An imbalance in oxidative stress and antioxidant defense mechanisms contributes to the development of ischaemic retinopathies such as diabetic retinopathy and retinopathy of prematurity (ROP). Currently, the therapeutic utility of targeting key transcription factors to restore this imbalance remains to be determined. We postulated that dh404, an activator of nuclear factor erythroid-2 related factor 2 (Nrf2), the master regulator of oxidative stress responses, would attenuate retinal vasculopathy by mechanisms involving protection against oxidative stress-mediated damage to glia. Oxygen-induced retinopathy (OIR) was induced in neonatal C57BL/6J mice by exposure to hyperoxia (phase I) followed by room air (phase II). dh404 (1 mg/kg/every second day) reduced the vaso-obliteration of phase I OIR and neovascularization, vascular leakage and inflammation of phase II OIR. In phase I, the astrocytic template and vascular endothelial growth factor (VEGF) expression necessary for physiological angiogenesis are compromised resulting in vaso-obliteration. These events were attenuated by dh404 and related to dh404’s ability to reduce the hyperoxia-induced increase in reactive oxygen species (ROS) and markers of cell damage as well as boost the Nrf2-responsive antioxidants in cultured astrocytes. In phase II, neovascularization and vascular leakage occurs following gliosis of Müller cells and their subsequent increased production of angiogenic factors. dh404 reduced Müller cell gliosis and vascular leakage in OIR as well as the hypoxia-induced increase in ROS and angiogenic factors with a concomitant increase in Nrf2-responsive antioxidants in cultured Müller cells. In conclusion, agents such as dh404 that reduce oxidative stress and promote antioxidant capacity offer a novel approach to lessen the vascular and glial cell damage that occurs in ischaemic retinopathies.
An imbalance in oxidative stress and antioxidant defense mechanisms contributes to the development of ischaemic retinopathies. However, the therapeutic utility of targeting key transcription factors to restore this imbalance remains to be determined.
Our findings in pre-clinical models of ischaemic retinopathy indicate that dh404, an activator of Nrf2, the master regulator of oxidative stress responses, attenuates retinal vasculopathy by mechanisms involving protection against oxidative stress-mediated damage to glial cells.
The potential significance of our findings is that the bolstering of antioxidant defense mechanisms may provide a powerful approach to attenuate the complex cellular events that lead to vision-threatening retinal diseases such as ROP and diabetic retinopathy.
Retinopathy of prematurity (ROP) and diabetic retinopathy are leading causes of vision impairment for which preventative treatments are limited [1,2]. The hallmark feature of these diseases is damage to the retinal microvasculature, which critically involves the participation of glial and inflammatory cell populations [3,4]. Oxidative stress is a major contributor to vascular and glial cell injury in the retina, arising when there is excessive production of reactive oxygen species (ROS) and suppressed antioxidant defense mechanisms [5,6]. Considerable attention has been given to reducing excess ROS levels in retinopathy [5,6]. However, less explored is the therapeutic utility of targeting key transcription factors involved in bolstering antioxidant defences to restore vascular and glial cell health in retinopathy.
The transcription factor nuclear factor erythroid-2 related factor 2 (Nrf2)/Kelch-like ECH-associated protein 1 (Keap1) pathway is the master regulator of antioxidant responses and is involved in the up-regulation of antioxidants including hemeoxygenase-1 (HO-1), NAD(P)H quinine oxidoreductase 1 (NQO1) as well as enzymes involved in glutathione biosynthesis, such as glutamate-cysteine ligase (GCL) and glutathione synthase (GSS). The increased expression of these genes protects against cellular stressors in tissues [7,8] including the retina [9–11]. The discovery of several small molecules capable of up-regulating Nrf2 responses in cell lines and tissues, including derivatives of the naturally occurring triterpenoid oleanolic acid , led to investigations into their use as potential therapeutics to limit oxidative stress . A potent synthetic derivative of oleanolic acid is 2-cyano-3,12-dioxo-oleana-1,9,-dien-28-oic acid (CDDO, also known as bardoxolone or RTA 401). Subsequent chemical modifications of CDDO resulted in the development of the triterpenoid dh404 (dihydro-CDDO-trifluoroethyl amide), designed to improve efficacy and minimize toxicity [14,15]. We recently demonstrated its strong antioxidant and anti-inflammatory properties in limiting two often-linked diabetic complications, diabetes-associated atherosclerosis and diabetic nephropathy .
