Protective Effect of Metformin against Hydrogen Peroxide-Induced Oxidative Damage in Human Retinal Pigment Epithelial (RPE) Cells by Enhancing Autophagy through Activation of AMPK Pathway

Age-related macular degeneration (AMD) is a leading cause of blindness with limited effective treatment. Although the pathogenesis of this disease is complex and not fully understood, the oxidative damage caused by excessive reactive oxygen species (ROS) in retinal pigment epithelium (RPE) has been considered as a major cause. Autophagy is essential for the degradation of cellular components damaged by ROS, and its dysregulation has been implicated in AMD pathogenesis. Therefore, strategies aiming to boost autophagy could be effective in protecting RPE cells from oxidative damage. Metformin is the first-line anti-type 2 diabetes drug and has been reported to stimulate autophagy in many tissues. We therefore hypothesized that metformin may be able to protect RPE cells against H2O2-induced oxidative damage by autophagy activation. In the present study, we found that metformin attenuated H2O2-induced cell viability loss, apoptosis, elevated ROS levels, and the collapse of the mitochondria membrane potential in D407 cells. Autophagy was stimulated by metformin, and inhibition of autophagy by 3-methyladenine (3-MA) and chloroquine (CQ) or knockdown of Beclin1 and LC3B blocked the protective effects of metformin. In addition, we showed that metformin could activate the AMPK pathway, whereas both pharmacological and genetic inhibitions of AMPK blocked the autophagy-stimulating and protective effects of metformin. Metformin conferred a similar protection against H2O2-induced oxidative damage in primary cultured human RPE cells. Taken together, these results demonstrate that metformin could protect RPE cells from H2O2-induced oxidative damage by stimulating autophagy via the activation of the AMPK pathway, supporting its potential use in the prevention and treatment of AMD.


Introduction
Age-related macular degeneration (AMD) is the leading cause of blindness in individuals over 50 years of age. It is a disease that affects the macula of the retina, causing a chronic and progressive vision loss [1]. Late AMD can be divided into neovascular (wet) and nonneovascular (dry) forms [1]. Currently, therapies such as antivascular endothelial growth factor (anti-VEGF) therapy have been proven to be effective in treating wet AMD [2]. However, dry AMD, which accounts for approximately 90% of AMD cases, still lacks an effective treatment.
Although the pathogenesis of dry AMD is complex, the degeneration of the aging retinal pigment epithelium (RPE) cells is widely considered to be the initial event [3]. The RPE consists of a single layer of epithelial cells that sustains the function of photoreceptor cells by supporting the phagocytosis of photoreceptor outer segments (POS), vitamin A metabolism, and the regeneration of visual pigments [4][5][6][7]. Its impairment results in a secondary degradation of photoreceptors and eventually leads to vision loss [3,8]. RPE cells are especially susceptible to ROS-induced oxidative damage. As a high energy-demanding tissue, RPE cells produce high levels of ROS derived from the oxygen metabolism [5,9]. ROS can also be generated by light or the phagocytosis of POS in RPE [5,10]. Additionally, studies have been showing that RPE cell impairment can lead to the accumulation of damaged organelles and various nonfunctional or toxic proteins, including lipofuscin, and promote the formation of drusen which is a typical characteristic of AMD [11].
