Sustained Upregulation of Endothelial Nox4 Mediates Retinal Vascular Pathology in Type 1 Diabetes

Diabetic retinopathy (DR) is a leading cause of blindness in working-age adults, characterized by progressive damage of retinal blood vessels and neurons resulting in retinal dysfunction, breakdown of blood-retinal barrier, loss of retinal capillaries, and ultimately retinal neovascularization following severe retinal ischemia (1). While the exact molecular mechanisms are not fully understood, increased oxidative stress has been shown to play a central role in vascular injury in DR (2–4). Activation of prooxidative pathways increase reactive oxygen species (ROS) production (5); dysfunction of the antioxidant system further promotes free radical buildup and exacerbates oxidative damage in diabetic retinas (6,7). In diabetic ECs, increased ROS are derived primarily from overactivated NADPH oxidases (Nox) and dysregulated mitochondrial respiration (2,8,9). Activation of Nox initiates mitochondrial damage resulting in a vicious cycle of ROS production in retinal ECs during diabetes (8). Targeting NADPH oxidases may therefore provide an effective approach for prevention and treatment of retinal vasculopathy in DR.

NADPH oxidase 4 (Nox4) is the predominant isoform of Nox in retinal ECs (9–12). Unlike other Nox family members, Nox4 constitutively produces hydrogen peroxide (H₂O₂), and this activity requires a single interaction partner, p22phox (13,14). As such, the only known mechanism controlling Nox4-mediated ROS production is the abundance of the enzyme or direct posttranslational modification (15). Previous work showed that Nox4 was upregulated in retinal ECs under diabetes conditions such as high glucose and hypoxia (9,11,16). Moreover, Nox4 levels were found substantially increased in rodent retinas with type 1 or type 2 diabetes or oxygen-induced retinopathy (9,10,12,17). Silencing Nox4 expression in the retina significantly reduced NADPH oxidase activity, decreased expression of vascular endothelial growth factor, reduced vascular permeability in diabetic mice (9), and mitigated retinal neovascularization formation in oxygen-induced retinopathy (12), suggesting a pivotal role of activation of retinal Nox4 in vascular damage. However, the cell type–specific role of endothelial Nox4 in DR pathogenesis remains elusive.

Using an EC-specific Nox4 knockout (Nox4^EC-KO) mouse model, we recently confirmed a role of endothelial Nox4 in pathological retinal angiogenesis (18). Herein, we seek to understand the implication of endothelial Nox4 in the development of retinal vascular pathology in DR. To this end, we generated a humanized transgenic mouse model that overexpresses human Nox4 in ECs (hNox4^EC-Tg). Using this model and Nox4^EC-KO mice with streptozotocin-induced diabetes, we investigated the impact of sustained Nox4 upregulation in ECs on retinal vascular integrity and pathology under diabetic and nondiabetic conditions in vivo. Furthermore, we explored the role and mechanism of Nox4 upregulation in mitochondrial dysfunction and bioenergetics, oxidative stress, and apoptosis in ECs.

A transgenic mouse model with EC-specific overexpression of human Nox4 gene (hNox4^EC-Tg) on a C57BL/6J background was generated by crossing of lox-Stop-lox-human Nox4 Tg (19) mice with Tie2-Cre mice (stock no. 008863; The Jackson Laboratory) (Fig. 1A). Genotyping was performed with tail DNA by PCR using appropriate primers (Supplementary Table 1). Overexpression of hNox4 in ECs was confirmed by RT-PCR in brain microvascular ECs (BMECs) and retinal microvessels (RMVs) from hNox4^EC-Tg mice. Generation of EC-specific Nox4 knockout (Nox4^EC-KO) mice has previously been described (18). Littermates were used as control. Both male and female mice were included in the study, and sex was balanced across experimental groups. The care, use, and treatment of all animals were approved by the Institutional Animal Care and Use Committees at the University at Buffalo and were in strict agreement with the Statement for the Use of Animals in Ophthalmic and Vision Research from the Association for Research in Vision and Ophthalmology (ARVO).