ROP can develop in preterm children following exposure to supplemental oxygen to assist breathing, and is reliably modelled in animals using an oxygen-induced retinopathy (OIR) protocol . OIR is a biphasic disease by which exposure of neonates to hyperoxia severely damages the astrocytic template in the retina necessary for physiological growth of the vasculature, and results in regions of vaso-obliteration. The second phase occurs when neonates are exposed to room air, which induces relative ischaemia within the retina and the subsequent production of pro-angiogenic mediators in neuro-glial cells and notably, macroglial Müller cells, and marked pathological neovascularization and vascular leakage . Contributing to retinopathy is inflammation, mediated in part by the resident immunocompetent cell of the retina, microglia [18,19].
Oxidative stress occurs in preterm children and infants with ROP , and promotes damage to astrocytes and Müller cells in the retina [21,22]. Importantly, these two glial cell types are a major source of antioxidants and have a fundamental role in resolving redox imbalance in the retina [9,22,23]. We therefore hypothesized that treatment of OIR using dh404, designed to bind to a specific cysteine residue on Keap1 and thereby induce Nrf2 activation, would lessen hyperoxia phase I-linked vaso-obliteration and improve hypoxia phase II-linked neovascularization and vascular leakage by bolstering the antioxidant capacity of astrocytes and Müller cells.
MATERIALS AND METHODS
An expanded section is available in Supplementary Material.
Animals were obtained from Alfred Medical Research and Education Precinct (AMREP) Animal Services, Melbourne, Victoria, Australia. Experimental procedures were approved by the AMREP Animal Ethics Committee (Applications E/1093/2011 and E/1121/2011). Equal numbers of male and female animals were evaluated in OIR studies and tissue culture experiments.
OIR was induced in C57BL/6J mouse pups according to our previous method [23,24]. Phase I of OIR was implemented by exposing pups at postnatal day (P) 7 with their mothers to 75% O2 ± 5% and 2% CO2 for 22 h/day until P12. Phase II of OIR occurred when mice were placed in room air from P12 until P18. Controls were in room air. To determine if an early or late treatment approach attenuated OIR, dh404 (Reata Pharmaceuticals) was administered every second day between P6 to P18 (early treatment) or P12 to P18 (late treatment) and at a dose of 1 mg/kg [20 μl in 100% DMSO, i.p.(intraperitoneally)] based on our studies showing anti-atherosclerotic and reno-protective effects in diabetic mice . Separate groups of OIR and control mice were untreated and received DMSO. Pups were killed by an i.p. injection of pentobarbital (120 mg/kg body weight; Virbac). As expected, the body weight of pups with OIR was reduced compared with room air controls . dh404 had no effect on body weight (Supplementary Table S1).
Retinal vaso-obliteration and neovascularization in OIR
Retinal wholemounts were prepared as previously described [23,24]. Retinas were stained with FITC-conjugated lectin Griffonia (Bandeiraea) simplicifolia BS-I isolectin (1:100, Sigma) to identify the vasculature. Vaso-obliteration was quantified in phase I OIR at P12 and in phase II OIR at P18 using the freehand line tool in ImageJ (v3.1, National Institutes of Health) and expressed as a percentage of total retinal area. Neovascular tufts were quantified at P18 over the entire retina and expressed as neovascular tufts per retina. Seven to eight mice and three litters per group were evaluated.
Astrocyte morphology and quantification in retina
To determine the association of astrocytes with the vasculature, retina were wholemounted and immunolabelled for glial fibrillary acidic protein (GFAP, 1:200, Dako) and the vasculature distinguished with isolectin as described above. For quantification, six non-overlapping fields spanning the central and mid areas of the retina where vaso-obliteration occurs were captured at ×400 magnification. Co-localized pixels between two immunofluorescent colours were traced using the Autoscan A tool in the image analysis software (AIS v6, analytical imaging software) and expressed as the percentage of dual GFAP and isolectin immunolabelling per field. Five to six mice per group were evaluated.
