Resveratrol reverses the adverse effects of bevacizumab on cultured ARPE-19 cells

Age-related macular degeneration (AMD) and proliferative diabetic retinopathy (PDR) are one of the major causes of blindness caused by neo-vascular changes in the retina. Intravitreal anti-VEGF injections are widely used in the treatment of wet-AMD and PDR. A significant percentage of treated patients have complications of repeated injections. Resveratrol (RES) is a polyphenol phytoalexin with anti-oxidative, anti-inflammatory and anti-proliferative properties. Hence, we hypothesized that if RES is used in combination with bevacizumab (BEV, anti-VEGF), it could reverse the adverse effects that precipitate fibrotic changes, drusen formation, tractional retinal detachment and so on. Human retinal pigment epithelial cells were treated with various combinations of BEV and RES. There was partial reduction in secreted VEGF levels compared to untreated controls. Epithelial-mesenchymal transition was lower in BEV + RES treated cultures compared to BEV treated cultures. The proliferation status was similar in BEV + RES as well as BEV treated cultures both groups. Phagocytosis was enhanced in the presence of BEV + RES compared to BEV. Furthermore, we observed that notch signaling was involved in reversing the adverse effects of BEV. This study paves way for a combinatorial strategy to treat as well as prevent adverse effects of therapy in patients with wet AMD and PDR.

Clinical dosage of BEV has been shown to mildly reduce proliferation, and with a higher concentration or with high glucose levels, it caused cytotoxicity in cultured RPE cells [17][18][19] . Clinical dosage of BEV upregulates CTGF leading to pro-fibrotic changes with increased loss of epithelial properties in cultured RPE cells resulting in induction of epithelial-mesenchymal transition (EMT) 20 . We have previously shown that a short exposure of clinical concentration of BEV in cultured human RPE cells reduces cell proliferation and phagocytosis with increased epithelial-mesenchymal transition (EMT) and transmembrane potential 7 . Results from animal and clinical studies have revealed the most complications of BEV treatment are vitreous hemorrhage, tractional retinal detachment, fibrotic membrane formation and retinal pigment epithelial tears 21,22,7,10 . There are also reports on macular atrophy occurring after repeated injections of anti-VEGF for wet AMD 23 . Clinical trials like ANCHOR, MARINA and CATT study have reported that 8-10% of patients on treatment with anti-VEGF agents develop dry AMD like phenotype with geographic atrophy [24][25][26][27] . Moreover, despite adequate treatment, there remains a cohort of ~40% and ~45% anti-VEGF non-responders with PDR and AMD respectively 28,29 . The above factors necessitate the need for alternatives as well as combinatorial therapy without compromising treatment efficacy.
We investigated the influence of RES, a stilbenoid natural polyphenol phytoalexin, as a potential protective agent. It is found in the skin of grapes, berries and peanuts and exerts its anti-oxidant, anti-inflammatory, anti-epithelial-mesenchymal transition and anti-proliferative roles through sirtuin 1 30,31 . RES has been used in the treatment of diabetic retinopathy and dry AMD due to its anti-angiogenic and enhanced phagocytic properties, respectively 32 . In a cell culture model RES inhibited EMT induced by TGF-β, thereby restoring the ZO-1 and α-SMA staining and reducing the expression of mesenchymal marker vimentin by suppressing Smad2 and Smad3 phosphorylation 33 . Studies have shown that impaired autophagy, a major driver for AMD can be restored in the presence of Resvega suggesting a plausible therapeutic window for treating AMD 34 . By regulating PCNA, p21, p27 and p38MAPK/MMP-9 expression, RES can block proliferation and migration in ARPE-19 cells 35 . Administration of dietary RES reduces inflammatory, senescence and oxidative stress markers in trabecular meshwork cells exposed to 40% O2 36 . Retinal ganglion cell death has been shown to be prevented by a dietary supplement of RES in an optic nerve crush mice model [37][38][39] . It has also been shown that RES, through activation of SIRT1, downregulated IL-17 in mononuclear cultures of PVR patients suggesting a protective mechanism of RES in DR progression 40 . Oxidative stress induced lens epithelial cultures were protected from apoptosis by SIRT1 inhibition of p53 pathway 41 . In cultured RPE cells, RES caused induction of mesenchymal-epithelial transition besides blocking TGF-β induced EMT, proliferation and migration by activation of SIRT1 31 . RES reduces the inflammatory response induced by endotoxins in uveitis and is effective in controlling retinopathy of prematurity by regulating nitric oxide levels in animal models 30,42 . RES can restore RPE phagocytosis property in UV-exposed RPE cultures through ion channels 43 . Apart from the activation of SIRT1, RES acts through the Smad2, Smad4 and PI3K/Akt/mTOR pathways 44 . RES has also been shown to activate Notch signaling in both ocular and non-ocular cells 45,46 .
