Ischemic injury leads to extracellular matrix alterations in retina and optic nerve

Retinal ischemia occurs in a variety of eye diseases. Restrained blood flow induces retinal damage, which leads to progressive optic nerve degeneration and vision loss. Previous studies indicate that extracellular matrix (ECM) constituents play an important role in complex tissues, such as retina and optic nerve. They have great impact on de- and regeneration processes and represent major candidates of central nervous system glial scar formation. Nevertheless, the importance of the ECM during ischemic retina and optic nerve neurodegeneration is not fully understood yet. In this study, we analyzed remodeling of the extracellular glycoproteins fibronectin, laminin, tenascin-C and tenascin-R and the chondroitin sulfate proteoglycans (CSPGs) aggrecan, brevican and phosphacan/RPTPβ/ζ in retinae and optic nerves of an ischemia/reperfusion rat model via quantitative real-time PCR, immunohistochemistry and Western blot. A variety of ECM constituents were dysregulated in the retina and optic nerve after ischemia. Regarding fibronectin, significantly elevated mRNA and protein levels were observed in the retina following ischemia, while laminin and tenascin-C showed enhanced immunoreactivity in the optic nerve after ischemia. Interestingly, CSPGs displayed significantly increased expression levels in the optic nerve. Our study demonstrates a dynamic expression of ECM molecules following retinal ischemia, which strengthens their regulatory role during neurodegeneration.


Results
ECM glycoproteins in the control and ischemic retina and optic nerve. First, we analyzed the expression pattern of the ECM glycoproteins fibronectin, α1-laminin, tenascin-C and -R in control (CO) and ischemic (I/R) retinae via qRT-PCR ( Fig. 1A-D). Regarding the mRNA expression pattern of the aforementioned glycoproteins, we observed a significant upregulation of fibronectin in the I/R group (1.41-fold, p = 0.032; Fig. 1A). In contrast, both α1-laminin (0.42-fold, p = 0.031; Fig. 1B) as well as tenascin-C (0.61-fold, p = 0.027; Fig. 1C) mRNA levels were significantly downregulated in I/R retinae in comparison to CO tissue. For tenascin-R no significant difference was observed between both groups (1.29-fold, p = 0.219; Fig. 1D).
To further analyze the distribution pattern of these glycoproteins, we labeled horizontal retinal sections with specific antibodies and analyzed their immunoreactivity by semi-quantitative area analyses ( Fig. 2A-L). Regarding fibronectin staining, specific signals were restricted to retinal blood vessels of control and ischemic retinae ( Fig. 2A,B). Analyses of fibronectin immunoreactivity revealed a significant area increase in ischemic (7.19 ± 1.45 area [%]/image; p = 0.016) in comparison to CO retinae (4.19 ± 1.64 area [%]/image; Fig. 2C). By Western blot analyses of retinal cell lysates, fibronectin was observed at > 250 kDa (Fig. 3A,B). In line with the qRT-PCR and immunohistochemical results, quantitative protein analyses verified a significantly increased protein level in I/R (0.66 ± 0.05; p < 0.001) compared to CO retinae (0.43 ± 0.07).
Tenascin-R immunoreactivity was specifically enriched in the OPL, INL, IPL and GCL (Fig. 2J,K). Here, our immunofluorescence analyses showed a significant upregulation of tenascin-R area in ischemic ( Fig. 2L). By Western blotting, tenascin-R was detected as two major bands at 160 and 180 kDa (Fig. 3G,H). Here, densitometric measurements , tenascin-C (C) and tenascin-R (D) mRNA expression using qRT-PCR in control (CO) and ischemic (I/R) retinae. Our results revealed significantly elevated levels for the glycoprotein fibronectin in I/R retinae, whereas α1-laminin and tenascin-C displayed a significantly reduced expression. No expression changes were observed for tenascin-R. Values are median ± quartile ± maximum/minimum. *p ≤ 0.05; n = 4/group. of total tenascin-R protein revealed comparable levels (CO: 0.32 ± 0.13; I/R: 0.24 ± 0.05; p = 0.223). Interestingly, a significant upregulation of the larger tenascin-R isoform was observed in I/R (0.44 ± 0.09; p = 0.007; data not shown) compared to CO retinae (0.22 ± 0.10). In contrast, reduced protein levels were observed for the low molecular weight isoform in both groups (CO: 0.41 ± 0.15; I/R: 0.05 ± 0.02; p < 0.001; data not shown), indicating an isoform-specific regulation of tenascin-R.