For induction of diabetes, 6- to 8-week-old mice received daily injections of 50 mg/kg i.p. streptozotocin (Sigma-Aldrich) in sterile 10 mmol/L sodium citrate (pH 4.5) for five consecutive days. Blood glucose levels were measured 1 week after the last injection, and mice with blood glucose >13.9 mmol/L were considered diabetic. Six months after diabetes induction, mice were sacrificed and retinas collected for immunohistochemical study.

Retinal imaging and fluorescein angiography (FA) were performed with MICRON III imaging system (Phoenix) (20). Mice were anesthetized with isoflurane with an anesthesia machine (RWD Life Science), and pupils were dilated. Retinal images were captured immediately after an intraperitoneal injection of 10% sodium fluorescein at 1 mL/kg body wt and continuously for 15 min. Retinal vessel tortuosity was defined as the integral of the curvature square along vessel path normalized by path length and was quantified following a published standard method (21). Five 160 × 120 pixels fields in an FA image were randomly selected in the midperipheral region and skeletonized with ImageJ software. The actual length of each branch and the imaginary straight length between two branch nodes were measured. We calculated vessel tortuosity by dividing the sum of actual branch lengths by the sum of straight lengths between branch nodes. Vessel tortuosity of each retina was obtained in averaging the values from five fields in an FA image. All measurements and quantifications were carried out in a masked manner.

Retinas were fixed in 4% paraformaldehyde, dehydrated in cold methanol, and permeabilized with 0.5% Triton X-100. Retinas were incubated with primary antibodies for 16 h at 4°C and corresponding secondary antibodies (Supplementary Table 4). ECs were labeled with isolectin GS-IB4 (Invitrogen). Numbers of acellular capillaries were counted in at least eight random microscopic fields (430 μm × 320 μm) in midperipheral region at superficial, intermediate, or deep layers. For visualization of IgG extravasation, retinas were incubated with Alexa Fluor 488 goat anti-mouse IgG antibody (Invitrogen) and isolectin GS-IB4. The numbers of leaky vessels were quantified in the whole retina.

Mouse BMECs were isolated and cultured as previously described (18,23,24). Isolated BMECs were cultured in low-glucose (5.5 mmol/L) DMEM with 20% FBS. In the first 48 h, 4 μg/mL puromycin was added to the medium to eliminate nonvascular cells. After 7 days of culture, cells reached 80% confluence and were subjected to desired treatment or analysis. For Seahorse experiments, isolated BMECs were seeded in collagen I–coated Seahorse XFe24 plate. For high-glucose experiment, BMECs from WT mice were treated with d-glucose, 30 mmol/L, or osmotic control (l-glucose, 30 mmol/L) in culture medium containing 5% FBS for 72 h prior to Seahorse analysis.

Primary BMECs were incubated with 5 μmol/L CellROX Deep Red (Invitrogen) in phenol red–free media at 37C° for 20 min. Cells were washed with PBS and imaged with a Zeiss LSM 510 Meta Confocal Microscope. Fluorescence intensities were analyzed with ImageJ software (25). Cells were outlined and mean fluorescence intensity were measured. Background fluorescence intensity was subtracted. For BMECs from each mouse, a minimum of five images was analyzed to generate an average of fluorescence intensity for statistical analysis.

BMECs grown in 35-mm glass-bottom dishes were loaded with 5 μmol/L MitoSOX Red (Invitrogen) in phenol red–free DMEM at 37C° for 20 min. After extensive washes, cells were examined with a Zeiss LSM 510 Meta Confocal Microscope. Fluorescence intensities were analyzed in 5–6 random microscopic fields in BMECs from each mouse with ImageJ software.