Retinal vascular leakage
Albumin levels in retina were measured using a mouse albumin ELISA kit (Bethyl Laboratories) and normalized to dry retinal weight . Six mice per group were evaluated.
Immunohistochemistry for Müller cell gliosis and microglia in retina
Three-micrometre paraffin sections of retina were incubated overnight at 4°C with antibodies to GFAP (1:500, Dako) to evaluate Müller cell gliosis and ionized calcium-carbamoyl adaptor protein-1 (Iba1, 1:1000, Wako) to identify microglia  and phosphorylated-ERK1/2 (pERK1/2, 1:400, #4370, Cell Signaling Technology). For quantification, 4 sections at least 60 μm apart were randomly selected from each eye. In each section, four non-overlapping fields spanning the entire retina were captured at ×40 magnification using a Spot digital camera (SciTech). Quantification of GFAP immunolabelling was performed in all layers of the retina and Iba1 immunolabelling in the inner retina (retinal surface to the inner plexiform layer) . Results are expressed as immunolabelling per field of retina. Five to six mice per group were evaluated.
Mice were perfused via the left atrium with PBS to clear non-adherent cells, and then rhodamine-concanavalin A (0.25 mg/kg, Vector Laboratories) to stain adherent cells and the endothelium . Eyes were fixed in 4% paraformaldehyde for 30 min and flatmounted. Non-overlapping images at 20× magnification were captured and the total number of adherent leucocytes counted. Five to six mice per group were evaluated.
Primary culture of rat astrocytes and Müller cells
Astrocytes were isolated from brain cortices of 5-day-old Sprague–Dawley rats and cultured according to previous protocols with minor modifications [26,27]. To mimic phase I OIR, astrocytes were exposed to hyperoxia in a Modular Incubator Chamber (MIC101, QNA International) and the chamber flushed for 4 min with 40% O2, 55% N2 and 5% CO2. The chamber was immediately sealed and placed in a 37°C incubator for 16 h. Müller cells were isolated from 6-day-old Sprague–Dawley rats and cultured as previously described [25,28,29]. To mimic phase II OIR, Müller cells were exposed to hypoxia in a Modular Incubator Chamber and the chamber flushed for 4 min with 0.5% O2, 94.5% N2 and 5% CO2. Astrocytes and Müller cells were treated with 0.1 μM dh404 for 16 and 72 h respectively. Using lactate dehydrogenase  it was determined that this dose of dh404 had no cytotoxic effects (Supplementary Figures S1 and S2). Comparisons were made to cells exposed to normoxia (21% O2) and either hypoxia or hyperoxia [treated with vehicle only (0.001% DMSO)]. Three independent experiments were performed. In each independent experiment, three samples of cells were studied and obtained from the retinas of one litter of pups comprising 10–11 rats. Thus, three separate litters of 30–33 rat pups were used resulting in an n of 9 and three independent experiments.
ROS levels were measured in retina and cultured cells as reported previously . Briefly, unfixed 10 μm cryosections were incubated with sterile PBS containing dihydroethidium (DHE, 5 μM, Sigma) for 30 min at 37°C. The intensity of staining was measured using the ImageJ program. From each animal, three sections were randomly selected and labelling intensity was measured in the inner retina (retinal surface to the end of the inner nuclear layer). Five animals per group were evaluated. Cultured cells were washed with HBSS and stained with 5 μM DHE in PBS at 37°C in 5% CO2 for 30 min. Cells were washed with HBSS and imaged immediately using an Eclipse TE2000 inverted fluorescence microscope (Nikon Instruments) and NIS-Elements software (version 2.20, Nikon Instruments). The fluorescence intensity of cells was measured using ImageJ software. For quantification, at least 15 randomly selected cells from three different fields from each dish were evaluated. The intensity of fluorescence in a blank space between two cells was measured as background intensity. n=9 comprising three samples in three independent experiments.
