Loss of the Extracellular Matrix Molecule Tenascin-C Leads to Absence of Reactive Gliosis and Promotes Anti-inflammatory Cytokine Expression in an Autoimmune Glaucoma Mouse Model

Previous studies demonstrated that retinal damage correlates with a massive remodeling of extracellular matrix (ECM) molecules and reactive gliosis. However, the functional significance of the ECM in retinal neurodegeneration is still unknown. In the present study, we used an intraocular pressure (IOP) independent experimental autoimmune glaucoma (EAG) mouse model to examine the role of the ECM glycoprotein tenascin-C (Tnc). Wild type (WT ONA) and Tnc knockout (KO ONA) mice were immunized with an optic nerve antigen (ONA) homogenate and control groups (CO) obtained sodium chloride (WT CO, KO CO). IOP was measured weekly and electroretinographies were recorded at the end of the study. Ten weeks after immunization, we analyzed retinal ganglion cells (RGCs), glial cells, and the expression of different cytokines in retina and optic nerve tissue in all four groups. IOP and retinal function were comparable in all groups. Although RGC loss was less severe in KO ONA, WT as well as KO mice displayed a significant cell loss after immunization. Compared to KO ONA, less βIII-tubulin+ axons, and downregulated oligodendrocyte markers were noted in WT ONA optic nerves. In retina and optic nerve, we found an enhanced GFAP+ staining area of astrocytes in immunized WT. A significantly higher number of retinal Iba1+ microglia was found in WT ONA, while a lower number of Iba1+ cells was observed in KO ONA. Furthermore, an increased expression of the glial markers Gfap, Iba1, Nos2, and Cd68 was detected in retinal and optic nerve tissue of WT ONA, whereas comparable levels were observed in KO ONA. In addition, pro-inflammatory Tnfa expression was upregulated in WT ONA, but downregulated in KO ONA. Vice versa, a significantly increased anti-inflammatory Tgfb1 expression was measured in KO ONA animals. We conclude that Tnc plays an important role in glial and inflammatory response during retinal neurodegeneration. Our results provide evidence that Tnc is involved in glaucomatous damage by regulating retinal glial activation and cytokine release. Thus, this transgenic EAG mouse model for the first time offers the possibility to investigate IOP-independent glaucomatous damage in direct relation to ECM remodeling.

Previous studies demonstrated that retinal damage correlates with a massive remodeling of extracellular matrix (ECM) molecules and reactive gliosis. However, the functional significance of the ECM in retinal neurodegeneration is still unknown. In the present study, we used an intraocular pressure (IOP) independent experimental autoimmune glaucoma (EAG) mouse model to examine the role of the ECM glycoprotein tenascin-C (Tnc). Wild type (WT ONA) and Tnc knockout (KO ONA) mice were immunized with an optic nerve antigen (ONA) homogenate and control groups (CO) obtained sodium chloride (WT CO, KO CO). IOP was measured weekly and electroretinographies were recorded at the end of the study. Ten weeks after immunization, we analyzed retinal ganglion cells (RGCs), glial cells, and the expression of different cytokines in retina and optic nerve tissue in all four groups. IOP and retinal function were comparable in all groups. Although RGC loss was less severe in KO ONA, WT as well as KO mice displayed a significant cell loss after immunization. Compared to KO ONA, less βIII-tubulin + axons, and downregulated oligodendrocyte markers were noted in WT ONA optic nerves. In retina and optic nerve, we found an enhanced GFAP + staining area of astrocytes in immunized WT. A significantly higher number of retinal Iba1 + microglia was found in WT ONA, while a lower number of Iba1 + cells was observed in KO ONA. Furthermore, an increased expression of the glial markers Gfap, Iba1, Nos2, and Cd68 was detected in retinal and optic nerve tissue of WT ONA, whereas comparable levels were observed in KO ONA. In addition, pro-inflammatory Tnfa expression was upregulated in WT ONA, but downregulated in KO ONA. Vice versa, a significantly increased anti-inflammatory Tgfb1 expression was measured in KO ONA animals. We conclude that Tnc plays an INTRODUCTION Glaucomatous neurodegeneration is characterized by a progressive loss of retinal ganglion cells (RGCs) and their axons, which form the optic nerve. The molecular mechanisms of RGC degeneration are still not fully understood. In addition to increased intraocular pressure (IOP), immunological processes, glial activation, and remodeling of extracellular matrix (ECM) constituents are associated with glaucoma. In regard to the immune system, studies also indicate an alteration in serum antibodies against various retinal proteins in glaucoma patients with a normal IOP (1)(2)(3). The connection between an immune response and glaucoma disease with the characteristic loss of RGCs was already demonstrated in an experimental autoimmune glaucoma (EAG) rat model. Here, glaucomatous damage was induced by immunization with ocular proteins (4,5). Furthermore, a pathological upregulation of specific ECM components could be demonstrated in this model (6). However, the relationship between a change in ECM components and glaucoma pathogenesis needs to be investigated.
The ECM consists of several molecules, including proteoglycans and glycoproteins and controls cellular key events such as adhesion, differentiation, migration, proliferation as well as survival (7)(8)(9)(10)(11)(12). During development of the central nervous system (CNS), tenascin-C (Tnc) is strongly expressed by radial glial cells, immature astrocytes, and oligodendrocyte precursor cells (OPCs). It regulates axonal as well as neurite outgrowth and glial cells differentiation (11,(13)(14)(15)(16)(17). In this regard, axonal growth is modulated by specific Tnc isoforms, which contain a different combination of fibronectin type III domains (18). Moreover, Tnc expression is associated with the stem cell niche and regulates stem cell maintenance in addition to differentiation (19).
In the diseased brain, Tnc is highly re-expressed (11). For example, Tnc immunoreactivity is directly linked to amyloidβ plaques in Alzheimer's disease patients (20). A lack of Tnc leads to reduced deposits of amyloid-β plaques and protects from Alzheimer's disease (21). Furthermore, an immunomodulatory impact of Tnc was described by regulating the Th17 cell differentiation and activation (22,23). Here, Tnc deficiency ameliorates experimental autoimmune encephalomyelitis in mice (24).
In the context of retinal neurodegenerative processes, Tnc also plays an important role (25). During retinal development, Tnc is found in the inner neuroblastic layer at embryonic day 13 (26). In the adult retina, Tnc is highly enriched in the outer and inner plexiform layers and is prominently expressed by amacrine and horizontal cells (27,28). Additionally, optic nerve astrocytes synthesize Tnc (29,30). ECM molecules can provide an inhibitory environment for neural regeneration and migration in the retina (30). A dramatic remodeling of ECM constituents was already described after ischemia and glaucomatous damage (25,31). An upregulation of Tnc has been noted in a glaucoma animal model (32) as well as in open-angle glaucoma patients (33).
Tnc is a key regulator of the immune system and plays an important role during neuroinflammation and glial response (34)(35)(36). Moreover, expression of Tnc by astrocytes is regulated by cytokines secreted by microglia (37,38). Microglia play an important role during neurodegenerative and neuroinflammatory processes (39). Their activation is characterized by an enhanced proliferation, migration, phagocytosis, and increased expression levels of neuroinflammatory molecules (40,41). The neurotoxic M1-subtype has an amoeboid morphology and releases pro-inflammatory signaling molecules, like tumor necrosis factor-alpha (TNF-α) and inducible nitric oxide synthase (iNOS) (42)(43)(44). In contrast, the M2-phenotype is characterized by a morphology with ramified processes and the expression of anti-inflammatory cytokines, such as the transforming growth factor-beta (TGF-β) (45)(46)(47).
In this study, we used a Tnc deficient EAG mouse model to further analyze the importance of Tnc during retinal neurodegeneration and neuroinflammatory outcomes. We immunized wild type (WT) and Tnc knockout (KO) mice with an optic nerve antigen (ONA) homogenate and examined retinal and optic nerve damage as well as macro-and microglial activity. Furthermore, we determined the expression pattern of pro-and anti-inflammatory cytokines. The present study was undertaken to address the role of Tnc in glaucomatous damage, retinal glial activation, myelination, and inflammatory cytokine release.