Lipid peroxidation was examined with the Click-iT Lipid Peroxidation Imaging Kit (Invitrogen). BMECs grown on glass coverslips were loaded with Clike-iT LAA in DMEM containing 5% FBS at a final concentration of 50 μmol/L for 30 min. Cells were washed with PBS and fixed with 4% paraformaldehyde, followed by permeabilization with 0.5% Triton X-100 and blockade with 1% BSA for 30 min. Five to seven were taken randomly in BMECs from each mouse with an Olympus Provis AX70 fluorescent microscope, and fluorescence intensities were quantified and averaged with ImageJ software.

BMECs grown on glass coverslips were loaded with ThiolTracker Violet (Invitrogen) working solution in Dulbecco’s PBS with calcium and magnesium (Invitrogen) in the 37°C cell culture incubator for 30 min. Cells were washed with PBS, fixed with 4% paraformaldehyde, and imaged with an Olympus Provis AX70 fluorescent microscope. Five to eight random images were captured in one BMEC culture from each mouse and fluorescence intensities were quantified with ImageJ software.

BMECs grown on glass coverslips were fixed and permeabilized with 0.1% Triton X-100. Apoptosis was examined with In Situ Cell Death Detection Kit, TMR red (Roche). Cell nuclei were stained with DAPI in VECTASHIELD mounting medium. Images were taken in 8–10 random fields in one BMEC culture from each mouse. The percentage of apoptotic cells was calculated with the number of TUNEL-positive cells divided by the total number of DAPI-labeled cells in the same microscopic field.

BMECs grown in 35-mm glass-bottom dishes were loaded with 100 nmol/L tetramethylrhodamine, ethyl ester (TMRE) (Invitrogen), in phenol red–free DMEM containing 2% BSA for 15 min. After extensive washes, live cells were photographed with a Zeiss LSM 510 Meta Confocal Microscope. For measurement of mitochondrial membrane potential (ΔΨm) with 5,5,6,6′-tetrachloro-1,1′,3,3′ tetraethylbenzimi-dazoylcarbocyanine iodide (JC-1) method, BMECs were loaded with 5 μmol/L JC-1 (Sigma-Aldrich) in phenol red–free DMEM with 2% BSA for 30 min at 37°C. After extensive washes, live cells in phenol red–free DMEM with 0.25% BSA were photographed with Zeiss LSM 510 Meta Confocal Microscope with two sets of filters to detect rhodamine (540/570 nm) and FITC (488/535 nm). The ratio of red to green fluorescence intensities was measured in six random fields in BMECs from one mouse, and five mice per group were used for the experiment.

BMECs were seeded and grown to 80% confluence on a Seahorse XFe24 plate. Cellular oxygen consumption rate (OCR) was quantified with a Seahorse XFe24 Analyzer (Agilent). Key parameters of mitochondrial respiratory function were measured with a Seahorse XF Cell Mito Stress Test Kit as previously described (26,27). Immediately after Seahorse experiments, cells were trypsinized and cell numbers were counted with Countess II cell counter (Life Technologies). Final results were normalized to the total number of viable cells in each well.

Total RNA was isolated with Trizol (Life Technologies). cDNA was synthesized with iScript cDNA Synthesis Kit (Bio-Rad Laboratories). RT-PCR was performed with primers listed in Supplementary Table 2. The gene products were run on an agarose gel and imaged with ChemiDoc MP Imaging System (Bio-Rad Laboratories).

Total DNA was extracted from BMECs with DNeasy Blood & Tissue Kit (QIAGEN). DNA concentration was measured with Quant-iT dsDNA Assay Kit (Invitrogen). Mitochondrial DNA copy number was assessed by PCR using primers for Cyt b, CO II, and β-actin (Supplementary Table 3).

Quantitative data are expressed as means ± SD. Statistical analyses were performed with unpaired Student t test when comparing two groups and two-way ANOVA for three or more groups. Statistical differences were considered significant at a P value of <0.05.

