Reactive oxygen species‐responsive mitochondria‐targeted liposomal quercetin attenuates retinal ischemia–reperfusion injury via regulating SIRT1/FOXO3A and p38 MAPK signaling pathways

Abstract Retinal ischemia–reperfusion (RIR) injury is involved in the pathogenesis of various vision‐threatening diseases. The overproduction of reactive oxygen species (ROS) is thought to be the main cause of RIR injury. A variety of natural products, including quercetin (Que), exhibit potent antioxidant activity. However, the lack of an efficient delivery system for hydrophobic Que and the presence of various intraocular barriers limit the effective retinal delivery of Que in clinical settings. In this study, we encapsulated Que into ROS‐responsive mitochondria‐targeted liposomes (abbreviated to Que@TPP‐ROS‐Lips) to achieve the sustained delivery of Que to the retina. The intracellular uptake, lysosome escape ability, and mitochondria targeting ability of Que@TPP‐ROS‐Lips were evaluated in R28 retinal cells. Treating R28 cells with Que@TPP‐ROS‐Lips significantly ameliorated the decrease in ATP content, ROS generation, and increase in the release of lactate dehydrogenase in an in vitro oxygen–glucose deprivation (OGD) model of retinal ischemia. In a rat model, the intravitreal injection of Que@TPP‐ROS‐Lips 24 h after inducing retinal ischemia significantly enhanced retinal electrophysiological recovery and reduced neuroinflammation, oxidative stress, and apoptosis. Que@TPP‐ROS‐Lips were taken up by retina for at least 14 days after intravitreal administration. Molecular docking and functional biological experiments revealed that Que targets FOXO3A to inhibit oxidative stress and inflammation. Que@TPP‐ROS‐Lips also partially inhibited the p38 MAPK signaling pathway, which contributes to oxidative stress and inflammation. In conclusion, our new platform for ROS‐responsive and mitochondria‐targeted drug release shows promise for the treatment of RIR injury and promotes the clinical application of hydrophobic natural products.