As revealed by immunohistochemistry, fibronectin and laminin showed a distinct cellular expression pattern (Fig. 5A,B,D,E). Especially in the ischemic condition, both tenascins displayed a more widely extracellular staining pattern in optic nerve tissue (Fig. 5G,H  Representative retinal cross-sections of control (CO) and ischemic (I/R) eyes stained using specific antibodies directed against the ECM glycoproteins fibronectin (A,B, red), laminin (D,E, green), tenascin-C (G,H, green) and tenascin-R (J,K, green) as well as the nuclear dye TO-PRO-3 (blue) shown as merged images. Small inserts display TO-PRO-3 (blue) and glycoprotein (green/red) staining separately. In both experimental groups, fibronectin as well as laminin displayed a blood vessel-associated staining. In addition, laminin was found in the ILM and in the GCL. Tenascin-C and -R staining was mainly localized in the IPL, OPL and GCL. Quantification revealed a significant increase in the fibronectin and tenascin-R staining area in ischemic retinae, whereas no significant changes were observed regarding the tenascin-C and laminin staining (C,F,I,L). Values are mean ± SEM ± SD. *p ≤ 0.05; **p ≤ 0.01; n = 5/group. GCL = ganglion cell layer, ILM = inner limiting membrane, INL = inner nuclear layer, IPL = inner plexiform layer, ONL = outer nuclear layer, OPL = outer plexiform layer. Scale bar = 50 μ m. CSPGs in the control and ischemic retina and optic nerve. Next, we analyzed expression levels of the CSPGs aggrecan, brevican and phosphacan/RPTPβ/ζ via qRT-PCR in CO and I/R retinae (Fig. 6A-D). Based on these analyses, we verified a significant downregulation of brevican expression levels (0.38-fold, p = 0.03; Fig. 6B). In contrast, comparable mRNA expression levels were seen for aggrecan (0.60-fold, p = 0.059; Fig. 6A), all RPTPβ/ζ-isoforms (RPTPβ/ζ CA: 1.04-fold, p = 0.884; Fig. 6C) as well as for the RPTPβ/ζ receptor variants (RPTPβ/ζ PTP1: 0.98-fold, p = 0.883; Fig. 6D) in both groups. Further, we evaluated the immunohistochemical staining pattern of the CSPGs aggrecan, brevican and phosphacan/RPTPβ /ζ in the CO and I/R retinae ( Fig. 7A-L). Concerning aggrecan and brevican, prominent immunoreactivity was localized in the IPL of the retinae (Fig. 7A,B,D,E). In addition, brevican immunostaining was prominently seen in the GCL, whereas aggrecan was also found in the IPL. Our statistical analyses revealed no significant changes for aggrecan (CO: 20.  Fig. 7F) staining area within the I/R group in comparison to the CO group. In contrast, quantitative analyses of total aggrecan protein, detected as two bands at > 100 and > 150 kDa, revealed a significantly reduced level in the I/R group compared to CO retinae (CO: 0.32 ± 0.02; I/R: 0.16 ± 0.03; p < 0.001; Fig. 8A,B). For brevican, prominent protein bands were detected at ~50 and > 100 kDa (Fig. 8C,D). Here, relative quantification verified comparable total protein levels in I/R and CO retinae (CO: 1.30 ± 0.05; I/R: 1.32 ± 0.07; p = 0.683; Fig. 8C,D).
