An increased reactivity of resident astrocytes in POAG is indicated by morphologic changes and changes in the expression pattern of Glial fibrillary acidic protein (GFAP), a specific marker for astrocytes. Therefore, we investigated the distribution of GFAP in tangential sections of the glial lamina of 1 month and 2 month old βB1-CTGF1 mice, to analyze astrocyte morphology and expression profile. 1 month-old βB1-CTGF1 mice showed no changes in IOP compared to their wildtype (WT) littermates (WT: 14.71 ± 1.68; TG: 15.11 ± 3.34; p = 0.75). In contrast, 2 month-old transgenic (TG) mice showed a significant increase in IOP compared to WT controls (WT: 15.34 ± 2.08; TG: 16.70 ± 1.47; p = 0.018). GFAP immunostaining in the glial lamina of 1 month old TG and WT shows no alteration either in the GFAP synthesis or astrocyte morphology (Figure 1A, upper panel). In 2 month-old TG animals, the immunoreactivity for GFAP was increased in the glia lamina, compared to WT littermates (Figure 1A, middle and lower panel). The morphological analysis of the glial lamina showed a thickening of the astrocytic processes and the open spaces between the astrocyte processes were diminished in the TG animals in comparison to WT controls. The schematic illustration in Figure 1B shows the region where cross sections of the glial lamina and ON and ONH samples for molecular analysis were obtained. Quantification of the GFAP stained area in the glial lamina showed a significant increase in the TG animals compared to WT littermates in 2 month-old animals, but not in 1 month-old mice (1 month-old: WT: 1 ± 0.16; TG: 1.03 ± 0.33; 2 month-old: WT: 1.00 ± 0.35; TG: 1.39 ± 0.34; Figure 1C). Real-time RT-PCR analyses were performed to determine the expression level of Gfap in the ONH, containing the unmyelinated part of the nerve, and in the ON, comprised of the remaining myelinated nerve of 1 month and 2 month old βB1-CTGF1 and WT mice (Figure 1B). The Gfap mRNA expression analyses of the 1 month old TG mice shows no changes either in the ON (WT: 1 ± 0.06; TG: 1.19 ± 0.24) or the ONH (WT: 1 ± 0.14; TG: 1.14 ± 0.15) in comparison to their WT littermates (Figure 1D). Interestingly, real-time RT-PCR experiments showed a dramatic increase of Gfap in the ONH of 2 month old TG animals, which have a significantly enhanced intraocular pressure compared to WT controls (Junglas et al., 2012) (WT: 1 ± 0.13; TG: 5.08 ± 1.62; Figure 1C). Intriguingly, the mRNA analyses of the ON show no change in the Gfap expression (WT: 1 ± 0.68; TG: 0.85 ± 0.45; Figure 1D).

It is suggested that axonal damage is associated with alterations of the biomechanical properties of the peripapillary sclera (Coudrillier et al., 2012), which translates the IOP changes to the glial lamina and ON axons (Kimball et al., 2014). As the 2 month old TG βB1-CTGF1 mice show intense changes in astrocytes morphology in the ONH (Figure 1A), we focused on the 2 month old animals for the analysis of fibronectin and filamentous actin in the peripapillary sclera (Figure 2A). The immunohistochemical analysis of fibronectin and its distribution in tangential sections of the glial lamina region showed a marked increase of fibronectin in the peripapillary sclera in TG in comparison to controls. The localization of fibronectin within the glial lamina was mostly around the blood vessels. More interestingly the intensity of the staining was enhanced in TG animals compared to WT mice (Figure 2A, upper panel). The quantification of fibronectin stained areas showed a significant increase in the peripapillary sclera region of 2 month old TG mice compared to WT littermates (WT: 1 ± 0.45; TG, 2.64 ± 0.24; Figure 2B). The observed changes were accompanied by a considerable increase in the amount of filamentous actin in the ON of βB1-CTGF1 compared to their WT littermates. The increase in phalloidin labeled actin was especially pronounced next to the peripapillary sclera (Figure 2A, lower panel). The measurement of the area of phalloidin labeled filamentous actin in the ONH could show an increase in the TG animals in comparison to WT littermates (WT: 1 ± 0.58; TG: 3.07 ± 0.11; Figure 2C).

