Prolyl-4-Hydroxylases Inhibitor Stabilizes HIF-1α and Increases Mitophagy to Reduce Cell Death After Experimental Retinal Detachment

Abstract

Purpose: This study investigated the neuroprotective effect against photoreceptor cell death using prolyl-4-hydroxylases inhibitor (PHI), an HIF-1α stabilizer, in experimental retinal detachment (RD).

Methods: RD was created in Brown Norway rats by subretinal injection of 1% sodium hyaluronate. FG-4592 (a PHI, 25 mg/kg) or a vehicle was administered every 2 days with retro-orbital injection. Photoreceptor death was evaluated by TdT-dUTP terminal nick-end labeling (TUNEL) assay 3 days after RD and by the thickness of the outer nuclear layer 7 days after RD. The mitophagy-related markers Hypoxia Inducible Factor 1α (HIF-1α), BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3), autophagy-related gene 5 (Atg5), microtubule-associated protein 1 light chain 3 beta (LC3B), and FUN14 domain containing 1 (FUNDC1) were detected by Western blot and immunofluorescence. Transmission electron microscopy was used to observe ultramicro-morphological changes. Mitochondrial damage was evaluated by the measurement of reactive oxygen species (ROS) by in situ ROS detection with dihydroethidium.

Results: The accumulation of HIF-1α and BNIP3 significantly increased after PHI treatment (P < 0.05), the pattern of Atg5 and LC3 changed, and FUNDC1 and LC3 were colocated. More autophagic vacuoles engulfing mitochondria were observed in transmission electron microscopy sections after PHI treatment when compared with the control. ROS significantly decreased in the PHI-treatment group (P < 0.05). This resulted in reduced TUNEL-positive photoreceptors 3 days after RD and an increased thickness of the outer nuclear layer 7 days after RD (P < 0.05).

Conclusions: HIF-1α stabilization as a result of PHI treatment, along with the enhancement of mitophagy, could provide protection against photoreceptor injury following RD, which might be mediated by excessive ROS generation.

Retinal detachment (RD), defined as the separation of the neurosensory retina from retinal pigment epithelium (RPE), is one of the primary causes of visual impairment. It can occur spontaneously or secondary to other diseases, such as age-related macular degeneration, diabetic retinopathy, or pathologic myopia.^1,2 Photoreceptor cell death is the major pathologic change subsequent to RD. The separation of the neurosensory retina from the underlying RPE reduces photoreceptor O₂ and nutrient supply, leading to an excessive generation of reactive oxygen species (ROS) and consequent induction of apoptosis and other types of cell death.^3–5 Clinically, the recovery of visual function remains unsatisfactory even after successful anatomical reattachment of the neurosensory retina to the underlying RPE, mainly because of the loss of photoreceptor cells.^6,7 Recent studies have identified that intrinsic prosurvival pathways are activated after RD and serve to counteract or mitigate the activation of cell death.^8–13 Zacks and his colleagues have demonstrated that prosurvival pathways, including the interleukin-6 pathway and the autophagy pathway, become activated after RD.^14–16 Furthermore, their team recently published results that the activation of autophagy was Hypoxia Inducible Factor 1 (HIF-1)–dependent on the early stage of RD.¹⁷ 

HIF-1 is a heterodimer composed of oxygen-sensitive HIF-1α and constitutively expressed HIF-1β subunits. Under normoxic conditions, the HIF-1α subunit is rapidly degraded by the ubiquitin–proteasome pathway following hydroxylation of proline residues by HIF prolyl-4-hydroxylases (prolyl hydroxylation domain protein, PHD).¹⁸ The functional activity of PHD is dependent on molecular oxygen. During hypoxia or ischemia, such as separation of the photoreceptor from RPE, PHD is less active, leading to the inhibition of HIF-1α degradation and causing HIF-1α to transfer to the nucleus where it recruits HIF-1β, inducing the expression of specific target genes, such as BNIP3. The HIF-1-dependent expression of BNIP3 is reportedly essential in mitochondrial autophagy, also called selective mitophagy.¹⁹ FUN14 domain containing 1 (FUNDC1) is a receptor for hypoxia-induced mitophagy, and its interaction with microtubule-associated protein 1 light chain 3 (LC3) is an important marker of mitophagy. In addition, the clearance of damaged mitochondria by mitophagy is necessary to prevent increased levels of ROS.^19–22 