promise for the treatment of RIR injury and promotes the clinical application of hydrophobic natural products. These physiological barriers pose a challenge in the treatment of posterior segment eye diseases. Retinal disease is difficult to be treated systemic or periocular routes of administration. 4 Conventional treatments mainly include surgery, laser treatment, eye drops, and intraocular injections (e.g., anti-VEGF) to inhibit disease progression and bypass the barriers bypass barriers as possible. 5 However, traditional drugs suffer from poor permeation, ineffective drug distribution, insufficient drug bioavailability, and poor long-term therapeutic effectiveness, resulting in the need for repeated intravitreal intervention. Therefore, new treatments for retina diseases are needed to improve patient compliance, enhance biocompatibility, and reduce side effects.
In recent decades, various herb-derived natural products have been reported to exhibit antioxidant or anti-inflammatory effects. 6 For example, Puerarin has been reported to have an anti-inflammatory effect in RIR injury by inhibiting the activation of TLR4/NLRP3 inflammasome. 7 However, in addition to the inflammation, the mechanisms of RIR injury also include oxidative stress and apoptosis. Therefore, the application of natural products with antioxidant and anti-inflammatory effects in alleviating RIR injury has more important potential clinical significance.
Quercetin (3,3 0 ,4 0 ,5,7-pentahydroxyflavonoid; Que) is a natural flavonoid product found in many vegetables and fruits. 8,9 Que has a range of pharmacological effects including anti-cancer, 10 anti-oxidant, 11 and anti-inflammatory 12 effects. It has been reported to treat ischemiareperfusion injury in multiple organs including the liver, 13 kidney, 14 heart, 15 and cerebrum. 16 Arikan et al. reported that intraperitoneal injection of Que dissolved in dimethyl sulfoxide (DMSO) attenuated the retinal thinning caused by RIR injury. 17 As mentioned earlier, the eye is sensitive and has many barriers that reduce the efficiency of intraperitoneal injection, and the hydrophobic nature of Que makes vitreous injection difficult.
The rapid development of nanomaterials science has resulted in the emergence of many nanomedicines for intravitreal injection, which can deliver the hydrophobic compounds and bypass the barriers. 18 Liposomes have shown promise as nanocarriers for the treatment of ocular diseases due to their biocompatibility, biodegradability, low toxicity, and efficient encapsulation of hydrophobic drugs. 19 Although liposome-based drugs for intravitreal injection have not yet been marketed, they have been reported to improve the intravitreal administration of drugs (e.g., vincristine). 18 Some nanomicelles and inorganic nanoparticles have also been reported for use in ocular diseases, these nanomaterials have more disadvantages compared with liposomes. For example, poly (lactic-co-glycolic acid) nanomicelles have problems such as relatively low drug loading, inappropriate release rate, and potential toxicity, 20 while inorganic nanoparticles have been reported disadvantages of low solubility and toxicity concerns. 21 Liposomes have numerous advantages as drug carriers, which can significantly increase the duration of drug treatment effects, as well as drug levels in the posterior segment of the eye. 22 However, one of the potential disadvantages of conventional liposomes is the burst release of nonspecific cargo in vivo. 23 In recent years, a variety of ROS-responsive materials have been synthesized and studied, including thiols, thioketals, hydroquinones, metallocenes, polypyridine ruthenium complexes, and thioethers. 24 Increasing studies reported that ROS-based nanomaterials can be used to treat myocardial ischemia-reperfusion injury. 25,26 In our previous study, we reported ROS-responsive lipids (Di-S-PC) composed of thioether phosphatidylcholines in which the fatty acid chain of a typical lipid is replaced by two tails with thioether linkages. 27 This ROSresponsive element can be applied in stimuli-responsive liposomes with good biocompatibility to obtain drug carriers targeting environments of oxidative stress. Thus, the design of ROS-responsive liposomes would specifically improve the efficacy of the drugs.
Emerging evidences suggested that a major early event in RIR injury is ROS-induced oxidative stress in the retina, which leads to retinal ganglion cell (RGC) loss, inflammation, and vascular dysfunction in posterior segment disorders. 28,29 Mitochondria are an important source of ROS in most mammalian cells. 30 Therefore, we can construct mitochondria-targeted nanomedicines to reduce ROS levels at earlier pathological stages. Triphenylphosphonium (TPP) is a positively charged lipophilic cation that preferentially accumulates in negatively charged mitochondria. Based on this feature, TPP has been conjugated on the surfaces of nanocarriers to develop a general mitochondria-targeted drug delivery system. Recently, TPP-based mitochondrial antioxidant delivery systems such as Mito-TEMPO, 31 MitoQ, 32 MitoC, 33 MitoE, 34 TPP-IOA, 35 and lipid-polymer hybrid nanoparticles 36 have shown cardioprotective and antitumor effects by reducing mitochondrial ROS accumulation. TPP-conjugated niacin protects against hydrogen peroxide (H 2 O 2 )-induced cytotoxicity and mitochondrial dysfunction by upregulating antioxidant-related genes in retinal pigment epithelial (RPE) cells. 37 However, no in vitro or in vivo studies have been reported on the development of mitochondria-targeted TPP-conjugated liposome drugs for the treatment of ocular diseases.
To the best of our knowledge, no ROS-responsive, mitochondriatargeted lipid-based nanosystem for the delivery of Que has been developed for the treatment of ocular diseases. In the present study, we developed a novel ROS-responsive liposomal quercetin nanoformulation (Que@ROS-Lips). The surfaces of the liposomes were functionalized by TPP to construct mitochondria-targeted Que@ROS-Lips (Que@TPP-ROS-Lips). We then evaluated the role of Que@TPP-ROS-Lips in RIR injury both in vitro and in vivo. The treatment of R28 cells with Que@TPP-ROS-Lips and the intravitreal injection of Que@TPP-ROS-Lips showed therapeutic effects in RIR injury in vitro and in vivo, respectively. Furthermore, mechanistic study showed that Que@TPP-ROS-Lips inhibited the activation of the mitogen-activated protein kinase (MAPK) signaling pathway, which can promote oxidative stress and inflammatory response (Scheme 1). These findings provide a new strategy for the treatment of RIR injury and promote the clinical application of Que.