Animals
Animals were housed under a 12 h light-dark cycle and had free access to chow and water. All procedures were approved by the animal care committee of North Rhine-Westphalia, Germany and performed according to the ARVO statement for the use of animals in ophthalmic and vision research. For the experiments, male and female littermates of 129/Sv WT and Tnc KO mice (48) were used at 6 weeks of age.
Immunization WT (WT ONA) and KO (KO ONA) mice were immunized intraperitoneally with ONA (1 mg/ml) mixed with incomplete Freund's adjuvants (FA; 50 µl) and 1 µg pertussis toxin (PTx; both Sigma Aldrich, St. Louis, MO, USA) as described (49). To generate the ONA homogenate, fresh bovine eyes were obtained from a local slaughterhouse (Schlachthaus Wuppertal, Germany). As previously described, optic nerves were cut off behind the optic nerve head, cleaned from surrounding tissue and the dura mater was removed. Nerves were pulverized in a cooled mortar and then suspended in phosphate-buffered saline (PBS) (5). A final concentration of 1 mg/ml was set. FA acted as an immunostimulatory and PTx was given to ensure the permeability of the blood retina barrier. Intraperitoneal PTxapplication was repeated 2 days after immunization. Booster injections containing half of the initial dose were given intraperitoneally 4 and 8 weeks after initial immunization. The control groups (WT CO; KO CO) were injected with 1 ml sodium chloride (B. Braun Melsungen AG, Melsungen, Germany), FA and PTx. Ten weeks after immunization, retinae, and optic nerves were explanted for immunohistochemistry, quantitative real time PCR (RT-qPCR), and Western blot analyses. For RT-qPCR and Western blot, retinal as well as optic nerve tissue of both eyes from one animal were pooled.

Intraocular Pressure Measurements
Before IOP measurement, mice were anesthetized with a ketamine/xylazine mixture (120/16 mg/kg). Both eyes were analyzed and 10 readings of each eye were averaged. IOP measurements were performed before immunization in WT and KO mice at 5 weeks of age with a rebound tonometer (TonoLab; Icare; Oy; Finland; n = 16/group) as previously described (50,51). After immunization, IOP was measured weekly in all four groups until the end of the study (n = 8/group).

Immunohistochemistry and Confocal Laser Scanning Microscopy
Eyes and optic nerves were dissected and fixed in paraformaldehyde (PFA) for 1 day, dehydrated in sucrose (30%) and embedded in Tissue-Tek freezing medium (Thermo Fisher Scientific, Cheshire, UK). Retinal cross-sections and optic nerve longitudinal sections (16 µm) were cut with a cryostat (CM3050 S, Leica) and transferred onto Superfrost plus object slides (Menzel-Glaeser, Braunschweig, Germany). First, slices were blocked with 1% bovine serum albumin (BSA; Sigma-Aldrich), 3% goat serum (Dianova, Hamburg, Germany), and 0.5 % Triton TM -X-100 (Sigma-Aldrich) in PBS for 1 hour (h) at room temperature (RT). Afterwards, the primary antibodies were diluted in blocking solution and incubated at RT overnight ( Table 1). Sections were washed 3 times in PBS and incubated for 2 h with adequate secondary antibody (Dianova, Hamburg, Germany; Table 1) solution without Triton TM -X-100. Cell nuclei were detected with TO-PRO-3 (1:400; Thermo Fisher Scientific). The retinal and optic nerve slices were analyzed with a confocal laser-scanning microscope (LSM 510 META; Zeiss, Göttingen, Germany). Two sections per slide, 4 images per retina (400x magnification), and 3 images per optic nerve (200x magnification) were captured (n = 4-5/group). In addition, a 630x magnification was used for colocalization staining in optic nerve sections with antibodies against CC1 (coiled-coil 1) and Olig2 (oligodendrocyte transcription factor 2). Accordingly, 4 images were taken per slide (n = 5/group). Laser lines and emission filters were adjusted using the Zeiss ZEN black software. Cropping of the images was done using Coral Paint Shop Pro X8 (Coral Corporation, Ottawa, Canada). Masked evaluation of the staining signal was performed with ImageJ software (ImageJ 1.51w, National Institutes of Health; Bethesda, MD, USA) as previously described (6,53). Images were converted into gray scales and the background was subtracted. Then, the lower and upper threshold values was determined for each image ( Table 2). The percentage of the area fraction was measured using an ImageJ macro as previously described (53). This analysis was performed for immunohistochemical stainings against βIII-tubulin, glial fibrillary acidic protein (GFAP), myelin basic protein (MBP), and Tnc. Cell countings were done for immunopositive Brn3a + cells in retinal cross-sections and for Olig2 + /CC1 + cells in optic nerve slices. Values were transferred to Statistica software and the WT CO group was set to 100% (V13.3; StatSoft (Europe), Hamburg, Germany).