Notch signaling is a developmentally regulated, conserved signaling pathway involved in angiogenesis and VEGF regulation, apart from its role in cell fate decision, cell-to-cell communication, proliferation and apoptosis. There are several studies implicating cross-talk between notch signaling and VEGF 47 . The stringency in the culturing conditions of RPE also has been shown to be decisive with respect to the expression of Notch signaling. We have shown that RPE cells cultured on de-epithelialized amniotic membrane down regulated Notch signaling when compared to controls 8 . We have previously depicted that a short pulse of BEV is sufficient enough to dampen Notch signaling activity, which plays an important role in epithelial-mesenchymal transition, proliferation, membrane potential and phagocytic properties of RPE 7 . Members of Notch signaling pathway, Notch 1, 4 and Delta like 4 play crucial roles during vasculogenesis and angiogenesis 48,49 .
In order to exploit the protective potential of RES, we investigated the response of human RPE cultures to RES and BEV alone and in combination. It's role in mitigating the adverse effects of anti-VEGF agents and on other properties of RPE cells such as secretion of VEGF, phagocytosis, proliferation, apoptosis, EMT induction and membrane potential have been studied.

VEGF expression and secretion patterns in RES and/or BEV treated ARPE-19 cells.
Gene expression of vegf a (p = 0.007), vegf r1 (p = 0.003) and vegf r2 (p = 0.002) was significantly decreased in cells treated with BEV when compared to untreated controls. There was no significant difference in the mRNA levels of vegf a, vegf r1 and vegf r2 between RES, BEV + RES and control groups. There was a significant difference in the mRNA levels of vegf a (p = 0.008), vegf r1 (p = 0.002), vegf r2 (p = 0.008) in BEV + RES treated cells when compared to BEV alone (Fig. 1A). We further looked at the cellular regulation of VEGF A, VEGF R1 and VEGF R2 in human RPE cells treated with BEV and RES alone or in combination. The results of the mean fluorescent intensity revealed a significant decrease in the levels of VEGF immunofluorescence staining in cells treated with BEV in comparison to untreated controls (p = 0.0002) and BEV + RES (p = 0.0002) treated cells ( Fig. 1B and C). Apart from the cellular VEGF levels, we looked at the secreted VEGF form in the cell culture supernatant. While controls showed the highest levels of VEGF compared to the treated cells. BEV neutralized the secreted VEGF whereas RES and RES + BEV treated cells partially neutralized the secreted VEGF levels. Secreted VEGF levels were significantly higher (p = 0.0004) in BEV + RES treated cultures compared to BEV alone treated cultures (Fig. 1D). Western blot analysis revealed results similar to that of immunofluorescence staining. BEV treated cells showed significantly lower amounts of VEGF A and VEGF R2 than the untreated controls (p = 0.0003; p = 0.002) and BEV + RES (p = 0.002; p = 0.002) treated cells, respectively ( Fig. 1E and F).