RNA was extracted from retina and cultured cells using the RNeasy mini kit (Qiagen), and 500 ng of RNA used. mRNA levels were normalized to 18s rRNA endogenous control and the relative fold difference in expression was calculated using the comparative 2−ΔΔCt method .
Protein levels of pro-angiogenic and pro-inflammatory factors
Vascular endothelial growth factor (VEGF) and monocyte chemoattractant protein-1 (MCP-1) were measured in retina by ELISA . Retina were homogenized on ice in 0.01 M sodium phosphate buffer (pH 9.5) containing protease/phosphatase inhibitor cocktail (1:100, Sigma). The total protein concentration was quantified using a colorimetric assay (Bio-rad Laboratories). Undiluted retinal lysates were assayed in duplicate using ELISA kits for mouse VEGF (#DY493, R&D systems) and rat MCP-1 (#555130 BD Biosciences). In cultured cells, rat VEGF protein levels were measured (#DY594, R&D systems).
Western blotting for ERK1/2, GFAP, HO-1, NQO1 and Nrf2
Retinas were isolated and snap frozen in liquid nitrogen until use. Cultured cells were harvested and processed immediately for protein extraction. Whole cell lysates from retinal samples and cultured cells were prepared using RIPA buffer (Sigma) containing 1:100 protease and phosphatase inhibitor cocktail (Sigma) with a homogenizer. Nuclear and cytosolic protein fractions were obtained using the NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo Scientific), according to the manufacturer's instructions. Total protein levels were quantified using the Bio-rad assay (Bio-rad Laboratories). Fifty to 75 μg of protein extracts were fractionated by SDS gel electrophoresis and transferred to PVDF membranes (Bio-rad Laboratories). Membranes were then incubated overnight at 4°C with the following antibodies: anti-ERK1/2 (1:1000, #4395, Cell Signaling Technology), anti-pERK1/2 (1:500, #4370, Cell Signaling Technology), anti-HO-1 (1:500, #ADI-SPA-895, Enzo Life Sciences) for cultured cells, anti-HO-1 (1:500, #ab13243, Abcam) for retinal lysates, anti-NQO1 (1:1000, #ab34172, Abcam), GFAP (1:2000, clone 6F2, Dako) and anti-Nrf2 antibody (1:500, clone D1Z9C, Cell Signaling Technology). Membranes were washed and stained with goat anti-rabbit IgG conjugated with HRP (horseradish peroxidase) for 1 h at room temperature. Membranes were developed using the ECL chemiluminescence kit (Thermo) and a film developer and then washed in TBS overnight and incubated with an anti-β-actin or anti-H3 antibody (1:1000, Cell Signaling Technology) as loading controls. Quantification was performed using the v22 Bio-rad Quantity One 1D analysis software and optical densities expressed as the ratio between corresponding protein and β-actin or H3. The specificity of the Nrf2 antibody has been tested in Nrf2 knockout mouse astrocytes treated with Nrf2 activators and shows high specificity for mouse Nrf2 . n=9 comprising three samples in three independent experiments.
All data were analysed using the GraphPad Prism Software (v.5). Normality was assessed by the Pearson, Shapiro–Wilk and Kolmogoro–Smirnov normality tests. Analyses were performed using a one-way ANOVA followed by appropriate post hoc analysis correcting for the number of comparisons and unpaired t-tests (parametric), or a Kruskal–Wallis test followed by Mann–Whitney U tests (nonparametric). Investigators were masked to the groups. A value of P<0.05 was considered significant.