Quantification of RGCs and Microglia in Retinal Flat-Mounts
Eyes were enucleated and fixed in 4% PFA at 4 • C for 1 h. The retinae were dissected from the eye and prepared as flat-mounts (n = 9/group). The tissue was fixed again in 4% PFA for 5 min and washed 3 times in PBS. Flat-mounts were blocked in 1% BSA, 3% donkey serum, and 2% Triton TM -X-100 in PBS at RT for 1 h. Next, incubation was performed with the RGC specific marker Brn3a [brain-specific homeobox/POU domain protein 3a; (54,55)] and microglia marker Iba1 [ionized calcium-binding adapter molecule 1; (56)] for 2 days at 4 • C. Following PBS washing (3 × 20 min), flat-mounts were incubated with secondary antibodies donkey anti-goat Cy3, donkey anti-rabbit Alexa Fluor 488, and TO-PRO-3 (1:400) in blocking solution without Triton TM -X-100 at RT for 2 h. Microscopic images were captured using Axio Zoom.V16 (Zeiss, Göttingen, Germany). Flat-mounts were divided into 16 quadrants (200 × 200 µm) and Brn3a + and Iba1 + cells were counted. Groups were compared using two-way ANOVA followed by Tukey's post-hoc test. The WT CO group was set to 100%.

Western Blotting
Retinal tissue (n = 5/group) was homogenized in 150 µl and optic nerve tissue (n = 5/group) in 100 µl lysis buffer (60 mM n-octyl-β-D-glucopyranoside, 50 mM sodium acetate, 50 mM Tris chloride, pH 8.0 and 2 M urea) containing a protease inhibitor cocktail (Sigma-Aldrich) on ice for 1 h. Prior to lysis, the optic nerve tissue was incubated in liquid nitrogen. Subsequently, all samples were centrifuged at 14.000 × g at 4 • C for 30 min and the supernatant was applied to determine the protein concentration. A BCA Protein Assay kit (Pierce, Thermo Fisher Scientific, Rockford, IL, USA) was used for retinal tissue. For optic nerves, the Qubit R Protein Assay kit (Life Technologies GmbH, Darmstadt, Germany) was used according to manufacturer's instructions. 4x SDS buffer was added to each protein sample (20 µg) and denaturized at 94 • C for 5 min. After separation via SDS-PAGE (10% gels, respectively, 4-12% polyacrylamide gradient gels), proteins were transferred to a polyvinylidene difluoride (PVDF) membrane (Roth, Karlsruhe, Germany) by Western blotting (1-2 h and 75 mA). Membranes were blocked (5% w/v milk powder in TRISbuffered saline (TBS) with 0.05% Tween R 20; TBST) at RT for 1 h and incubated with the primary antibody ( Table 3) in blocking solution at 4 • C overnight. On the next day, membranes were washed with TBST and incubated with horseradish peroxidase (HRP) coupled secondary antibodies (  Germany) was applied to develop the membrane (mixed 1:1 for 5 min). Finally, protein immunoreactivity was detected with a MicroChemi Chemiluminescence Reader (Biostep, Burkhardtsdorf, Germany). Band intensity was analyzed using ImageJ software and normalized to a corresponding reference protein (actin/vinculin). The normalized values of the Western blot results were given in arbitrary units (a.u.).

RNA Isolation, cDNA Synthesis, and RT-qPCR
Retinal tissue was homogenized by trituration using a pipette (n = 5/group

Statistical Analysis
Immunohistological, Western blot, IOP, and ERG data of control WT (WT CO) and KO (KO CO) as well as ONA-immunized WT (WT ONA) and KO (KO ONA) were analyzed by twoway ANOVA followed by Tukey's post-hoc test using Statistica software (V13.3; StatSoft (Europe), Hamburg, Germany). Results of IOP measurements were presented as mean ± standard error mean (SEM) ± standard deviation (SD). ERG recordings, immunohistochemical, and Western blot data were shown as mean ± SEM. Analyses of Tnc protein levels in WT mice via immunohistochemistry and Western blot were analyzed via Student t-test and presented as mean ± SEM. For RT-qPCR results, groups were compared using the pairwise fixed reallocation and randomization test (REST © software) and were presented as median ± quartile ± minimum/maximum (57).

No Changes in IOP and Retinal Functionality in the EAG Mouse Model
IOP measurements were performed before immunization in 5-week-old WT (WT CO) and KO (KO CO; Figure 1A). After immunization, IOP was measured weekly in control and immunized WT (WT ONA) and KO (KO ONA) animals until the end of the study. At 5 weeks of age (−1), we observed no significant differences in the IOP of WT CO (9.8 ± 0.2 mmHg) and KO CO (9.7 ± 0.1 mmHg; p = 1.0). Furthermore, no changes in the IOP were found in control and immunized groups throughout the study (Supplementary Table 1).
To determine possible retinal function deficits, induced by ONA-immunization, we performed ERG recordings of control and immunized WT and KO mice. Under scotopic conditions, a-wave responses arise from rod-photoreceptors, while b-waves represent the rod bipolar and Müllerglia cell response. In all four conditions no significant differences were observed between control and immunized WT and KO animals (Figures 1B,C; Supplementary Table 2). Therefore, we concluded that photoreceptor and bipolar cell function was not affected in this EAG mouse model.

