RNA Interference of GADD153 Protects Photoreceptors from Endoplasmic Reticulum Stress-Mediated Apoptosis after Retinal Detachment

Background Apoptosis of photoreceptors plays a critical role in the vision loss caused by retinal detachment (RD). Pharmacologic inhibition of photoreceptor cell death may prevent RD. This study investigated the role of GADD153 that participates in endoplasmic reticulum (ER) stress-mediated apoptosis of photoreceptor cells after RD. Methods Retinal detachment was created in Wistar rats by subretinal injection of hyaluronic acid. The rats were then randomly divided into four groups: normal control group, RD group, GADD153 RNAi group and vehicle group. RNA interference of GADD153 was performed using short hairpin RNA (shRNA). Expressions of GADD153 mRNA and protein were examined by RT-PCR and Western blotting analysis, respectively. GADD153 protein distribution in the retinal cells was observed using immunofluorescence confocal laser scanning microscopy. Apoptosis of retinal cells was determined by TdT-mediated fluorescein-16-dUTP nick-end labeling (TUNEL) assay. Results Lentivirus GADD153 shRNA with the most effective silencing effect was chosen for in vivo animal study and was successfully delivered into the retinal tissues. GADD153 mRNA and protein expressions in GADD153 RNAi group were significantly lower than those in the RD group. Silencing of GADD153 by RNAi protected photoreceptors from ER stress-induced apoptosis. Conclusion ER stress-mediated pathway is involved in photoreceptor cell apoptosis after RD. GADD153 is a key regulatory molecule regulating ER-stress pathways and plays a crucial role in the apoptosis of photoreceptor cells after RD.


Introduction
Acute or chronic detachment of the retina from the retinal pigment epithelium (RPE) surface is the leading cause of vision loss in patients with diabetic retinopathy, pathological myopia, posterior eye trauma and age related macular degeneration. Retinal detachment (RD) results in not only the separation of the photoreceptor cell layer from the apical surface of the RPE but also the expansion of the interphotoreceptor space. Photoreceptor cell death by apoptosis, which could be observed immediately after RD, plays a critical role in visual loss. Therefore, new insights into the mechanisms underlying photoreceptor cell apoptosis in RD would be of clinical interest and could lead to new treatments.
Growth arrest DNA damage-inducible gene 153 (GADD153), also known as C/EBP homologous protein, plays a vital role in ER stress-induced apoptosis. It has been proven to be involved in the pathogenesis of various diseases, including diabetes [1], brain ischemia [2,3] and neurodegenerative disease [4]. Over expression of GADD153 and microinjection of GADD153 protein have been reported to cause cell cycle arrest and/or apoptosis [5][6][7][8]. Previously we found that the expression of GADD153 was temporally and spatially associated with the apoptosis of photoreceptor cells, suggesting the involvement of ER stress-mediated pathway in the apoptosis of photoreceptor cell after RD [9]. Recently, researchers found CHOP2/2 mice exhibited reduced apoptosis in response to ER stress and GADD153-deficient cells were resistant to ER stress-induced apoptosis [10,11]. Considering that GADD153 has been found as a key molecule in ER stress pathway, it would be of interest to know whether interference of GADD153 could protect photoreceptor from apoptosis in RD.
In order to further confirm whether GADD153 participates in ER stress-mediated apoptosis of photoreceptor cells after RD, in this study, we suppressed GADD153 expression by injecting lentivirus GADD153 shRNA into the subretinal space, and observed the apoptosis of photoreceptor cells after RD.

Materials and Methods
Generation of Lentivirus GADD153 shRNA Lentiviral vectors encoding shRNAs against GADD153, or lentiviral vectors without encoding GADD153 shRNA were synthesized by Telebio Biomedical Co., Ltd (Shanghai, China). Vector particles were prepared by Lentivirus Expression Systems. Three constructed lentivirus shRNAs targeting different sites of GADD153 and a negative control lentivirus shRNA were transduced in HEK 293T cells (CRL11268, American Type Culture Collection, Rockville, MD) to test the efficacy [12,13] (Table. S1). The lentivirus GADD153 shRNA (LV-GADD153-sh) with the best silencing efficacy was selected for in vivo experimental RD study. ( Figure S1, Figure S2).