Next, we used an antibody against the 473HD epitope, a particular chondroitin sulfate glycan, specifically localized on the secreted splice variant phosphacan as well as RPTPβ /ζ long 20,21 . As demonstrated by previous studies and in this study, 473HD immunoreactivity is restricted to Müller glia cells (Fig. 7G,H) 13,22,23 . Due to the downregulation of the RPTPβ /ζ long isoform within the adult retina, 473HD immunoreactivity mainly reflects the expression of the secreted phosphacan isoform. Analyses of the 473HD staining area showed no significant alteration between both groups (CO: 10.43 ± 0.55 area [%]/image; I/R: 9.76 ± 0.94 area [%]/image; p = 0.206; Fig. 7I), which is in line with the observed comparable mRNA levels revealed by qRT-PCR analyses. Western blotting using the 473HD antibody revealed a protein expression at > 150 kDa. Quantitative analyses revealed comparable levels in CO and I/R retinae (CO: 0.66 ± 0.18; I/R: 0.78 ± 0.18; p = 0.32; Fig. 8E,F).
We also evaluated the mRNA and protein levels of the CSPGs aggrecan, brevican as well as of phosphacan/ RPTPβ /ζ -isoforms in the optic nerves of CO and ischemic eyes (Figs 9A-D and 10A-L).
No significant difference in the aggrecan mRNA level was observed (0.84-fold, p = 0.378; Fig. 9A). On the other hand, extracellular aggrecan immunostaining was significantly increased in the I/R optic nerves (CO: 6.02 ± 9.  Discussion ECM remodeling upon retinal damage was reported by numerous previous studies. Nevertheless, little information is available regarding the significance of ECM remodeling following ischemic retinal injury. In the present study, we subsequently analyzed the expression and distribution pattern of several extracellular glycoproteins and CSPGs in the retina and optic nerve of an I/R rat model via qRT-PCR and immunohistochemistry. Additionally, retinal protein levels were quantified via Western blot analyses. Our results demonstrate that each ECM molecule displays an unique spatial expression and protein regulation, which reflects their potential functional role during ischemic degeneration. , tenascin-C (G,H, green) and tenascin-R (J,K, green) as well as the nuclear dye TO-PRO-3 (blue) shown as merged images. Small inserts display TO-PRO-3 (blue) and glycoprotein (green/red) staining separately. Fibronectin as well as laminin staining was restricted to single cells within the optic nerve. In contrast, both tenascin proteins showed an extracellular staining pattern throughout the optic nerve. Analyses of the staining area verified a significant increase of laminin and tenascin-C in ischemic optic nerves. No significant increase was observed for fibronectin and tenascin-R (C,F,I,L). Values are mean ± SEM ± SD. *p ≤ 0.05; **p ≤ 0.01; n = 5/group. Scale bar = 50 μ m.
Scientific RepoRts | 7:43470 | DOI: 10.1038/srep43470 Dysregulation of ECM glycoproteins under ischemic conditions. We first monitored the expression pattern of the ECM glycoproteins fibronectin, laminin, tenascin-C and tenascin-R. For fibronectin, a significantly increased mRNA as well as protein level was noted in the ischemic retina. No regulation was seen in the optic nerve. As revealed by others and our study, prominent fibronectin staining is mainly restricted to retinal blood vessels of the inner retina 24,25 . ECM constituents display a central functional importance during vascular development and neovascularization 26 , which is associated with severe retinal ischemia. Abnormalities in the ECM of the retinal microvasculature are common, e.g. in diabetic retinopathies 27 . In patients with diabetic retinopathy, fibronectin was overexpressed in retinal microvessels. Here, it was speculated that increased fibronectin synthesis and deposition by microvascular cells may modify cell-matrix interaction with functional consequences relevant to retinal damage 28 . Due to these and our findings, we assume that fibronectin upregulation in the ischemic retina reflects substantial blood vessel remodeling, sprouting and/or neovascularization. The precise function of fibronectin during retinal angiogenesis is still largely unknown. It's crucial importance in angiogenesis is underscored by knockout mice that exhibit severe vascular defects 29 . Stenzel et al. proposed that retinal astrocytes represent a major cellular source of fibronectin. Moreover, they provide evidence that its binding to vascular endothelial growth factor (VEGF) is important for retinal angiogenesis 30 . Under pathological conditions, astrocytes of the human glaucomatous optic nerve head display an enhanced fibronectin expression upon transforming growth factor-β2 (TGF-β 2) treatment 31 . Laminin also represents a major component of retinal vascular basement membranes 32 . In addition, high