Significant RGC Loss Following Immunization
Previous studies of an EAG rat model showed a significant reduction of RGC numbers 4 weeks post immunization with ONA (4,58). Additionally, an upregulation of Tnc was found before significant RGC loss occurred (6). In our EAG mouse model, we detected unaltered Tnc protein levels 10 weeks after immunization (Supplementary Table 3, Supplementary Figure 1). Based on these findings, immunohistochemical stainings of RGCs were performed with an antibody against Brn3a, which specifically detects RGCs (Figure 2, Supplementary Figures 2A,B). Interestingly, no significant differences were detected between control and immunized KO in horizontal cross-sections (KO CO: 92.2 ± 3.9% Brn3a + cells vs. KO ONA: 83.7 ± 8.7% Brn3a + cells, p = 0.57, Figures 2A,B).
To further characterize the RGC population, we counted Brn3a + cells in retinal flat-mounts ( Figure 2C). We determined the total amount ( Figure 2D) as well as the number of Brn3a + cells within the central ( Figure 2E) and peripheral ( Figure 2F) area of the retina. A significant reduction in the total number was observed in immunized WT compared to both control genotypes (WT ONA: 80.3 ± 1.5% Brn3a + cells vs. WT CO: 100.0 ± 2.0% Brn3a + cells, p < 0.001, KO CO: 98.2 ± 3.3% Brn3a + cells, p < 0.001). Also, a significant loss of RGCs was detected in KO ONA mice (86.9 ± 3.1% Brn3a + cells) compared to KO CO (p = 0.02) and WT CO (p = 0.01). A comparable percentage of Brn3a + cells was also observed in the central retina. So, immunized WT and KO animals showed a significant decline of RGCs compared to the corresponding control groups (WT ONA: 82.7 ± 1.7% Brn3a + cells vs. WT CO: 100.0 ± 2.5% Brn3a + cells, p < 0.001 and KO ONA: 86.3 ± 3.7% Brn3a + cells vs. KO CO: 97.8 ± 2.9% Brn3a + cells, p = 0.03). No significant differences were found between both immunized genotypes (p = 0.80). Furthermore, a decrease in the RGC density was verified in the peripheral area. Retinae of the WT ONA group (77.0 ± 1.8% Brn3a + cells) displayed a loss of about 25% RGCs compared to WT CO (100.0 ± 1.7% Brn3a + cells, p < 0.001). A significant reduction was also found in the comparison of KO CO and KO ONA (KO CO: 99.0 ± 4.1% Brn3a + cells vs. KO ONA: 87.1 ± 2.8% Brn3a + cells, p = 0.02). However, KO ONA showed a decrease of about 15 % in the peripheral part compared to WT CO group (p = 0.01).
Although not statistically significant, we found a trend to a weaker RGC damage in immunized KO compared to WT ONA.

Extenuated Macroglial Reactivity After Immunization in KO Mice
Our results showed a decrease in the RGC number 10 weeks after immunization in WT and KO animals. Next, we investigated, if this glaucomatous neurodegeneration is associated with an altered macroglial response. Therefore, we analyzed the immunoreactivity of GFAP + astrocytes in retinal cross-sections. GFAP stained astrocytes were mainly localized in the ganglion cell layer (GCL; Figure 4A). The GFAP + area was increased in WT ONA (167.22 ± 18.61% GFAP + area) compared to the control groups (WT CO: 100.00 ± 9.28% GFAP + area, p = 0.01 and KO CO: 105.81 ± 4.54% GFAP + area, p = 0.02, Figure 4B). Interestingly, no changes in the GFAP signal area were found in KO ONA (142.49 ± 8.19% GFAP + area) compared to KO CO (p = 0.18) as well as to WT CO (p = 0.11). The statistical comparison of both immunized genotypes showed no significant differences (p = 0.46).
Then, we evaluated GFAP protein levels via Western blot. For GFAP, a prominent band was detected at 50 kDa ( Figure 4C).
We also analyzed the mRNA expression of Gfap in retinae via RT-qPCR (Figures 4E-G, Supplementary Table 5). Analysis revealed comparable levels of Gfap in KO CO and WT CO (1.4fold, p = 0.11, Figure 4E). A significant increase of Gfap mRNA expression levels was observed in WT ONA (WT CO vs. WT ONA: 1.7-fold, p = 0.04, Figure 4F), whereas no differences could be detected in KO ONA (KO CO vs. KO ONA: 1.2-fold, p = 0.40, Figure 4F). The expression was comparable in both immunized genotypes (1.0-fold, p = 0.99, Figure 4G).
For GFAP, a thread-like staining pattern was observed in optic nerve slices (Figure 5A). The evaluation of the GFAP immunoreactivity in optic nerve sections also showed no  Figure 5B). Moreover, a nearly doubled GFAP intensity was observed in WT ONA (196.70 ± 13.60% GFAP + area) compared to the corresponding control group (p = 0.04).
Finally, the RT-qPCR results of the optic nerve tissue showed no changes in Gfap expression between both control groups (WT CO vs. KO CO: 1.1-fold, p = 0.54, Figure 5E, Supplementary Table 5). In line with the immunohistochemical results, we found a slightly enhanced mRNA level in WT ONA (WT CO vs. WT ONA: 1.4-fold, p = 0.07, Figure 5F), whereas the KO ONA animals exhibited a reduction of Gfap expression (KO CO vs. KO ONA: 0.5-fold, p < 0.05). Interestingly, in a direct comparison of the two immunized groups, Gfap expression was significantly reduced in KO animals (WT ONA vs. KO ONA: 0.4-fold, p = 0.02, Figure 5G).
In summary, we concluded that Tnc deficiency resulted in a diminished macroglial reaction during retinal and optic nerve degeneration in the EAG mouse model.