Animals and Experimental RD
All the experiments were humanely performed in accordance with the Statement of Association for Research in Vision and Ophthalmology for the Use of Animals in Ophthalmic and Vision Research, and the protocols were approved by the Shanghai First People's hospital institutional review board. A total of 124 male Wistar rats (weighing 180-220 g, 7-8 weeks old) were supplied by the Laboratory Animal Center of the institute, and were divided into four groups: normal control group (n = 4), RD group (n = 40), LV-GADD153-sh+RD group (n = 40), and vehicle (Lentiviral vector) +RD group (n = 40).
Retinal detachment was created by subretinal injection of 10 mg/ml sodium hyaluronate (Bausch & Lomb Freda, China) as  Figure 1, the location of the GADD153 staining should be in yellow color in the outer segments. The yellow color were observed in the ONL 2 days after detachment and disappeared under GADD153 RNAi. There were no GADD153-positive cells in retinas of the normal control group. GADD153-positive cells in ONL increased after retinal detachment, most GADD153-positive cells were located in the ONL layer (GADD153 distribution was shown with white arrows). Compared with retinal detachment group, GADD153-positive cells decreased in RNA interference group (LV-GADD153-sh) but not in Vehicle group. (B) RT-PCR semi-quantification of GADD153 mRNA expression in RD rats under shRNA, vehicle or nontreatment. GADD153 mRNA increased in RD groups with the highest expression found at 2 to 4 days after detachment and then decreased. The change of mRNA in vehicle group was the same as that in the non-treated RD group. The RNA interference group (LV-GADD153-sh) showed a very low level of GADD mRNA. (C) Signal intensities of GADD153 protein were measured by Bandscan analysis and the ratios of GADD153 protein expression to b-actin in each sample were given. West blotting revealed similar changes of protein expression at different time points after RD. The RNA interference group (LV-GADD153-sh) showed a very low level of GADD protein. doi:10.1371/journal.pone.0059339.g001 described previously [14][15][16][17]. Briefly, the rats were anesthetized with an intraperitoneal (i.p.) injection of 10% chloral hydrate; a 30gauge needle was inserted into the subretinal space via an external trans-scleral trans-choroidal approach. 1% sodium hyaluronate was gently injected into the subretinal space to enlarge the retinal detachment (50 ml each makes 50% RD) [18].

In vivo Delivery of Vectors
To deliver the vectors into the eye, the animals were anesthetized with an i.p. injection of 10% chloral hydrate. Subretinal injection was performed under a microscopeand under sterile conditions according to the method reported previously [19]. In brief, a small incision was made in the sclera at approximately 1.5 mm posterior to the limbus with a 30-gauge needle. The needle was slowly introduced into the vitreous cavity, then to subretinal space by producing a small hole in the nasal peripheral retina. A single 5 ml vector suspension containing 10 8 TU/ml of LV-GADD153-sh or lentiviral vector was administered in 30 s. Two weeks after delivering the vectors, RD was created as described above.

RT-PCR Analysis
Two weeks after delivering the vectors, samples of rat retina collected at the time points of day1, 2, 4, and 7 after RD were quickly frozen in liquid nitrogen. Total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized from 2 mg total RNA in 20 ml reaction mixture using a RT-PCR kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Samples of cDNA were subjected to GADD153 amplification with b-actin used as a housekeeping gene. The primers used were as following: GADD153, 59ATGGCAGCT-GAGTCTCTGCC39 and 59TGATGGTGCTGGGTA-CACTCC39. Semi-quantitative analysis was carried out by measuring the intensity abundance of PCR products in gel photographs using Gene Genius Image software.

Western Blotting Analysis
Two weeks after delivering the vectors, samples were collected on 1, 2, 4 and 7 days after RD. Samples were run on 4% to 12% SDS polyacrylamidegel electrophoresis and transferred onto nitrocellulose membranes (Whatman, Maidstone, UK). After being blocked with 3% nonfat dried milk, the membrane was reacted with GADD153 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA). The density of the signal was quantified using Bandscan43 software and the protein expression levels were normalized for b-actin (1:50000; Sigma-Aldrich Chemical Co., St Louis, Mo, USA). Immunofluorescence Examination GADD 153 immunofluorescence examination was performed on sections obtained from the retina. Sections were treated with 0.5 mg/ml trypsin and each section was incubated for 30 min in phosphate-buffered saline (PBS) containing 5% skim milk to block non-specific binding, followed by incubation with antibodies against GADD153 (1:50, Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4uC. Sections were then incubated with Envision goat-anti-rabbit IgG HRP polymer (Dako, Carpinteria, CA) for 1 h at room temperature. Signals were amplified with TSA (tyramide signal amplification) biotin system (PerkinElmer, Boston, MA). The sections were stained with propidium iodine (PI) for 15 min to illustrate retinal cell distribution and to clear retinal layers such as outer nuclear layer (ONL), inner nuclear layer (INL) and ganglion cell layer (GCL). Laser-scanning confocal microscopy was used to observe and count the number of GADD153positive cells and PI stained cells in ONL at 1.0-1.5 mm from the optic disc. The percentages of GADD153 positive cells/PI stained cells were calculated by LSM 510 Expert Mode SP2 software.

