SOCS1-Derived Peptide Administered by Eye Drops Prevents Retinal Neuroinflammation and Vascular Leakage in Experimental Diabetes

Current treatments for diabetic retinopathy (DR) target late stages when vision has already been significantly affected. Accumulating evidence suggests that neuroinflammation plays a major role in the pathogenesis of DR, resulting in the disruption of the blood-retinal barrier. Suppressors of cytokine signaling (SOCS) are cytokine-inducible proteins that function as a negative feedback loop regulating cytokine responses. On this basis, the aim of the present study was to evaluate the effect of a SOCS1-derived peptide administered by eye drops (2 weeks) on retinal neuroinflammation and early microvascular abnormalities in a db/db mouse model. In brief, we found that SOCS1-derived peptide significantly reduced glial activation and neural apoptosis induced by diabetes, as well as retinal levels of proinflammatory cytokines. Moreover, a significant improvement of electroretinogram parameters was observed, thus revealing a clear impact of the histological findings on global retinal function. Finally, SOCS1-derived peptide prevented the disruption of the blood-retinal barrier. Overall, our results suggest that topical administration of SOCS1-derived peptide is effective in preventing retinal neuroinflammation and early microvascular impairment. These findings could open up a new strategy for the treatment of early stages of DR.


Introduction
Diabetic retinopathy (DR) remains a leading cause of blindness in developed countries. At present, treatments for DR (i.e., laser photocoagulation, intravitreal injections of anti-VEGF agents or corticosteroids) are administered in advanced stages of the disease and, therefore early therapeutic strategies based on a better understanding of the pathogenesis of DR can be envisaged as a more efficient therapeutic approach [1].
DR has been classically defined as a microvascular complication of diabetes. However, a growing body of evidence indicates that retinal neurodegeneration is an early event in the pathogenesis of

Müller Glial Cells Activation
Glial fibrillar acidic protein (GFAP) expression was mainly confined to the retinal ganglion cell layer (GCL) in non-diabetic mice (control db/+). The diabetic mice treated with vehicle presented significantly higher GFAP expression than non-diabetic mice matched by age (Figure 1). It should be noted that 100% of diabetic mice presented a GFAP score ≥3.

. Müller Glial Cells Activation
Glial fibrillar acidic protein (GFAP) expression was mainly confined to the retinal ganglion cell layer (GCL) in non-diabetic mice (control db/+). The diabetic mice treated with vehicle presented significantly higher GFAP expression than non-diabetic mice matched by age ( Figure 1). It should be noted that 100% of diabetic mice presented a GFAP score ≥3.
The administration of the SOCS1 peptidomimetic resulted in a significant decrease of reactive gliosis, the GFAP score being ≤3 in all cases ( Figure 1).

Microglial Activation
The presence of Iba-1 positive cells in the retina, a marker of activated microglia, was higher in retinas from diabetic mice treated with vehicle than in non-diabetic control mice. In diabetic mice treated with SOCS1 peptidomimetic the number of Iba-1 positive cells was significantly reduced in comparison with diabetic mice treated with vehicle ( Figure 2). The administration of the SOCS1 peptidomimetic resulted in a significant decrease of reactive gliosis, the GFAP score being ≤3 in all cases ( Figure 1).

Microglial Activation
The presence of Iba-1 positive cells in the retina, a marker of activated microglia, was higher in retinas from diabetic mice treated with vehicle than in non-diabetic control mice. In diabetic mice treated with SOCS1 peptidomimetic the number of Iba-1 positive cells was significantly reduced in comparison with diabetic mice treated with vehicle ( Figure 2).

Apoptosis
The percentage of apoptotic cells in diabetic mice was significantly higher in comparison to that observed in retinas from age-matched non-diabetic controls in all retinal layers ( Figure 3).
Diabetic mice treated with SOCS1-derived peptide presented a significantly lower rate of apoptosis than diabetic mice treated with vehicle. This result was observed in all retinal layers. No differences in the percentages of apoptotic cells between diabetic mice treated with SOCS1-derived peptide and non-diabetic mice were observed.

