestingly, human-IgG was scarcely co-localized with GFAP in the retina in eyes exposed to the AQP4 antibody positive sera at 7D. 3.3. Marked inflammatory response in the optic nerve after exposure to AQP4 seropositive serum To evaluate inflammation in the optic nerve, the CD11-positive cells in the optic nerve were counted at 7D and 14D. CD11 is a maker of mainly macrophages, monocytes and granulocytes and excessive staining of CD11 could be considered as the existence of inflammation. CD11-positive cells which had long branched processes were present in the native optic nerve, whereas massive increase of CD11-positive cells was observed in the optic nerve of AQP4þ group as shown in Fig. 3A. The cell densities in AQP4þ and control group were 259  87/mm2 and 125  55/mm2 at 14D, respectively. The ratio of the CD11-positive cell density in AQP4þ group to control group was 1.5  0.29 (at 7D) and 2.1  0.69 (at 14D) (Fig. 3B). CD11-positive cells were significantly increased in AQP4þ group than control group (N ¼ 6, P < 0.001) at 7D and 14D. Complement CD5 positive cells were present in the optic nerve following treatment with AQP4þ serum at 7D (Fig. 3C). It is believed that complement-dependent cytotoxicity might also be involved. Haematoxylin and eosin stain revealed that massive cell invasion in the optic nerve treated with AQP4þ serum compared to the control and normal optic nerve without any treatments (Fig. 3D). 3.4. The disorganized and reduced NF expression after exposure of AQP4 seropositive serum on the optic nerve We evaluated the change of optic nerve fibers both by the fluorescence intensity in the NF immunohistochemistry and by gene expression. The intensity of fluorescein was calculated in ES and BS.

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

65

Fig. 2. Localization of human-IgG in the optic nerve and retina treated with the serum from AQP4 seropositive NMO-patients. The optic nerve (A) and retina (B) were stained with antibodies against GFAP, NF or MBP (green) and anti-human-IgG (red). Nuclei are stained with Hoechst (blue). Scale bar ¼ 100 mm. The images indicate that human-IgG was colocalized with the astrocytic marker GFAP, but not the oligodendrocytic marker MBP in optic nerve. Human-IgG immunoreactivity was also detected at astrocytes in retina.

The preliminary experiments revealed that there were not apparent changes in the treated optic nerve sections at 3D. Although the overall NF fluorescence intensity did not change at 7D among groups (ES; P ¼ 0.33, BS; P > 0.5), disrupted NF immunoreactivity was also observed in some portions of the optic nerve in the AQP4þ group. As shown in Fig. 4B, NF immunoreactivity was reduced and became punctate in pattern at 14D. The NF fluorescence intensity significantly decreased at 14D compared with control group in both ES (P ¼ 0.005) and BS (P ¼ 0.013) (Fig. 4, C and D). Real-time RT-PCR revealed that the gene expression of NF in the optic nerve with AQP4 seropositive serum did not show the significant reduction at 7D (0.67  0.57, P ¼ 0.056) and 14D (0.87  0.81, P ¼ 0.225) (Fig. 4E). One possible reason for this may be due to the small amount and the large variability of the optic nerve sample volume among eyes. We also evaluated the NF gene expression in the retina. Nerve fiber in the optic nerve is formed by the expansion of retinal nerve fiber from RGC. The assessment of NF in retina could be considered as an evaluation method of axonal damage. As shown in Fig. 4F, the expression of NF was not changed between AQP4þ and control group (P > 0.5) at 7D, but at 14D the expression in AQP4þ group was significantly decreased (N ¼ 6) (0.60  0.14, P ¼ 0.027). Therefore, AQP4 seropositive serum damaged optic and retinal nerve fibers. After the treatment these degenerations started from at least 7 days and were apparent at 14 days.

3.5. The loss of RGCs after exposure of AQP4 seropositive serum on the optic nerve We evaluated the effect of AQP4 seropositive serum by the number of RGCs. RGCs labeled with fluorogold were quantified on whole-mounted retinas at 14D in this study. Microglial cells (spindle-shaped, smaller than RGCs) were also detected in retinas treated with the serum from NMOSDs-patients. They were distinguished from RGCs by their shape, and were excluded from the cell counts. As shown in Fig. 5, the RGC density in the AQP4þ group (1013  263/mm2) was significantly decreased compared with AQP4 (1846  173/mm2) and control group (2038  248/mm2) (N ¼ 4, p ¼ 0.002). The RGC density in normal retinas without any optic nerve-treatments was 2086  211/mm2 (N ¼ 4). There was no significant difference between AQP4 and normal group (P ¼ 0.13). These results revealed that AQP4 seropositive serum produced RGC death. 3.6. Increased GFAP expression in the retina after exposure to AQP4 seropositive serum GFAP expression in retina could be evaluated as a stress marker after axonal injury at optic nerve. (Chen and Weber, 2002; Chidlow et al., 2005; Kanamori et al., 2005; van Adel et al., 2005) As shown in Fig. 6, increased GFAP immunoreactivity that spanned the retina