Epithelial-mesenchymal transition (EMT) is influenced by BEV and RES. Expression of EMT
genes was studied in untreated and treated (BEV, RES, BEV + RES) cells Epithelial markers (e-cadherin), mesenchymal markers (vimentin, fibronectin, a-smooth muscle actin) and extracellular matrix markers (collagen I and collagen IV) were used to investigate the effect of BEV and RES on EMT.Gene expression profile of coll I, coll IV and vimentin were significantly increased in BEV treated cells in comparison to untreated controls (p ≤ 0.05), RES (p ≤ 0.05) and BEV + RES (p ≤ 0.05) treated cells. Fibronectin and α-smooth muscle actin (α− sma) mRNA showed elevated expression though it was not significant in comparison to controls. On the contrary, e-cadherin expression was significantly low in cells treated with BEV compared to controls (p = 0.007) and BEV + RES (p = 0.003) treated cells ( Fig. 2A). Immunofluorescence staining with ZO-1 and F-ACTIN showed a well demarcated cellular boundary in controls, RES and BEV + RES treated cells but was absent in BEV treated cells. There was a marked appearance of the cell margin staining by ZO-1 in cells treated with BEV + RES which was not present in the BEV treated cells (Fig. 2B i-iv). Actin cytoskeleton status was evaluated with Alexa Fluor 488 phalloidin staining. Controls as well as RES treated cells revealed lesser stress fibers compared to BEV treated cells. The cell shape was more elongated and mesenchymal-like in BEV treated cells. The cellular morphology of BEV + RES treated cells was similar to control s (Fig. 2B v-viii). Scratch assay was performed to revalidate the Figure-1. Effects of BEV and/or RES on VEGF and its receptors. Bar graph shows the relative gene expressions of vegf and its receptors r1 and r2 with respect to mRNA levels of gapdh in ARPE-19 cells treated with BEV, RES and BEV + RES (A). Representative immunofluorescent images showing the VEGF (green) and DAPI (blue) staining in ARPE-19 cells after various treatments (B). Mean fluorescent intensity measured using Image J software and represented graphically (C). Graphical representation of the secreted VEGF concentration was measured using sandwich ELISA (D). Representative western blot analysis for VEGF, VEGF-R2 and GAPDH in ARPE-19 cells incubated with BEV, RES, BEV + RES and untreated cultures (E). Graphical representation of the quantified western blot analysis with respect to GAPDH protein expression (F). Statistical analysis was performed using student's t-test (**P < 0.01, P*** < 0.005) (n ≥ 3). Scale bar = 5 µm. Representative western blot analysis for the EMT markers VIMENTIN and E-CADHERIN in ARPE-19 cells exposed to BEV, RES and BEV + RES and untreated controls (E). Graphical representation of densitometric analysis of western blot quantification represented with respect to the expression levels of GAPDH (F). Statistical analysis was performed using student's t-test (*P < 0.05,**P < 0.01, ***P < 0.005) (n ≥ 3). Scale bar = 5 µm.
EMT status of the treated cells. A 61.2% closure of the cells in the control group was noted that was significantly lower (p = 0.0005) than the closure in BEV (74.2%) treated cells. Cells treated with BEV + RES showed 59.4% closure that was similar to that of untreated controls. There was a significant upregulation in the percentage of closure in BEV treated cells compared to controls (p = 0.0005) and BEV + RES (p = 0.015) treated cells ( Fig. 2C and D). Western blot densitometric analysis revealed a significant upregulation in the expression of VIMENTIN and a concurrent significant reduction in the levels of E-CADHERIN in BEV treated cells in comparison to controls (p = 0.0001; p = 0.0002) and BEV + RES (p ≤ 0.005) treated cells ( Fig. 2E and F).

Effects of BEV and RES on cell proliferation. Cell proliferation was assayed in cells after treatment with
BEV, RES and BEV + RES. A brief exposure to BEV resulted in a significant decrease in the expression of pcna (p = 0.007), ki67 (p = 0.002), cdc20 (p = 0.0006) and cyclin d1 (p = 0.003) in comparison to controls. There was a significant difference in the mRNA levels of Ki67 (p = 0.025), Cdc20 (p = 0.006) and Cyclin D1 (p = 0.006) in cells treated with BEV in comparison to cells treated with BEV + RES (Fig. 3A). Western blot analysis revealed that BEV treated cells had a significantly (p = 0.0003) decreased CYCLIN D1 protein expression compared to controls. In the presence of BEV + RES, the protein level of CYCLIN D1 further decreased (p = 0.0001) in comparison to cells treated with BEV alone ( Fig. 3B and C). BrdU labeling analysis showed a significant decrease in BrdU positivity in cells treated with BEV in comparison to controls (p = 0.0003). A significant decrease was observed in cells treated with BEV + RES in comparison to those treated with BEV (p = 0.005) alone ( Fig. 3D and E). Furthermore, immunofluorescence staining with Ki67 showed a significant decrease in nuclear staining positivity in cells treated with BEV in comparison to untreated (p = 0.003) cells. There was no significant difference in the percentage of Ki67 positivity between BEV treated compared to BEV + RES treated cells (Fig. 3F,G).