Autophagy is a protective mechanism designed for the degradation and removal of different cellular components, including those damaged by ROS, supporting cellular renovation and homeostasis [5]. The autophagic process begins with the formation of a sequestering membrane, termed as phagophore, to form an autophagosome that subsequently engulfs the intracellular components and carry them to lysosomes for degradation [12]. Autophagy initiation is controlled by the activation of ULK-1 complex I (contains ULK1, FIP200, and ATG13) and of Beclin1 complex II (contains the proteins p150 and Atg14L and class III phosphatidylinositol 3-phosphate kinase Vps34). Following amino acid withdrawal, ULK-1 was shown to phosphorylate Beclin1, and this phosphorylation step is crucial for the function of Beclin1 in autophagy [13]. The active ULK-1 and Beclin1 complexes relocalize to the site of autophagosome initiation, the phagophore, where they both contribute to the recruitment of different downstream autophagy components [14]. The phagophore formation is followed by the elongation stage of autophagy essential for the expansion of the autophagosomal membrane which requires ATG5-ATG12 and the microtubule-associated protein light chain 3 (LC3-ATG8) conjugation systems [15]. The autophagosomal formation is denoted by the combination of cytosolic LC3-I with phosphatidylethanolamine on the surface of the autophagosomal membrane forming LC3-II protein, which is visible as discrete puncta on immunofluorescence analysis. Being proportional to the number of autophagic vesicles, LC3 is considered an autophagosome marker molecule [16]. The LC3-binding protein p62, also known as sequestosome 1 (SQSTM1), binds to the LC3-II protein located on the inner membrane of the autophagosome and then is degraded in autophagolysosomes. The inhibition of autophagy is accompanied by the accumulation of p62 protein [17]. Many lines of evidences suggest that autophagy impairment is associated with a variety of diseases, such as diabetes and cancer, infectious diseases, and neurodegenerative diseases, including AMD [18,19]. Specifically, autophagy has been shown to be crucial for RPE homeostasis and visual cycle, as evidenced by the decreased POS responses to light stimuli in mice with essential autophagy gene Atg5-deficient RPE cells [20]. In AMD, impaired autophagic capacity occurs simultaneously with the accumulation of dysfunctional mitochondria and toxic proteins that ultimately lead to RPE cell damage or apoptosis [7,21]. While autophagy inhibition in ARPE-19 cells by 3-MA or by the knockdown of ATG7 or Beclin1 was reported to increase its susceptibility to H 2 O 2 -induced oxidative damage, autophagy upregulation by rapamycin was shown to protect the cells from H 2 O 2 -induced injury [7]. Therefore, strategies aimed at upregulating autophagy may be helpful for protecting RPE cells from oxidative damage and preventing or reversing the progression of AMD.
Metformin, a biguanide derivate widely used in the treatment of type 2 diabetes, has been shown to stimulate autophagy in a variety of tissues by activating AMP-activated protein kinase (AMPK) [22][23][24]. AMPK is an energy sensor that maintains the cellular energy homeostasis by regulating several important cellular processes, such as cell growth and proliferation, apoptosis, and autophagy [22,25]. AMPK can activate autophagy by inhibiting the mammalian target of rapamycin (mTOR) [26]. In addition, AMPK activation also promotes autophagy by directly phosphorylating ULK1, the protein kinase that initiates autophagy [27]. We have previously reported that metformin could protect neuronal cells from H 2 O 2 -induced oxidative damage by activating AMPK [28]. Given the crucial role of autophagy in AMD pathogenesis, we therefore hypothesized that metformin may confer a similar protection in RPE cells by stimulating autophagy. In the present study, we aimed to investigate whether metformin could protect RPE cells from H 2 O 2induced oxidative damage and whether the protective effects are associated with the upregulation of autophagy promoted by AMPK activation.

Oxidative Medicine and Cellular Longevity
(1 × 10 4 cells/well). After treatment, cells were incubated with a TUNEL reaction mixture (50 μL) at 37°C for 60 min in the dark condition. Cells staining red were positive for TUNEL staining and considered apoptotic. The apoptotic index was measured according to the following formula: apoptotic index ð%Þ = ðapoptotic cells/total cellsÞ × 100%.
2.8. Hoechst 33342 Staining. The DNA binding dye Hoechst 33342 was used to measure cell apoptosis. The cells were seeded into 96-well plates (1 × 10 4 cells/well). After treatment, cells were washed with PBS and fixed with 4% formaldehyde for 15-20 min at room temperature. The cells were then incubated with 10 μg/mL of Hoechst 33342 for 2 min at room temperature. After this period, the cells were washed with PBS and observed under EVOS FL Imaging System (Thermo Fisher Scientific, Waltham, MA, USA). Cells exhibiting fragmented nuclei or condensed chromatin were scored as apoptotic cells.