levels of this glycoprotein are associated with the ILM and the GCL. Indeed, laminin plays a key role in RGC survival and reduced expression levels have been associated with glaucoma and optic nerve damage [33][34][35] . Our results indicate a downregulation of α1-laminin mRNA level in the ischemic retina, while a comparable regulation of laminin was observed on protein level. Retinal laminin degradation was previously reported in an I/R mouse model 36 . In this model, laminin-β 1-integrin signaling and activation of the focal adhesion kinase were shown to be essential for the survival of RGCs. In contrast, matrix metalloproteinase-9 upregulation and consecutive laminin degradation lead to decreased levels of β 1-integrin in RGCs and a reduced expression of the pro-survival factor Bcl-xL. Furthermore, agonists that maintain β 1-integrin-activation can prevent RGC death. This was demonstrated in a mouse model of hypoxia-stimulated proliferative retinopathy. Here, a synthetic agonist peptide of the receptor acts protective against retinal ischemia by inhibiting hypoxia-induced neovascularization 37 . In conclusion, retinal downregulation of α1-laminin in our ischemic model is in accordance with previous findings. Nevertheless, we found no alteration on protein level, which might indicate a different regulation of various laminin-chains. Following ischemia, significantly increased endothelial staining was observed in the optic nerve, which seems to correspond to the observed enhanced laminin levels.
Our results revealed a significantly decreased mRNA expression for the glycoprotein tenascin-C in the I/R retina. Although no reduction was found for total tenascin-C protein levels, our Western blot analyses verified a significant downregulation of the 250 kDa tenascin-C band. In the retina, amacrine and horizontal cells are the cellular source of tenascin-C and it is enriched in the plexiform synaptic layers 38 . In general, tenascin-C is a main structural component of synaptic sites 39,40 . In response to ischemic injury, the retina shows signs of structural alterations and neuronal remodeling, also defined as injury-induced plasticity, possibly to preserve or regain some of its neuronal connections 41 . Due to our findings, tenascin-C dysregulation in the synaptic strata might indicate I/R-inflicted damage and synaptic reorganization. Moreover, its downregulation might also reflect the impact on amacrine cells, which was previously reported in this model 19 . Remodeling of tenascin-C was also reported in other tissues following ischemia. For instance, it is dynamically expressed following hepatic or myocardial ischemia/reperfusion injury [42][43][44] . Dysregulation of this glycoprotein was also monitored after cerebral ischemia 45,46 .
In the I/R optic nerve, we could verify a significant increase of tenascin-C staining area. Due to these findings, we assume that optic nerve astrocytes, which represent a main source of this glycoprotein, respond to the ischemic damage 22,47 . With ongoing central nervous system development tenascin-C is progressively downregulated, although pronounced re-expression is monitored following neurodegeneration or injury 48,49 . Previous studies exploring the significance of this glycoprotein demonstrate a dysregulation following glaucomatous damage. In a rat glaucoma model of ocular hypertension, tenascin-C levels were enhanced in the optic nerve head 50 . Elevated levels were also associated with reactive astrocytes in optic nerve heads of primary open-angle glaucoma patients 51 . Here, it was speculated that tenascin-C might act protective to RGC axons by providing a barrier for blood-derived factors that may cause further tissue damage. In this context, it might have a neuroprotective role in ischemic optic nerve tissue. Regarding glaucomatous damage, we also reported an upregulation of tenascin-C in an intraocular pressure-independent autoimmune glaucoma model 52 .
Our analyses revealed a significant larger tenascin-R staining area in the ischemic retina. Via quantitative RT-PCR of mRNA levels no regulation was observed. Also, via Western blotting comparable levels of the total tenascin-R protein were found in both groups. But, in line with the immunohistochemical data, the larger (180 kDa) tenascin-R isoform was significantly upregulated. In contrast, the small (160 kDa) isoform was significantly downregulated following retinal ischemia. These findings strongly indicate an isoform-specific regulation of tenascin-R under ischemic conditions. Both isoforms can be distinguished via their number of fibronectin-type III repeats (8 or 9) and the tendency to form dimers and trimers, respectively, although their significance is not well understood yet 53,54 .