Apoptosis
The percentage of apoptotic cells in diabetic mice was significantly higher in comparison to that observed in retinas from age-matched non-diabetic controls in all retinal layers ( Figure 3).

ERG Abnormalities
The amplitude of a-wave, b-wave and oscillatory potentials (Ops) was significantly lower in diabetic mice treated with vehicle when compared with non-diabetic mice at several flash intensities ( Figure 4). Concurrently, their implicit time significantly increased in diabetic mice when compared with non-diabetic mice at several flash intensities. Treatment with SOCS1 peptidomimetic was able to ameliorate these functional abnormalities induced by diabetes ( Figure 4). Diabetic mice treated with SOCS1-derived peptide presented a significantly lower rate of apoptosis than diabetic mice treated with vehicle. This result was observed in all retinal layers. No differences in the percentages of apoptotic cells between diabetic mice treated with SOCS1-derived peptide and non-diabetic mice were observed.

ERG Abnormalities
The amplitude of a-wave, b-wave and oscillatory potentials (Ops) was significantly lower in diabetic mice treated with vehicle when compared with non-diabetic mice at several flash intensities ( Figure 4). Concurrently, their implicit time significantly increased in diabetic mice when compared with non-diabetic mice at several flash intensities. Treatment with SOCS1 peptidomimetic was able to ameliorate these functional abnormalities induced by diabetes ( Figure 4).   (C) OPs Amplitude and Implicit time in the experimental groups. Results are mean ± SD. * p < 0.05 (db/db treated with vehicle vs. the other groups). # p < 0.05 (db/db treated with vehicle vs. db/db treated with MIS1). (D) Oscillatory potential traces in response to low, medium and high stimulus intensities (10 cd·s·m-2, 40 cd·s·m-2 and 80 cd·s·m-2) in a representative non-diabetic mouse (green), a db/db mouse treated with vehicle (red) and a db/db mouse treated with MIS1 eye-drops (blue).

SOCS1 Treatment Prevents Vascular Leakage
Higher extravasation of albumin was observed in db/db mice treated with vehicle in comparison with non-diabetic animals. Treatment with SOCS1-derived peptide prevented albumin leakage in db/db/mice ( Figure 5A,B). The beneficial effect of the topical treatment with SOCS1 peptidomimetic on vascular leakage was clearly shown by using the Evans Blue method ( Figure 5C). 0.05 (db/db treated with vehicle vs. the other groups). # p <0.05 (db/db treated with vehicle vs. db/db treated with MIS1). (D) Oscillatory potential traces in response to low, medium and high stimulus intensities (10 cd·s·m-2, 40 cd·s·m-2 and 80 cd·s·m-2) in a representative non-diabetic mouse (green), a db/db mouse treated with vehicle (red) and a db/db mouse treated with MIS1 eye-drops (blue).

SOCS1 Treatment Prevents Vascular Leakage
Higher extravasation of albumin was observed in db/db mice treated with vehicle in comparison with non-diabetic animals. Treatment with SOCS1-derived peptide prevented albumin leakage in db/db/mice ( Figure 5A,B). The beneficial effect of the topical treatment with SOCS1 peptidomimetic on vascular leakage was clearly shown by using the Evans Blue method ( Figure 5C).

Glutamate Metabolism
Glutamate levels (µmol/g protein) in the diabetic retinas were higher than in the non-diabetic retinas ( Figure 6A). In diabetic mice treated with SOCS1 peptidomimetic glutamate concentration was lower than in diabetic mice treated with vehicle, but without reaching statistical significance.
immunofluorescence images of vascular permeability assessed by Evans blue dye (red) leakage in retinal whole mounts. Lower panel: Confocal immunofluorescence images with Spec3 by FV1000, Olympus. Scale bars, 20 µm. For quantification 3 mice from each group were analyzed.