FG-4592 is a small-molecule Prolyl-4-Hydroxylases Inhibitor (PHI) that can stabilize HIF-1α in normoxic conditions. Earlier reports have demonstrated that administration of its analogs can induce an accumulation of HIF-1α and consequent neuroprotection in the central nervous system.^18,23 However, it is unknown whether the administration of FG-4592 can stabilize HIF-1α in photoreceptors after RD. Thus, we hypothesized that administration of FG-4592 can stabilize HIF-1α, enhance mitophagy, and reduce the generation of ROS, leading to protection of the photoreceptor from damage after RD. 

In this study, we observed the histological changes in an RD animal model after PHI treatment. The result indicated that PHI treatment could enhance mitophagy and attenuate retinal histopathologic damage in the photoreceptor layer after RD. 

Methods

Animal Experiments

All of the experiments were performed in accordance with the Statement of Association for Research in Vision and Ophthalmology for the Use of Animals in Ophthalmic and Vision Research. The protocols were approved by the Shanghai General Hospital institutional review board. 

Brown-Norway rats were provided by the Laboratory Animal Center at the institution, and retinal detachments were created in the left eyes of male rats (8–10 weeks; 180–250 g) as described previously.⁹ Briefly, the rats were anesthetized with 1% sodium pentobarbital (Sigma-Aldrich Chemical Co., St. Louis, MO, USA), and the pupils were dilated with 0.5% tropicamide and 0.5% phenylephrine hydrochloride eye drops (Santen Pharmaceutical Co. Ltd, Osaka, Japan). In each rat, the sclera was punctured approximately 1.5 mm posterior to the limbus with a 30-gauge needle with special caution taken to avoid damaging the lens. The needle was slowly advanced into the vitreous cavity through the sclerotomy and then to the subretinal space by producing a small hole in the peripheral retina. Sodium hyaluronate (10 mg/mL, LG Life Sciences, Jeollabuk-do, Korea) was slowly injected until almost two-thirds of the neurosensory retina detached from the underlying RPE. RD was confirmed for each rat by a surgical microscope. The blebs of RD were created and remained essentially stable for 7 days. All of the procedure steps were performed on the left eyes of the control group, except for the introduction of the subretinal injector and injection of the sodium hyaluronate. A total of 102 rats were used, and all animal experiments contained at least three rats per group. 

FG-4592 (Selleck Chemicals, Houston, TX, USA) was administrated as in previous studies.^24,25 FG-4592 was dissolved in 20% dimethyl sulfoxide (DMSO) and diluted with 0.9% sodium chloride (NaCl). The left eye of each rat was retro-orbitally injected with 25 mg/kg FG-4592 every 2 days, and an equal volume of 20% DMSO diluted with 0.9% NaCl was administrated with retro-orbital injection and served as the control. All of the detection time points were selected at 4 to 6 hours after retro-orbital injection. 

Western Blot Analysis

The retinae (attached, 1 day, 3 days, 5 days, and 7 days after RD) were homogenized and lysed with buffer containing 50 mM hydroxymethyl(Tris)-aminomethane(Hcl), 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, and a protease inhibitor tablet. Samples were run on 8% to 12% 2-hydroxyethyl (Bis)-hydroxymethyl (Tris) gel electrophoresis and transferred onto polyvinylidene difluoride (PVDF) membranes (0.2-mm pores). After blocking with 3% nonfat dried milk, the membranes were incubated overnight with primary antibody HIF-1α (Novus Biologicals, Littleton, CO, USA), LC3 (Sigma-Aldrich Chemical Co.), BNIP3, autophagy-related gene 5 (Atg5), and β-actin (Cell Signaling Technology, Boston, MA, USA). The blotted membranes were then incubated for 60 minutes at room temperature with a horseradish peroxidase (HRP)-labeled secondary antibody. Immunoreactive bands were visualized by enhanced chemiluminescence (ECL) and detected with an Amersham Imager 600 (GE, Fairfield, CT, USA). A minimum of three rats were used for each condition. 