66

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

Fig. 3. Evaluation of inflammation in the optic nerve. A: The optic nerve sections at 14 days after the treatment were stained with antibodies against CD11 (green). Nuclei are stained with Hoechst (blue). Scale bar ¼ 100 mm. B: The ratio of CD11-positive cell density in AQP4þ group to the control group at 7D and 14D. C: The optic nerve sections at 7D were stained with antibodies against complement C5 (red). D: Haematoxylin and eosin stain revealed that massive cell invasion in the optic nerve treated with AQP4þ serum (No treatment: normal optic nerve without any treatments, control: treated with control serum).

and corresponded to Müller cells was observed in the retinal sections from the eye with AQP4þ group compared with the control group at 14D. The gene expression of GFAP in AQP4þ group at 14D (1.83  0.76, N ¼ 6) was increased than control group (N ¼ 6) (P ¼ 0.01). These data suggested that astrocytic activation in the retina secondarily occurred, which was accompanied by axonal loss. 4. Discussion We demonstrated that the direct administration of the serum from NMOSDs-patients for rat optic nerve produced loss of AQP4 and GFAP and followed by the loss of optic nerve fibers and inflammatory cell infiltration. The exposed area of optic nerve had

similar histological features of human NMO-lesions (Misu et al., 2007). We confirmed the neural cell loss that was accompanied by the axonal loss and glial activation in the retina. Our study would be the first report showing optic neuritis model directly using NMO-IgG in vivo. Evidence is accumulating that AQP4 antibody is pathogenic itself and is able to induce astrocyte death in vitro and ex vivo using optic nerve and spinal cord (Hinson et al., 2007; Kinoshita et al., 2009b; Sabater et al., 2009; Marignier et al., 2010; Zhang et al., 2011). Previous in vivo studies also revealed that AQP4 antibody reduced immunoreactivities of AQP4 and GFAP (Bradl et al., 2009; Kinoshita et al., 2009a; Saadoun et al., 2010). Furthermore, the intra-cerebral injection of IgG from NMO-patients enabled us the longitudinal observation in NMO-pathology and revealed that

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

67

Fig. 4. Evaluation of neurofilament damage in nerve and retina. A: Areas of rat optic nerve for immunohistochemical analyses of neurofilament (NF) (green). Inserted squares indicate areas that were subjected to quantitative analyses. EA, an area where patients’ sera were applied; ES, an area 600 mm proximal to the eyeball side from EA, BS; an area 600 mm distal to the eyeball from EA. Scale bar ¼ 100 mm. B: Representative photos of immunohistochemistry with antibodies against NF (green) and AQP4 (red) at EA. Nuclei are stained with Hoechst (blue). Scale bar ¼ 100 mm. AQP4 seropositive serum from NMO-patient (AQP4þ) leads to disorganized NF staining. C and D: Quantitative analyses of NF fluorescence intensity of ES (C) and BS (D) in the optic nerve at 14D (N ¼ 4) (arbitrary unit). The AQP4þ group shows significantly reduced NF expression compared with control group. E and F: The expression of NF gene in optic nerve (E) and retina (F) examined by real-time PCR. Compared with AQP4 group, NF of AQP4þ group did not changed in optic nerve and 7D retina, but was significantly decreased in 14D retina (N ¼ 6, mean  SD). *, P < 0.05.

NMO-IgG produced loss of AQP4 expression, subsequent loss of astrocytes and extensive demyelination (Saadoun et al., 2010). In this study, we found that astrocytic AQP4 expression in the restricted area of the optic nerve was initially reduced, followed by the extensive loss of the RGCs and their axons beyond the injured site. Primary astrocyte injury and initiation of an inflammatory cascade are believed to secondarily damage oligodendrocytes and neurons in NMO. This model reproduced similar pathogenesis in NMO-patient’s spiral cord in which acute axonal injury and delayed neural cell death were previously demonstrated. The key of this in vivo model was how patient’s serum could be invaded well into rat optic nerve. As shown in Fig. 2, the immunoreactivity of human-IgG with TRITC implied that the serum pervasively invaded and docked with astrocytes. Optic nerve sheath which is composed of a stiff membrane extending from sclera and jointing to bone membrane of optic canal has a barrier between the intra- and outside of the optic nerve. When optic nerve fibers were directly labeled by a dye to quantitate RGC