Glutamate Metabolism
Glutamate levels (µmol/g protein) in the diabetic retinas were higher than in the non-diabetic retinas ( Figure 6A). In diabetic mice treated with SOCS1 peptidomimetic glutamate concentration was lower than in diabetic mice treated with vehicle, but without reaching statistical significance.
GLAST, the main glutamate transporter expressed by Müller cells, accounts for at least 50% of glutamate uptake in the mammalian retina [23]. GLAST content was downregulated in retinas from diabetic mice treated with vehicle, but in diabetic mice treated with SOCS1 peptidomimetic, GLAST downregulation was prevented ( Figure 6B,C). GLAST, the main glutamate transporter expressed by Müller cells, accounts for at least 50% of glutamate uptake in the mammalian retina [23]. GLAST content was downregulated in retinas from diabetic mice treated with vehicle, but in diabetic mice treated with SOCS1 peptidomimetic, GLAST downregulation was prevented ( Figure 6B,C).

Expression of Proinflammatory Cytokines and VEGF
SOCS1 peptidomimetic treatment was able to prevent the overexpression (mRNA and protein) of IL-1β induced by diabetes ( Figure 7A-C).
Although the reduction in mRNA levels of TNF-α and IL-6 after MIS1 treatment was non-statistically significant, a significant reduction of IL-6 content and perivascular TNF-α deposition was observed in mice treated with SOCS1 peptidomimetic in comparison with vehicle ( Figure 8A-C).
Regarding VEGF, SOCS1 peptidomimetic treatment significantly decreased the overexpression of VEGF induced by diabetes ( Figure 8D).  Although the reduction in mRNA levels of TNF-α and IL-6 after MIS1 treatment was non-statistically significant, a significant reduction of IL-6 content and perivascular TNF-α deposition was observed in mice treated with SOCS1 peptidomimetic in comparison with vehicle ( Figure 8A-C).
Regarding VEGF, SOCS1 peptidomimetic treatment significantly decreased the overexpression of VEGF induced by diabetes ( Figure 8D).

SOCS1 Peptide Suppresses STAT Activation and Downstream Gene Expression in Human Retinal Pigment Epithelial Cells
To confirm the in vivo results, we analyzed the impact of SOCS1 peptide in ARPE-19 cells exposed to conditions mimicking the diabetic milieu. Confocal microscopy studies revealed that SOCS1 peptide prevented the phosphorylation and nuclear translocation of STAT1 and STAT3 proteins induced by IFNγ ( Figure 9A) and TNFα ( Figure 9A). Real-time PCR demonstrated that ARPE-19 cells responded to either cytokines or hyperglycemia by upregulating proinflammatory genes such as cytokines (IL-1β and IL-6) and chemokines (CCL2, CCL5) as well as VEGF, an established inducer of vascular leakage ( Figure 9C-E). Furthermore, pretreatment of cells with SOCS1 peptide resulted in a significant reduction in the mRNA expression of those genes ( Figure  9C-E).

SOCS1 Peptide Suppresses STAT Activation and Downstream Gene Expression in Human Retinal Pigment Epithelial Cells
To confirm the in vivo results, we analyzed the impact of SOCS1 peptide in ARPE-19 cells exposed to conditions mimicking the diabetic milieu. Confocal microscopy studies revealed that SOCS1 peptide prevented the phosphorylation and nuclear translocation of STAT1 and STAT3 proteins induced by IFNγ ( Figure 9A) and TNFα ( Figure 9A). Real-time PCR demonstrated that ARPE-19 cells responded to either cytokines or hyperglycemia by upregulating proinflammatory genes such as cytokines (IL-1β and IL-6) and chemokines (CCL2, CCL5) as well as VEGF, an established inducer of vascular leakage ( Figure 9C-E). Furthermore, pretreatment of cells with SOCS1 peptide resulted in a significant reduction in the mRNA expression of those genes ( Figure 9C-E).