Immunofluorescence Staining

Immunofluorescence staining was performed 3 days after RD. The eyes (n = 3, per group) were enucleated, embedded in OCT compound, and frozen in liquid nitrogen. The retinas were vertically sectioned in a cryostat set at 10 μm and mounted on microscope slides. After fixation in acetone, nonspecific binding was blocked by 5% skim milk for 1 hour. Subsequently, sections were incubated with rabbit anti-rat-LC3B (Sigma-Aldrich Chemical Co.) and goat anti-rat-FUNDC1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at 4°C overnight. FITC-conjugated goat anti-rabbit IgG and Texas Red-conjugated mouse anti-goat IgG were used as the secondary antibodies and incubated at room temperature for 1 hour. At the end of the process, DAPI was used to counterstain the nuclei. Images of the retinas were taken with an upright fluorescence microscope. 

Transmission Electronic Micrographs of the Photoreceptor Cell

Transmission electronic microscopy (TEM) was performed 3 days after RD as previously described.²⁶ For electron microscopy examination, the eyes remained immersed in 4% glutaraldehyde (0.1 mol/L phosphate buffer, pH 7.4) for 24 hours at 4°C. The posterior part of the retina was used as the specimen for TEM. Photoreceptor cells were photographed and subjected to quantification of autophagosomes, amphisomes, and autolysosomes and to evaluate mitochondrion, the membranous disc, and the nucleus of the photoreceptor in a masked manner. 

Autophagosomes (also referred to as initial autophagic vacuoles, AVi) have a double membrane that is usually at least partly visible as two parallel membrane bilayers that contain cytosol and/or organelles that look morphologically intact. Amphisomes can sometimes be identified by the presence of small internal vesicles inside the autophagosome/autophagic vacuole (AV). Late or degradative autophagic vacuoles and autolysosomes (AVd) usually have only one limiting membrane and contain cytoplasmic material and/or organelles at various stages of degradation.²⁷ 

In Situ ROS Detection

Dihydroethidium (DHE; 5 μM; Sigma-Aldrich Chemical Co.) was used as described previously to detect superoxide in fresh-frozen eye sections (10 μm) 3 days after RD.^28–30 Sections were viewed under a fluorescence microscope (excitation 546 nm, detection 590 nm), and all images were taken with uniform exposure settings (100 ms). The fluorescence intensity of retinal sections was measured with ImageJ software (National Institutes of Health, Bethesda, MD, USA). A minimum of three rats were used for each condition. 

TdT-Mediated Fluorescein-16-dUTP Nick-End Labeling (TUNEL) Assay

Eyes (n = 3, per group) were immediately enucleated 3 days after RD, fixed with 10% formalin, and embedded in paraffin. Transverse sections (5-um thick) were prepared, and the sections around the maximal height of the retinal detachment were selected. Apoptotic cells were detected by TUNEL assay. The assay was performed using the In Situ Cell Death Detection Kit, Fluorescein (Roche Molecular Biochemicals, Mannheim, Germany), according to the manufacturer's protocol. Retinal sections were counterstained with DAPI to reveal cell nuclei. The number of TUNEL-positive cells was counted in the outer nuclear layer (ONL) in a masked fashion. The area of the ONL was also measured by ImageJ software, and the TUNEL-positive cell density in the ONL was calculated. The average TUNEL-positive cell density at two parts of the retina was calculated as the representative TUNEL-positive photoreceptor cell density of the section.^31,32 