Decreased Demyelination After ONA-Immunization in KO Mice
Our study demonstrates RGC degeneration in WT and KO animals after immunization. Furthermore, we noted that Tnc deficiency resulted in a diminished macroglial response. Finally, we analyzed the impact of ONAimmunization on oligodendroglia in optic nerve tissue. The oligodendrocytes appear in two different populations, as immature OPCs and as myelinating, mature oligodendrocytes. To analyze both oligodendrocyte populations separately, an immunohistochemical colocalization staining was performed using the markers Olig2 and CC1. Olig2 is expressed by oligodendrocytes of all stages (59). In contrast, CC1 is only expressed by mature oligodendrocytes (60). Colocalization identified double positive cells as mature and single Olig2 + cells as immature oligodendrocytes. Immunohistochemical stainings revealed fewer Olig2 + cells in the WT ONA group compared to the other groups. Interestingly, there were more Olig2 + cells in KO ONA than in WT ONA ( Figure 6A). 76.3 ± 1.6% Olig2 + cells were found in WT ONA, which indicates a significant oligodendrocyte loss over 25% compared to control WT (100.0 ± 3.5% Olig2 + cells, p < 0.001, Figure 6B). The number of Olig2 + cells was also significantly decreased in WT ONA compared to KO CO (p = 0.04). No differences were observed between both Tnc deficient groups (KO CO: 90.2 ± 2.4% Olig2 + cells vs. KO ONA: 95.2 ± 4.9% Olig2 + cells, p = 0.71). Most interestingly, we verified significant differences between both immunized groups (p < 0.01). The number of double-positive (Olig2 + /CC1 + ) mature oligodendrocytes was clearly reduced in WT ONA compared to all other groups ( Figure 6A). The statistical evaluation demonstrated only 64.7 ± 5.8% Olig2 + /CC1 + cells in WT ONA, whereas KO ONA exhibited 107.8 ± 8.9% Olig2 + /CC1 + cells in optic nerve slices (p = 0.002, Figure 6C). Also, immunized WT showed a significant loss of mature oligodendrocytes compared to WT CO (100.0 ± 8.3% Olig2 + /CC1 + cells, Data were analyzed using two-way ANOVA followed by Tukey's post-hoc test and present as mean ± SEM in (B,D). For RT-qPCR results, groups were compared using the pairwise fixed reallocation and randomization test and were shown as median ± quartile ± minimum/maximum in (E-G). *p < 0.05. n = 5/group. Scale bar = 20 µm. ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. p = 0.01) and KO CO (118.4 ± 3.5% Olig2 + /CC1 + cells, p < 0.001).
To consolidate the immunohistochemistry results, we analyzed the Olig2 protein level in optic nerves by Western blot analyses ( Figure 6D). For Olig2, we observed two bands at 32 and 50 kDa. A slight but not significantly decrease of the band intensity was found in WT ONA (0.36 ± 0.10 a.u.) compared to the corresponding control group (0.86 ± 0.41 a.u., p = 0.41, Figure 6E). Equal Olig2 protein levels were observed in KO CO (0.62 ± 0.10 a.u.) and KO ONA (0.67 ± 0.12 a.u., p = 0.99). Missing of Tnc resulted in an equal Olig2 protein level in both immunized groups (p = 0.76).
In conclusion, we found a significant decrease in mature as well as immature oligodendroglia in WT after immunization. Remarkably, immunized Tnc deficient mice showed no demyelination.