To address the question, whether CCN2/CTGF is altered related to increased astrocyte reactivity and enhanced IOP, the immunoreactivity of CCN2/CTGF in the glial lamina was analyzed. A faint staining throughout the entire ON tissue with no obvious preference for glial or neuronal tissue is observed in TG and WT mice (Figure 3A). Immunoreactivity against CCN2/CTGF shows no changes in the glial lamina of 1-month old TG mice, compared to WT littermates (Figure 3A). Furthermore, quantification of CCN2/CTGF stained area in the glial lamina sections could not identify any variation (WT: 1.00 ± 0.05; TG: 0.98 ± 0.10; Figure 3B). 2 month old WT mice showed the same distribution for CCN2/CTGF as observed for the 1 month old animals. Interestingly, the intensity of the immunoreactivity for CCN2/CTGF was increased in the 2 month old animals compared to WT animals (Figure 3A). Measurements of CCN2/CTGF stained area in the ONH show a significant increase in the TG mice in comparison to WT littermates (WT: 1.00 ± 0.46; TG: 1.63 ± 0.16; *p = 0.028; Figure 3B). mRNA analysis of Ccn2/Ctgf in the 1 month old animals showed similar results, which were seen for GFAP. Ccn2/Ctgf mRNA expression is not changed either in the ON (WT: 1 ± 0.10; TG: 1.21 ± 0.14) or in the ONH (WT: 1 ± 0.18; TG: 1.06 ± 0.14; Figure 3C) of 1 month old animals. Interestingly, real time RT-PCR experiments showed a huge increase of Ccn2/Ctgf in the ONH of TG animals compared to WT littermates (WT: 1 ± 0.12; TG: 10.04 ± 1.66; **p = 0.006; Figure 3C) in the 2 month old animals. Intriguingly, the mRNA analyses of the ON showed no change in the Ccn2/Ctgf expression (WT: 1 ± 0.18; TG: 1.33 ± 0.29; Figure 3C).

We further investigated the CCN2/CTGF expression in the human ON under healthy and glaucomatous conditions. Therefore, we analyzed the CCN2/CTGF immunoreactivity in two healthy and four glaucomatous human ON (Figure 4). In the healthy human ON we found a weak, but consistent CCN2/CTGF signal in the entire ON, with no obvious preference for glial or neuronal tissue. Additionally, we could not reveal any difference of the CCN2/CTGF synthesis related to different regions of the ON (Figures 4A,B, upper panel). The border area (Figure 4A, upper panel) showed a similar CCN2/CTGF immunoreactivity as the central part of the healthy ON (Figure 4B, upper panel). To investigate whether an altered CCN2/CTGF synthesis can be observed under glaucomatous conditions, we analyzed the CCN2/CTGF immunoreactivity of four human glaucomatous ON in comparison to the healthy human ON. We could find an intense increase of CCN2/CTGF immunoreactivity in the border area of the glaucomatous ON compared to the healthy ON. This finding is consistent in all four analyzed glaucomatous ON (Figure 4A). In contrast we could not observe changes in the CCN2/CTGF intensity in the central area of the glaucomatous ON, except one of the glaucomatous ON which shows a slight increase (Figure 4B, lower panel, right ON).