2.9. Flow Cytometry. Flow cytometry was used to assess cell apoptosis. Briefly, D407 cells were trypsinized after the treatment, washed twice with 1x PBS, and then centrifuged at 1000 rpm for 5 min and resuspended in Annexin V-FITC 1x binding buffer (195 μL). The cells were supplemented with Annexin V-FITC (5 μL) and incubated in the dark at 37°C for 20 min. After this period the cells were centrifuged at 1000 rpm for 5 min, resuspended in 1x binding buffer (190 μL) and incubated with propidium iodide (PI) (10 μL) in the dark for 5 min. Apoptosis was quantified using the flow cytometer. Cell Quest Pro software (BD AccuriC6, BD, USA) was used for the analysis of the apoptosis condition.
2.10. Immunocytochemistry. After appropriate treatment, D407 cells were fixed with 4% paraformaldehyde (PFA) at room temperature for 15 min and then washed with 1x PBS for three times. The cells were then incubated with 1x PBST (containing 0.25% Triton X-100) for 10 min and blocked with 3% BSA for 1 h at room temperature. Afterwards, cells were incubated with the primary antibody in PBS containing 1% BSA at 4°C overnight. The following day, the cells were washed 3 times with PBS for 5 min and incubated with the secondary antibody for 1 h at room temperature in the dark. Nuclei were counterstained with one drop of antifade mounting medium with DAPI (P0131, Beyotime), and the images were acquired with a Nikon A1 confocal microscope.

Measurement of the Autophagic Flux. Monitoring
Autophagy flux is important for the analysis of autophagy. A tandem fluorescent-tagged LC3 (mRFP-GFP-LC3) virus vector was used to investigate the autophagic flux based on the different pH stability of mRFP and GFP fluorescent proteins. While mRFP is pH insensitive, GFP is degraded in the acidic environment of autophagolysosomes. Cells with no autophagy and cells containing autophagosomes exhibit yellow fluorescence due to the coexpression of mRFP and GFP. Upon fusion of autophagosomes and lysosomes to form autophagolysosomes, the acidic lysosomal environment quenches the acid-sensitive GFP, while RFP is not affected, causing autophagolysosomes to exhibit a red fluorescence. Therefore, red fluorescence can be used to measure the autophagic flux. Approximately 5 × 10 4 D407 cells were seeded in 12-well plates covered with coverslips and grown in DMEM medium with 10% FBS until 60% confluence. Diluted virus stock solutions in blank medium (20,40, and 80 multiplicity of cellular infection (MOI)) were added to the cells. After 12 h of infection, the medium was changed to complete culture medium (DMEM medium with 10% FBS) and the cells were cultured for additional 48 h before adding the drugs. After treatment, the cells were fixed in 4% PFA for 10-15 min at room temperature. Fixed cells were then washed twice with 1x PBS. The coverslips were mounted on glass slides using VECTASHIELD® Mounting Medium without DAPI (H-1000). The images were acquired using a Nikon A1 confocal microscope.
2.12. Western Blotting. Cells were harvested and lysed with ice-cold RIPA lysis buffer. Protein concentration was determined by the BCA method. Samples with an equal amount of proteins (50 μg) were separated by 10% SDS-PAGE and then transferred to a PVDF membrane. After blocking with 3% BSA for 1 h, the membranes were incubated with the selective primary antibodies at 4°C overnight. The next day, the primary antibody was washed with 1x TBST for three times and incubated with a secondary antibody for another 2 h. The intensity of the bands was quantified using ImageJ software.
2.13. LC3B And AMPK Silencing by shRNA. Gene silencing by shRNA was performed as previously described [29]. Before the experiment, transfection complexes were made by mixing shRNA with Lipo2000 and Opti-MEM followed by incubation for 15 min at room temperature. Obtained transfection complexes were added drop-wise onto cell cultures. The cells were then moved to a 37°C incubator and maintained for 6 h. After this period, the medium was replaced by complete medium. Cells were collected for further experiments 48 h after transfection.