RES and BEV affect RPE phagocytosis.
A short exposure of BEV resulted in significantly reduced expressions of mertk (p = 0.0007; p = 0.0004), whereas a significant upregulation was noted in mRNA levels of lc3a (p = 0.002; p = 0.006), and p62 (p = 0.0003; p = 0.0001) mRNA compared to controls and RES + BEV treated cells, respectively. The reduction in the mRNA expression of mertk in the presence of BEV was restored partially in the presence of BEV + RES. However, there was no difference in the mRNA levels of lc3b in controls compared to RES + BEV treated cells (Fig. 4A). ARPE-19 cells treated with BEV, showed increase in the LC3-I, LC3-II and p62 levels compared to untreated or controls in Western blot. Contrarily, cells treated with BEV + RES showed decreased levels of LC3-I, LC3-II and p62 (Fig. 4B). RPE phagocytosis was studied using FITC labeled latex-beads in ARPE-19 cells treated with BEV, RES and BEV + RES. Phagocytosis assay was performed by quantifying the number of FITC tagged opsonized beads using flow cytometry. In BEV treated cells (5.1%) the percentage of opsonized beads was significantly lower than untreated control (7.45%, p ≤ 0.005) as well as BEV + RES (7.46%) treated cells (p ≤ 0.005, Fig. 4C and D). Immunofluorescence assay for phagocytosis was performed by counting the number of cells with opsonized beads. The results revealed a significant increase in the number of cells with beads in BEV + RES (p = 0.0002) treated cells compared to BEV treated cells. The number of cells with opsonized beads were also significantly low in BEV treated cells compared to controls (p ≤ 0.005, Fig. 4E and F).
Effects of BEV and RES on membrane potential. Since the membrane potential status of RPE is crucial for its proper functionality we looked at the effect of the BEV, RES and BEV + RES on cell polarization. Cultures exposed to BEV alone resulted in depolarization of the cell membrane and to RES alone caused cell membrane hyperpolarization. A combination of BEV + RES resulted in a significant decrease in the depolarization status of cells compared to BEV treated cells (p = 0.0003). BEV treated cells showed a higher depolarization in comparison to controls (p ≤ 0.005, Fig. 5A and B).

Effect of BEV and RES on Notch 4 and Dll 4 signaling.
We have previously shown that BEV exposed RPE cells show lower expression and signaling that is mediated by notch 4 and dll 4 7 . Gene expression of notch 4 and dll 4 were significantly lower in cells treated with BEV in comparison to controls (p = 0.03; p = 0.009) and cells treated with BEV + RES (p = 0.02; p = 0.003) (Fig. 6A). Immunofluorescence staining for NOTCH 4 and DLL4 showed that the mean fluorescent intensity of cells treated with BEV were significantly lower than those with BEV + RES (p = 0.001; p = 0.0005) and controls (p = 0.0008; p = 0.007). Additionally, there was no detectable difference in the mean fluorescent intensity in NOTCH 4 and DLL 4 levels in controls when compared to RES and BEV + RES treated cells ( Fig. 6B and C). Furthermore, western blot analysis depicted a significantly lower expression of NOTCH 4 (p = 0.004) and DLL 4 (p = 0.002) in cells exposed to BEV in comparison to cells treated with BEV + RES ( Fig. 6D and E). In an attempt to look at the status of downstream target of Notch signaling, the cells (treated and untreated) were immunoblotted for levels of HES1. Cells treated with BEV + RES showed significantly (p = 0.02) higher levels of HES1 in comparison to BEV alone treated cells ( Fig. 6D and E). There was no significant difference in the levels of NOTCH 4, DLL4 and HES 1 in controls compared to cells treated with RES and BEV + RES.