Decreased Number of Microglia and Declined Microglial Response in KO ONA Mice
Neurodegeneration is often accompanied by reactive microgliosis. In order to analyze the microglia population in the EAG mouse model and the effects of immunization, we performed immunohistochemical staining of retinal flat-mounts using an Iba1 antibody (Figure 8A, Supplementary Table 4). The number of Iba1 + cells in the total as well as in the central and peripheral area of the retina was evaluated (Figures 8B-D). A significant increase in microglia numbers was detected in WT ONA (123.0 ± 2.4% Iba1 + cells) compared to control WT and KO in the total retina (WT CO: 100.0 ± 2.9% Iba1 + cells, p < 0.001 and vs. KO CO: 102.3 ± 5.7% Iba1 + cells; p = 0.002, Figure 8B). No differences were found between both control groups (p = 0.97). Remarkably, a significantly lower number of Iba1 + cells was observed after immunization in KO ONA (KO ONA: 84.5 ± 2.7% Iba1 + cells) compared to WT ONA (p < 0.001), WT CO (p = 0.03), and KO CO (p < 0.01). Also, in the central part of the retina 20% more Iba1 + cells were detected in WT ONA (122.5 ± 2.9% Iba1 + cells) compared to the corresponding control group (WT CO: 100.0 ± 3.5% Iba1 + cells, p < 0.01, Figure 8C). Immunized WT  also showed significantly more microglial cells compared to control (p < 0.001) and immunized KO mice (p < 0.001). In KO CO, we counted 98.1 ± 5.8% Iba1 + cells, whereas KO ONA only has 84.0 ± 2.9% Iba1 + cells (p = 0.08). A reduced microglia number was noted in KO ONA compared to WT CO (p = 0.04). Equal numbers of microglial cells were seen in both control groups (p = 0.99). Similarly, the number of microglia in WT ONA (123.6 ± 3.4% Iba1 + cells) was significantly enhanced compared to WT CO (100.0 ± 3.2% Iba1 + cells, p = 0.002, Figure 8D) and KO CO (107.2 ± 6.4% Iba1 + cells, p < 0.05) in peripheral regions of the retinal. Also, the number of Iba1 + cells was lower in KO ONA (85.1 ± 3.1% Iba1 + cells) compared to KO CO (p < 0.01) and WT CO (p = 0.08) in the periphery. Additionally, a significantly reduced microglial response was detected in immunized KO and WT animals (p < 0.001). Regarding the quantification of Iba1 + cells in the periphery, both control groups had similar cell counts (p = 0.63).
In the next step, RT-qPCR was used to investigate whether microglia have reactive phenotypes. Besides Iba1, we also examined the markers Nos2 and Cd68 in retinal tissue (Figures 8E-G, Supplementary Table 5). No differences could be detected in the Iba1 expression between control WT and KO mice (1.3-fold, p = 0.2, Figure 8E). However, KO CO mice showed significantly elevated levels of Nos2 (1.7-fold; p = 0.013) and Cd68 (1.3-fold; p = 0.017) compared to WT CO. The comparison of immunized and non-immunized WT illustrated a significantly increased expression of Iba1 (1.5fold, p = 0.048) as well as of the reactive markers Nos2 (1.4fold, p = 0.021) and Cd68 (1.3-fold, p = 0.032, Figure 8F). Interestingly, comparable expression levels of these microglial markers were found in immunized KO and KO CO (p Quantification of Iba1 + microglia in control and immunized WT and KO animals in the total, central and peripheral retina (n = 9/group). WT ONA group exhibited clearly more microglia. In contrast, KO ONA animals displayed fewer Iba1 + cells. (E) RT-qPCR analyses (n = 5/group) of the relative Iba1, Nos2, and Cd68 mRNA expression showed a significant increase of Nos2 and Cd68 in KO CO compared to WT CO retinae. No differences were observed for Iba1 mRNA expression. (F) Compared to WT CO, a significant upregulation of Iba1, Nos2, and Cd68 levels were found in WT ONA. No significant changes were detected regarding the expression levels of these markers in KO ONA compared to KO CO. (G) After immunization, a comparable mRNA level of Iba1, Cd68, and Nos2 was detected in KO ONA compared to WT ONA. Data were analyzed using two-way ANOVA followed by Tukey's post-hoc test and present as mean ± SEM in (B-D). For RT-qPCR, groups were compared using the pairwise fixed reallocation and randomization test and were shown as median ± quartile ± minimum/maximum in (E-G). *p < 0.05; **p < 0.01; ***p < 0.001. Scale bar = 50 µm.
In line with the RT-qPCR results of retinal tissue, we found a similar expression pattern of microglial markers in the optic nerve of control and immunized WT and KO mice (Supplementary Table 5, Supplementary Figure 2).
In summary, WT ONA animals showed a significantly increased microglia infiltration and glial marker expression, indicating an increased microglial response. Remarkably, a significantly reduced invasion and reactivity of microglia were observed in KO ONA, suggesting that Tnc signaling is an important modulator of microglia in glaucomatous neurodegeneration.

Altered Expression Pattern of Pro-and Anti-inflammatory Cytokines in Immunized WT Compared to Immunized KO Animals
In our study, we noted a reactive gliosis and an increased microglial response after immunization in WT mice. Interestingly, these effects could not be detected in Tnc deficient animals.
Next, we analyzed the pro-and anti-inflammatory responses of the microglial phenotypes in retinae and optic nerves (Figure 9, Supplementary Table 5). Here, Tnfa was used to study M1 pro-inflammatory microglia, while Tgfb1 is expressed by M2 anti-inflammatory microglia. RT-qPCR experiments revealed comparable mRNA level of Tnfa (1.4-fold, p = 0.132) and Tgfb1 (1.4-fold, p = 0.071) in KO CO mice compared to WT CO mice ( Figure 9A). After ONA-immunization, Tnfa was significantly upregulated in WT and interestingly downregulated in KO compared to the corresponding control groups (WT CO vs. WT ONA: 1.7-fold, p = 0.026 and KO CO vs. KO ONA: 0.4-fold, p = 0.031, Figure 9B). Statistical comparable Tgfb1 mRNA levels were found in WT CO and WT ONA (1.0-fold; p = 0.807) as well as in KO CO and KO ONA (0.9-fold, p = 0.415; Figure 9B). The evaluation of both immunized genotypes showed a significant reduction of Tnfa (0.5-fold, p = 0.036) and a significant increase of Tgfb1 (1.2-fold, p = 0.005) after immunization in KO mice ( Figure 9C).
Finally, we examined, which microglial subtypes are altered due to an increased microglial reactivity in the optic nerve tissue of immunized and non-immunized WT and KO mice (Figures 9D-F, Supplementary Table 5). Equal mRNA levels of Tnfa (1.4-fold, p = 0.07) and Tgfb1 (0.9-fold, p = 0.659) were seen in WT and KO controls ( Figure 9D). A significantly enhanced Tnfa expression (2.1-fold, p = 0.008) and an unchanged Tgfb1 expression (0.9-fold, p = 0.71) were detected in WT ONA compared to WT CO ( Figure 9E). No changes of these markers were found in KO CO in comparison to KO ONA mice (Tnfa: 0.8-fold, p = 0.443 and Tgfb1: 0.9-fold, p = 0.575). In line with the RT-qPCR results of retinal tissue, we found a slightly reduced Tnfa (0.6-fold, p = 0.101) and an unaltered Tgfb1 (0.8-fold, p = 0.297) expression between both immunized groups ( Figure 9F).
In conclusion, missing Tnc resulted in a reduced mRNA level of pro-inflammatory Tnfa, but an enhanced expression of anti-inflammatory Tgfb1. The increased expression of the proinflammatory cytokine in WT after immunization points toward an enhanced presence of reactive M1 microglia.