dh404 reduced retinal vasculopathy in OIR
In room air controls, retinal vascularization was normal at P18 and not affected by early or late treatment with dh404 (Figure 1A). As expected, in untreated OIR mice, vaso-obliteration was present in the central retina and neovascularization in the mid-peripheral retina (Figure 1A). Early and late treatment with dh404 markedly reduced retinal vaso-obliteration (Figures 1A and 1B) and neovascularization (Figures 1A and 1C). Early treatment was more effective than late treatment, and we found a reduction in Nrf2, NQO1 and HO-1 in phase I OIR (Supplementary Figure S3), but all factors were increased in phase II OIR. Of particular note was the increased nuclear accumulation of Nrf2 in phase II OIR, consistent with the known activation and translocation into the nucleus of this transcription factor . These findings suggest that the effectiveness of early dh404 treatment was due to the enhancement of antioxidant genes. Consistent with this idea, dh404 increased NQO1 and HO-1 to control levels in phase I OIR (Supplementary Figure S4). Subsequent studies were performed in mice administered early dh404 treatment.
dh404 reduced retinal vaso-obliteration and neovascularization in phase II of OIR
In OIR, dh404 attenuated hyperoxia-induced vaso-obliteration and restored VEGF levels and contacts between astrocytes and blood vessels
In OIR, retinal vaso-obliteration occurs in response to hyperoxia and is often maximal in phase I, although it persists in phase II . In phase I OIR at P12, hyperoxia caused robust vaso-obliteration of approximately 40%, which was reduced with dh404 (Figures 2A and 2B). VEGF is a survival factor for blood vessels and its suppression during phase I OIR contributes to vaso-obliteration . In phase I OIR, VEGF mRNA and protein levels in retina were reduced compared with room air controls, and restored by dh404 treatment (Figures 2C and 2D). Vaso-obliteration in phase I OIR is also influenced by astrocytes as they provide a template for physiological angiogenesis [33,34]. As expected in untreated room air controls, GFAP immunolabelling revealed astrocytes were star-like shaped cells associated with blood vessels (Figure 2E). High-powered images showed numerous branch-like extensions of astrocyte endfeet closely encircling retinal vessels (Figure 2E). In room air controls, dh404 did not influence the morphology of retinal astrocytes or their contact with blood vessels. However, in phase I OIR at P12, astrocytes appeared reactive with ramified processes and fewer endfeet extensions encircling blood vessels, suggesting that the astrocytic syncytium in the retina was compromised. Treatment of OIR mice with dh404 restored the integrity of the astrocytic template at P12 such that astrocytes were in contact with retinal blood vessels (Figure 2E). Quantification revealed that in OIR mice at P12, the percentage of dual GFAP and isolectin immunolabelling was reduced compared with controls, and increased by dh404 treatment (Figure 2F).
dh404 reduced hyperoxia-induced vaso-obliteration and restored VEGF levels and the contact between astrocytes and the vasculature in phase I OIR
dh404’s protection of astrocytes in hyperoxia is related to a robust antioxidant response
To gain further insight into the protection afforded by dh404 on astrocytes in phase I OIR, we studied primary cultures of rat brain astrocytes. In astrocytes exposed to hyperoxia, ROS levels were increased compared with normoxia controls, and reduced with dh404 treatment (Figures 3A and 3B). Furthermore, dh404 increased the expression of key Nrf2-responsive genes, NQO1 (Figure 3C), GSS (Figure 3D) and HO-1 (Figures 3E and 3F) in both hyperoxia and normoxia. Hyperoxia increased the mRNA levels of cell injury markers such as GFAP (Figure 3G), tumour necrosis factor α (TNFα) (Figure 3H) and inducible nitric oxide synthase (iNOS) (Figure 3I) compared with normoxia untreated controls. dh404 reduced the hyperoxia-induced expression of these factors and had no effect on cells grown under normoxia (Figures 3G–3I).