number, a certain degree of incision or injection of the dye into the optic nerve was fundamentally needed (Arkin and Miller, 1988; Ishii et al., 2003; Kudo et al., 2006). The incision of the sheath enabled the dye placed on the optic nerve to invade into the entire nerve and to label all RGCs in retina. The incision of optic nerve sheath has been adapted for such labeling-method or as a sham operation for the transection in past numerous studies. Hence, we believed that the incision of the optic nerve did not damage fibers in significant level. In this study, the number of surviving RGCs in rats treated with AQP4 or control serum did not decrease in comparison with that of normal rats and the fact supported our experimental approach. Unfortunately, this procedure could not be performed in mice because of their small size of the optic nerve and operative difficulty. Also, the direct injection into the optic nerve was impractical in mice because the optic nerve was so small that a needle might damage nerve fibers in some degree. As far as we knew, histological examinations using autopsy have not been reported in the optic nerve with NMO. Histopathological

68

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

Fig. 5. Representative photographs of fluorogold-labeled retinal ganglion cells (RGCs) in whole-mounted retina, 14 days after the treatment with control serum (A), serum from the patient with idiopathic optic neuritis (AQP4) (B) and serum from AQP4 seropositive NMO-patient (AQP4þ) (C). Arrowheads indicate microglia. Scale bar ¼ 50 mm. D: Density of surviving RGCs in whole-mounted retina. The direct administration of NMO-patient’s serum on the optic nerve decreased RGCs compared with normal retinas without any treatments on optic nerve, the control and AQP4 group (mean  SD). *, P < 0.05.

features of optic neuritis in NMO remain unclear, but that would be similar with the NMO-lesion in spinal cord. However, some difference between the responses in the optic nerve and spinal cord to neuroinflammation induced by MOG-immunization (Collongues et al., 2012). Also, the difference between the types of edema regulated by AQP4 channel in the two organs suggested that the physiological function of the bloodebrain barrier might differ from that of the spinal cord (Saadoun and Papadopoulos, 2010). Hence, it is needed to originally investigate the mechanism of NMO-lesion in the optic nerve. Only two experimental studies using optic nerve for NMO have been performed. Both ex vivo studies revealed that NMO-IgG could damage astrocytes in the optic nerve. Zang et al. demonstrated that recombinant NMO-IgG with human complement reduced AQP4 immunoreactivity (Zhang et al., 2011). In another study, Marignier et al. showed IgG from NMO-patients without human complement damaged oligodendrocytes in the optic nerve (Marignier et al., 2010). These ex vivo study excellently revealed the pathogenesis of NMO-IgG in the optic nerve tissue, but had some limitations for evaluating the association of glial cells with neural components, time-course of inflation and functional assessment. In optic neuritis, the reduction of visual function usually precedes structural damage. Our in vivo model will provide more information in optic neuritis in NMO from the standpoint of visual function. Furthermore, the advantage of this model was the regional corresponding of nerve fiber in the optic nerve and RGC in the retina. Not likely intra-cerebral method, initial damage in our model could be produced in a place (optic nerve) so far from an organ existing neural cell body (retina). This allows us to minimize the effect of direct damage on the tissue accompanied with the experimental procedures. Furthermore, the qualification of RGC density is well established and has been frequently used. This model using rat optic nerve should be useful for evaluating treatments for NMO. Regarding the optic nerve damage, a significant loss of the immunofluorescence of GFAP and AQP4 could be observed in the group treated with AQP4 seropositive serum. However, these

changes did not reach significant difference in real-time PCR assay. Previous studies for the pathogenesis of NMO-IgG performed similar approach based on immunostaining and did not demonstrate the results of gene expression. The experimental strategy

Fig. 6. GFAP expression as a stress marker in the retina. Rat retinal sections were stained with antibodies against GFAP (green). Nuclei are marked with Hoechst (blue). Retinal astrocytes expressed GFAP in normal status. The retina in AQP4þ group increased GFAP expression in Müller cells. Scale bar ¼ 100 mm.