Assessment of Retinal ONL Damage

The eyes (n = 3, per group) were enucleated 1 week after RD and were embedded in paraffin. The sections were then stained with hematoxylin and eosin. Retinal histoarchitecture was evaluated by light microscopy. The ONL thickness on day 7 after experimental RD was determined by ImageJ software. The thickness of the inner nuclear layer (INL) was usually used as a control to account for possible differences in angles of sectioning and to allow for intersample comparison. The thickness ratio of the ONL to INL was calculated in a blind manner. Areas of abnormal retinal morphology were excluded.^10,31 

Statistical Analysis

The data were presented as the mean ± standard deviation. Analyses were performed using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA). If the data showed normal distribution and variances were equal, Student's t-test or one-way analysis of variance were used to analyze the means of each group. If the data did not meet normal distribution or the variances were not equal, the Mann-Whitney U or Kruskal-Wallis tests were used. A P value less than 0.05 was considered statistically significant. 

Results

Expression of HIF-1α and BNIP3 Increased Under Retinal Detachment

The expressions of HIF-1α and BNIP3 were detected after RD at different time points. Proteins of HIF-1α and BNIP3 were detected by Western blot (Fig. 1A). HIF-1α protein levels post-RD peaked at days 3 and 5 post-RD (P < 0.05), which was paralleled by BNIP3. This suggests that HIF-1- and BNIP3-dependent autophagy were activated after RD, consistent with the results of Shelby¹⁷ (Fig. 1B). 

Figure 1

 

Detection of HIF-1α and BNIP3 protein by Western blot. (A) Retina were dissected from control and RD eyes at different time points and assessed for HIF-1α and BNIP3 proteins. In all tissues, the control protein β-actin was observed. HIF-1α and BNIP3 proteins peaked on days 3 and 5 after RD. (B) The signal intensities of the HIF-1α and BNIP3 proteins were measured by ImageJ software (National Institutes of Health, Bethesda, MD, USA) and shown as the mean ± standard deviation. Data represent the ratios of target protein antibody staining to β-actin (*P < 0.05, n = 3 rats per time point). Att, attached retina; 1D, 1-day detached retinas; 3D, 3-day detached retinas; 5D, 5-day detached retinas; 7D, 7-day detached retinas.

PHI Treatment Enhanced the HIF-1-Related Mitophagy Pathway

We found that the level of HIF-1α protein detected by Western blotting significantly increased after PHI treatment on day 3 post-RD (P < 0.05); a higher expression of BNIP3 when compared with the control was also found on day 3 post-RD (P < 0.05) (Fig. 2). 

Figure 2

 

Protein levels of HIF-1α, BNIP3, Atg5, and LC3 detected by Western blot analysis of retina 3 days after RD. (A) Western blot showing that PHI treatment raised levels of HIF-1α and BNIP3, increased conversion of LC3-I to LC3-II, and reduced Atg5 cleaving. β-actin served as the control. (B) Densitometry ratios of HIF-1α/β-actin, BNIP3/β-actin, and full-length Atg5/cleaved Atg5 and LC3-II/LC3-I are shown. Animals of the PHI-treatment group showed a significant increase in HIF-1α/β-actin, BNIP3/β-actin, and full-length Atg5/cleaved Atg5 and LC3-II/LC3-I when compared with animals of the vehicle-treated group (*P < 0.05, n = 3 rats per group).

The ratio of LC3-II to LC3-I is regarded as a marker for autophagy; LC3 migrates as two bands on Western blotting: LC3-I (its inactive form) and LC3-II (its active form). LC3-II is a marker of autophagy and is indicative of autophagosome formation. FUNDC1 is located at the mitochondrion, and its colocalization with punctate LC3, from a diffused pattern, indicates increased mitophagy.²⁰ Our Western blotting assay showed that the ratio of LC3-II to LC3-I of the PHI-treatment group was significantly higher than that of the vehicle-treatment group 3 days after RD (P < 0.05) (Fig. 2). Immunofluorescence demonstrated that there were more punctate LC3 forms in the PHI-treatment group than in the vehicle-treatment group, and the colocalization of punctate LC3 and FUNDC1 was detected in the PHI-treatment group 3 days after RD (Fig. 3). 