2.14. Statistical Analysis. All the results are presented as the mean ± SD for triplicates. Data was analyzed using Graph-Pad Prism 5.0 statistical software (GraphPad software, Inc., San Diego, CA, USA). Statistical analysis was carried out using one-way ANOVA followed by Tukey's multiple comparison post hoc test, with p < 0:05 being considered statistically significant.       (Fig. S1). According to the results, 80 MOI had the highest virus infection efficiency and this dilution was used to treat the cells before the treatments. The autophagic flux was measured by comparing the abundance of autophagosomes labeled with yellow and red fluorescence. As shown in Figure 3(a), 1 mM metformin induced an increase in the number of red light spots indicative of the stimulatory effect of metformin on the autophagic flux in D407 cells. When autophagy occurs, the cytoplasmic form LC3-I combines with phosphatidylethanolamine to form LC3-II, which is a marker of autophagosome formation. As an isoform of LC3, LC3B is widely used as a molecular marker of autophagy. Accordingly, Western blot analysis confirmed that metformin treatment increased the level of LC3B-II expression (Figure 3(b)). Fur-ther confirmation was achieved through the assessment of the effect of metformin treatment in LC3B-II and p62 proteins levels and in ULK-1 and Beclin1 phosphorylation. As shown in Figures 3(c) and 3(d), metformin treatment promoted the increase of LC3B-II levels and of ULK-1 and Beclin1 phosphorylation in a time-and dose-dependent manner. In contrast, p62 levels were decreased by metformin treatment in a similar manner. Taken together, these findings are indicative that metformin is able to activate autophagy.

Autophagy Inhibition Blocked the Protective Effect of
Metformin. In order to investigate whether autophagy is involved in the metformin-mediated protection against H 2 O 2 -induced oxidative damage, we used 3-MA and CQ to inhibit the formation of functional autophagosomes and the autophagosome-lysosome fusion, respectively [30,31]. As shown in Figure 4 LC3B in the cells exposed to H 2 O 2 damage, which is indicative of its ability to stimulate the autophagic flux. Acute H 2 O 2 treatment was previously reported to induce autophagy as a protective attempt to prevent oxidative damage in RPE cells [7]. Consistent with this finding, our results showed that treating D407 cells with 200 μM H 2 O 2 for 24 h resulted in an increase of the number of LC3 puncta per cell that was further strengthened by 1 mM metformin pretreatment. p62 is a substrate of autophagy widely used as a marker of autophagic degradation [32]. A recent study has shown that oxidative stress upregulated p62 gene expression to assist selective autophagy, serving as an early protective response [33]. In line with this, we found that H 2 O 2 induced the elevation of p62 levels in D407 cells. In addition, p62 levels were significantly decreased upon metformin treatment. 3-MA pretreatment reversed metformin-induced increase of LC3B expression and decrease of p62 expression, whereas CQ pretreatment resulted in the accumulation of both LC3B and p62, representing the blockade of the autophagic flux. In addition, inhibition of autophagy by either 3-MA or CQ blocked metformin-induced protection against H 2 O 2 -induced cell viability loss (Figures 4(b) and 4(c)). To further confirm these findings, we proceeded to the knockdown of the key autophagic proteins Beclin1 or LC3B by using shRNA. As expected, the knockdown of either gene abolished the protective effect of metformin on H 2 O 2 -induced cell viability loss, ROS production, and loss of mitochondrial membrane potential, indicating the essential role of autophagy activation in metformin-mediated protective effects (Figures 4(d)-4(h)).

AMPK Signaling Is Involved in the Protective Effect of
Metformin against H 2 O 2 -Induced Cell Damage. As metformin has been shown to activate AMPK-dependent autophagy in many tissues, we investigated whether metformin could stimulate AMPK phosphorylation in D407 cells. As shown in Figures 5(a) and 5(b), metformin treatment caused a significant upregulation of AMPK phosphorylation levels in a dose-dependent manner, starting at 1 mM. Similarly, metformin also time-dependently stimulated AMPK phosphorylation in D407 cells (Figures 5(c) and 5(d)). Additionally, we performed immunocytochemistry to detect the   Oxidative Medicine and Cellular Longevity levels and subcellular localization of phosphorylated AMPK (P-AMPK). As presented in Figure 5(e), P-AMPK is mainly located in the cytoplasm in both basal and H 2 O 2stimulating conditions. Exposure to metformin promoted the increase of P-AMPK levels in both the cytoplasm and the nucleus, being more prominent in the latter, when cells were treated with or without H 2 O 2 . These findings are indicative that metformin is able to stimulate AMPK activation in D407 cells.