Y. Matsumoto et al. / Experimental Eye Research 119 (2014) 61e69

using gene analysis might be less sensitive to detect significant changes and other experimental procedures would be needed to verify the findings from multiple aspects. The role of human complement in the combination with AQP4 antibody was demonstrated in some studies using purified NMOIgG to produce NMO-like lesion (Hinson et al., 2008; Saadoun et al., 2010; Zhang et al., 2011). However, AQP4 antibody acts as a specific agonist or alters antigen density in HEK cells per se (Waters et al., 2008). Given the application of patients’ sera without any modification such as filtration, it is highly likely that co-existing human complements may also play a role in the development of the neurodegenerative event in this study (Hinson et al., 2008; Waters et al., 2008; Matsushita et al., 2009). Although the exact concentration of AQP4 antibody in this study was unclear, NMOlike lesion in the optic nerve could be produced by the NMO-patient’s serum. In conclusion, the serum from NMOSDs-patients initiated NMOlike lesion in the rat optic nerve. This model will facilitate the investigations for assessing optic nerve degeneration and seeking for novel treatments in optic neuritis of NMO. Acknowledgments The authors thank Ms. Keiko Hayashibe and Dr. Akiko Miki for experimental assistance. This work was supported by by JSPS KAKENHI grant number 23592568 (M.N.) and 23791983 (A.K.) from the Japanese Government, the Suda Memorial Foundation (A.K.), the Mishima Memorial Foundation (A.K.) and the Santan Pharmaceutical Founder Commemoration Ophthalmic Research Fund (A.K.). References Arkin, M.S., Miller, R.F., 1988. Mudpuppy retinal ganglion cell morphology revealed by an HRP impregnation technique which provides Golgi-like staining. J. Comp. Neurol. 270 (2), 185e208. Bennett, J.L., Lam, C., Kalluri, S.R., et al., 2009. Intrathecal pathogenic antiaquaporin-4 antibodies in early neuromyelitis optica. Ann. Neurol. 66 (5), 617e629. Bradl, M., Misu, T., Takahashi, T., et al., 2009. Neuromyelitis optica: pathogenicity of patient immunoglobulin in vivo. Ann. Neurol. 66 (5), 630e643. Chen, H., Weber, A.J., 2002. Expression of glial fibrillary acidic protein and glutamine synthetase by Muller cells after optic nerve damage and intravitreal application of brain-derived neurotrophic factor. Glia 38 (2), 115e125. Chidlow, G., Casson, R., Sobrado-Calvo, P., et al., 2005. Measurement of retinal injury in the rat after optic nerve transection: an RT-PCR study. Mol. Vis. 11, 387e396. Collongues, N., Chanson, J.B., Blanc, F., et al., 2012. The Brown Norway opticospinal model of demyelination: does it mimic multiple sclerosis or neuromyelitis optica? Int. J. Dev. Neurosci. 30 (6), 487e497. Hinson, S.R., Pittock, S.J., Lucchinetti, C.F., et al., 2007. Pathogenic potential of IgG binding to water channel extracellular domain in neuromyelitis optica. Neurology 69 (24), 2221e2231. Hinson, S.R., Roemer, S.F., Lucchinetti, C.F., et al., 2008. Aquaporin-4-binding autoantibodies in patients with neuromyelitis optica impair glutamate transport by down-regulating EAAT2. J. Exp. Med. 205 (11), 2473e2481. Hu, P., Pollard, J., Hunt, N., et al., 1998. Microvascular and cellular responses in the optic nerve of rats with acute experimental allergic encephalomyelitis (EAE). Brain Pathol. 8 (3), 475e486. Ishii, Y., Kwong, J.M., Caprioli, J., 2003. Retinal ganglion cell protection with geranylgeranylacetone, a heat shock protein inducer, in a rat glaucoma model. Invest. Ophthalmol. Vis. Sci. 44 (5), 1982e1992. Jarius, S., Wildemann, B., 2010. AQP4 antibodies in neuromyelitis optica: diagnostic and pathogenetic relevance. Nat. Rev. Neurol. 6 (7), 383e392.

