Cytoprotective Effects of Natural Highly Bio-Available Vegetable Derivatives on Human-Derived Retinal Cells

Retinal pigment epithelial cells are crucial for retina maintenance, making their cytoprotection an excellent way to prevent or slow down retinal degeneration. In addition, oxidative stress, inflammation, apoptosis, neovascularization, and/or autophagy are key pathways involved in degenerative mechanisms. Therefore, here we studied the effects of curcumin, lutein, and/or resveratrol on human retinal pigment epithelial cells (ARPE-19). Cells were incubated with individual or combined agent(s) before induction of (a) H2O2-induced oxidative stress, (b) staurosporin-induced apoptosis, (c) CoCl2-induced hypoxia, or (d) a LED-autophagy perturbator. Metabolic activity, cellular survival, caspase 3/7 activity (casp3/7), cell morphology, VEGF levels, and autophagy process were assessed. H2O2 provoked a reduction in cell survival, whereas curcumin reduced metabolic activity which was not associated with cell death. Cell death induced by H2O2 was significantly reduced after pre-treatment with curcumin and lutein, but not resveratrol. Staurosporin increased caspase-3/7 activity (689%) and decreased cell survival by 32%. Curcumin or lutein protected cells from death induced by staurosporin. Curcumin, lutein, and resveratrol were ineffective on the increase of caspase 3/7 induced by staurosporin. Pre-treatment with curcumin or lutein prevented LED-induced blockage of autophagy flux. Basal-VEGF release was significantly reduced by lutein. Therefore, lutein and curcumin showed beneficial protective effects on human-derived retinal cells against several insults.


Introduction
Age-related macular degeneration (AMD) is a leading cause of vision loss in the population over 65 [1]. Its pathogenesis involves a complex interaction of metabolic, functional, genetic, and environmental factors. The mechanisms associated with AMD combine oxidation, inflammation, hypoxia, and neovascularization of retinal tissues [1]. Abnormalities affect functional interrelated tissues, i.e., photoreceptors, retinal pigment epithelium (RPE), Bruch's membrane, and choriocapillaries. However, the impairment of RPE cell function is an early and crucial event in the molecular pathway LED-induced Autophagy blockage ARPE-19 cells were seeded as 1.8 mL per well of a suspension at 1.10 5 cells/mL in 24-well plates. Two days later, cells were untreated or treated with lutein at 3.5 µM or curcumin at 3.4 µM for 24 hours. Fresh cell culture media was replaced with 1.8 mL of media without any treatment. Cells were washed before adding fresh cell culture media. Then, they were kept in the dark or exposed to blue LED for 4 hours. Twenty-four hours later, autophagic flux was evaluated.
Autophagy analysis by immunoblot: Cells were either treated or untreated with bafilomycin A1 (LC Laboratories), an inhibitor of vacuolar-type H + ATPases, at 150 nM for 2 h prior to cell lysis and protein extraction. Whole-cell protein extracts were prepared by using RIPA lysis and extraction buffer (Thermo Fisher Scientific). Briefly, proteins were separated on 4%-15% Mini-PROTEAN®TGXTM precast protein gels (Bio-Rad), transferred to nitrocellulose membrane, blocked for 2 h in phosphate-buffered saline (PBS) solution containing 5% nonfat dry milk, probed overnight at 4 • C with primary antibodies directed against LC3B (Sigma L7543, 1/1000) or Actin (Sigma A5060, 1/1000), and for 2 h at room temperature with secondary HRP-coupled antibodies (DAKO P0448). After membrane revelation using the ECL detection kit (Bio-Rad), quantification was performed with the Image Lab software (Bio-Rad).
Quantification of Vascular endothelial growth factor (VEGF) by enzyme-linked immunosorbent assay (ELISA). The supernatants of lutein, resveratrol or curcumin pre-treated or untreated cells were collected and concentration of VEGF was determined using the Human VEGF ELISA Kit (KHG0111; Invitrogen Novex®) according to the manufacturer's instructions. Briefly, media were centrifuged at 1400 rpm at 4 • C, and 50 µL of supernatant was added to 50 µL of a standard diluent into a 96-well plate pre-coated with a polyclonal anti-VEGF antibody. Hu-VEGF biotin conjugate, followed by a Streptavidin-HRP solution, was added, and after 30 minutes at room temperature, the wells were washed before adding the chromogen. After dark incubation for 30 min at room temperature, the stop solution was added, and absorbance was read at 450 nm.
Immunocytochemistry staining. Cells were seeded at 1.8 mL per well into a 24-well plate containing glass cover plates for 48 h and then treated with curcumin, lutein, or curcumin + lutein for 24 hours. Nuclei were labeled with DAPI (Sigma-Aldrich) at a dilution of 1/500, and F-actin was labeled using Phalloidin Alexa Fluor 568 (Invitrogen). Images were obtained using a confocal microscope ((Leica SP5; Leica Corp., Wetzlar, HE, Germany) and analyzed with Image J software. Results indicate means ± SEM. Analysis of variance (ANOVA) was performed using Statistica (StatSoft, Inc). If ANOVA was significant, multiple comparisons were conducted to assess pairs of mean values, and differences between groups were studied using the post hoc Turkey test (significance set for p = 0.05). For autophagic data, statistical analyses were performed with GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). The non-parametric Mann-Whitney test was used to compare results between cells with and without bafilomycin A1. The non-parametric Kruskal-Wallis test was used for multiple comparisons between conditions (untreated cells, and cells treated with lutein and curcumin). P values ≤ 0.05 were considered as statistically significant. There is one symbol for p < 0.05, two symbols for p < 0.01, three symbols for p < 0.001, and four symbols for p < 0.0001.