Figure 3

 

LC3 staining (green) and FUNDC1 (red) staining on day 3 post-RD. More punctate LC3 staining was demonstrated after PHI treatment than with vehicle treatment; punctate staining was accompanied by colocalization of FUNDC1 and LC3. DAPI (blue) was used to highlight the outer nuclear layer. The arrow indicates punctate-like structures, and the asterisk indicates colocalization of FUNDC1 and LC3. OS, outer segments; IS, inner segments.

FUNDC1-induced mitophagy is dependent on Atg5, and a ratio of full-length Atg5 (33 kDa) to cleaved Atg5 (24 kDa) indicates the level of autophagy/apoptosis, which was shown to be significantly elevated in the PHI-treatment group than in the vehicle-treatment group 3 days after RD (P < 0.05) (Fig. 2). 

TEM was employed to further investigate which type of autophagy was enhanced after RD combined with PHI treatment: selective autophagy-mitophagy or nonselective autophagy? It was demonstrated with TEM that AVi, lots of amphisomes, and AVd were all identified on day 3 post-RD when combined with PHI treatment. In addition, characteristics of mitophagy were found in the PHI-treatment group that indicated that the contents of AVi and AVd were mainly mitochondria. In contrast, only a few AVi could be found in the vehicle-treatment group. We found that the mitochondria in cytoplasm, the membranous disc, and the nucleus of the photoreceptor in the PHI-treatment group were more intact than those in the vehicle-treatment group (Fig. 4). 

Figure 4

 

TEM images of mitophagy in the retina on day 3 post-RD. (A) One autophagosome or AVi of the retina of the PHI-treatment group is shown. The AVi of mitophagy could be identified by its contents (morphologically intact mitochondria) and the limiting membrane that was partially visible as two bilayers separated by a narrow electron-lucent layer. (B) One autolysosome or AVd of the retina of the PHI-treatment group is shown. The AVd of mitophagy could be identified by its contents of partially degraded mitochondria. (C, D) The mitochondria in the cytoplasm of the photoreceptors in the retina of the PHI-treatment group were more intact and showed less swelling than those in the vehicle-treatment group. mi, mitochondrion.

PHI Treatment Decreased Generation of ROS

ROS accumulation was evaluated with DHE staining in fresh-frozen eye sections, and the results showed that DHE fluorescence was most intense in the photoreceptor layer on day 3 post-RD, and it decreased significantly in the retinas of the PHI-treatment group when compared with those of the vehicle-treatment group (P < 0.05) (Fig. 5). 

Figure 5

 

DHE imaging of ROS accumulation in a fresh-frozen retinal section on day 3 post-RD. (A) DHE fluorescence weakened after PHI treatment when compared with that in vehicle treatment. Fluorescence was particularly prominent in the outer nuclear layer after RD. (B) Quantitative analysis of DHE fluorescence intensity showed a marked decrease in ROS accumulation in the retina of the PHI-treatment group (*P < 0.05, n = 3 rats per group).

PHI Treatment Decreased TUNEL-positive Photoreceptor Cells

We assessed photoreceptor death 3 days post-RD using TUNEL staining, which detects DNA fragmentation in apoptotic or necrotic nuclei. The time point was selected because TUNEL-positive photoreceptors are mostly found on day 3 post-RD in the rat RD model.³ PHI treatment showed a significant decrease in TUNEL-positive cells on day 3 post-RD when compared with the vehicle-treatment group, indicating a neuroprotection effect of PHI treatment (P < 0.05) (Fig. 6). 