Since TGF-β2 induces a reactive phenotype in human ONH astrocytes in vitro and promotes the synthesis of GFAP and ECM proteins (Fuchshofer et al., 2005; Kirschner et al., 2021), we wanted to know, whether cultured murine ON astrocytes react in the same manner. The TGF-β2 induced ECM synthesis is dependent on CCN2/CTGF (Fuchshofer et al., 2005), so we wanted to investigate the direct effects of CCN2/CTGF in murine ON astrocytes in vitro. For this reason, we established a murine ON astrocyte cell culture (see Supplementary Methods S1; Supplementary Figure S1). Astrocytes were treated either with 1 ng/ml TGF-β2, 50 ng/ml and 100 ng/ml CCN2/CTGF and were compared to untreated control cells. Treatment with 1 ng/ml TGF-β2 led to marked increase in GFAP immunoreactivity in murine ON astrocytes in comparison to untreated control cells. Treatment with 50 ng/ml CCN2/CTGF caused a comparable reaction to that seen after TGF-β2 treatment (Figure 5A). The evaluation of the effect of both growth factors on the actin cytoskeleton showed a significant increase in α-actinin protein synthesis after CCN2/CTGF treatment, which is necessary for actin filament crosslinking and therefore for the regulation of cell adhesion and motility (Figure 5B; 50 ng/ml: 2.91 ± 0.91; 100 ng/ml: 2.29 ± 0.53). Treatment with 1 ng/ml TGF-β2 did not lead to significant changes in α-actinin synthesis, compared to untreated control cells (Figure 5B; 1.71 ± 0.46). To evaluate the effects of TGF-β2 and CCN2/CTGF on astrocyte migration, scratch assays were carried out. Scratches were measured before and 12 h after the treatment and the migration area was calculated. After 12 h the treatment with 1 ng/ml TGF-β2 resulted in a significantly increased migration rate, in comparison to untreated cells (Figure 5C; 142.28 % ± 23.78%). Treatment with 50 ng/ml CCN2/CTGF (118.07 % ± 11.71%) or 100 ng/ml CCN2/CTGF (134.81 % ± 4.40%) also increased the migration rate significantly, compared to controls (Figure 5C).

Furthermore, the direct effect of CCN2/CTGF on ECM protein expression was investigated in murine ON astrocytes in vitro. Immunocytochemical staining, Western blot analyses and Real time RT-PCR analyses were performed for different ECM components. Immunoreactivity of fibronectin and tropoelastin showed a marked increase after treatment with TGF-β2 and CCN2/CTGF (Figure 6A). To analyze the ECM induction in a molecular approach, we investigated additionally one of the several collagen types, collagen 3a1 (Col3a1). We wanted to analyze a fibrillar collagen component. In the murine glial lamina region, the immunohistochemical staining against collagen type III and collagen type I showed an intense signal in this region (May and Lutjen-Drecoll, 2002). We focused on collagen type III as in remodeling processes collagen type III is primarily build before the stronger collagen type I (Suda et al., 2016). Murine ON astrocytes were treated again with 1 ng/ml TGF-β2, or 50 ng/ml CCN2/CTGF and 100 ng/ml CCN2/CTGF for 24 h. Treatment with TGF-β2 or CCN2/CTGF led to a significantly higher expression of mRNA for Col3a1, compared to untreated control cells (1 ng/ml TGF-β2: 2.30 ± 0.72, 50 ng/ml CCN2/CTGF: 2.47 ± 0.70, 100 ng/ml CCN2/CTGF: 3.07 ± 1.03; Figure 6B). In comparison to control, fibronectin was elevated significantly after treatments on mRNA in comparison to untreated control cells (1 ng/ml TGF-β2: 1.64 ± 0.17, 50 ng/ml CCN2/CTGF: 1.81 ± 0.44, 100 ng/ml CCN2/CTGF: 2.26 ± 0.87; Figure 6B). Finally, mRNA for tropoelastin was also induced significantly after the treatments with TGF-β2 or CCN2/CTGF, compared to untreated control cells (1 ng/ml TGF-β2: 2.72 ± 0.66, 50 ng/ml CCN2/CTGF: 2.74 ± 0.73, 100 ng/ml CCN2/CTGF: 2.99 ± 08; Figure 6B). The results from Western blot analyses detecting collagen type III, fibronectin and tropoelastin supported the mRNA data. Collagen type III was significantly upregulated by 1.5-fold after TGF-β2 treatment and 2-fold after CCN2/CTGF in comparison to untreated control cells (1 ng/ml TGF-β2: 1.47 ± 0.2, 50 ng/ml CCN2/CTGF: 2.08 ± 0.83, 100 ng/ml CCN2/CTGF: 1.92 ± 0.29; Figure 6C). Furthermore, fibronectin was significantly elevated after the treatments with TGF-β2 and CCN2/CTGF compared to untreated cells (1 ng/ml TGF-β2: 2.62 ± 0.48, 50 ng/ml CCN2/CTGF: 2.87 ± 0.56, 100 ng/ml CCN2/CTGF: 2.89 ± 0.74; Figure 6C). Finally, western blot experiments to detect tropoelastin, the soluble precursor of elastin showed a high and significant upregulation after the treatments in comparison to untreated control cells (1 ng/ml TGF-β2: 2.69 ± 0.53, 50 ng/ml CCN2/CTGF: 1.70 ± 0.27, 100 ng/ml CCN2/CTGF: 2.87 ± 0.4; Figure 6C).