Calibration of H 2 O 2 Effects
Depending on the experiment, plates had to be seeded with 100 or 300 µL of cell suspension, resulting in a different number of cells to start with. Therefore, the first step consisted of evaluating the sensitivity of cells to H 2 O 2 in these different conditions (Figure 1). The effect of H 2 O 2 was quantified by MTT measuring the metabolic activity and neutral red for cell survival [39]. The concentration of H 2 O 2 leading to metabolic activity reduction of 50% was 560 ± 16 µM for 300 µL suspensions (6.10 4 cells) and 610 ± 20 µM for 100 µL (2.10 4 cells) (Boltzmann Growth/Sigmoidal function) (Figure 1). Therefore, in the following experiment we used 600 µM H 2 O 2 to induce oxidative stress.  (Figure 1). Therefore, in the following experiment we used 600 µM H2O2 to induce oxidative     H2O2. In addition, we can observe that only a concentration of the agent leading to a reduction of 218 basal activity was significantly able to significantly reduce the effect of H2O2, suggesting a correlation 219 between a reduced basal metabolic activity and a resistance to oxidative stress.

225
Curcumin and lutein concentrations corresponded to the one affecting cell sensitivity to H2O2 226 (curcumin at 0.34 or 3.4 µM and lutein at 3.5 µM) and for resveratrol, we used the highest 227 concentration (43.6 µM). In order to calculate the reduction of metabolic activity induced by H2O2 we 228 subtracted the metabolic activity of cells with H2O2 to the one of cells without H2O2 for each treatment 229 ( Figure 4). The results showed that pretreatment with curcumin + lutein ± resveratrol significantly 230 limited (p < 0.01) the reduction of metabolic activity induced by H2O2. However, there was no These results showed that curcumin and lutein reduced the effect of oxidative stress induced by H 2 O 2 . In addition, we can observe that only a concentration of the agent leading to a reduction of basal activity was significantly able to significantly reduce the effect of H 2 O 2 , suggesting a correlation between a reduced basal metabolic activity and a resistance to oxidative stress.