In order to understand the plausible signaling pathway involved, we investigated the phosphorylation status of kinase pathway in absence and presence of RES, BEV and RES + BEV. PathScan ® ELISA was performed for phosphor proteins, p-MEK1/2, p44/42 MAPK (Thr202/Tyr204), pSAPK/JNK (Thr183/Tyr185), pAkt (Ser473) and pAkt (Thr308). ELISA results revealed that there was significant decrease in the phosphorylation status in of pMEK1/2(Ser217/221) and p44/42 MAPK (Thr202/Tyr204) (Fig. 6F). The ELISA results were validated using Western blot for phosphorylated MEK1/2 and phosphorylated p44/42 MAPK (Thr202/Tyr 204) which showed decreased levels of phosphor MEK1/2 in ARPE-19 cells incubated with RES and RES + BEV compared to cells with BEV alone (Fig. 6G(i)). Similar results were also observed on phosphorylated p44/42 MAPK. In the presence of RES and BEV + RES there is decrease in the phosphorylated levels of p44/42 MAPK in comparison to that of BEV treated cells (Fig. 6G(ii)). These findings suggest that the restoration effects of RES treatment in ARPE cells is probably an outcome of the decreased phosphorylation levels of MEK1/2 and p42/44 MAPK.

Discussion
Intravitreal injections of anti-VEGF agents is the most preferred therapeutic intervention for wet AMD, RVO and PDR, despite some severe adverse effects and need for repeated injections 5 . It remains popular due the lack of alternatives to address pathological neo-vascularization. One of the major limitations of anti-VEGF therapy  Statistical analysis was performed using student's t-test (***P < 0.005) (n ≥ 3). Scale bar = 5 µm.
is that it neutralizes all the available secreted VEGF, which is largely a homeostatic function of the RPE and its secretome 1,9 . This leads to deleterious effects on the vascular health of the retina as VEGF is essential for maintaining a viable retinal vascular architecture 10,50,51 .
Anti-VEGF agents such as BEV areknown to induce altered phagocytosis and enhance EMT in RPE cells both on short-term and long-term exposure 7,52 . BEV exposure in RPE cells has been shown to induce pro-fibrotic changes via EMT induction 20 . Clinical trials using intravitreal anti-VEGF injections have shown adverse effects in wet-AMD patients 27 . The loss of visual acuity was associated with changes in the RPE akin to geographic atrophy 53 . Hence, it is important to investigate combinatorial treatment strategies to prevent such complications. In such an effort, we compared the effect of anti-VEGF (BEV) with and without RES.
RES, a poly phenol found in red wine, has been shown to restore normal function in dry AMD in animal and cell culture studies 54 . RES has also been approved by the US Food and Drug Administration (FDA) as a dietary supplement. It is widely used in ophthalmology both for clinical as well as experimental studies using animal models and in vitro culture systems. Though RES reduces VEGF levels, it does not have the same efficacy as anti-VEGFs. It also maintains phagocytic properties in cultured RPE cells in adverse treatment conditions 55 . BEV reduces VEGF levels but also affect the normal cellular physiology of RPE cells 7 . RES has also been shown to have anti-oxidant, anti-inflammatory, anti-aging, anti-angiogenic properties 32,42 . Most of its functions are driven by sirtuin 1(SIRT1), that deacetylates transcription factors and other proteins 56 . Moreover, Resvega a dietary supplement containing resveratrol in combination with vitamins is developed based on the findings of Age-related eye disease study (AREDS), reduced the probability of developing dry AMD 57 . On the other aspect, most of the adverse effects of BEV treatment has been attributed to the intracellular accumulation of BEV 14 . It could be hypothesized that such an accumulation of BEV might induce oxidative stress to the cell machinery thereby leading to defective physiological functions of the RPE. The anti-oxidant role of RES has been well demonstrated in several studies 32,42,54 . With this background, we thus investigated the possible role of RES in reversing the adverse effects of anti-VEGF (BEV), thereby providing an option for a combined therapy of anti-VEGF and RES in wet AMD and PDR patients. Such a combinatorial therapy may prevent geographic atrophy and other complications such as RPE tear, fibrosis, drusen formation and so on.