Discussion
In the present study we provide evidence that a SOCS1 peptidomimetic, administered by eye drops, has dual beneficial effects in terms of neuroprotection and the inhibition of vascular leakage. The neuroprotective action consisted of a significant reduction of both glial (macroglia and microglia) activation and apoptosis. Moreover, a significant improvement of electroretinogram (ERG) parameters was observed, thus revealing the clear impact of the histological findings on global retinal function. The inhibition of vascular leakage was clearly demonstrated by the SOCS1 peptidomimetic because the disruption of the BRB, which is an essential step in the development of diabetic macular edema, was prevented.
Many proinflammatory cytokines such as IL-1β, IL-6, IL-8, MCP-1, and TNF-α are upregulated in the diabetic retina [12,13]. It should be noted that the mean levels of IL-8 and MCP-1 within the vitreous fluid of diabetic patients have been found in the same range as those reported in pleural effusions of patients with pneumonia or tuberculosis and they correlated with proliferative DR (PDR) activity [24]. In addition, the increased vitreous levels of IL-6 and IL-8 correlated with the progression of PDR and the outcome of vitreous-retinal surgery [25]. Furthermore, the activation of the JAK/STAT signaling pathway by means of proinflammatory cytokines can lead to angiogenesis either by triggering angiogenic factors, such as VEGF, or through alternative pathways [26][27][28]. These findings underscore inflammation as crucial in the pathogenic events that lead to both early and advanced stages of DR.
The present results define a hitherto unknown therapeutic effect of SOCS1 on the diabetic retina, and are in good agreement with our previous reports on the efficacy of SOCS1 to prevent the onset and progression of diabetic nephropathy, a microvascular complication of diabetes that shares common pathogenic mechanisms with DR [20,21]. Thus, we found that intrarenal delivery of adenovirus expressing SOCS1 to diabetic rats significantly improved renal function and reduced renal lesions associated with diabetes, such as mesangial expansion, fibrosis, and the influx of macrophages [20]. Furthermore, we also demonstrated that the intraperitoneal administration of the SOCS1 peptidomimetic herein tested in an experimental mouse model of diabetes ameliorated STAT activity and resulted in reduced serum creatinine levels, albuminuria, and improvement of renal histologic changes over time. Notably, mice treated with the SOCS1 peptidomimetic also exhibited reduced leukocyte levels in the kidney and decreased expression levels of proinflammatory and profibrotic markers that were independent of glycemic and lipid changes [21]. SOCS1 inhibits lymphocyte recruitment into the retina and rats and mice with targeted over-expression of SOCS1 in the retina are partially protected from experimental autoimmune uveitis [29]. Furthermore, it has been shown that topical administration of a SOCS1 peptidomimetic suppresses uveitis and confers protection from ocular pathology during experimental autoimmune uveitis [30]. These beneficial effects were produced through direct inhibition of immunopathology mediated by T lymphocytes and by protecting resident ocular cells from apoptosis during chronic intraocular inflammation [31].
As commented, inflammation plays a major role in the pathogenesis of DR [3,[10][11][12]. In fact, corticosteroids (intravitreal injections or implants) are indicated in advanced stages of DME. However, when the early stages of DR are the therapeutic target, it would be inconceivable to recommend an aggressive treatment such as intravitreal injections. In this regard, it has been reported that topical administration of corticosteroids and NSAIDs could be useful for treating DME [32,33]. However, a lack of effect in reducing retinal thickness after one year of topical administration of nepafenac was also reported [34].
Besides being a non-invasive route, the advantage of topical administration is that it ensures low systemic absorption, thus minimizing the associated systemic effects. In this regard, the negligible level of systemic absorption of a SOCS1 peptide after its topical administration has been demonstrated by the absence of inhibitory effects on peripheral immune responses [31].
Regarding the mechanisms involved in the beneficial effect of the SOCS1 peptidomimetic, we found that it reduces the upregulation of IL-1β, IL-6, TNF-α, and VEGF induced by diabetes. It should be noted that IL-1β is a pivotal inflammatory cytokine because it is able to activate nuclear factor-κB, the transcription factor which in turn governs the production of other proinflammatory cytokines [35] and plays a significant role in the degeneration of retinal capillaries induced by diabetes [36]. TNF-α downregulates tight junction proteins of endothelial cells and it is also required for VEGF-induced leakage, thus participating in the breakdown of the BRB, which is the main pathogenic factor of DME [37]. We found that TNF-α expression in diabetic mice treated with vehicle was mainly detected in the perivascular areas of the inner retina. Supporting this finding, it has been reported that in the early stages of DR, advanced glucose end products (AGEs) stimulate perivascular microglial cells to produce TNF-α [38]. It has recently been described how TNFα alone induces small-molecule permeability of the BRB in vitro, whereas the combination of TNFα, IL-1β, and VEGF induces permeability to large molecules [39]. In addition, we found that SOCS1 peptidomimetic suppressed STAT activation in RPE cells, thus downregulating the expression of IL-1β, IL-6, CCL2, and CCL5, as well as the expression of VEGF. The combination of anti-inflammatory and anti-VEGF action of SOCS1 peptidomimetic observed in the present study is an extra-value in targeting two essential events in the pathogenesis of DR. However, the design of our study does not permit us to know whether the effect on VEGF is directly mediated by the peptidomimetic SOCS1 or by the downregulation of proinflammatory cytokines.
The extracellular accumulation of glutamate plays a key role in the excitotoxicity, which plays a crucial role in neuron death induced by diabetes [2]. GLAST is the main transporter of extracellular glutamate and, therefore, GLAST downregulation induced by diabetes results in high extracellular glutamate levels, thus contributing to neurodegeneration [2]. In the present study, we provide evidence that SOCS1 avoids the increase in retinal glutamate levels and that this is associated with a dramatic inhibition of diabetes induced GLAST downregulation. Therefore, the preservation of GLAST seems one of the main mechanisms by which SOCS1 peptidomimetic decreases the levels of extracellular glutamate. It is worth mentioning that we found that a GLAST mediated reduction of glutamate accumulation is a common mechanism of drugs with associated anti-inflammatory activity [5,6,8,40]. In this regard, inflammatory cytokines such us TNF-α have been shown to interact with glutamate pathways in several important ways, including decreasing the expression of glutamate transporters on relevant glial elements and increasing the release of glutamate from astrocytes in the brain [41]. Therefore, the inhibition of TNF-α induced by SOCS1 peptidomimetic could also participate in reducing extracellular glutamate levels. However, specific experiments to confirm this hypothesis are needed.
The primary neuroinflammatory action of SOCS1 peptidomimetic suggests that it could be more effective when inflammation is a predominant event in the pathogenesis of DR. Recently there has been an increasing interest in the determination and validation of non-invasive imaging retinal parameters, as possible biomarkers of a local retinal "inflammatory condition" in DR and DME [42]. The most important imaging modalities have been spectral domain (SD)-OCT and fundus autofluorescence (FA). These imaging biomarkers include: subfoveal neuroretinal detachment and hyperreflective retinal spots/foci evaluated on SD-OCT, and foveal hyperautofluorescence evaluated on FA. These methods could be useful for identifying a subset of the diabetic population in which inflammation plays an essential pathogenic role in DR. In addition, they could be useful in prognosis and in predicting treatment response. This is important in the era of personalized or precision medicine.
The present study has several limiting factors such as the lack of dose-efficacy studies, formulation stability analyses, and specific assessments addressed to evaluate potential ocular and systemic toxicity. Therefore, further research aimed at evaluating these important issues is needed. In conclusion, our results suggest that blocking retinal neuroinflammation by topical administration of SOCS1-derived peptide could be a new strategy for the treatment of DR. However, specific clinical trials aimed at evaluating its safety and effectiveness are needed.