Effect of Combination of Natural Agents on Cell Susceptibility to Oxidative Stress Induced by H 2 O 2
Further, the susceptibility to H 2 O 2 was analyzed in cells pretreated with a combination of these natural agents (curcumin + lutein or curcumin + lutein + resveratrol) ( Figure 4) in order to evaluate a synergistic or antagonist effect. The concentrations used were based on the previous experiments. Curcumin and lutein concentrations corresponded to the one affecting cell sensitivity to H 2 O 2 (curcumin at 0.34 or 3.4 µM and lutein at 3.5 µM) and for resveratrol, we used the highest concentration (43.6 µM). In order to calculate the reduction of metabolic activity induced by H 2 O 2 we subtracted the metabolic activity of cells with H 2 O 2 to the one of cells without H 2 O 2 for each treatment ( Figure 4). The results showed that pretreatment with curcumin + lutein ± resveratrol significantly limited (p < 0.01) the reduction of metabolic activity induced by H 2 O 2 . However, there was no synergistic or antagonist effect, since the results of the combined treatment were similar to that of the single treatment ( Figure 2).

254
IC80 were determined from dose-response data. IC50 was 3.6 ± 9 µM for cell metabolic activity and 255 6.1 ± 0.7 µM for cell viability. IC80 was 0.7 ± 0.2 µM for metabolic activity and 1.1 ± 0.1 µM for cell 256 viability. In further experiments, we will choose to use the dose of 1 µM of staurosporin to limit 257 drastic effects.

Calibration of Staurosporin Effects
Cell metabolic activity and cell viability were evaluated on staurosporin-treated ARPE-19 cells by an MTT or neutral red assay, respectively ( Figure 5). As expected, staurosporin induced a dose-dependent decrease in cell metabolic activity associated with a reduction of cell viability. IC50 and IC80 were determined from dose-response data. IC50 was 3.6 ± 9 µM for cell metabolic activity and 6.1 ± 0.7 µM for cell viability. IC80 was 0.7 ± 0.2 µM for metabolic activity and 1.1 ± 0.1 µM for cell viability. In further experiments, we will choose to use the dose of 1 µM of staurosporin to limit drastic effects.
Second, we studied the effect of pretreatment on ARPE19 susceptibility to staurosporin. Staurosporin significantly reduced cell survival (Figure 6c; 1µM staurosporin-concentration 0) and induced a significant activation of caspase 3/7 (Figure 6d, 1µM staurosporin-concentration 0) Curcumin, lutein, or resveratrol pre-treatment did not reduce activation of caspase 3/7 induced by staurosporin. However, we observed that staurosporin did not significantly affect survival of cell pretreated with curcumin at doses of 0.34 and 3.4 µM, nor with lutein at all tested doses, but induced a significant decrease of survival in cells pretreated with resveratrol.
These results suggest that curcumin and lutein, but not resveratrol protects human retinal cells from apoptosis induced by staurosporin. In addition, this protection does not involve the caspase 3/7 pathway. It should be noted that high concentrations of curcumin (27 µM) could have toxic effects and lead to a large caspase 3/7 activation (Figure 6c) associated with cell death (Figure 6d).      (Figure 7b). Therefore, pretreatment with a combination 300 of curcumin + lutein reduced cell death independently from caspase 3/7 activity. There were no 301 synergistic or antagonist effects between lutein and curcumin (compared to Figure 6).
Thereafter, we evaluated the effect of combination of natural agents on caspase 3/7 activity and cell death induced by staurosporin in ARPE-19 cells. Staurosporin had no significant effect on survival in cells pretreated with curcumin + lutein. However, these pretreatments did not reduce caspase activation induced by staurosporin (Figure 7b). Therefore, pretreatment with a combination of curcumin + lutein reduced cell death independently from caspase 3/7 activity. There were no synergistic or antagonist effects between lutein and curcumin (compared to Figure 6).

309
Autophagy is a dynamic process. To accurately measure autophagy, we performed an autophagic vacuoles [40]. In ARPE-19 cells not exposed to blue LED, we observed that lutein induced 314 a slight but non-significant increase in the amount of LC3-II compared to untreated cells, both in 315 conditions without and with bafilomycin A1 (Figure 8a and b). No significant effect of curcumin on 316 autophagy was observed in these conditions (Figure 8a and b).