dh404 prevented the hyperoxia-induced oxidative stress imbalance and cell injury in cultured astrocytes
In OIR, dh404 reduced ROS, pro-angiogenic factors, Müller cell gliosis and vascular leakage
Phase II OIR features damage to Müller cells and increased production of pro-angiogenic factors that stimulate neovascularization and vascular leakage . ROS levels as measured by DHE labelling were increased in all retinal layers compared with room air controls, and quantified in the inner retina where pre-retinal neovascularization develops and Müller cell nuclei reside (Figure 4A). Treatment with dh404 did not alter DHE labelling in the inner retina of room air controls, but reduced labelling in OIR (Figures 4A and 4B). As expected, expression of the pro-angiogenic factors, VEGF and erythropoietin, were increased in OIR compared with room air controls (Figures 4C–4E). Treatment with dh404 did not alter the expression of these factors in room air controls, but effectively reduced their levels in OIR. pERK1/2 is a downstream mediator of these pro-angiogenic factors , and the increase in the ratio of pERK1/2 to total ERK which occurred in the retina in OIR was reduced with dh404 treatment (Figure 4F). These findings were verified with immunohistochemistry for pERK1/2 demonstrating an increase in immunolabelling in Müller cells in OIR compared with controls, which was reduced with dh404 treatment (Figure 4G). Müller cell damage in OIR is reliably measured with GFAP . GFAP immunolabelling is confined to the retinal surface in room air controls; however, in OIR it extended throughout Müller cell processes (Figures 4H and 4I). dh404 treatment had no effect on GFAP immunolabelling in room air controls, but reduced immunolabelling in OIR (Figures 4H and 4I). In addition, vascular leakage was significantly elevated in OIR and reduced by dh404 treatment (Figure 4J).
dh404 reduced ROS and pro-angiogenic mediators and improved gliosis of Müller cells and vascular leakage in OIR
dh404’s protection of hypoxia-induced damage to Müller cells is related to an antioxidant response and accompanied by a reduction in pro-angiogenic and pro-inflammatory factors
To further evaluate the protection afforded by dh404 on Müller cells in phase II OIR, we studied primary cultures of Müller cells. Hypoxia increased ROS as detected by DHE labelling, which was reduced by dh404 (Figures 5A and 5B). Furthermore, dh404 increased the gene expression of Nrf2-responsive genes, NQO1 (Figure 5C), glutamate-cysteine ligase modifier (GCLM) (Figure 5D) and HO-1 (Figures 5E and 5F) in both normoxia and hyperoxia. Hypoxia increased the mRNA levels of GFAP (Figure 5G), VEGF (Figure 5H) and MCP-1 (Figure 5I) compared with normoxia untreated controls and dh404 reduced their expression (Figures 5G–5I).
dh404 reduced the hypoxia-induced oxidative stress imbalance and cell injury in cultured Müller cells
In OIR, dh404 reduced retinal inflammation
Retinal inflammation contributes to the development of OIR [18,25]. We evaluated leucocyte adherence to the vasculature and found few adherent leucocytes in room air controls. However, in OIR, leukostasis was greatly increased (Figures 6A and 6C), and reduced by dh404. Microglia are resident immune cells of the retina, and when activated, release pro-inflammatory cytokines . Microglia can be readily detected with Iba1 immunolabelling (Figure 6B). In untreated and dh404 treated room air controls, few microglia were present in the inner retina. In OIR, Iba1-positive microglia were increased in the inner retina and present in neovascular blood vessels (Figure 6B). In OIR, dh404 reduced Iba1-positive microglia, but immunolabelling remained elevated compared with controls (Figure 6D). In OIR, mRNA levels of the pro-inflammatory mediators, intracellular adhesion molecule-1 (ICAM-1), vascular cellular adhesion molecule-1 (VCAM-1) and TNFα (Figures 6E–6G), as well as the protein levels of MCP-1 in retina were increased compared with room air controls (Figure 6H). Treatment with dh404 effectively reduced ICAM-1, VCAM-1, TNFα and MCP-1 levels in OIR (Figures 6E–6H).