3.6. The Activation of Autophagy and the Protective Effect of Metformin Decreased after Inhibition of AMPK Signaling Pathway. In order to investigate if AMPK signaling is required for the metformin-mediated autophagy upregulation and its protective effect in D407 cells, we proceeded to the inhibition of AMPK using compound C (AMPK inhibi-tor). The results revealed that compound C reversed both the metformin-induced upregulation of LC3B levels and downregulation of p62 levels (Figures 6(a) and 6(b)). Moreover, AMPK inhibition blocked the effect of metformin in the phosphorylation levels of ULK-1 and Beclin1 (Figure 6(b)), suggesting that metformin-induced autophagy in D407 cells is AMPK-dependent. Further assessment of AMPK-mediated autophagy in the protective effect of metformin revealed that upon AMPK inhibition metformin failed to protect D407 cells from H 2 O 2 -induced cell viability loss (Figure 6(c)). In addition, metformin reversion of H 2 O 2induced LDH release was blocked by compound C treatment ( Figure 6(d)). To confirm this finding, we proceeded to the knockdown of AMPK in D407 cells. The knockdown efficiency of different shRNA is shown in Figure 6(e), and shAMPK3 was used in following experiments as it was who     (Figure 7(b)). Metformin pretreatment significantly reversed H 2 O 2 -induced cell viability loss and this protective effect peaked at the concentration of 0.5 mM (Figure 7(b)). In addition, compound C also blocked the protective effect of metformin against H 2 O 2 -induced cell viability loss, suggesting that metformin-mediated protection is dependent of the AMPK pathway (Figure 7(c)). Taken together, our results demonstrate that metformin has a similar protective effect in primary RPE cells.

Discussion
Although the pathogenesis of AMD is complex, RPE degeneration caused by oxidative damage has been considered a major cause of this condition. The retina is a high oxygenconsuming tissue in which redox homeostasis is particularly important [9]. Antioxidant capacity decreases with the increase of age, and the excessive ROS accumulation can lead to RPE cell damage and drive the pathogenesis of AMD [34]. Therefore, agents with antioxidant properties may confer protection against RPE degeneration induced by oxidative damage. In fact, yttrium oxide nanoparticles, which are free radical scavengers, were shown to prevent photoreceptor death in a light-damage model [35]. Additionally, dietary plant-derived antioxidants such as anthocyanins were also reported to be able to protect RPE cells from light-induced oxidative damage [36]. In the present study, we found that metformin significantly reversed H 2 O 2 -induced cell viability loss, LDH release, Δψm loss, and cell apoptosis in D407 cells. As the relevant concentrations of metformin showed no cytotoxicity, we propose that metformin treatment may serve as a safe therapeutic option for preventing and treating AMD.
Autophagy is critical for protection towards oxidative damage, and diminished autophagy has been implicated in a variety of diseases including AMD. In RPE cells, inhibition of autophagy was shown to increase ROS generation and to exacerbate oxidative stress-induced mitochondria damage and cell death [7]. Thus, interventions aimed at enhancing the autophagic flux stand as a promising strategy to prevent and reverse the progression of AMD. In RPE cells, stimulation of autophagy by rapamycin was found to protect ARPE-19 cells from a lethal dose of H 2 O 2 [7]. Despite being a potent autophagy activator, many lines of evidences have shown that rapamycin has many adverse effects, such as diabetic-like symptoms, gastrointestinal disorders, and increased risk of infection [37,38]. Therefore, it is important to search for other alternatives with fewer side effects.