Assessment of Retinal Outer Nuclear Layer Damage
The eyes (n = 3) were enucleated one week after RD and were embedded in paraffin. The eye wall sections were then stained with hematoxylin and eosin (H&E). The retinal histoarchitecture was evaluated by light microscopy. The ONL thickness was determined by Image-Pro Plus software on day 7 after experimental RD. Five sections were selected from the same quadrant in each eye. The thickness was measured at 10 points in each section by independent observers.

Statistical Analysis
Statistical analysis was performed using SPSS (SPSS 13.0, SPSS Institute Inc., Chicago, IL, USA). The results were presented as means 6 SD. The One-Sample Kolmogorov-Smirnov test was used to measure whether these data were in a normal distribution. Statistical significance was calculated with a one factor ANOVA test. A P value less than 0.05 was considered statistically significant.

GADD153 RNA Interference Resulted in Decreased GADD153 mRNA and Protein Contents in vivo
The experimental RD was induced two weeks after subretinal injection of the LV-GADD153-sh or vectors 5610 5 TU. We firstly examined GADD153 mRNA expressions from whole retina at 1 day, 2 day, 4 day and 7 day after RD by using RT-PCR. The GADD153 mRNA was hardly found expressed in normal control retinal tissues. It increased as early as 1 day after experimental RD. The expressions of GADD153 mRNA in RNAi group significantly decreased at different time points after RD compared with those in the RD group and vehicle group.
Temporal observation of GADD153 expression using Western blotting revealed an increase of GADD153 protein level, and immunofluorescence microscopy demonstrated that positive staining was mainly located in the nucleus and confined only to the ONL at different time points after RD (Figure 1). The expressions of GADD153 protein in LV-GADD153-shRNA group significantly decreased compared with those in the RD group (P,0.05), while the GADD153 protein expressions in the vehicle group showed no significant decrease.

GADD153 RNA Interference Decreased ONL Damage and Protected Retina
In RD rats, ONL thickness of retina rapidly and significantly decreased after 7 days. The ONL structure changed slightly. Thickness of layers below the outer plexiform layer were never altered after RD. The ONL of retinas exposed to LV-GADD153sh was substantially thicker than that from control eyes injected with LV vectors or untreated RD eyes. After 7 days of RD, the average thickness of ONL preservation of eyes that received LV-GADD153-sh was 4663 mm compared with 3965 mm in the eyes that received LV-vector or 3964 mm in untreated RD eyes (ANOVA; P,0.001; Figure 2).

GADD153 RNA Interference Decreased TUNEL-positive Photoreceptor Cells and Protected Retina
We assessed photoreceptor death after RD using TUNEL staining, which detects DNA fragmentation in apoptotic or necrotic nuclei. Hardly any TUNEL-positive cells were found in the normal control group. In the RD group, TUNEL-positive photoreceptor cells appeared on day 1 (7.96%). The number of TUNEL-positive photoreceptor cells peaked on day 2 (20.32%) and day 4 (19.9%), and then decreased on day 7 (4.72%). In GADD153 RNAi group, the number of TUNEL-positive photoreceptor cells also peaked from day 2 (10.53%) to day 4 (9.24%), then attenuated to 1.54% on day 7. Compared with the RD group, the apoptotic cells in GADD153 RNAi group significantly decreased at different time points after RD. The injection of vehicle did not reduce the number of TUNELpositive cells in ONL at all time points (Figure 3).