dh404 reduced retinal inflammation in OIR
The principal findings of the present study are that a pharmacological approach to bolster the transcription factor Nrf2, ameliorates oxidative stress imbalance in glial cells and has a substantial protective effect on the pathogenesis of OIR. Specifically, we demonstrated the potential therapeutic utility of an Nrf2 activator to reduce the vision-threatening features of OIR, namely vaso-obliteration, neovascularization and vascular leakage. We uncovered key cellular mechanisms by which this occurred in the hyperoxia-driven phase I of OIR and the hypoxia-mediated phase II of OIR. The OIR model largely recapitulates ROP, a disease of retinal development that in phase I features damage to the astrocytic template necessary to support physiological angiogenesis . Our finding that blood vessels exhibited increased coverage by astrocytes in phase I OIR following dh404 treatment, suggests suppression of excess ROS and an increase in Nrf2-responsive antioxidant genes in astrocytes may contribute to the reduction in vaso-obliteration. In phase II OIR, Müller cells have a key role in promoting neovascularization and vascular leakage by their increased production of pro-angiogenic and pro-permeability factors. The protective effect of dh404 on this vasculopathy was most likely in part due to the prevention of damage to Müller cells via dh404’s rectification of hypoxia-induced oxidative stress imbalance and a reduction in VEGF and erythropoietin. Finally, the ability of dh404 to attenuate retinal leukostasis and reduce Iba1-positive microglia and pro-inflammatory factors highlights the powerful anti-inflammatory actions of pharmacological activation of Nrf2.
Current treatments for ROP such as laser photocoagulation target the neovascular and end-stage of the disease, and although moderately beneficial, are associated with significant sequelae including compromised visual acuity . Hence, preventative treatments that influence the fundamental pathological pathways involved in ROP are desired. In this respect, the observation that the extent of vaso-obliteration in phase I OIR predicts the degree of neovascularization in phase II OIR , indicates that early treatment may be of most benefit. This was the situation with dh404, which although reducing neovascularization when administered at the commencement of the neovascular phase II of OIR, was more efficacious when treatment began at phase I OIR. The effect of early dh404 treatment on vaso-obliteration was particularly striking and confirmed in a separate set of animals evaluated when vaso-obliteration is maximal in phase I OIR. Validation of dh404’s ability to reduce vaso-obliteration and hence promote re-vascularization in phase I OIR was shown with an increase in the mRNA and protein levels of the potent angiogenic factor, VEGF. To our knowledge, this is the first report detailing the utility of pharmacologicaql activators of Nrf2 in OIR. Our findings are in agreement with studies utilizing Nrf2 knockout mice, which exhibited exacerbated vaso-obliteration in OIR . Moreover, of relevance is evidence that Nrf2 has angiogenic capacity by promoting blood vessel sprouting via VEGF's induction of VEGF receptor 2-PI3K/Akt in endothelial cells in the developing retina .
Previous studies evaluating Nrf2 in OIR and retinal development have largely been concerned with Nrf2’s direct effect on the vasculature [11,37]. Here, we explored other mechanisms by which Nrf2 influences vasculopathy by focusing on glial cell populations that have a critical role in supporting the retinal vasculature [3,21]. In retinal development, astrocytes by their physical contact with the vasculature and mediation of the extracellular assembly of matrices provide a template for physiological angiogenesis . In phase I OIR, which temporally coincides with retinal development, hyperoxia results in injury to astrocytes and vaso-obliteration [33,34,39]. Indeed, strategies to restore astrocytes are suggested as an approach to treat OIR . Our finding that in phase I OIR, dh404 prevented the dissociation of astrocyte endfeet from retinal blood vessels indicated that amelioration of oxidative stress imbalance in astrocytes may assist in reducing vaso-obliteration. Although the factors influencing the viability of astrocytes in OIR are not fully understood, oxidative stress is a candidate due to the high rate of oxidative metabolism in ROP resulting from an inadequate ability to scavenge ROS due to low levels of antioxidants . Here, by using cultured astrocytes exposed to hyperoxia we determined that dh404’s up-regulation of Nrf2-responsive antioxidant genes, NQO1, GSS and HO-1, and reduction in ROS, limited oxidative stress and maintained redox homeostasis. These findings, together with dh404’s reduction of the cell injury markers, GFAP, TNFα and iNOS, in astrocytes suggest that amelioration of oxidative stress imbalance via Nrf2 activation is central to the viability of astrocytes.