Metformin is a widely used antidiabetic drug that acts on common pathways and confers beneficial effects on a spectrum of age-related disorders. There is clinical evidence of metformin's ability to induce autophagy or prevent oxidative stress/autophagy-related diseases, including AMD. In fact, metformin was shown to be able to protect retinal photoreceptors and RPE from oxidative stress in preclinical mouse models. Moreover, its use was also reported to reduce the likelihood of developing AMD [39]. In a different study, metformin improved the neurological function and oxidative stress status of acute stroke patients with type 2 diabetes [40]. Still, the mechanisms underlying these effects remain poorly understood. In this study, we found that metformin is able to upregulate autophagy in D407 cells, as evidenced by the increased LC3B expression and decreased p62 expression. Moreover, autophagy inhibition, with either 3-MA or CQ, blocked metformin-induced changes in LC3B and p62 levels, as well as its protective effect against H 2 O 2 -mediated cell viability loss and apoptosis. The metformin protective effect was also inhibited by the knockdown of the two essential autophagic genes Beclin1 and LC3B, suggesting that Metformin has been proven to be a potent AMPK activator, and its autophagy-inductive effect has been associated with AMPK activation in many tissues [41]. Consistent with these findings, our observations showed that metformin was able to stimulate AMPK phosphorylation in a time-and dose-dependent manner. The subcellular localization of AMPK is crucial for its modulatory activity on downstream targets. By immunocytochemistry staining, we found that P-AMPK was predominately located in the cytoplasm, in both basal-and H 2 O 2 -stimulating conditions. Exposure to metformin promoted the increase of P-AMPK expression in the cytoplasm and nucleus in both conditions. In addition, both pharmacological (compound C) and genetic (shRNA) inhibitions of AMPK abolished the autophagy-stimulating and protective effects of metformin, indicating that they are AMPK-dependent. Interestingly, our findings on P-AMPK nuclear localization appear to contradict those of a study reporting that although oxidative stress triggered by diethyl maleate (DEM) induced total-AMPK nuclear accumulation, the AMPK phosphorylation and the nuclear/cytosol ratio of p-AMPK were reduced [42]. One possible explanation for this discrepancy may be the use of different cell lines and reagents to induce oxidative stress. Whereas the previous study tested the effect of DEM on HeLa cells, we used H 2 O 2 to induce oxidative stress in D407 cells. Furthermore, although metformin-dependent AMPK activation has been reported to stimulate autophagy and to have a neuroprotective role in many studies, some exceptions have also been reported. For example, in cerebral ischemia, AMPK overactivation by a large dose of metformin treatment led to prolonged astrocytic glycolysis and subsequently to exacerbated stroke injury [43][44][45]. In addition, metformin-mediated AMPK activation inhibited autophagy in cancer cells [46,47]. Another study also demonstrated that AMPK activation triggered by glucose starvation was able to prevent both autophagosome formation and lysosome acidification [48]. The precise mechanisms and consequences of metformin-regulated autophagy in different cell types and conditions are an interesting question that needs further investigation.
In our study, we found that metformin could protect D407 cells from H 2 O 2 -induced oxidative damage by stimulating autophagy via the activation of AMPK pathway. In addition, metformin also conferred a similar protection in primary cultured human RPE cells. In summary, our study demonstrates that metformin is able to protect RPE cells against H 2 O 2 -induced oxidative damage, and this effect is attributed to its ability to enhance autophagy by activating the AMPK pathway ( Figure 8). Our findings reveal the underlying mechanisms of the antioxidant and autophagymodulating effects of metformin in RPE cells and support its potential use in AMD prevention and treatment.

Data Availability
The data used to support the findings of this study are included within the article.  Figure 8: Metformin proposed the mechanism of action. Metformin activates the AMPK pathway, which further stimulates autophagy, promotes the binding of autophagosomes and lysosomes to ensure the smooth flow of autophagic flux, and prevents the cell damage caused by hydrogen peroxide. 12 Oxidative Medicine and Cellular Longevity Fig.s1: The optimal infection efficiency of the RFP-GFP-LC3 lentiviral is 80MOI. D407 cells were infected with RFP-GFP-LC3 lentivirus with different MOI (20, 40, and 80). Images were taken after 48 h by using a Nikon A1. (Supplementary  Materials)