Discussion
Our previous study showed that GADD153 was present in the retina of an RD model, accompanied by apoptosis of photoreceptor cells [9].In this study we observed that silencing GADD153 using RNAi technology could inhibit the apoptosis of photoreceptor cells and protect retina in experimental RD. These findings strongly suggested that the ER stress pathway activated by GADD153 was involved in the apoptosis of photoreceptor cells after RD.
Expression of GADD153 was at a very low level in the normal retina of rats, and it increased on day 2 and day 4 after RD, which was consistent with the peak of photoreceptor cell apoptosis. GADD153-positive cells were confined to the ONL, where most apoptotic cells could also be found, and this was consistent with the result of our pervious study [9]. In the present study, we targeted GADD153 expression using RNAi technology (Table S1). We induced RD model after two weeks of Lentivirus injection. It is in agreement with the reported transfection time of lentivirus in retinal pigment epithelium [20]. In our pretest experiment, the greatest knockdown effect was achieved with the use of current LV-GADD153-shRNA sequence, and it demonstrated the suppression of GADD153 levels by 70% in vitro. In the present in vivo study, we were able to safely deliver lentivirus vector to the subretinal space in a rat model, with no apparent adverse effects caused by either the viral vector or the procedure itself. We were also able to show that the vector successfully delivered the gene product within the retina using RT-PCR and west blotting.
GADD153 positive cells and apoptotic photoreceptor cells were observed simultaneously on day 1. The number of GADD153 positive cells and apoptotic photoreceptor cells peaked on day 2 to day 4, and then decreased on day 7 in RD group, vehicle group and LV-GADD153-sh group. The number of apoptotic photoreceptor cells and the expression of GADD153 in GADD153 RNAi group significantly decreased at different time points after RD compared with those in RD group. The number of GADD153 positive cells and apoptotic cells in ONL both decreased in GADD153 RNAi group. These results suggested that the decrease of retinal GADD153 expression was caused by the specific silencing effect of GADD153 shRNA, leading to inhibition of photoreceptor cell apoptosis to a great extent. It is indicated that after RD, the expression of GADD153 increased in the photoreceptor cells, which then caused cell cycle arrest and DNA damage, finally leading to the cell apoptosis. When GADD153 expression was inhibited, the ER stress pathway was widely blocked, possibly through GADD153 targeting the downstream molecules such as TRB3, GADD34, and finally attenuated the apoptosis. These suggest that ER stress-mediated pathway participated in the apoptosis of photoreceptor cells after RD. To evaluate the neuroprotection of GADD153 RNAi against RD-induced cell death, we measured the thickness of the ONL. There are several methods available to measure the ONL thickness, such as the ONL thickness normalized to retinal thickness [21][22][23][24], measurement of the ONL thickness from the optic disc at the same intervals on the photographs [25], measurement along four different meridians to make color-coded topographic maps [26] and acquirement an average from directly measurement in each H&E staining retina section [27][28][29][30][31]. We chose to measure ONL thickness in the sections under H&E staining because it is also a method frequently used in retinal studies. Our data showed the ONL is 4663 mm and 44.0363 mm, respectively, in LV-GADD153-shRNA treated group and in normal group 7 days after detachment. There is no significant difference between the ONL thicknesses in normal and LV-GADD153-shRNA treated eyes. However, the ONL thicknesses in both of the above two groups are thicker than that in the RD control eyes (3964 mm, P,0.001, Figure 2). This trend indicated the protective effect of GADD153 RNAi in experiment retinal detachment. GADD153, as a key marker for ER stress, plays a crucial role in photoreceptor cell apoptosis after RD.
Interestingly, there were reports showing that some agents can protect retina from the damage after RD without interfering GADD153 expression, which indicates that multi-pathways take part in retinal damage after RD. In this study, we found that shRNA to GADD153 dramatically reduced photoreceptor cells apoptosis after RD. These findings encouraged us to hypothesize that GADD153 RNAi could have a protective role by blocking ER stress-induced photoreceptor cell death. However, whether it takes part in other pathway of photoreceptor apoptosis is still unclear. Considering that ER stress-induced apoptosis is a critical step in the pathogenesis of many neurodegenerative diseases, silencing GADD153 can function as a survival factor for neurons [32][33][34][35][36]. Chen et al showed that CHOP partially mediated ER stressinduced neuronal death and suggested that suppression of CHOP activation may contribute to BDNF-mediated neuroprotection during ER stress responses [37]. In another case, it was found that CHOP knockdown prevented perturbations in the AKT/ FOXO3a pathway in response to ER stress-induced apoptosis in neuronal cells [38].
In summary, this study demonstrates that ER stress-mediated pathway should be considered as a mechanism of inducing apoptosis in photoreceptor cells after RD. GADD153 is the key regulatory molecule regulating the ER Stress pathway, thus plays a crucial role in the photoreceptor cell apoptosis after RD. The findings in the present study may cast new lights on the therapeutic strategies aiming at rehabilitation and reconstruction of visual function after RD, as photoreceptor cells may be protected from RD-induced damage by subretinal injection of GADD153 RNAi. Gene therapy might be a good adjuvant to present treatments for complex types of RD. Table S1 Three alternative sequences of GADD153 shRNA and the control construction. All the alternative sequences of GADD153 shRNA and the control construction are shown. (DOC)