DISCUSSION
Glaucoma involves a progressive degeneration of RGCs and their axons leading to visual field loss (62)(63)(64). Developing glaucoma is often associated with elevated IOP, but RGC damage can also occur without IOP changes. Previous studies provided evidence that an altered immune response is involved in glaucoma pathology (2,(65)(66)(67)(68). In addition, a remodeling of ECM constituents was found in several retinal neurodegenerative diseases, including glaucoma (31-33, 69, 70).
In glaucoma pathology the mechanisms are currently poorly understood, especially the relationship between RGC loss and the role of the immune system as well as ECM molecules. Hence, we characterized glaucomatous damage associated with the absence of the ECM glycoprotein Tnc in an IOP-independent EAG mouse model for the first time. Therefore, we immunized WT and KO mice with ONA to induce retinal damage and analyzed IOP, retinal functionality, RGC degeneration, glial activation, and proand anti-inflammatory cytokine expression.
In a previous study, we already successfully transferred the EAG glaucoma model from rats to mice (49). Here, glaucomatous neurodegeneration was observed 6 weeks post ONA-immunization. However, in order to provoke a possible more robust glaucomatous macro-and microgliosis in our transgenic Tnc EAG mouse model, the duration of our study was extended to 10 weeks post immunization. Accordingly, we observed significantly increased GFAP levels and a higher number of Iba1 + infiltrating microglia cells in the WT, suggesting an induction of glaucomatous gliosis.
Our analyses revealed that the IOP of WT and KO stayed in normal ranges. Previous studies of the EAG animal model also showed no alteration in the IOP (49,58). Comparable IOP in control and immunized animals points to the fact that the EAG model can be considered a suitable model for normal tension glaucoma.
Analyses of retinal functionality via ERG recordings revealed no differences in a-and b-wave responses in control and immunized WT and KO mice, which indicates that photoreceptor cells as well as bipolar and Müller glia cells are not affected in our model.
We found a significant loss of Brn3a + RGCs in both genotypes following immunization. Interestingly, immunized KO mice displayed ∼15% more RGCs in retinal flat-mounts compared to immunized WT mice. Moreover, a comparable number of RGCs was found in retinal cross-sections of KO ONA animals. We also verified a severe optic nerve damage by diminished βIII-tubulin staining in immunized WT, while no alterations were found in the KO ONA mice. Previous studies of the EAG rat model showed that antibody deposits are accompanied by apoptotic RGC death (71). Furthermore, activation of the complement system via the lectin pathway seems to trigger retinal degeneration in this model (72). In our present study, we demonstrate that Tnc deficiency results in an extenuated loss of RGCs. Accordingly, we suggest that beside immunological alterations, Tnc-mediated signaling pathways are involved in glaucoma pathology.
In an EAG rat model a significant loss of RGCs could be detected 22 days after immunization (71). While an early upregulation of Tnc and its interaction partner RPTPβ/ζ/phosphacan was observed already at 7 days (6). Additionally, we noted an increased Tnc immunoreactivity in the retina 14 days post immunization. However, we did not reveal any differences regarding the Tnc protein level at later Here, the pro-inflammatory factor was downregulated and the anti-inflammatory cytokine was significantly upregulated in KO mice after immunization. (D-F) Quite similar expression patterns of both examined cytokines were noted in optic nerve tissue. Only, a significantly enhanced Tnfa expression was detected in WT ONA compared to WT CO (E). Groups were compared using the pairwise fixed reallocation and randomization test and values were shown as median ± quartile ± maximum/minimum. *p < 0.05; **p < 0.01. n = 5/group. points in time, namely 28 days after immunization. Therefore, we assume that alterations of Tnc levels occurred at an early point in time, shortly after immunization and then returned to normal values. Due to the fact that glaucomatous degeneration in both rodent EAG models is temporally comparable, we also assume that Tnc induction takes place at similar points in time. However, the precise timeline of Tnc induction should be determined in a future follow-up project. Ten weeks after immunization, we observed Tnc immunoreactivity in horizontal and amacrine cells as well as in the outer and inner plexiform layers (27,28). The quantification of the Tnc signal intensity revealed comparable protein levels in control and immunized WT mice. This is in line with results from the EAG rat model, were no differences in the Tnc protein levels were noted at later points in time (6).
In regard to the temporal induction, Tnc expression levels were enhanced 3 days post injury and returned to normal levels 7 and 14 days after the induction of brain laser lesion (73). In the CNS, Tnc is re-expressed under pathological conditions and represents an important component of the glial scar (11,25,36,(74)(75)(76). In Alzheimer's disease, Tnc immunoreactivity is highly associated with amyloid-β plaques (20). In multiple sclerosis, Tnc has an immunomodulatory function and loss of Tnc protects from experimental autoimmune encephalopathy (24). In regard to brain injury it was demonstrated that Tnc isoforms that contain the fibronectin type III domains B and D as well as the smallest splice variants, were specifically upregulated after cortical lesion (77). Future studies should focus on comprehensive analyses of specific Tnc isoforms in our glaucoma model. Due to these and our results, we conclude that Tnc induction occurred at early points in time and returned to normal levels at later points in time after immunization with ocular antigens. We speculate that upregulation of Tnc may serve as an early indicator for neurodegenerative processes before retinal damage is detectable. Finally, we assume that Tnc is a key mediator of inflammatory responses in our glaucoma mouse model.
Studies reported that Tnc is involved in the pathogenesis of CNS autoimmunity and astrocytes are a main Tnc source (24,29,30,(34)(35)(36). Astrocytes are the major glial cells in the optic nerve head and provide neurotrophic as well as mechanical support for RGC axons. Furthermore, astrocytes influence survival and functionality of neurons (78,79). Neuroinflammatory processes, characterized by altered functional properties and distribution of glial cells, appear to have an obvious function in glaucomatous optic neuropathy (80)(81)(82). In addition, an enhanced astrocyte reactivity in response to inflammatory signals is directly regulated by the ECM (83). Furthermore, RGC degeneration is accompanied by reactive astrogliosis, which results in a higher GFAP expression (51,58,(84)(85)(86). In our study, we observed an upregulated Gfap expression and an increased GFAP immunostaining in retinal and optic nerve tissue of immunized WT mice, whereas absence of Tnc led to a missing reactive gliosis. Therefore, severe glaucomatous damage induced by immunization seems to be triggered by Tnc expressing reactive astrocytes in the WT condition. We assume that the increase of astrocytic Tnc results in an enhanced glial cell proliferation, infiltration and cytokine release.