To investigate the effect of increasing substratum stiffness on mouse ON astrocytes, cells were cultured on Poly dimethylsiloxane (PDMS) substratum with different E-moduli (10 kPa, 30 kPa, 60 kPa) (see Supplementary Methods S2; Supplementary Figure S2). First, the actin stress fibers were labeled with phalloidin to assess if astrocytes can sense the changes in stiffness of their surrounding matrix and visualize the effect on the actin cytoskeleton structure. We could observe an intensification of the labeled F-actin filaments with increasing stiffness (Figure 7A, upper panel). Additionally, the increased substratum stiffness caused an increase in longitudinally-oriented actin stress fibers. Astrocytes grown on a substratum with the stiffness of 60 kPa contained numerous longitudinally arranged actin stress fibers (Figure 7A, upper panel), which were longer and thicker compared to those grown on substratum with 10 kPa or 30 kPa (Figure 7A, upper panel). Next, we performed immunocytochemial staining against CCN2/CTGF and GFAP. CCN2/CTGF immunoreactivity was increased in astrocytes grown on stiffer substratum, compared to cells cultured on 10 kPa (Figure 7A, middle panel). Similar results were seen for GFAP, which was more intense in astrocytes grown on 30 or 60 kPa compared to those cultured on 10 kPa (Figure 7A, lower panel). To further evaluate the effect of increasing substratum stiffness on murine ON astrocytes we performed western blot analyses against CCN2/CTGF, GFAP and Vimentin. Western blotting experiments of proteins from astrocytes grown on a substratum with an E-modulus of 60 kPa showed an increase in the amounts of CCN2/CTGF (60 kPa: 3.48 ± 1.02), compared with proteins from astrocytes cultured on substrata with an E-modulus with 10 kPa (Figure 7B). In cells cultured on 30 kPa enhanced, but not significant increase for CTGF was detected (30 kPa: 3.05 ± 0.92).In proteins from cells grown on substratum with a 30 kPa or 60 kPa stiffness, we observed significantly increased amounts of GFAP compared to those from cells grown on 10 kPa (30 kPa: 1.65 ± 0.18; 60 kPa: 2.27 ± 0.64; Figure 7C). Vimentin protein was also significantly increased when cells were grown on the 60 kPa substratum compared to astrocytes plated on 10 kPa (60 kPa: 2.58 ± 0.12; Figure 7D). In cells cultured on 30 kPa enhanced, but not significant, increase for Vimentin was detected (30 kPa: 2.45 ± 0.85).

Transgenic βB1-CTGF1 mice were generated as described previously (Junglas et al., 2012) and compared to wildtype mice. Mice were housed under standardized conditions of 62% air humidity and 21°C room temperature. Feeding was ad libitum. Animals were kept at a 12 h light/dark cycle (6 am–6 pm). All procedures conformed to the tenets of the National Institutes of Health Guidelines on the Care and Use of Animals in Research, the EU Directive 2010/63/E and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by the local authorities (54-2532.1-44/12; Regierung Oberpfalz, Bavaria, Germany).