Figure 6

 

TUNEL (green) and DAPI (blue) staining of representative sections on day 3 post-RD. (A) Fewer TUNEL-positive cells were detected in the PHI-treatment group than in the vehicle-treatment group. (B) TUNEL-positive cell density (cells/mm²) in the outer nuclear layers is shown, suggesting that TUNEL-positive cell density significantly decreased after PHI treatment. (*P < 0.05, n = 3 rats per group).

PHI Treatment Decreased ONL Damage and Protected the Retina

To verify if PHI treatment would relieve photoreceptor damage after RD, we investigated the histopathologic changes of the retina 7 days post-RD when disorders of the photoreceptor layer are obviously observed in the rat RD model. The ratio of ONL to INL was significantly higher in the PHI-treatment group when compared with that of the vehicle-treatment group, again demonstrating neuroprotection from PHI treatment (P < 0.05) (Fig. 7). 

Figure 7

 

Analysis of the ratio of ONL to INL in the retinas 7 days after RD. (A) Representative histopathologic images of retinal sections from vehicle-treated rat eyes and PHI-treated rat eyes 7 days post-RD and from attached rat eyes. (B) PHI treatment led to better preservation of the ratio of ONL to INL than vehicle treatment (*P < 0.05, n = 3 rats per group).

Discussion

In this study, we found that PHI treatment significantly induced the accumulation of HIF-1α and upregulated the HIF target gene, BNIP3, enhancing mitophagy along with subsequently decreasing the generation of ROS. This intervention attenuated retinal histopathologic damage in the photoreceptor layer after RD. 

Several reports have identified that both cell survival and cell death pathways are activated soon after photoreceptor–RPE separation.^{1,3,8–12,32,33} Zacks and his colleagues demonstrated that some prosurvival pathways become activated after RD and HIF-1-dependent autophagy is activated at an early stage of RD.^14–17,34 In this study, we confirmed that HIF-1-dependent autophagy was activated after RD. Using Western blotting, we found that protein levels of HIF-1α and BNIP3 from detached rat retinas peaked on days 3 and 5 post-RD and thereafter decreased. This was similar to the results of Shelby¹⁷ except for a slightly delayed peak. Different strains of experimental animals may be responsible for this discrepancy.³¹ 

As previously shown in stroke and hypoxia models, the systemic administration of PHI protects the brain from ischemic injury.^18,23 In the current study, retro-orbital administration of PHI was selected instead of systemic administration to improve intraocular bioavailability and to reduce systemic toxicity. The time point to compare HIF-1α protein levels between the PHI-treatment group and the vehicle-treatment group was chosen at 3 days post-RD because the level of hypoxia is the most severe in the rat RD model at that time.^1,35,36 Our results showed that PHI treatment improved the expression of HIF-1α under hypoxia secondary to RD. This was not in line with the results of Trollmann,¹⁸ whose studies showed that HIF-1α stabilization was maximal under hypoxia. A possible reason for this discrepancy is that the extent of hypoxia in RD was distinct from that in their brain disease models. 

BNIP3 is an important HIF-1 target gene. It protects the cell from cell death during damage by displacing Beclin1 from Bcl-2/Beclin1 or Bcl-XL/Beclin1 complexes and activating mitophagy, a metabolic adaptation for survival that is able to control ROS production and DNA damage.^17,19,21,22 Our findings revealed that the expression of BNIP3 was elevated secondary to stabilization of HIF-1α after PHI treatment, suggesting that BNIP3-related mitophagy was intensified. 

FUNDC1 has previously been demonstrated to be a receptor of selective mitophagy in response to hypoxia by specifically interacting with LC3-II²⁰; therefore, an increase in the colocalization and interaction between FUNDC1 and LC3-II indicates intensive mitophagy.²⁰ LC3 plays a critical role in both autophagosome membrane biogenesis and target recognition. Upon autophagic stimuli, LC3-I is converted into LC3-II; therefore, the LC3-II to LC3-I ratio on Western blotting is the primary biochemical marker of general autophagy activation and autophagosome formation.²⁷ The increased level of LC3-I conversion to LC3-II can also be identified by examining the staining pattern of LC3 on immunohistochemistry. A punctate pattern changed from a diffused pattern indicates autophagy activation.²⁷ In our study, the increased ratio of LC3-II to LC3-I was obvious on our Western blots after PHI treatment. Our immunohistochemistry showed that the diffused pattern was changed to the punctate pattern of LC3, along with the presentation of colocalization of FUNDC1 and LC3, indicating enhancement of mitophagy. 