69

Kanamori, A., Catrinescu, M.M., Kanamori, N., et al., 2010a. Superoxide is an associated signal for apoptosis in axonal injury. Brain 133 (9), 2612e2625. Kanamori, A., Catrinescu, M.M., Mahammed, A., et al., 2010b. Neuroprotection against superoxide anion radical by metallocorroles in cellular and murine models of optic neuropathy. J. Neurochem. 114 (2), 488e498. Kanamori, A., Naka, M., Fukuda, M., et al., 2009. Latanoprost protects rat retinal ganglion cells from apoptosis in vitro and in vivo. Exp. Eye Res. 88 (3), 535e541. Kanamori, A., Nakamura, M., Nakanishi, Y., et al., 2005. Long-term glial reactivity in rat retinas ipsilateral and contralateral to experimental glaucoma. Exp. Eye Res. 81 (1), 48e56. Kinoshita, M., Nakatsuji, Y., Kimura, T., et al., 2009a. Neuromyelitis optica: passive transfer to rats by human immunoglobulin. Biochem. Biophys. Res. Commun. 386 (4), 623e627. Kinoshita, M., Nakatsuji, Y., Moriya, M., et al., 2009b. Astrocytic necrosis is induced by anti-aquaporin-4 antibody-positive serum. Neuroreport 20 (5), 508e512. Kitley, J., Leite, M.I., Nakashima, I., et al., 2012. Prognostic factors and disease course in aquaporin-4 antibody-positive patients with neuromyelitis optica spectrum disorder from the United Kingdom and Japan. Brain 135 (Pt 6), 1834e1849. Kudo, H., Nakazawa, T., Shimura, M., et al., 2006. Neuroprotective effect of latanoprost on rat retinal ganglion cells. Graefes Arch. Clin. Exp. Ophthalmol. 244 (8), 1003e1009. Lennon, V.A., Kryzer, T.J., Pittock, S.J., et al., 2005. IgG marker of optic-spinal multiple sclerosis binds to the aquaporin-4 water channel. J. Exp. Med. 202 (4), 473e477. Lennon, V.A., Wingerchuk, D.M., Kryzer, T.J., et al., 2004. A serum autoantibody marker of neuromyelitis optica: distinction from multiple sclerosis. Lancet 364 (9451), 2106e2112. Marignier, R., Nicolle, A., Watrin, C., et al., 2010. Oligodendrocytes are damaged by neuromyelitis optica immunoglobulin G via astrocyte injury. Brain 133 (9), 2578e2591. Matsushita, T., Isobe, N., Matsuoka, T., et al., 2009. Aquaporin-4 autoimmune syndrome and anti-aquaporin-4 antibody-negative opticospinal multiple sclerosis in Japanese. Multi. Scler. 15 (7), 834e847. Milici, A.J., Carroll, L.A., Stukenbrok, H.A., et al., 1998. Early eosinophil infiltration into the optic nerve of mice with experimental allergic encephalomyelitis. Lab. Invest. 78 (10), 1239e1244. Misu, T., Fujihara, K., Kakita, A., et al., 2007. Loss of aquaporin 4 in lesions of neuromyelitis optica: distinction from multiple sclerosis. Brain 130 (Pt 5), 1224e 1234. Misu, T., Fujihara, K., Nakamura, M., et al., 2006. Loss of aquaporin-4 in active perivascular lesions in neuromyelitis optica: a case report. Tohoku J. Exp. Med. 209 (3), 269e275. Naka, M., Kanamori, A., Negi, A., et al., 2010. Reduced expression of aquaporin-9 in rat optic nerve head and retina following elevated intraocular pressure. Invest. Ophthalmol. Vis. Sci. 51 (9), 4618e4626. Nielsen, S., Nagelhus, E.A., Amiry-Moghaddam, M., et al., 1997. Specialized membrane domains for water transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. J. Neurosci. 17 (1), 171e180. Ratelade, J., Bennett, J.L., Verkman, A.S., 2011. Intravenous neuromyelitis optica autoantibody in mice targets aquaporin-4 in peripheral organs and area postrema. PLoS One 6 (11), e27412. Saadoun, S., Papadopoulos, M.C., 2010. Aquaporin-4 in brain and spinal cord oedema. Neuroscience 168 (4), 1036e1046. Saadoun, S., Waters, P., Bell, B.A., et al., 2010. Intra-cerebral injection of neuromyelitis optica immunoglobulin G and human complement produces neuromyelitis optica lesions in mice. Brain 133 (Pt 2), 349e361. Sabater, L., Giralt, A., Boronat, A., et al., 2009. Cytotoxic effect of neuromyelitis optica antibody (NMO-IgG) to astrocytes: an in vitro study. J. Neuroimmunol. 215 (1e 2), 31e35. Takahashi, T., Fujihara, K., Nakashima, I., et al., 2007. Anti-aquaporin-4 antibody is involved in the pathogenesis of NMO: a study on antibody titre. Brain 130 (Pt 5), 1235e1243. van Adel, B.A., Arnold, J.M., Phipps, J., et al., 2005. Ciliary neurotrophic factor protects retinal ganglion cells from axotomy-induced apoptosis via modulation of retinal glia in vivo. J. Neurobiol. 63 (3), 215e234. Waters, P., Jarius, S., Littleton, E., et al., 2008. Aquaporin-4 antibodies in neuromyelitis optica and longitudinally extensive transverse myelitis. Arch. Neurol. 65 (7), 913e919. Zhang, H., Bennett, J.L., Verkman, A.S., 2011. Ex vivo spinal cord slice model of neuromyelitis optica reveals novel immunopathogenic mechanisms. Ann. Neurol. 70 (6), 943e954.