In the retina, tenascin-R is associated with unmyelinated fasciculated RGC axons, although horizontal cells are the major cellular source of its transcripts. Consistently, an enrichment of this extracellular protein is evident in the OPL 55,56 , its functional importance in horizontal cells is not known. Nevertheless, based on the enhanced staining of tenascin-R upon retinal ischemia, horizontal cells seem to react to retinal damage.
In general, tenascin-R represents a well-defined repellent, growth-inhibiting ECM component of optic nerve fibers in several species [57][58][59][60] . In this context, the epidermal growth factor family member CALEB, which is dynamically regulated after optic nerve lesion, represents a favorable interaction partner of tenascin-R during RGC axon regeneration 61 . In the optic nerve, tenascin-R is restricted to the myelinated part. Here, mainly oligodendrocytes contain its transcripts 55 . In addition, spots of increased labeling of this glycoprotein can be found in nodes of Ranvier. Nevertheless, since CO and I/R optic nerves had comparable levels of tenascin-R, we assume that tenascin-R has a minor functional importance in the ischemic optic nerve.
Especially proteoglycans of the lectican family display enhanced expression levels in the ischemic optic nerve. Additionally, we focused on the dysregulation of specific proteoglycans in the I/R retina and optic nerve. Proteoglycans can bind to several other ECM molecules and cell surface receptors and play a pivotal role in CNS, including the retina 62 . Enhanced CSPG levels are associated with pathological conditions in the CNS and represent major constituents of the glial scar. Moreover, CSPGs exert growth-inhibitory effects on axonal regeneration 13 . Aberrant expression of proteoglycans was previously reported in numerous retinal diseases, such as retinitis pigmentosa, age-related macular degeneration and myopia [63][64][65][66] . Yet, little is known about their potential role in the ischemic retina and optic nerve. The upregulation of decorin in a rat ischemia/ reperfusion model indicates its contribution to damage and repair processes in the injured retina 18 . Additionally, Inatani et al. reported an upregulation of the proteoglycan neurocan in the retina following transient ischemia 17 . In a nerve crush model, CSPGs inhibit optic nerve regeneration 67 . Nevertheless, no reports exist regarding the expression of the CSPGs aggrecan, brevican and phosphacan/RPTPβ /ζ in the ischemic retina and optic nerve. In the present study, we provide first evidence for a dysregulation of these CSPGs in a retinal ischemia animal model. Although the investigated CSPGs showed a minor dysregulation in the ischemic retina, a prominent upregulation was observed in the optic nerve, suggesting a re-expression of the investigated proteins following nerve degeneration.
In our study, a significantly increased aggrecan immunoreactivity was found in the ischemic optic nerve, while reduced protein levels were noted in the ischemic retina. Aggrecan expression was previously investigated in a dystrophic rat model 68 . Here, no retinal dysregulation was found in comparison to non-dystrophic rats. As our analyses revealed significantly reduced levels, we speculate that this CSPG is specifically involved in the ischemic retinal degeneration process or the reorganization of the retina.
Nagel and colleagues observed remodeling of aggrecan in individual neurons of a focal cerebral ischemia model, suggesting it plays a role in neuronal reorganization 69 . This might explain the observed upregulation of aggrecan in the ischemic optic nerve. As shown recently, aggrecan inhibits growth of axonal fibers in an optic nerve crush model in vivo 70 . Therefore, under ischemic conditions, aggrecan represents a favorable candidate that inhibits axonal regeneration.