The IC50 of RES differs for various cell lines 58,59 . In choroidal endothelial cells, IC50 for RES was determined to be 26 µM at the 96 hr time point 60 . In our study with ARPE-19 cells, IC50 for RES was calculated to be 625.3 µM at 48 hr incubation (Suppl Fig. 1A). Moreover, vital staining (Trypan blue) performed on BEV, RES (100 μM) and BEV + RES treated cells and untreated controls did not show any difference in the percentage of cell viability (Suppl Fig. 1B). MTT assay performed on treated and untreated ARPE-19 cells also did not show any difference in the percentage of cell viability (Suppl Fig. 1C). There are several studies reporting no toxicity on treating ARPE-19 cells with 100 μM of RES and incubation for 48 hrs 31,35,61 . Hence, we used 100 μM of RES in combination with a clinical concentration (0.25 mg/ml) of BEV in ARPE-19 cells for the study. Furthermore, we investigated the cellular apoptosis levels by looking for activated caspase levels by FLICA immunofluorescence staining. There was no difference in the FLICA positivity between the treated and untreated cultures. (Suppl Fig. 2). In the presence of RES there was a significant down regulation (p = 0.002) of the mRNA levels of vegf r2 compared to controls. A similar significant decrease (p = 0.005) in the mRNA levels of cells treated with BEV + RES was also noted  compared to the controls. It has been shown that RES downregulates VEGFR2 phosphorylation and its further activation in cultured endothelial cells 62 . Apart from mRNA expression, VEGF A as well as VEGF R2 are significantly decreased in comparison to the controls. The anti-angiogenic properties of RES are mitigated via the VEGF R2. There was no significant decrease in the mRNA levels of vegf r1 in treated cells and it remained unaffected in comparison to the untreated cells. There are others studies which have also shown that VEGF-R1 remains unaffected by RES 63 .
There was a significant difference in the mRNA levels of VEGF A between the controls and RES (p = 0.043) as well as RES + BEV (0.005) treated cells. Nagineni  Our data shows that in the presence of RES, there is a significant decrease in the proliferation of cultured RPE cells as shown by BrdU, Ki67 and CYCLIN D1 positivity. Western blot and ELISA analysis show that there is decrease in phosphorylated Erk and MEK in cells treated with RES. This finding are similar to what has been shown by other's wherein the dampening of RPE proliferation has been attributed to be via Erk and MEK pathways. We as well as others have shown previously that phagocytosis of RPE cells is reduced in the presence of BEV treatment 7,13,14 . It has been shown that inhibition of autophagy associated with reduced phagocytic activity in ARPE-19 cells treated with 3-methyladenine 69 . We were further interested on assessing the levels of LC3 and p62 in ARPE cells treated with or without BEV, RES respectively. Here we show that in the presence of BEV with RES, the phagocytic property of RPE is restored to the controls levels. ARPE-19 cells treated with BEV, showed increase in the LC3-II and p62 levels compared to untreated or controls suggesting that blockage of autophagy degradation. Conversely, cells treated with BEV and RES showed decreased levels of LC3-II and p62. There could be a possibility that RES mediated induction of autophagy enhances the autophagic lysosomal clearance in ARPE cells treated with BEV 34 . Ferguson et al., have shown a convergence of phagocytosis with autophagy, wherein there is conversion of LC3 I subunit to II, enabling a proper degradation of phagocytosed photoreceptor outer segments 70 . Though our results provide a clue of a possible interaction of phagocytosis and autophagy, further studies are needed to investigate in depth the signaling crossover of phagocytosis and autophagy using pharmacological activators and blockers of autophagy. With the expression of LC3 protein regulation, it can be well envisaged that the phagocytosis of RPE is LC3-associated phagocytosis with involvement of non-canonical autophagy 71 . Grandbarbe et al., have shown that activated Notch signaling can induce phagocytosis in microglia cell 72 . Activated Notch signaling has been shown to activate autophagy in gliobalstoma cells 73 .