Astrocytes were cultured either on 6-well plates and treated with 1 ng/ml TGF-β2, 50 ng/ml CCN2/CTGF or 100 ng/ml CCN2/CTGF, or cultured on PMDS substrata with different E-moduli (10kPa, 30 kPa, 60 kPa). The concentration for CCN2/CTGF treatment were chosen in accordance to our previous findings (Junglas et al., 2009).Before enucleation of the eyes, mice were anesthetized with CO₂ and euthanized by atlanto-occipital dislocation. First the anterior eye segment, the lens and retina were removed, and the ON was cleaned from muscles and connective tissue. The sclera was cut away from the ONH and potentially remaining parts of sclera and retina were removed. Second the ON tissue was separated from the ONH. Both tissue ON and ONH were processes independently for following experiments. Total RNA was extracted with TriFast™ (Peqlab, Erlangen, Germany) according to the manufacture’s recommendations. cDNA was prepared from total RNA using the qScript™cDNA Synthesis Kit (Quanta Biosciences, Gaithersburg, United States) according to the manufacturer’s introductions. Real-time RT-PCR was performed on a BioRad iQ5 Real-Time PCR Detection System (BioRad, München, Germany) with the temperature profile as follows: 50 cycles of 20 s melting at 94°C, 10 s of annealing at 60°C and 20 s of extension at 60°C. All primers were purchased from Invitrogen and extended over exon-intron boundaries: 5′-ccc​ctg​gaa​tct​gtg​aat​c-3′ (msCol3a1 forward), 5′-tga​gtc​gaa​ttg​ggg​aga​at-3′ (msCol3a1 reverse), 5′-gct​gct​gct​aag​gct​gct​aa-3′ (msElastin forward), 5′-agc​acc​tgg​gag​cct​aac​tc-3′ (msElastin reverse), 5′-cgg​aga​gag​tgc​ccc​tac​ta-3′ (msFN forward), 5′-cga​tat​tgg​tga​atc​gca​ga-3′ (msFn reverse), 5′-tct​gca​agt​aca​cgg​tcc​ag-3′ (msGnb2l forward), 5′-gag​acg​atg​ata​ggg​ttg​ctg-3′ (msGnb2l reverse), 5′-tga​cct​gga​gga​aaa​cat​taa​ga-3′ (msCcn2/Ctgf forward), 5′-agc​cct​gta​tgt​ctt​cac​act​g-3′ (msCcn2/Ctgf reverse), 5′-tcg​aga​tcg​cca​cct​aca​g-3′ (msGfap forward), 5′-gtc​tgt​aca​gga​atg​gtg​atg​c-3′ (msGfap reverse), 5′-gct​gcc​atc​tgt​ttt​acg​g-3′ (msRpl32 forward), 5′-tga​ctg​gtg​cct​gat​gaa​ct-3 (msRpl32 reverse) or Taqman^® probes were used (Rpl32: Mm02528467_g1; Gfap: Mm01253033_m1; Ccn2/Ctgf: Mm01192933_g1 from ThermoFisher Scientific, Darmstadt, Germany). RNA that was not reverse transcribed served as negative control. For relative quantification of the experiments, Gnb2l and Rpl32 were used as a housekeeping gene. BioRad iQ5 Optical System Software (version 2.0) was used for analysis and ΔΔct-method was applied for normalization.

Proteins were isolated following the RNA isolation according to the manufacturer’s instructions. The proteins were dissolved in 1% SDS containing protease (Serva Electrophoresis GmbH, Heidelberg, Germany) and phosphatase inhibitor (Sigma-Aldrich, Taufkirchen, Germany). Protein concentration was determined by the bicinchoninic acid assay (Interchim, Montluçon Cedex, France). Proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Roche, Mannheim, Germany). Western Blot analysis was performed with specific antibodies as described previously (Fuchshofer et al., 2005). Specific antibodies were used as follows: goat anti-α-actinin (1:500), rabbit anti-Col III (1:250; Santa Cruz Biotechnology, Cat# sc-28888, RRID:AB_2082354), goat anti-CTGF (1:500; Santa Cruz Biotechnology, Cat# sc-14939, RRID:AB_638805), rabbit anti-fibronectin (1:500; Santa Cruz Biotechnology, Cat# sc-9068, RRID:AB_2105699), chicken anti-GFAP (1:10,000; LSBio (LifeSpan), Cat# LS-B4775-50, RRID:AB_10803257), rabbit anti-tropoelastin (1:250, Elastin Products Company, Owensville, MO, United States), goat anti-vimentin (1:500, Sigma-Aldrich Cat# V4630, RRID:AB_477619, Taufkirchen, Germany), donkey anti-goat (1:2000, Bethyl Laboratories Inc., Montgomery, TX, United States, Cat# A50-201P, RRID:AB_66756), goat anti-rabbit (1:5,000, Cell Signaling Technology, Danvers, MA, United States, Cat# 7074, RRID:AB_2099233), rabbit anti-chicken (1:2,000, Bethyl Laboratories Inc., Montgomery, TX, United States, Cat# A30-207P, RRID:AB_67387). Chemiluminescence was detected on a LAS 3000 imaging workstation (Fujifilm, Düsseldorf, Germany). GAPDH (rabbit anti-GAPDH-HRP, 1:5,000, Cell Signaling Technology, Cat# 3683, RRID:AB_1642205) or α-tubulin (rabbit anti-α-tubulin, 1:2500, Rockland Immunochemicals Inc., Gilbertsville, United States, Cat# 600-401-880, RRID:AB_2137000) were used as loading controls to normalize the signal intensity of the Western blots. The intensity of the bands detected by Western blot analysis was determined using appropriate software (AIDA Image analyzer software, Raytest).