The data sets and EC-specific Nox4 transgenic mice generated during the current study are available from the corresponding author on reasonable request.

EC-specific human Nox4 transgenic mice (hNox4^EC-Tg, here referred to as TG) was generated as shown in Fig. 1A. Littermate hNox4^−/−, Tie2 cre⁺ mice were used as wild-type (WT) control. We performed RT-PCR to determine hNox4 mRNA expression in BMECs and RMVs derived from WT and TG mice. As expected, hNox4 mRNA was detected only in BMECs from TG mice but not in WT cells (Fig. 1B). Similarly, hNox4 mRNA was only observed in RMVs from TG mice (Fig. 1C). Furthermore, intracellular ROS content (Fig. 1D) in BMECs from TG mice was significantly higher than in WT cells.

To determine the impact of Nox4 overexpression in ECs on retinal vascular development, we analyzed retinal vascular network in WT and TG mice at age of postnatal day 26. We found no significant difference in vessel lengths, vascular junctions, and acellular capillaries (identified as Col IV–positive and IB4-negative vessels) in all three layers of retinal vasculature between WT and TG mice (Supplementary Fig. 1A–D). These findings suggest that the elevated Nox4 expression in ECs does not affect the development of retinal vasculature.

Next, we examined the long-term effect of Nox4 overexpression on retinal vascular integrity in WT and TG mice at ages 10–12 months. Compared with WT mice, TG mice exhibited increased retinal blood vessel tortuosity (Fig. 1E) and vessels with focal leakage (Fig. 1F). The majority of leakage sites appeared to be localized to retinal capillaries in the deep layer of vascular network (Fig. 1F). Numbers of acellular capillaries were also significantly increased (Fig. 2A and B) and vascular densities were reduced in the retina of TG mice (Fig. 2C). Furthermore, pericyte coverage was markedly reduced in retinal vasculature in TG mice (Fig. 2D and E). Together, these changes support that sustained overexpression of Nox4 in ECs leads to vascular damage in the retina.

To explore the mechanism by which overexpression of Nox4 leads to retinal vascular pathology, we determined the effect of Nox4 overexpression on EC apoptosis, mitochondrial ROS generation, and oxidative stress. We found a significantly higher number of TUNEL-positive apoptotic cells in BMECs derived from TG mice (Fig. 3A and E), which was consistent with an increased activation of caspase-3 in RMVs from TG mice (Supplementary Fig. 2). There was a significant increase in mitochondrial superoxide (O₂⁻) in BMECs derived from TG mice (Fig. 3B and F). In contrast, the levels of glutathione (GSH), a major thio compound that regulates cellular redox status, were drastically reduced in Nox4-overexpressing cells (Fig. 3C and G). Lipid peroxidation was also increased in BMECs from TG mice (Fig. 3D and H). These results indicate that overexpression of Nox4 increases mitochondrial ROS generation and oxidative stress resulting in apoptosis of ECs.

To assess whether Nox4 overexpression influences mitochondrial activity, we measured ΔΨm in BMECs from WT or TG mice using TMRE or JC-1 staining. TMRE is a cell-permeable fluorescent probe that labels active mitochondria (28). JC-1 is a membrane potential–dependent dye, which exists as a monomer that fluoresces green when ΔΨm is low and forms J-aggregates that fluoresce red when ΔΨm is high (29). The ratio of red to green fluorescence is considered a more accurate measurement of ΔΨm in different cells in that it is independent of mitochondrial mass (30). Compared with WT controls, BMECs from TG mice demonstrated a significant decrease in TMRE fluorescence (Fig. 4A) and a reduction in red-to-green fluorescence ratio after JC-1 staining (Fig. 4B), suggesting decreased ΔΨm in Nox4-overexpressing ECs.