317
Then, we analyzed the effect of blue LED exposure on autophagy. We observed a 2.7-fold higher 318 accumulation of LC3-II in ARPE-19 cells exposed to blue LED comparatively to non-exposed cells (p 319 < 0.05), in a bafilomycin A1-free condition (Figure 8c and d). This result suggested that blue LED

Effects of Pretreatment on the Alteration of Autophagy Flux Induced by Blue LED
Autophagy is a dynamic process. To accurately measure autophagy, we performed an autophagy flux assay by quantifying accumulation of LC3-II protein in cells treated or untreated with bafilomycin A1, a compound that blocks the autophagy flux. Briefly, accumulation of LC3-II in cells with bafilomycin A1, compared to cells without bafilomycin A1 indicates the production of new autophagic vacuoles [40]. In ARPE-19 cells not exposed to blue LED, we observed that lutein induced a slight but non-significant increase in the amount of LC3-II compared to untreated cells, both in conditions without and with bafilomycin A1 (Figure 8A,B). No significant effect of curcumin on autophagy was observed in these conditions ( Figure 8A,B).
Then, we analyzed the effect of blue LED exposure on autophagy. We observed a 2.7-fold higher accumulation of LC3-II in ARPE-19 cells exposed to blue LED comparatively to non-exposed cells (p < 0.05), in a bafilomycin A1-free condition ( Figure 8C,D). This result suggested that blue LED could induce autophagy. However, comparison of the LC3-II amounts in cells with and without bafilomycin A1 revealed that the autophagy flux was altered by cell exposure to blue LED ( Figure 8C,D). Indeed, whereas a significant increase in the amount of LC3-II was observed in ARPE-19 cells not exposed to blue LED with bafilomycin A1 in comparison to those without bafilomycin A1, the amount of LC3-II remained similar with and without bafilomycin A1 in cells exposed to blue LED.
The pre-treatment of cells with lutein, but not with curcumin, before blue LED exposure allowed for the maintenance of autophagy flux, as shown by the significant increase in the amount of LC3-II in cells exposed to blue LED with bafilomycin A1 in comparison to those without bafilomycin A1 ( Figure 8C,D).

337
The pre-treatment of cells with lutein, but not with curcumin, before blue LED exposure allowed 338 for the maintenance of autophagy flux, as shown by the significant increase in the amount of LC3-II 339 in cells exposed to blue LED with bafilomycin A1 in comparison to those without bafilomycin A1     Effect of lutein and curcumin on autophagy. ARPE-19 cells were treated with lutein at 3.5 µM and curcumin at 3.4 µM for 24 h before being exposed or not to blue LED for 4 h. Bafilomycin A1 (Baf) was added 2 h before protein extraction. Immunoblot analyses were performed using anti-LC3B and anti-Actin antibodies. Quantification of LC3-II and Actin was done. (A,C) are representative immunoblots. (B,D) are the corresponding ratios of LC3-II/Actin normalized to that obtained for cells that were untreated, non-exposed to blue LED and without Baf, defined as 1.0. Data are means ± SEM of at least 3-4 independent experiments. For statistical analysis, comparisons were performed between untreated cells with Baf and cells treated with lutein or curcumin, with Baf by using the non-parametric Mann-Whitney test. Comparisons were also performed between cells without Baf or cells with Baf, in the different conditions (untreated, or treated with lutein or curcumin), by using the non-parametric Kruskal-Wallis test. * p < 0.05.
Altogether, these results show that blue LED exposure induces a blockade of the autophagy flux, and suggests that lutein could prevent this alteration.

Effects of Pretreatment on VEGF Secretion in Basal Conditions
ARPE-19 cells were pretreated with curcumin, lutein, resveratrol, or a mixture of 2 or 3 of these natural agents for 24 h before collecting cell supernatants for VEGF quantification (Figure 9). When an ANOVA statistical analysis was performed, no significant difference was observed between groups. This is probably because of inter-test variations. However, when we compared the effect of each molecule individually, we observed that cell treatment with lutein significantly (p = 0.0035) reduced the basal level of VEGF secreted in culture medium, and that cell treatment with resveratrol at 3.4 µM significantly (p = 0.002) increased it. No significant effect on VEGF level was observed for cells treated with curcumin at 3.5 µM. It should be noted that the decrease in VEGF amount observed in cells treated with resveratrol at the highest concentration is related to a reduced cell number. The combination of curcumin 3.4 µM + lutein 3.5 µM significantly (p = 0.04) decreased VEGF secretion.