Earlier studies showed that FUNDC1-induced mitophagy is dependent on Atg5 because knockdown of Atg5 completely blocks mitophagy induced by FUNDC1.²⁰ In addition, Atg5 could regulate the shift from protective autophagy to apoptosis, indicating its dual role in both pathways, and recent studies have showed that this dual function was regulated by the proteolysis of Atg5. Upon lethal stress, the 33-kDa full-length Atg5 protein is cleaved by calpains to remove the C terminus, generating a 24-kDa fragment, which loses its autophagy-inducing activity and instead acquires a pro-apoptotic one. Therefore, the ratio of full-length Atg5 to cleaved Atg5 indicates the level of autophagy and apoptosis.³⁷ In our study, the ratio of full-length Atg5 to cleaved Atg5 was significantly increased after PHI treatment, indicating that the extension of autophagy correlated with the reduction of apoptosis, which contributed to the attenuation of photoreceptor damage. 

TEM is an important method for monitoring autophagy and can be used to monitor both nonselective and selective autophagy, such as mitophagy. In the case of mitophagy, the mitochondria comprise a prominent component of AVs, whereas bulk cytoplasm is essentially excluded. In contrast, during nonselective autophagy, the content of the AV is morphologically identical to the cytoplasm.²⁷ Through TEM, we found that more AVs, mainly consisting of mitochondria, appeared after PHI treatment, indicating increased mitophagy. Moreover, we detected that mitochondria outside the AV were less damaged after PHI treatment when compared with the vehicle treatment, suggesting that mitophagy could preserve mitochondria by clearing damaged mitochondria. 

The clearance of damaged mitochondria by mitophagy is necessary to prevent increased levels of ROS, which play a critical role in photoreceptor death after RD. Their reduction has been demonstrated to be associated with a neuroprotective effect on photoreceptors after RD.^4,19,38 Our DHE results confirmed the reduction of ROS secondary to intensive mitophagy after PHI treatment. 

Reduced TUNEL-positive photoreceptors 3 days post-RD and an elevated thickness of the photoreceptor layer 7 days post-RD because of a decreased ROS level were demonstrated by our TUNEL assay and histological section analyses. These findings suggest that PHI treatment could protect photoreceptors from damage after RD by reducing ROS generation. 

Compared with previous studies, to further study the role of HIF-1α in RD, we administrated PHI by retro-orbital injection to stabilize HIF-1α, which was proven effective in decreasing photoreceptor cell death after RD by reducing ROS generation. The way of drug delivery involved in our study may provide a potential therapy in the clinical application of photoreceptor protection after RD. Moreover, we confirmed that the type of autophagy enhanced by the stabilizing of HIF-1α in RD was mainly selective autophagy (mitophagy) rather than nonselective autophagy, providing new insights into the role of autophagy participating in photoreceptor cell death after RD. 

By using FG-4592, the current study provided evidence that PHI is a promising therapeutic agent that protects photoreceptors from apoptosis and tissue damage after RD. The reduction of ROS secondary to the clearance of damaged mitochondria by HIF-1-dependent mitophagy was the mechanism of this neuroprotection. Therefore, PHI may provide us with a new agent to minimize cell death after RD. 

Acknowledgments

Supported by the National Science Fund for Distinguished Young Scholars of China (81425006), the National Natural Science Foundation of China (81400413), and the Xuzhou Technology Program (XM13B077). The authors alone are responsible for the content and writing of the paper. 

Disclosure: H. Liu, None; H. Zhu, None; T. Li, None; P. Zhang, None; N. Wang, None; X. Sun, None