In phase II OIR, exposure to room air provides the ischaemic stimulus in the retina to promote neovascularization, highly influenced by pro-angiogenic Müller cell-derived factors such as VEGF  and erythropoietin , which exert their effects via activation of ERK1/2 (ERK-1 or p44MAPK and ERK-2 or p42MAPK) . In our study, the reduction in retinal neovascularization following dh404 treatment was accompanied by a decrease in VEGF and erythropoietin and phosphorylation of ERK1/2 in whole retina and Müller cells, highlighting the potent anti-angiogenic effects of dh404 in OIR. A consideration when interpreting these findings is that dh404’s marked reduction of vaso-obliteration in phase I OIR may have had a feed forward effect to blunt the extent of neovascularization in phase II OIR. However, the ability of dh404 to improve oxidative stress imbalance in cultured Müller cells exposed to hypoxia, concomitant with a reduction in cellular injury, as shown for GFAP, as well as lower VEGF expression, suggests that Nrf2 activation has direct protective effects on Müller cells that are relevant to OIR.
Inflammation contributes to ischaemic retinopathies by promoting glial cell and vascular injury [18,19]. The attraction and adhesion of leucocytes to the microvasculature is mediated in part by the increased expression of intercellular and vascular adhesion molecules (ICAM-1 and VCAM-1) and chemokines such as MCP-1 [18,41]. In addition, the increased presence and activation of microglia, the resident immunocompetent cells of the retina, in OIR suggests a heightened inflammatory state. TNFα is of particular importance as its release from inflammatory cells as well as astrocytes may injure neighbouring retinal cell populations and compromise the integrity of the blood retinal barrier [42,43]. Treatment with dh404 restricted the adherence of leucocytes to the vasculature and Iba1-positive microglia in OIR and also lowered pro-inflammatory mediators implicated in OIR pathogenesis. Our experiments in cultured astrocytes and Müller cells under conditions mimicking OIR revealed that dh404 lessened the expression of inflammatory factors in these cell types and therefore attenuated their pro-inflammatory state. Overall, these findings extend the benefits of Nrf2 activation to include dampening of the retinal inflammation that occurs in OIR.
In conclusion, we demonstrated the importance of rectifying Nrf2-mediated oxidative stress imbalance in glial cell populations that have a critical role in the pathogenesis of OIR. These findings do not rule out the possibility that dh404 may have direct protective effects on the retinal vasculature. Our results highlight the potential therapeutic utility of Nrf2 activators to improve some diseases  which may extend to vision-threatening retinal pathologies such as ROP and diabetic retinopathy.
CONFLICT OF INTEREST
Colin Meyer and Keith Ward are employees of Reata Pharmaceuticals, Inc., and report personal fees from Reata Pharmaceuticals, Inc., during the conduct of this study. Judy de Haan reports financial support from Reata Pharmaceuticals, Inc., for this study.
Devy Deliyanti, Judy de Haan and Jennifer Wilkinson-Berka designed the experiments, critically evaluated the results and wrote the manuscript. Devy Deliyanti and Jae Young Lee performed the experiments. Jennifer Wilkinson-Berka prepared the figures. Colin Meyer, Keith Ward and Steven Petratos critically evaluated the manuscript.
This work was supported by the Multiple Sclerosis Research Australia Postgraduate Scholarship (to J.Y.L.); the Trish Multiple Sclerosis Research Foundation [grant number #12-060 (to J.Y.L.)]; the National Multiple Sclerosis Society Project [grant number RG4398A1/1 (to S.P.)]; the Reata Pharmaceuticals, Inc. (to J.B.dH.); and the Victorian Government's Operational Infrastructure Support Program.
glutamate-cysteine ligase modifier
glial fibrillary acidic protein
intracellular adhesion molecule-1
inducible nitric oxide synthase
Kelch-like ECH-associated protein 1
monocyte chemoattractant protein-1
NAD(P)H quinine oxidoreductase 1
nuclear factor erythroid-2 related factor 2
retinopathy of prematurity
reactive oxygen species
tumour necrosis factor α
vaSculars cellular adhesion molecule-1
vascular endothelial growth factor
J.L. Wilkinson-Berka and J.B. de Haan are joint senior authors.