Effects of Combined Pretreatment on VEGF Secretion in Hypoxia-Induced by CoCl 2
Hypoxia was induced by incubating ARPE19 cells with 200 µM of CoCl 2 for 4 hours. CoCl 2 had no significant effect on cell viability or caspase 3/7 activity (Figure 10b,c) but induced a significant increase in VEGF release (Figure 10a) in untreated cells. Pretreatment significantly reduced basal VEGF release and increased caspase activity without affecting cell viability. CoCl 2 did not increase VEGF release in pretreated cells. In the above, curcumin + lutein significantly reduced the release of VEGF induced by CoCl 2 .

Discussion
Inflammation, oxidative stress, and angiogenesis has been shown to highly contribute to AMD development or progression. More recently, autophagy has also been suggested as a contributing factor in AMD by playing a role in the regulation of RPE cell death [15]. Although no pharmaceutical medicines are available to treat AMD, natural agents with anti-inflammatory, anti-oxidant, and anti-angiogenic properties, such as omega-3 fatty acids, vitamins, and carotenoids have been reported to exert a protective effect against AMD [3,4]. As a consequence, numerous natural agents have been recommended as nutritional supplements in the prevention of ocular diseases [41]. However, only a few studies exist on the impact of these compounds at the cellular and molecular levels and their mode of action to protect retinal cells.
In the present study, we analyzed the ability of nutritional supplements to protect human retinal pigment epithelial cells (ARPE-19 cells) against several harmful stressors (oxidative stress, induced cell death, hypoxia, and autophagy alteration induced by blue LED). The chosen natural agents were lutein, curcumin, and resveratrol. Lutein was chosen because its intake was suggested in nutritional recommendations [7], resveratrol because it has anti-oxidative [17][18][19] and anti-inflammatory effects [20,21] and can modulate autophagy [22], and curcumin because its beneficial effects had been reported in several neurodegenerative diseases [29][30][31][32][33][34] and its therapeutic potential in ophthalmology has been reported [35]. The doses of natural agents used in this study were selected according to their oral bioavailability and the conceivable amount in the blood circulation after oral administration. Curcumin has been shown to have low bioavailability [42,43]. FloraGlo formulation of lutein enhances bioavailability [44]. Resveratrol is well-absorbed, but its bioavailability is low [45]. Cells were treated at concentrations ranging from 0 to 27.17 µM for curcumin, 0 to 70 µM for lutein [46], and 0 to 43.9 µM for resveratrol [47].
We first evaluated the effect of lutein, curcumin, and resveratrol in basal conditions (no stressors). It had been shown in several studies that lutein and resveratrol reduced cell proliferation [48][49][50][51]. However, in our study, cells were treated when they reached confluency, and so the proliferation was very limited. As previously described, no deleterious effects of lutein [52] and resveratrol [48] were observed. For curcumin, we noticed a decrease in metabolic activity. This decrease could have resulted from a reduced number of cells, since curcumin had been shown to decrease cell proliferation at 50-70 µM [53] or a slowdown of metabolic activity per cell. The second hypothesis was confirmed, since curcumin did not affect cell survival. Consequently, curcumin reduced basal metabolic activity without modifying the cell number. This reduction in metabolic activity can contribute to the protective effects of curcumin. Indeed, it has been suggested that reduction in energy usage induced by hypothermia would provide protection during ischemia [54]. In addition, the results obtained here highlight the fact that metabolic activity and cell number do not always directly correlate, and so a metabolic activity assay, such as MTT, has to be used with caution for cell viability.
Further, we showed that lutein, resveratrol, and curcumin agents increased basal caspase 3/7 activity. Such activation has been previously reported in lung cancer-derived cell lines [55]. Interestingly, recent studies demonstrated that caspases can exert both apoptotic and non-apoptotic activities, and their activation does not exclusively lead to cell death [56][57][58]. Indeed, caspases are involved in regeneration, wound healing, proliferation, and in cell motility [59,60]. Such a phenotype could explain why we observed caspase activation without cell death.