Since ΔΨm is a central component in oxidative energy metabolism (31), we determined mitochondrial respiratory function and ATP production using Seahorse Extracellular Analyzer. We found that both maximal respiration and spare respiratory capacity of mitochondria were significantly reduced in BMECs from TG mice, while basal respiration and ATP production were not altered compared with WT controls (Fig. 4C and D). Loss of ΔΨm and reduced spare respiratory capacity may contribute to increased apoptosis in Nox4-overexpressing ECs. To further evaluate the impact of Nox4 overexpression on mitochondrial damage, we measured mtDNA copy numbers, which is an important component of overall mitochondrial health. We found no difference in mtDNA copy numbers in Nox4-overexpressing BMECs compared with WT cells (Fig. 4E).

Previously, we showed that Nox4 is upregulated in retinal ECs under diabetic conditions (9). To elucidate whether Nox4 upregulation in ECs contributes to vascular pathology in DR, we examined acellular capillary formation in the retina of Nox4^EC-KO (referred to as KO) and WT mice at 6 months after streptozotocin induction of diabetes. No significant differences were observed in blood glucose and body weight between diabetic WT and KO mice (Supplementary Table 6). As shown in Fig. 5A–C, the numbers of acellular capillaries in diabetic WT mice were significantly increased compared with nondiabetic mice. These changes were largely blunted in diabetic KO mice (Fig. 5A–C). No difference was observed between nondiabetic KO mice and nondiabetic WT mice. Furthermore, retinal vascular permeability was drastically increased in diabetic WT mice but not in diabetic KO mice (Fig. 5D). These results suggest that Nox4 deficiency in ECs attenuates diabetes-induced retinal vascular pathology.

To determine whether Nox4-mediated retinal vascular pathology was attributable to endothelial apoptosis and oxidative stress, we isolated BMECs from 6-month-diabetic and nondiabetic WT and KO mice. We found that the number of apoptotic cells was significantly increased in BMECs from diabetic WT mice compared with nondiabetic mice; the increase in apoptosis was completely abolished in BMECs from diabetic KO mice (Fig. 6A and B). Likewise, mitochondrial superoxide generation was drastically increased in diabetic WT BMECs, which was blunted in diabetic KO cells (Fig. 6C and D). No significant difference in GSH levels was observed between groups, although there was a trend in increase of GSH in KO BMECs (Fig. 6E and F). Consistent with increased mitochondrial ROS, diabetic WT BMECs exhibited significantly higher lipid peroxidation compared with nondiabetic cells, and this change was ameliorated in diabetic KO BMECs (Fig. 6G and H). These findings suggest that increased Nox4 expression contributes to oxidative damage and apoptosis of diabetic ECs.

To explore whether targeting Nox4 could prevent diabetes-induced mitochondrial dysfunction, we measured ΔΨm and mitochondrial respiration function in BMECs from 6-month-diabetic WT and KO mice. As shown in Fig. 7A–D, ΔΨm was dramatically reduced in diabetic WT BMECs compared with nondiabetic mice, which was not observed in diabetic KO BMECs. This finding suggests that deletion of Nox4 prevented diabetes-induced mitochondrial damage in ECs. To further determine the impact of Nox4 downregulation on mitochondrial bioenergetics, we measured OCR in BMECs from WT or KO mice after treatment with high glucose for 72 h. We found that high glucose treatment reduced ATP production but did not alter basal and maximal respiration in WT BMECs (Fig. 7E–H). Interestingly, basal respiration and ATP production were found to be increased in KO BMECs compared with WT cells under normal glucose conditions, and were reduced after high glucose treatment. Taken together, these data suggested that Nox4 is implicated in mitochondrial bioenergetics and contributes to mitochondrial injury in diabetic ECs.