Animals
The neuroprotective effect of SOCS1 peptidomimetic was tested in a db/db mouse model. This mouse carries a mutation in the leptin receptor gene and is a model for obesity-induced type 2 diabetes.
Our group previously reported that the db/db mouse reproduces the features of the neurodegenerative process that occurs in the human diabetic eye [26] and has been used for testing the capacity of several drugs in reducing neuroinflammation and vascular leakage [6,8,40].
A total of 20 male db/db (BKS.Cg-Dock7m +/+ Leprdb/J) mice aged 8 weeks were purchased from Charles River Laboratories, Inc. In addition, 10 non-diabetic db/+ (BKS.Cg-Dock7m + Leprdb/+) mice matched by age were used as a control group. The animals were randomly housed under tight environmental conditions of humidity (60%), temperature (20 • C) and cycles of 12 h/12 h light/darkness. They had free access to filtered water and "ad libitum" food. SOCS1 peptidomimetic (10 mg/mL; 5 µL twice/daily) (n = 10) or vehicle (PBS; 5 µL twice/daily) (n = 10) eye drops were administered directly onto the superior corneal surface of each eye using a micropipette in 8 weeks-old mice. Ten non-diabetic mice matched by age served as the control group. The treatment (MIS1 or vehicle) was administered twice daily for 15 days. On day 15, the drop of MIS1 or vehicle was administered approximately two hours prior to necropsy. Mice were euthanized by cervical dislocation and the eyes were immediately enucleated. The dose of SOCS1 was selected based on our previous experience using the intraperitoneal route for treating diabetic nephropathy [21].
For the Evans Blue assay, a set of animals (n = 3 per group) were intravenously injected in the tail with a solution of Evans Blue (E2129 SIGMA) (17 mg/Kg body weight, in concentration 5 mg/mL dissolved in saline solution sterile). Immediately after injection, the animals visibly turned blue, confirming dye uptake and distribution. After 120 min, the mice were euthanized by cervical dislocation and the eyes were enucleated. The retinas of each animal were isolated, weighed and rapidly protected from light. Flat-mounted slides were obtained, and cover slipped with a drop of mounting medium Prolong Gold antifade reagent (Invitrogen, Carlsbad, CA, USA; Thermo Fisher Scientific, Waltham, MA, USA). Digital images from different random fields of all retinas were acquired using a confocal laser scanning microscope (FV1000; Olympus, Hamburg, Germany) at ×20 using the 561-nm laser line, and each image was recorded with identical beam intensity at the size of 1024 pixels × 1024 pixels. For quantitative analysis of the albumin-bound Evans Blue, Z-stacks retinal images (step size 1.16 µm) of different regions of the vascular tree were acquired. To remove the eye artefacts caused by the sample procedures, projections of the middle Z-plane images were analyzed with the magic wand tool (tolerance 120, eight pixels connected) of Fiji software (https://fiji.sc).
This study was approved by the Animal Care and Use Committee of VHIR (Vall d'Hebron Research Institute, CEEA 75/15, September 2015)). All the experiments were performed in accordance with the tenets of the European Community (86/609/CEE) and ARVO (Association for Research in Vision and Ophthalmology).