The transmembrane potential is readout of the blood-retinal barrier function and its integrity 7,74 . We had previously shown that a short exposure to BEV induced RPE membrane depolarization 7 . Our present experiments showed that there was a significant decrease in the depolarization of transmembrane potential in the presence of RES when compared to cells incubated with BEV alone. This suggests a protective role of RES in restoring the membrane integrity of blood-retinal barrier in the presence of BEV. Notch signaling is highly conserved developmentally regulated signaling pathway 47 . We previously reported that in a short exposure of BEV, NOTCH 4 and DLL 4 were blocked. Pinchot et al., using a high throughput screening found that Notch signaling is activated by RES 45 . Gidfar et al., have shown that RES can activate Notch signaling in meibomian gland epithelial cells 46 . Here, we show that in the presence of RES, Notch signaling gets activated. This activation along with dephosphorylation of Erk 1/2 and MEK has been shown to be major drivers for the functional restoration of RPE cells treated with RES + BEV compared to those treated with BEV alone.
Further studies using in-vitro diseased models such as culturing the cells under oxidative stress of hypoxia/ hyperoxia in the presence and absence of RES would provide insight into its clinical application. Additional work using animal disease models can reconfirm our findings and may provide a tool in the future to pave way for a combinatorial therapy for ocular neo-vascular diseases.
The advantage of RES is primarily its non-toxic nature. It is FDA approved and already a part of several clinical trials (NCT02625376, NCT02321189, NCT02321176). Our results provide an impetus to further investigate the beneficial role of RES in combination with BEV for patients with neo-vascular diseases. This study was done with a short term single pulse of BEV. It would be of interest to determine the role of RES in reversing the adverse effects of BEV on long-term repeated usage. Experiments with animal models could provide evidence of additional secondary effects of combination treatment with BEV + RES. Our findings can pave way for a newer treatment strategy for patients with AMD and PDR that can reduce the complications of anti-VEGF treatment regime. Partial reversal of the adverse effects of BEV by RES suggests an additional role of RES beyond its anti-oxidative properties.

Materials and Methods
Cell culture and reagents. The    IL, USA) using with the image Quant LAS 500 gel documentation/chemiluminescence detector (GE Healthcare Life Science, Uppsala, Sweden). The list of primary antibodies and secondary antibodies used for western blot are provided in Table 2.
Transmembrane potential assay. Membrane potential assay was carried out on cultured ARPE-19 cells treated with BEV, RES and BEV + RES and the analysis were done as mentioned earlier 7 . In brief, after 48 hours of various treatments, the cells were trypsinized and incubated with 20 nM of Bis-(1, 3-Dibutylbarbituric Acid) Trimethine Oxonol [DiBAC4(3)] for 30 mins at 37 °C. The intensity of the dye intake was analyzed in flow cytometry in the FL1 channel using the BD CellQuestTM Pro software.
Scratch assay. ARPE-19 cultures were grown to confluence in 35 mm tissue culture dishes. Using a sterile 1000 μl tip, cultures were scraped in a straight line to create a scratch. Care was taken to create scratches of similar width in the control as well as treatment groups. Cultures were washed with 1XPBS twice to remove detached and floating cells and cellular debris. The cells were then treated with BEV, RES and BEV + RES, wherein BEV treatment was given for 2 hours followed by RES treatment for 22 hours. The treated culture dishes were placed in the tissue culture incubator at 5% CO 2 and 37 °C. After 24 hours, the cells were observed using a phase contrast microscope and photographed with referenced matching points. In the images acquired, the distance between the scratches were measured using the Image J 1.48 version software (NIH, Bethesda, MD, USA).

Statistical analysis.
All experiments were performed in triplicate and results of three independent experiments were used for statistical analysis. Data were represented as the mean ± SD and were analyzed with the Student's t-test. Student's t-test was calculated between control sample and BEV treated, BEV and BEV + RES treated samples. Significance value denoted, p* < 0.05, ** < 0.01, *** < 0.005. The specific p-value is mentioned when the p ≥ 1 × 10 -4 , and when the p ≤ 1 × 10 -4 , it is stated as p ≤ 0.005.
Data Availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information files).