Retinal vascular dysfunction associated with progressive EC injury and cell loss are major characteristics of retinal pathologies in DR; however, the underlying mechanisms have not been fully elucidated. Our current study provides strong evidence supporting a critical role of endothelial Nox4 in DR pathogenesis. Using a humanized EC-specific Nox4 TG mouse line, we have shown that sustained overexpression of Nox4 in ECs leads to retinal vascular abnormalities, including vessel tortuosity, focal leakage, and acellular capillary formation, mimicking pathological changes in DR. These changes were associated with increased ROS production, mitochondrial dysfunction, and increased cell death in ECs. Furthermore, EC-specific knockout of Nox4 decreased mitochondrial ROS generation, alleviated mitochondrial damage, reduced EC apoptosis, and protected retinal blood vessels from damage in an animal model of diabetes. Collectively, our study identified targeting endothelial Nox4 as a potential approach for maintaining mitochondrial function, restoring redox balance, and reducing oxidative damage of retinal blood vessels to prevent DR progression.

Activation of NADPH oxidase and dysfunction of mitochondria are considered major contributors to ROS in diabetic ECs (8). Previous studies have shown that activation of Nox2 in diabetic retina proceeds and initiates mitochondrial damage, which, in turn, results in EC apoptosis (32). Herein, we demonstrate that overexpression of Nox4 in ECs is sufficient to induce mitochondrial damage, indicated by increased mitochondrial superoxide generation, decreased ΔΨm, and impaired mitochondrial respiration. In contrast, knockout of Nox4 gene in ECs prevented diabetes-induced mitochondrial superoxide generation and mitochondrial damage in ECs. These findings indicate a close interplay between Nox4 activation and mitochondrial dysfunction in ECs. In support, previous studies have shown that Nox4-derived ROS reduces the activity of mitochondrial proteins and promotes apoptotic cell death in cardiac myocytes (33,34). In addition, mitochondrial ROS triggers collapse of mitochondrial transmembrane potential, which coincides with a burst of mitochondrial ROS generation, followed by cytochrome c release and caspase activation (35,36). Through inducing mitochondrial dysfunction and ROS generation, sustained Nox4 upregulation contributes to apoptosis and loss of ECs in DR.

Although the exact mechanism by which Nox4 activation induces mitochondrial damage is not fully understood, it is possible that ROS, primarily H₂O₂, generated by Nox4 triggers oxidative damage to mitochondrial proteins, in particular key enzymes in the mitochondrial electron transport chain (ETC) complex (37). This, in turn, results in enhanced ETC uncoupling, impaired mitochondrial respiration, and increased production of superoxide, resulting in a vicious cycle of mitochondrial ROS generation. In addition, overactivation of Nox4 can cause an imbalance between the production of free radicals and their scavenging by the antioxidant defense system. One major system that decomposes H₂O₂ decomposition into O₂ and H₂O is the GSH redox system (38). As a carrier of an active thiol group, GSH directly interacts with reactive oxygen/nitrogen species and is critical for maintaining the redox homeostasis (39). We found that levels of GSH were drastically reduced in Nox4-overexpressing ECs. In addition, Nox4 was shown to inhibit the activation of Nrf2 (40), a master regulator of redox genes including glutamate cysteine ligase catalytic subunit (GCLC), which is an important enzyme in GSH synthesis (39,41). In DR, epigenetic changes, such as altered H3K4 methylation of the antioxidant responsive element 4 (ARE4) region of GCLC gene, reduce the Nrf2 binding at GCLC-ARE4 and decrease GCLC expression, resulting in insufficient GSH production and increased oxidative damage in the mitochondria (42). In addition, inhibition of Nrf2 by Nox4 activation can lead to dysfunction of mitophagy, reduce the capacity of eliminating damaged mitochondria, and exacerbate mitochondrial disruption in cells (43).