| Synthesis of ROS-responsive lipids
A two-step procedure similar to previously studies 27,38 was used to construct ROS-responsive lipids (Di-S-PC). Briefly, to a solution of thioglycolic acid (0.56 g, 6.01 mmol) in 25% KOH solution in methanol, octadecane bromide (2 g, 6.01 mmol) was added and stirred vigorously at room temperature for 48 h. After acidification to pH 1 with aqueous HCl (0.1 M), the solution was extracted with ethyl acetate, dried over anhydrous NaSO 4 , and purified by column chromatography on silica gel (hexane/EtOAc, 3:1) to yield 0.86 g C18 S COOH as a white solid (41% yield). 1

| Preparation and characterization of liposomes
Que@ROS-Lips and other liposomes were formulated using a thin-film hydration method. 39

| Lysosome escape and mitochondria-targeting ability
After nanoparticles enter cells, their localization within the cells is closely related to their ability to escape the lysosomes. 41 Thus, we evaluated the ability of Que@TPP-ROS-Lips to escape lysosomes by confocal laser scanning microscopy (CLSM). The cells were incubated with fluorescent-labeled liposomes (Lyso-Tracker red) for various times (2, 4, 6, 8, and 16 h). After 30 min, the experiment was terminated by rinsing the cells three times with PBS, which removed the Lyso-Tracker dye. The cells were then fixed with paraformaldehyde (4%, v/v), and the nuclei were stained with DAPI.

| In vitro oxygen-glucose deprivation model and detection of cell activity
An in vitro oxygen-glucose deprivation (OGD) model was constructed based on the method detailed by Roth et al. 5 Briefly, the cells were cultured in an anoxic, serum-free environment for 24 h. The cells were then returned to normal culture conditions for 18 h to simulate RIR.
The ATP levels were measured by CellTiter-Lumi™ Plus luminescent cell viability assay (Beyotime, Beijing, China). The cell membrane integrity of the R28 cells was evaluated by lactate dehydrogenase (LDH) release assay (Beyotime).

| Analysis of ROS content
The ROS in treated R28 cells were analyzed in vitro using an ROS assay kit (Beyotime). Briefly, the cells were incubated with liposomes

| Measurement of glutathione level
The intracellular reduced glutathione (GSH) levels were assessed after the 24-h treatment of OGD-injured R28 cells with Que@TPP-ROS-Lips using GSH and GSSG Assay Kit (Beyotime) according to the manufacturer's suggestion.
The Yantai University Committee for the Care and Use of Laboratory Animals authorized the animal protocols, which followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals in Research. All rats were housed under a 12-h dark/light cycle and given unlimited access to water and food. The rats were anesthetized with intraperitoneal injections of ketamine (100 mg/kg) and xylazine (7 mg/kg) prior to intravitreal injection.

| RIR injury model and intravitreal injection
RIR injury was established in rats using a previously published procedure. 29 The conjunctival sac of each anesthetized rat was cleaned with 0.3% ofloxacin eye drops. Tropicamide eye fluid was used to dilate the pupils, proparacaine hydrochloride was used as a local anesthetic, and tobramycin was used to prevent infection after sur-

| Preparation of frozen retina sections and immunofluorescent staining
The model rats were euthanized at 7 days after the intravitreal injection of liposomes. Immediately after the eyeball was removed, it was fixed in 4% paraformaldehyde for 2 h at 4 C, dehydrated overnight with 30% sucrose solution, embedded in an optimal cutting temperature compound (OCT) embedding bottom box (17 Â 17 Â 5 mm), and stored at À80 C. Slices with thicknesses of 8 μm were created using a frozen slicer at À22 C. The slices were placed on adhesive slides and dried at room temperature for more than 30 min before proceeding to the next step. For immunofluorescent staining, the retinal sections were permeabilized with enhanced immunostaining permeabilization buffer (Beyotime) for 20 min, blocked with blocking buffer (Beyotime), and incubated overnight at 4 C with primary antibodies: Class III β-TUBULIN (Tuj1 mAb) (Beyotime), Rabbit anti-IBA1/AIF-1, rabbit anti-FOXO3A and rabbit anti-SIRT1(Cell Signaling Technology).
After rinsing three times with PBS, the sections were incubated with secondary antibodies at room temperature for 2 h, rinsed three times with PBS, and counterstained with DAPI for 5 min. The slides were then sealed with cover glass for observation by CLSM.