Following retinal I/R, we noted enhanced brevican levels in the optic nerve. Within the myelinated axons of the optic nerve, this lectican family member co-localizes with the ECM molecules phosphacan and tenascin-R at the perinodal Ranvier nodes 71,72 . Regarding these findings and the observed upregulation of brevican following ischemia, we assume that ischemic damage might lead to a reorganization of the nodal matrix. A possible reorganization and associated functional consequences should be investigated in future studies. Our findings indicate a downregulation of brevican mRNA expression levels in the ischemic retina, although no regulation was observed  (473HD antibody, G,H, red), RPTPβ /ζ -isoforms (J,K, red) as well as the nuclear dye TO-PRO-3 (blue) shown as merged images. Small inserts show TO-PRO-3 (blue) and CSPG (green/red) staining separately. The investigated CSPGs showed immunostaining throughout the optic nerve tissue. A dotted-like pattern was observed for 473HD immunostaining. Our evaluation revealed a significant increase in stained area for aggrecan, brevican and 473HD in ischemic optic nerves. No differences were found regarding phosphacan/ RPTPβ /ζ staining area (C,F,I,L). Values are indicated as mean ± SEM ± SD. *p ≤ 0.05; **p ≤ 0.01; n = 5/group. Scale bar = 50 μ m. on protein level. Specifically, reduced levels of brevican and proteolysis were described post hypoxic-ischemic brain injury in the hippocampal matrix 73 . A marked reduction of brevican occurs around a phase of progressive cell death and injury. In a neonatal hypoxic-ischemic injury model, a severe decrease of brevican was observed in the cortex and hippocampus shortly after injury. Markedly, elevated levels were found at later points in time localizing to degenerated cells within and in close association with the lesion core 74 . Interestingly, reduced brevican levels were also noted in the contralesional site of the striatum 75 .
Stronger phosphacan expression was reported in the degenerated optic nerve following partial transection 76 . In addition, we recently verified a transient upregulation of phosphacan in the optic nerve of an experimental autoimmune glaucoma model 52 . This is in agreement with our current findings, which reveal a significant upregulation of phosphacan/RPTPβ /ζ in the ischemic optic nerve. In the optic nerve, glial cells represent the cellular source of phosphacan/RPTPβ /ζ . The increased expression within the ischemic optic nerve might indicate that glia cells respond to this damage. Indeed, after acute ischemia/reperfusion activation of glia cells was reported 19,77 . Contrary to the observed upregulation of phosphacan/RPTPβ /ζ in the optic nerve, we noted a significantly reduced staining area in the ischemic retina. Here, we assume that retinal degeneration and accompanied gliosis precede optic nerve degeneration. In addition, Western blot analyses revealed a significantly decrease of the intermediate protein band after ischemia, which indicates a shift in the expression of RPTPβ /ζ isoforms, proteolytic products or the glycosylation pattern.

Tenascin-C and phosphacan/RPTPβ/ζ show a corresponding expression pattern after ischemic damage.
In nervous tissue phosphacan/RPTPβ /ζ represents a well characterized interaction partner of a variety of cell surface receptors, adhesion molecules, growth factors as well as ECM molecules, including tenascin-C 78,79 . Most importantly, our study revealed that phosphacan/RPTPβ /ζ and tenascin-C exhibit a corresponding expression profile in the investigated ischemic tissues. While tenascin-C and phosphacan/RPTPβ /ζ protein levels were differentially regulated in the ischemic retina, both proteins displayed a significantly increased immunoreactivity within the ischemic optic nerve. Concerning these findings, it is tempting to speculate that possibly divergent signaling of both interaction partners seems to depend on the tissue, cellular source as well as the level and time point of ischemic damage.