VEGFs are proteins secreted by a variety of cells, primarily acting on endothelial structures to initiate and promote angiogenesis during development and in pathological conditions, such as AMD and diabetic retinopathy [61][62][63]. Anti-VEGF therapies are presently one of the main focuses for the therapeutic management of AMD [64][65][66]. Among the natural agents tested, lutein (3.5 µM) was the only one that reduced the basal amount of VEGF secreted by ARPE-19 cells. In addition, we showed that in our conditions, neither lutein nor curcumin were able to induce a significant activation of autophagy under normal conditions. Autophagy is crucial for homeostasis maintenance in RPE cells, particularly by its role in the degradation of damaged proteins and organelles in the regulation of the inflammatory response and in the fight against oxidative stress [15,67].
Thereafter, we evaluated the susceptibility of cells pretreated with lutein, curcumin, or resveratrol to stress conditions. Curcumin, lutein, but not resveratrol reduced oxidative stress induced by H 2 O 2 and apoptosis induced by staurosporin. The dose of curcumin that allows for protection of ARPE-19 cells from death induced by staurosporin was much lower than those used in primary rat hippocampal neuron cultures [68]. Nevertheless, neither curcumin nor lutein had effects on staurosporin-induced caspase 3/7 activity. These results confirm that caspase activation might not be directly associated with induction of apoptotic cell death. In the present study, we showed that exposure to blue LED induces accumulation of LC3-II in ARPE-19 cells. This result is in line with data published by others reporting accumulation of LC3-II in retinal cells in response to blue LED exposure [69][70][71]. However, by using an autophagy flux assay, we were able to show that the autophagy flux was altered in ARPE-19 cells exposed to blue LED. Given the crucial role played by autophagy in the RPE, alteration of the autophagy flux could have deleterious consequences for the cells, including accumulation of toxic compounds (i.e., vacuoles containing undigested materials, oxidized compounds, aggregates, and damaged organelles) and uncontrolled inflammation, which are conditions contributing to retinal degeneration. Data from the literature indicate that lutein and curcumin can modulate autophagy [12,38,72]. Our data suggest that lutein, and potentially curcumin can prevent autophagy flux alteration induced by blue LED exposure. We have to keep in mind that autophagy can have a dual role [73] Recently, it has been shown that autophagy-associated cell death was induced in ARPE-19 and primary human RPE (hRPE) cells by serum deprivation and oxidative stress by H 2 O 2 , and that the clearance of these cells induced inflammation, suggesting this mechanism could contribute to AMD [74]. Therefore, it would be interesting to evaluate the effects of curcumin and lutein in this mechanism.
Most of the time, dietary supplements are composed of a complex mixture. Therefore, we wanted to investigate the interactions between molecules leading to antagonistic or synergistic effects. The combination of curcumin and lutein decreased metabolic activity (previously observed with curcumin) and limited oxidative stress induced by H 2 O 2 and toxic effects of staurosporin, and prevented VEGF release induced by CoCl 2 . The association of curcumin (0.34 or 3.4 µM) and lutein (3.5 µM) with or without resveratrol did not show toxic effects, but no antagonistic or synergistic effect between agents could be found. Nevertheless, we did observe a complementary protective effect.

Conclusions
In this study, we analyzed the ability of lutein, curcumin, and resveratrol to protect human retinal pigment epithelial cells (ARPE-19 cells) against several harmful stressors (oxidative stress, induced cell death, hypoxia, and autophagy alteration induced by blue LED) known to be involved in retinal degeneration. We could not find evidence of any protective effect of resveratrol (in the range of 0.001 to 21.9 µM). Curcumin had higher protective effects against H 2 O 2 -induced oxidative stress than lutein. At the concentration tested, lutein and curcumin had a similar protective effect against staurosporine-induced cell death. Pretreatment with lutein, but not curcumin was found to prevent blue LED-induced alteration of autophagy. In addition, whereas curcumin had no effect on basal VEGF release from human-derived retinal cells, lutein reduced it. Therefore, because of the complementary effect of each compound, the association of lutein and curcumin is beneficial to protect human retinal pigment epithelial cells. The combination of curcumin and lutein should reduce the effect of oxidative stress on metabolic activity, apoptotic cell death, and VEGF secretion induced by hypoxia. Funding: This research was funded by a grant from Densmore, Monaco.