We noted a reduced population of mature oligodendrocytes and OPCs after immunization in WT mice. Also, a reduced MBP immunoreactivity demonstrated a demyelination of the optic nerve in the WT condition. Furthermore, no evidence of a decreased oligodendroglia density and demyelination was detected after ONA-immunization in KO animals. Interestingly, the fraction of mature Olig2/CC1 double immunopositive oligodendrocytes was significantly increased in the KO, in agreement with findings that Tnc inhibits the maturation of oligodendrocytes (17,87). In addition, a strong expression of Tnc in the optic nerve head inhibits the migration of oligodendrocytes from the optic nerve into the retina (88). Based on our results and the mentioned studies, we suggest that Tnc has an impact on demyelination processes. Missing of Tnc leads to no alteration in oligodendroglia but has a protective effect on myelination of optic nerve fibers in our EAG mouse model.
Microglia are the main resident immune cells of the CNS and play a crucial role during retinal neurodegeneration (89). They change their morphology to a reactive ameboid cell type after injury and neuroinflammatory changes in the retina occur during glaucomatous damage (81). Microglial activation is a very early event in glaucoma, often before significant loss of RGCs takes place (47,90,91). Analyses of microglial cells showed that Tnc deficiency results in a diminished microglial response characterized by a reduced number and reactivity. Furthermore, we detected enhanced levels of the antiinflammatory factor Tgfb1 and a decreased expression of the proinflammatory Tnfa. This is consistent with the study by Piccinini, which demonstrated that the deletion of Tnc reduced TNF-α production (92). In the retina, TGF-β signaling has an important role in neuronal differentiation and survival (93). In microglia, TGF-β regulates homeostasis and lack of TGF-β signaling promotes retinal degeneration (94,95). Additionally, TGF-β inhibits pro-inflammatory cytokines and regulates proliferation and activity of microglia (96). Our results indicate that the loss of Tnc signaling in microglia may promote the neuroprotective M2-subtype, resulting in a weaker RGC loss as well as axonal fiber damage in immunized KO animals.
The WT condition exhibited unaltered Tgfb1 expression and it is known that Tnc is able to bind this growth factor (97). In a pilocarpine seizure model it was shown that TGF-β signaling is accompanied by Tnc upregulation (98). Here, an increase of the Tgfb1 level occurred shortly after pilocarpine application, whereas the rise of Tgfb1 was less pronounced over time. Based on this date, we suggest that TGF-β dependent signaling in WT mice is associated with an early upregulation of Tnc and becomes less efficient at later points in time.
An enhanced microglial infiltration and marker expression revealed an increased glial response in WT post immunization in our study. Moreover, we found a significantly increased expression of the pro-inflammatory cytokine Tnfa. In glaucoma, a glia-derived neuronal death was described through TNF-α (99)(100)(101)(102). The indirect neurotoxicity of this pro-inflammatory factor is based on TNF receptor 2-mediated activation of microglia, whereas blocking of TNF-α results in an extenuated microglial response (101). Moreover, TNF-α does not directly induce RGC death but leads to an upregulation of the microglial Fas ligand, where the membrane-bound form elicits RGC apoptosis (103,104). However, in an early phase after injury a neuroprotective effect of TNF-α was found in an optic nerve crush animal model (105).
Microglia are involved in time-dependent astrocytic reactivity and this cross-talk is mediated by cytokines such as TNFα (106)(107)(108). Previous studies noted that resting microglia switched to a M1-like phenotype, which can lead to neurotoxic effects by producing high levels of pro-inflammatory cytokines (43,109,110). Based on this, we speculate that microglia are the first glial cells that react after immunization as a driving force of reactive gliosis in the EAG mouse model. Our investigations of immunized WT and Tnc deficient mice lead to the assumption that Tnc promotes immunomodulatory processes of the neurotoxic M1-subtype in turns of higher activity. This is in line with previous reports that Tnc supports the activity of M1-microglia (36,111).
Astrocytes as well as microglia express the toll-like receptor 4 (TLR4) and its expression is maintained by the extracellular environment (112,113). Tnc activates TLR4 via its fibrinogenlike globe domain and thereby regulates the production of pro-inflammatory cytokines, such as IL-6, IL-8, and TNF-α (114,115). In primary microglia, Tnc induced the synthesis of Tnfa and regulated the expression of iNOS via TLR4 signaling (38). Besides microglia, activated macrophages can also produce TNF-α (116). In human macrophages, specific alternatively spliced Tnc isoforms stimulate IL-6 and TNF-α release via TLR4 activation (117). In an experimental model of retinal degeneration, infiltration of peripheral monocytes occurred very early on, before microglia activation could be observed (118). Accordingly, we conclude that enhanced Tnfa expression, 10 weeks after immunization, is caused by Tnc-induced microglial TLR4 activation. Furthermore, we speculate that microglial TNF-α release and Tnc itself may regulate astrocytic Tnc expression, which results in a subsequently harmful astrogliosis in glaucomatous WT mice.
Our study could show that Tnc signaling pathways modulate both microglia and astrocyte response during glaucomatous damage. This might lead to a feedback loop by which increased levels of M1-microglial factors impact astrocytic Tnc release and vice versa. Finally, we propose that Tnc acts as an endogenous TLR4 ligand and represents an alarmin in our glaucoma model. However, it is possible that other receptors, for instance integrins, have an impact on inflammatory effects of Tnc (119,120). In this regard, also other ECM components and Tnc interaction partners might be involved. Here, e.g., the extra domain A of fibronectin was recently described to elevate IOP through TLR4 signaling in a TGFβ2-induced ocular hypertension mouse model (121). Hence, future studies should focus on the downstream signaling cascade of TLR4 in our IOP-independent EAG mouse model.

CONCLUSION
Taken together, our study demonstrated that Tnc influences glial response, migration, and inflammation during glaucomatous damage. This model is ideally suited for a better understanding of the molecular mechanisms between retinal neurodegeneration and ECM remodeling in order to develop future therapeutic options.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Landesamt für Natur, Umwelt und Verbraucherschutz, North Rhine-Westphalia.