| Evaluation of retinal function
All electroretinography (ERG) procedures were performed in the dark room under dim red-light illumination (>650 nm). 43 ERG recordings were conducted for four groups: sham, saline, TPP-ROS-Lips, and Que@TPP-ROS-Lips. The rats were dark-adapted overnight before receiving intraperitoneal 1% pentobarbital (3 ml/kg) and 50% isoflurane (250 μl/kg) anesthesia. Tropicamide eye fluid was used to dilate the pupils under dim red-light illumination. Following anesthesia and mydriasis, stainless-steel subdermal needle electrodes were implanted as the ground (at the tail) and reference (beneath individual eyelids) electrodes. The cornea was kept moist by applying 0.1% sodium hyaluronate eye drops (Santen Pharmaceutical) to the recording gold electrodes. The animals were placed on a thermal platform that was kept at 37 C. Under dark-adapted conditions, the intensity of white light stimulation for scotopic ERG was initially adjusted to 0.01 cdÁsÁm À2 , and flash ERG recordings were obtained concurrently from both eyes.
The intensity was then increased to 3.0 cdÁsÁm À2 . At the above luminance levels, each recording was averaged three times.

| Cytokine measurement by ELISA
Retinal tissue was homogenized with a low-temperature homogenizer (ServiceBio, Wuhan, China) and supernatants were then collected for

| Safety evaluation
In the section of normal rats, normal rats were intravitreal injection of Que@TPP-ROS-Lips. The rats were euthanized after 7 or 14 days, and the eyeballs were harvested for hematoxylin and eosin (H&E) staining to observe retinal changes. Frozen sections were used to detect the protein expression levels of β-III-TUBULIN and BRN3A in the retina 7 days after administration. ERG was used to detect the safety of the retinal function after 7 days of administration. In the section of Safety assessment for pathological conditions with RIR impairment, orbital blood was taken to detect the contents of alanine transferase and aspartic transferase in plasma. The heart, liver, spleen, lung, kidney, and other major organs were harvested for H&E staining to observe changes in each major organ.

| Statistical analysis
All data are presented as the mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism 6 (La Jolla, USA).
Treatment-related differences were evaluated by one-way analysis of variance (ANOVA) followed by Dunnett's test (for comparisons between different concentrations and the vehicle control) or two-way ANOVA followed by Tukey's multiple comparison test (for

| Preparation and characterization of Queloaded liposomes
In the present study, we conjugated liposomes with Di-S-PC lipids for ROS response. In addition, TPP and Cy5.5 dye were used to conjugate in the liposomes for targeting mitochondria and Lip-tracing, respectively. A schematic of Que@TPP-ROS-Lips is shown in Figure 1a

| Que@TPP-ROS-Lips attenuates decreased cell viability and oxidative stress in R28 cells subjected to OGD
In this study, the efficacy of Que@TPP-ROS-Lips in vitro was examined using the OGD model, a widely reported model for ischemiareperfusion injury in multiple organs. 5,[49][50][51] During the establishment of the OGD model, if the re-oxygenate culture time after hypoxia is too long, the cells will be close to normal physiological conditions without any treatment (data not shown). Therefore, we pretreated the cells before establishing the OGD model as this mode of administration would ensure the duration of action of Que@TPP-ROS-Lips without making the re-oxygenation time too long in vitro. Therefore, to investigate the role of Que@TPP-ROS-Lips in an in vitro OGD model, R28 cells pretreated with PBS for 24 h were subjected to OGD. As shown in Figure 3a Taken together, Que@TPP-ROS-Lips significantly inhibit OGDinduced cytotoxicity and oxidative stress in R28 cells. This observation of anti-oxidative stress activity of Que@TPP-ROS-Lips is congruent with a previous RIR injury study in which the role of SOD nanoformulations in RIR injury was attributed to their antioxidant effect. 29 The findings of our study indicated that Que@TPP-ROS-Lips inhibit the early pathological process of RIR injury and this encouraged further evaluation of Que@TPP-ROS-Lips following ischemic injury in rodent models. In the current study, compared to normal retinas, retinas exposed to RIR for 7 days showed a significant reduction in IPL thickness. Moreover, the administration of Que@TPP-ROS-Lips significantly alleviated this decrease in thickness (Figure 4a). Consistent with these findings, the H&E staining images indicated that the loss of RGCs induced by RIR injury was significantly attenuated after Que@TPP-ROS-Lips administration (Figure 4a right panel). In addition, TUNEL staining indicated that Que@TPP-ROS-Lips administration significantly decreased RIR injury-induced apoptosis in retinal cells (Figure 4b).

| Que@TPP-ROS-Lips administration following retinal ischemia in vivo attenuates ischemic damage
ERG, which is used to monitor retinal function and disease progression, 58   and macroglia by measuring the levels of their markers (IBA1 and GFAP, respectively). As we hypothesized, RIR injury increased the levels of IBA1 and GFAP. Notably, Que@TPP-ROS-Lips effectively prevented the upregulation of these two proteins (Figure 6a-d). In parallel, this upregulation was confirmed by CLSM analysis of retinal whole mounts ( Figure S5).