Astrocytes were plated on 35 mm µ-dishes (ibidi, Martinsried, Germany) with different E-moduli (10 kPa, 30 kPa, 60 kPa) at a density of 20000 cells per dish for phalloidin labeling, GFAP and CCN2/CTGF staining (Cells were cultured 7 days and starved under serum free conditions 24 h before fixation). For the characterization of the isolated ON astrocytes and for analysis of the effect of CCN2/CTGF treatment on the ECM synthesis, cells were seeded on coverslips and treated with either 1 ng/ml TGF-β2, 50 ng/ml CCN2/CTGF or left untreated as controls for 24 h under serum free conditions. Cells were washed twice with PBS, fixed with 4% (w/v) paraformaldehyde (PFA) for 5 min and washed again three times with PBS. Specific antibodies for staining were used as follows: rabbit anti-CTGF (1:200, Genetex, Irvine, United States, Cat# GTX26992, RRID:AB_369067), rabbit anti-GFAP (1:100, Dako, Hamburg, Germany, Agilent Cat# Z0334, RRID:AB_1001338), rabbit anti-tropoelastin (1:20, Elastin Products Company, Owensville, MO, United States), rabbit anti-fibronectin (1:100, Santa Cruz, Dallas, TX, United States, Cat# sc-9068, RRID:AB_2105699), Alexa Fluor 488-conjugated anti rabbit IgG (1:1,000, Invitrogen, Darmstadt, Germany, Thermo Fisher Scientific Cat# A-11070, RRID:AB_2534114) and Cy3™ goat anti rabbit (1:2000, Jackson Immuno Research Europe Ltd., Suffolk, United Kingdom, Cat# 111-165-144, RRID:AB_2338006). Phalloidin-rhodamin (1:1,000, 1 h at room temperature, Sigma-Aldrich, Taufkirchen, Germany) was used for labeling the actin cytoskeleton. As a control for unspecific binding of secondary antibodies, negative controls were performed. Finally, 4,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA, United States) diluted 1:10 in PBS was added to counterstain nuclear DNA and the immunofluorescence was visualized using a Zeiss Axio Imager fluorescence microscope (Carl Zeiss AG, Göttingen, Germany).

Astrocytes were grown to confluence in 6-well plates. 24 h prior to treatments, cells were starved under serum-free conditions. Three equal scratches were made in the cell layer of each well. To suppress proliferation, cells were treated with 5 μg/ml Mitomycin C (Sigma-Aldrich). Cells were treated with 1 ng/ml TGF-β2, 50 ng/ml CTGF or 100 ng/ml CTGF, untreated cells were used as control. Pictures were taken directly before (0 h) and 12 h after the treatment. For each time point, length and region of the scratches were measured using Axiovision 4.8 (Carl Zeiss AG). Region of migration (ΔA) was counted by subtracting the region at 12 h (At) from the region at 0 h (A0). To correct the results, they were then divided by the length of the scratch 0 h (A0-At)/s = ΔA/s. Migration assays were carried out several times. Results were allocated to an arithmetic mean.