Mitochondria are major subcellular machinery for bioenergetics in ECs. Previous work suggests that endogenous Nox4 represses mitochondrial bioenergetics and biogenesis via decreasing the activation of Nrf2 (40). Furthermore, deficiency of Nrf2 impairs mitochondrial respiration by lowering the availability of substrates for complex I (NADH) and II (FADH2) (44). In the current study, we demonstrated that enhanced expression of Nox4 in ECs reduced the maximal respiration and spare respiratory capacity. The defects in spare respiratory capacity may in turn lead to a failure to fulfill the increased energy demand in cells under stress conditions such as diabetes resulting in apoptosis and cell death (45). In addition, high glucose treatment reduced ATP production in ECs, which was prevented in Nox4-deficient cells. Deletion of Nox4 also increased mitochondrial oxygen consumption and reserve capacity in murine and human lung fibroblasts (40). These findings suggest that restoring mitochondrial bioenergetics likely contributes to the beneficial effect of Nox4 inhibition on improving EC survival during diabetes. It should be noted that vascular ECs are in direct contact with blood containing oxygen and nutrients. In diabetic conditions, these cells are often exposed to increased or fluctuating levels of glucose in the circulation. This leads to activation of NADPH oxidase including Nox4 and increased ROS production (9,46), and excess ROS, in turn, compromises the mitochondrial ETC function resulting in reduced ATP production (37). In addition, in diabetic ECs mtDNA undergo hypermethylation, which decreases the transcription of mtDNA-encoded ETC complex genes, contributing to ETC dysfunction (2,47). Notably, microvascular ECs contain twice the volume of mitochondria compared with ECs in large vessels and are highly sensitive to mitochondrial crisis (48,49). Thus, mitochondrial dysfunction contributes to EC and vascular injury in DR.

In summary, results from our study highlight an important role of endothelial Nox4 in mediating diabetes-induced mitochondrial damage, lipid peroxidation, and apoptosis in vascular ECs (Fig. 8). We have shown that sustained upregulation of Nox4 in ECs leads to focal vascular leakage and capillary loss, recapitulating major pathological changes in DR. These findings extend our previous work, where we demonstrated that Nox4 downregulation in the retina reduces oxidative stress and mitigates vascular leakage in diabetic animals (9). In addition, we show that genetic deletion of endothelial Nox4 alleviates diabetic vasculopathy, at least in part, via protecting mitochondria, restoring redox homeostasis, and preventing EC apoptosis. In line with this, a recent genome-wide association study identified a single nucleotide polymorphism in Nox4 gene (rs3913535) and two nearby single nucleotide polymorphisms (rs10765219 and rs11018670) as risk factors for severe nonproliferative DR and proliferative DR in patients with type 2 diabetes (50). Together, these findings suggest a pathogenic role of sustained activation of endothelial Nox4 in advancing vascular damage in DR. However, whether targeting Nox4 could provide beneficial effects in slowing down, stopping, or reversing the progression of vascular damage after DR onset remains to be determined. In addition, the potential contributions of Nox4 activation in nonvascular retinal cells such as Müller cells or RPE cells to DR pathogenesis should be further investigated in vivo.

Acknowledgments. The authors thank Dr. Xiaolan Yi (University of Texas Health Science Center at San Antonio) for her assistance and technical advice on transgenic mice. H.E.A. is deceased.

Funding. This work was supported by National Eye Institute/National Institutes of Health grants EY019949, EY025061, and EY030970 to S.X.Z., research grant NGR G2019302 from the BrightFocus Foundation to S.X.Z. and an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology, the State University of New York at Buffalo.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. X.T. designed and performed the experiments, analyzed data, and wrote the manuscript. J.W. performed the experiments, analyzed data, and revised the manuscript. Y.C. participated in experiments, helped with data analysis, and revised the manuscript. H.E.A. helped with study design and provided lox-Stop-lox-human Nox4 Tg mice. J.J.W. and S.X.Z. conceived and designed the study, analyzed data, and wrote and revised the manuscript. All authors reviewed and approved the final version of the manuscript. S.X.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at Association for Research in Vision & Ophthalmology 2019 Annual Meeting, Vancouver, Canada, April 28–May 2 2019.