Electroretinogram
Full field electroretinography (ERG) recordings were measured using the Ganzfeld ERG platform (Phoenix Research Laboratories, Pleasanton, CA, USA) following ISCEV (International Society for Clinical Electrophysiology of Vision) recommendations [43]. Three ERG components were assessed in terms of amplitude and timing: a-wave, b-wave, and oscillatory potentials (OPs). We added up OPs amplitudes (ΣOPs amplitude) and implicit time (ΣOPs implicit time) for the first 5 OPs.

Immunofluorescence Analysis
For immunohistochemical analysis, animals were intraperitoneally injected with 0.2 mL of anesthesia (proportion of solution was 1 mL ketamine and 0.3 mL xylazine) and transcardially perfused with p-formaldehyde 4%. The ocular globes were immediately enucleated, fixed in p-formaldehyde 4% for 6 h and embedded in paraffin blocks. The paraffin blocks were cut along the eye axis with a microtome and the sections of 4 µm were mounted on positive charged slides and stored at 4 • C. The sections used for immunofluorescence analyses (6 sections per retina) were obtained from central retina. Paraffin-embedded sections were deparaffinized in xylene and rehydrated in a graded ethanol series. The sections were fixed in ice-cold acid methanol for 1 minute and washed with 0.01 M phosphate buffered saline (PBS) at pH 7.4. Subsequently, antigen retrieval was performed. The sections were immersed in an antigen retrieval solution (sodium citrate 10 mM, pH 6.0) and heated in a pressure cooker at 150 • C for 4 min. They were blocked by the corresponding blocking solution for 1 h at room temperature and were incubated overnight at 4 • C with the corresponding primary antibodies described in Table 1. Next day, they were washed twice in PBS, and incubated for 1 h in darkness at room temperature along with secondary antibodies sections (Alexa 488 or Alexa 594; 1:600, Molecular Probes (Eugene, OR, USA). The sections were washed in PBS, counterstained with Hoechst (1:500) and coverslipped using a mounting solution (Prolong Gold antifade reagent Invitrogen, Thermo Fisher Scientific). Images were acquired with a confocal laser scanning microscope (FV1000; Olympus). Five fields (three corresponding to the central and two to the peripheral retina) from each section were analyzed. The same locations and number of fields were measured in all retinas. The fluorescence intensity of the images was quantified by ImageJ.