Methods for securing human tissue were humane, included proper consent and approval, and complied with the Declaration of Helsinki. Human ON were obtained from two healthy and four glaucomatous eyes. Human ON were embedded in paraffin and 6 µm sections were deparaffinized and washed in water. For immunohistochemistry ON sections were pretreated with proteinase K followed by blocking with 2% bovine serum albumin (BSA), 0.2% cold water fish gelatin (Sigma-Aldrich). Sections were incubated with rabbit anti-CTGF (1:50, novusbio, Centennial, CO, United States, Cat# NB100-724SS, RRID:AB_921075) at 4°C overnight. Afterwards sections were washed three times with 0.1 M phosphate buffer, followed by the incubation with Alexa Flour ^® 647 donkey anti-rabbit (1:200, Thermo Fisher Scientific Cat# A-31573, RRID:AB_2536183) for 2 h at room temperature. As a control for unspecific binding of secondary antibodies, negative controls performed. After washing three times with 0.1 M phosphate buffer, the slides were mounted using the DakoCytomation fluorescent mounting medium with DAPI 1:10 (Dako). Slides were dried overnight at 4°C before microscopy. Eyes of βB1-CTGF1 transgenic (TG) and wildtype (WT) mice were enucleated and fixed in 4% (w/v) paraformaldehyde (PFA) for 24 h. The eyes were equilibrated in 10%, 20% and 30% sucrose, embedded in Tissue Tek optimal cooling temperature compound (Sakura Finetek Europe B.V., Zoeterwounde, Netherland), and stored at −20°C. Frozen 12 µm sections were cut on the cryostat. Tangential sections of the glial lamina were obtained. After blocking with 1% BSA, 0.2% cold water fish skin gelatin (Sigma-Aldrich), 0.1% Triton-X-100 in 0.1 M phosphate buffer, frozen sections were incubated with rabbit anti-GFAP (1:1,000, Dako Denmark A/S,Glostrup, Denmark, Agilent Cat# Z0334, RRID:AB_1001338), rabbit anti-CTGF (1:400, Genetex Inc.,, Irvine, CA, United States, Cat# GTX26992, RRID:AB_369067), rabbit anti-fibronectin (1:500, Agilent Cat# A024502, RRID:AB_578510), chicken anti-GFAP (1:2000, LSBio (LifeSpan), Cat# LS-B4775-50, RRID:AB_10803257) or with phalloidin-rhodamin (1:500, 1 h at room temperature, Sigma-Aldrich) to label the actin cytoskeleton at 4°C overnight. Afterwards, tissue sections were washed three times with 0.1 M phosphate buffer, followed by an incubation with the secondary antibody, Alexa Flour ^® 488 goat anti rabbit (1:1,000, Life Technologies, Carlsbad, CA, United States, Thermo Fisher Scientific Cat# A-11070, RRID:AB_2534114), Alexa Flour ^® 488 goat anti chicken (1:1,000, Thermo Fisher Scientific Cat# A-11039, RRID:AB_2534096) or Cy3™ goat anti rabbit (1:2,000, Jackson Immuno Research Europe Ltd., Suffolk, United Kingdom, Cat# 111-165-144, RRID:AB_2338006) for 1 h at room temperature. As a control for unspecific binding of secondary antibodies, negative controls were performed. After washing three times with 0.1 M phosphate buffer, the slides were mounted using the DakoCytomation fluorescent mounting medium with DAPI 1:10 (Dako). Slides were dried overnight at 4°C before microscopy.

The images were analyzed using ImageJ’s built-in measuring feature (Wayne Rasband, formerly National Institutes of Health, Bethesda, MD, United States). The surface area of the ONH section was calculated and the amount of area emitting fluorescent signal for GFAP, CCN2/CTGF or filamentous actin within the outlines of the ONH section was determined by a standardized macro routine consisting of ImageJ’s color threshold plugin and particle analyzer. The resulting values were used to calculate the percentage of area within the ONH section. For fibronectin quantification, the surface area of the peripapillary sclera was quantified and the same methods of measurement that were mentioned above were used. These measurements were performed for each individual specimen after calibrating ImageJ with the scale bar. The resulting data was analyzed using SPSS (IBM, Armonk, NY, United States).

Western Blot and real-time RT-PCR was repeated at least three times with RNA and protein extract from mouse ON astrocytes and ON/ONH tissue. Each real-time RT-PCR analysis was performed in triplicates. The data is represented as mean ± SEM or otherwise stated in the figure legends. Statistical analysis of data was performed by two-tailed t-test compared to theoretical mean of 1 (normalized control, indicated in the figure legends).