Conclusion
In sum, we monitored a contribution of ECM remodeling in an I/R rat model. Our findings suggest that ECM glycoproteins and CSPGs display a unique expression profile and might play a role in ischemic retina and following optic nerve degeneration. Additional studies are necessary to delineate the functional processes underlying I/R injury. Still, our findings offer novel insights how ECM molecules contribute to ischemic damage.   Ischemia/reperfusion (I/R). Ischemia/reperfusion was performed as described previously 19,80 . Briefly, animals were anesthetized using a mixture of ketamine (0.65 ml), xylazine (0.65 ml) and vetranquil (0.2 ml). The right eyes were dilated using 5% tropicamide (Pharma Stulln GmbH, Stulln, Germany) followed by topical anesthesia using conjuncain EDO (Bausch & Lomb GmbH, Berlin, Germany) and a subcutaneous injection of carprofen (0.  Immunohistochemistry and confocal laser scanning microscopy. Eyes were enucleated, fixed in 4% paraformaldehyde (PFA), cryo-protected and embedded in Tissue-Tek freezing medium (Thermo Fisher Scientific, Cheshire, UK). Retinal tissue-sections (10 μ m) were cut using a cryostat (Thermo Fisher Scientific, Walldorf, Germany) and collected onto Superfrost plus object slides (Menzel-Glaeser, Braunschweig, Germany). For immunohistochemistry, retinal cross-sections were dried and rehydrated. Cross-sections were blocked for 1 h at room temperature in blocking solution containing 1% normal goat or donkey serum (both Dianova, Hamburg, Germany), 1% w/v bovine serum albumin (BSA; Sigma-Aldrich) and 0.5% Triton-X-100 (Sigma-Aldrich) in PBS. All primary antibodies were diluted in blocking solution and were applied at room temperature overnight ( Table 2). After washing in PBS, adequate secondary antibodies were applied and incubated at room temperature for 2 h. Fluorescent images (four images per two retinae; three images per three optic nerves) were captured by using a confocal laser-scanning microscope (LSM 510 META; Zeiss, Göttingen, Germany). Laser lines and emission filters were optimized using the Zeiss LSM Image Browser software. Staining signal areas were analyzed using ImageJ software (ImageJ 1.47t, National Institutes of Health, Bethesda, MD, USA) as described previously 52,81 . Briefly, photos were transferred into greyscale pictures. Then, background subtraction and upper and lower threshold were determined for each staining individually ( Table 3). The percentage of each staining was determined and values were transferred to Statistica software (V 12; Statsoft, Tulsa, OK, USA) for statistical evaluation.

SDS-PAGE and Western blotting.
Control and ischemic retinal tissue was homogenized in 200 μ l 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). The protein homogenate was centrifuged at 14.000 x g at 4 °C for 30 min. Afterwards the supernatant was used for determination of protein concentration using a BCA Protein Assay kit (Pierce; Thermo Fisher Scientific, Rockford, IL, USA) following the manufacturer's instructions.
Next, 4x SDS sample buffer was added to each protein sample (20 μ g). Then, samples were denaturized at 95 °C for 5 min and separated by SDS-PAGE (4-10% polyacrylamide gradient gels). Via Western blotting separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes (Roth, Karlsruhe, Germany). Membranes were incubated in blocking solution (5% w/v milk powder in TRIS-buffered saline (TBS) and Tween 20, TBST) at room temperature for 1 h. Primary antibodies were diluted (Table 4) in blocking solution and applied on PVDF membranes overnight. After washing in TBST, appropriate horseradish peroxidase (HRP)-or biotin-coupled secondary antibodies (Table 4) were diluted in blocking solution and applied. Following incubation at room temperature for 1 h, membranes were washed. For protein detection, an ECL Substrate (Bio-Rad Laboratories GmbH, München, Germany) was mixed 1:1, added to the membranes and incubated for 5 min. Afterwards, protein immunoreactivity was documented using a MicroChemi Chemilumiscence Reader (Biostep, Burkhardtsdorf, Germany). Protein intensities were measured using ImageJ software. The intensity of the protein levels was normalized to the reference protein α -tubulin (Table 4). Here, each blot was re-probed.

Statistical analyses.
Immunohistochemical and Western blot data from control and I/R groups were analyzed by using the unpaired Student's t-test and presented as mean ± standard error mean (SEM) ± standard deviation (SD). Data of qRT-PCR were presented as median ± quartile ± minimum ± maximum and analyzed by a pairwise fixed reallocation and randomization test (REST software). For all statistical analyses values of p ≤ 0.05 were considered significant.  Table 4. List of primary antibodies and appropriate secondary antibodies to analyze ECM glycoproteins and CSPGs in the retina of control and ischemic eyes via Western blotting. kDa = kilodalton.