Analysis for Glial Activation
Glial activation was evaluated by fluorescence microscopy using specific antibodies against GFAP (Glial fibrillar acidic protein). To evaluate the degree of glial activation, we used a scoring system based on the extent of GFAP staining [44] and previously used by our group [26] (Table 2).
Microglial activation was evaluated based on Iba-1 staining and analyzed by a semiquantitative score. Semiquantitative assessment of microglial activation in the retina: (-) absence of positive cells for Iba-1/power field (20×); (+) scattered, 1-3 cell/power field (20x); (++) moderate, 4-10 cells/ power field (20×). Müller cell endfeet region/GCL plus a few proximal processes 3 Müller cell endfeet plus many processes, but not extending to ONL 4 Müller cell endfeet plus processes throughout with some in the ONL 5 Müller cell endfeet plus lots of dark processes from GCL to outer margin of ONL

Apoptosis Assessment
Apoptosis was evaluated using the TUNEL (Terminal Transferase dUTP Nick-End Labeling) Dead-end Fluorometric System method (Promega Corporation, Madison, WI, USA). Briefly, sections of retina were permeabilized by incubation at room temperature for 1 min with 20 µg/mL Proteinase K solution, freshly prepared. The sections were washed in PBS, counterstained with Hoechst (1:500) and cover slipped using a mounting solution (Prolong Gold antifade reagent Invitrogen, Thermo Fisher Scientific). Apoptotic cells were identified by green fluorescein -12-dUTP -labeled DNA with 488 nm as excitation wavelength and the range for detecting positive cells in the confocal laser scanning microscope was 515-565 nm (green). Lastly, the number of green positive cells and total cells (in blue) was counted.

Measurement of Retinal Vascular Permeability
Retinal vascular permeability was examined by assessing the albumin leakage from blood vessels into the retina by both albumin immunofluorescence analyses and the Evans blue-albumin method (ex-vivo). The Evans blue-albumin method was used as previously described with some modifications [45][46][47]. Digital images from different random fields of all retina were acquired with a confocal laser scanning microscope (FV1000, Olympus. Hamburg, Germany) at 20× using the 561 nm laser line and each image was recorded with identical beam intensity at a size of 1024 × 1024 pixels.

Measurements of Cytokine Expression
Total RNAs were extracted from retinas using Trizol reagent (Invitrogen) according to the manufacturer's instructions. The quality and concentration of RNAs were determined with a Nanochip (Agilent Technologies, Madrid, Spain) and a Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific). cDNA reactions were performed in a 2720 Thermal Cycler with High Capacity kit reagents (Applied Biosystems, Foster City, CA, USA). Real-Time PCRs were carried out using SYBR Green PCR Master Mix (Applied Biosystems) in the 7.900 HT Sequence Detection System with 384-well optical plates. Each sample was assayed in duplicate, and negative controls were included.

Statistical Analysis
The results are expressed as mean ± SD. Statistical comparisons were performed with Student s unpaired t test. When multiple comparisons were performed, one-way ANOVA followed by Bonferroni's test was used. The Fisher's exact test was used to analyze categorical variables. Levels of statistical significance were set at p <0.05.

OPs
Oscillatory potentials SOCS Suppressors of cytokine signaling STATs Signal transducers and activators of transcription TUNEL Terminal Transferase dUTP Nick-End Labeling VEGF Vascular endothelial growth factor