RIR injury increases the secretion of inflammatory factors
including TNF-α and IL-1β. 5 Thus, we assessed the expressions of these two cytokines in retinal tissues after intravitreal injection of Que@TPP-ROS-Lips using ELISA. We found that RIR injury increased the levels of TNF-α and IL-1β by approximately 2.37-and 2.16-fold, respectively, compared with the sham group (Figure 6e,f).
In the Que@TPP-ROS-Lips group, the levels of TNF-α and IL-1β were only 1.49 and 1.56 times those of the sham group, respectively (Figure 6e,f). Overall, these results demonstrate that the intravitreal injection of Que@TPP-ROS-Lips was effective at reducing inflammation associated with RIR injury.

| Safety and distribution of Que@TPP-ROS-Lips in vivo
Evaluation of the safety of the nanosystem is crucial for potential clinical applications. 61  conditions (with RIR injury). As shown in Figure S6A,B, the contents of glutamic aspartate transaminase and alanine aminotransferase in plasma were within the normal ranges, without intergroup differences.
Moreover, the H&E staining images of the main visceral organs (heart, liver, spleen, lung, and kidney) showed that treatment with Que@TPP-ROS-Lips did not cause noticeable histological changes ( Figure S6C). Taken together, these results indicate that Que@TPP-

ROS-Lips has no obvious toxic effects in vivo.
After observing the functional neuroprotection provided by Que@TPP-ROS-Lips in the ischemic retina without apparent toxic effects, we assessed the distribution of Que@TPP-ROS-Lips in the retina. As shown in Figure 7f, Que@Cy5.5-TPP-ROS-Lips were primarily localized in the ganglion cell layer and persisted in the retina for at least 2 weeks after intravitreal injection. This is congruent with previous findings of liposomes 62,63 and suggests that Que@TPP-ROS-Lips can transport, in a sustained-release manner, the hydrophobic natural product Que to the retina to exert its effect. docking was performed to screen therapeutic targets of Que. Que was docked into the binding site of FOXO3A (Figure 8a,b). FOXO3A, a transcription factor, has been reported to reduce oxidative stress in ischemia-reperfusion models by enhancing the activity of superoxide dismutase. 25 The phenyl group of Que was located at the hydropho-   (Figure 8c,d). Notably, Que@TPP-ROS-Lips reversed the OGD-induced decrease in FOXO3A expression along with the decreases in SIRT1 and SOD1 expression (Figure 8e). This finding is consistent with the previously reported regulation of SIRT1/FOXO3A, a pathway that down-regulates oxidative stress, by Rg3 in cardiac ischemia. 64 This finding was confirmed by our results in the in vivo RIR injury model (Figure 8e). Since the results of TUNEL staining showed that Que@TPP-ROS-Lips can attenuate the apoptosis of retinal cells caused by RIR injury, we also detected apoptosis-related proteins in the OGD model. We found that the cleaved form of cleaved RIR injury activates the p38 signaling pathway, and inhibition of the p38 signaling pathway can attenuate ischemic injury [65][66][67] Thus, we investigated whether Que@TPP-ROS-Lips regulates p38 activation in RIR injury. In an OGD model, R28 cells were treated with Que@TPP-ROS-Lips, and the activation of p38 was monitored. Elevated p38 activation was observed in the OGD model, and this elevation was attenuated by the addition of Que@TPP-ROS-Lips (Figure 8f,g). As expected, the in vivo data were consistent with the in vitro data, as evidenced by immunofluorescence staining to detect p38 phosphorylation (Figure 8h).
Taken together, these data suggest that Que released from Que@TPP-ROS-Lips targets FOXO3A to suppress oxidative stress caused by RIR injury. In addition, the Que-mediated inhibition of p38 activation may also have contributed to this effect. We therefore infer that our targeted liposomes for efficient delivery of Que at least partially inhibited RIR injury-induced p38 activation, which in turn directly or indirectly prevented oxidative stress and inflammation. In future studies, we will continue to explore the relationships between p38 activation, the SIRT1/FOXO3A signaling pathway, and ROS in RIR injury.

| CONCLUSION
In this study, we (1) engineered the first ROS-responsive, TPP-targeting