A total of 48 3 month-old adult male Sprague-Dawley rats weighing 300–350 g (Tianjin Medical University Animal Center, Tianjin, China) were used in the present study. Animals were fed a standard rodent diet with a normal light-dark cycle. All animals were cared for with the approval of the Laboratory Animal Research Committee at Tianjin Medical University Eye Center. All experimental procedures involving animals were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Hydrogen was dissolved and saturated in 0.9% normal saline (NS) for 6 h under a high pressure of 0.4 MPa using hydrogen-producing apparatus (GCH-3000; Tianjin Tongpu Analytical Instrument Tech Co., Ltd., Tianjin, China) according to the method reported by Cai et al (17). The saturated HRS was stored at 4°C under normal pressure in an aluminum bag. HRS was freshly prepared and sterilized by γ radiation 24 h prior to the experiment, ensuring an effective concentration of 0.6 mM. The content of hydrogen in NS was confirmed through gas chromatography as described previously by Ohsawa et al (14).

All surgery was performed under general anesthesia with intraperitoneal administration of chloral hydrate (300 mg/kg). The anterior chamber of the left eye was cannulated with a 30-gauge needle connected to a reservoir containing NS. Intraocular pressure was increased to 130 mmHg for 60 min and ocular ischemia was confirmed by interruption of the ocular circulation, as described by Sun et al (2). Thereafter, the cannula was immediately retracted and the adequacy of retinal reperfusion was confirmed visually by ophthalmoscopy. Rats that did not recover from retinal perfusion 3 min after the end of the ischemic period and those with lens injury, which prevents retinal ganglion cell (RGC) death and promotes axonal regeneration (19), were excluded from the investigation. Following surgery, levofloxacin ophthalmic solution 0.3% (Cravit^®; Santen Pharmaceutical Co., Ltd., Osaka, Japan) was dropped into the conjunctival sac of the left eye to prevent infection. The model of IR injury was fully induced 24 h after reperfusion.

All the rats were classified into the following four groups: IR injury with HRS group, IR injury with NS group, HRS-treated control group and NS-treated control group. The animals in the IR injury with HRS and NS groups were peritoneally injected with HRS or NS (5 ml/g) at the beginning of reperfusion. Consecutive HRS or NS treatment was administered daily until the rats were sacrificed with overdose anaesthesia. The rats in the HRS-treated and NS-treated control groups were peritoneally injected with HRS or NS (5 ml/kg) at the same time points without initial reperfusion. As molecular hydrogen can easily diffuse into tissues and penetrate the cellular membrane, HRS (5 ml/kg) was injected peritoneally only at the beginning of reperfusion. Furthermore, consecutive HRS, compared with the equivalent quantity of NS, was administered daily until the rats were sacrificed.

Freshly enucleated eyeballs (n=6) were fixed in 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS) for 30 min. Subsequently, corneal paracentesis was performed using a 25 G needle. Following that, the entire eye cups were further fixed overnight and then processed for paraffin embedding. Sections (4 µm) were cut along the vertical meridian of the eye, 2 mm away from the optic nerve head and mounted on pre-coated glass slides. Following deparaffinization and rehydration, sections were stained with hematoxylin and eosin. Retinal damage was assessed by measuring the thickness of the retina and cell loss in the ganglion cell layer (GCL). The retinal thickness is defined as the total width between the inner limiting membrane to the interface of the outer plexiform layer and the outer nuclear layer (20). The number of cells in the GCL was calculated under a light microscope (magnification, x40) (BX51; Olympus Corporation, Tokyo, Japan). Three different sections were randomly selected and each section was measured at four different points between 1 and 2 mm from the optic disc. The average value was appointed as the thickness of retina and the number of cells in the GCL of the eye.

Cryosections (n=6) were fixed with 4% paraformaldehyde in 0.1 M PBS (pH 7.4) for 20 min. Cell death was detected by a TUNEL assay using the In Situ Cell Death Detection kit, POD (Roche Diagnostics Deutschland GmbH, Mannheim, Germany) and color-developed with 3,3′-diaminobenzidine (DAB) substrate (Roche Diagnostics Deutschland GmbH) according to the manufacturer’s instructions. The number of TUNEL-positive cells in the GCL was counted at x400 magnification for each section using a light microscope (BX51; Olympus Corporation).

Eyes (n=6) in optimal cutting temperature compound were cut into 6 µm sections (RM2165; Leica, Wetzlar, Germany) following being flash frozen in liquid nitrogen. The cryosections were fixed in ice-cold acetone for 10 min, washed in PBS three times and blocked with 1% bovine serum albumin (BSA) in PBST for 30 min at room temperature. Subsequently, primary goat anti-8-hydroxyguanine (8-OHdG, a classical DNA oxidative product) polyclonal antibody (1:100; cat. no. ab10802; Abcam, San Francisco, CA, USA) in 1% BSA in PBST were added to the samples and incubated in a humidified chamber overnight at 4°C. Following three washes in PBS, the sections were incubated with tetramethyl rhodamine isothiocyanate-conjugated rabbit anti-goat IgG secondary antibody (1:200; cat. no. BA1091; Beijing Zhongshan Golden Bridge Biotechnology Co., Ltd., Beijing, China) for 1 h in the dark at 37°C, followed by 4′,6-diamidino-2-phenylindole staining (0.1 µg/ml; Sigma-Aldrich, St. Louis, MO, USA) and another three washes in PBS. Coverslides were mounted and the immunoreactivity of 8-OHdG was detected under a fluorescence microscope (DFC500; Leica Microsystems, Renens, Switzerland).

Total retinal tissue lysates (n=6) were prepared by the addition of 500 µl sodium dodecyl sulfate (SDS) buffer [250 mM Tris (pH 6.8), 10% SDS, 500 mM dithiothreitol, 50% glycerol, 0.5% bromophenol blue and a 1:100 protease inhibitor cocktail (cat. no. P8340; Sigma-Aldrich)]. Total tissue extracts (20 µg) were separated on a 12% SDS polyacrylamide gel. Proteins were transferred onto an Immobilon-P membrane (Millipore, Billerica, MA, USA) and blocked with 5% skim milk in TBST (TBS containing 0.05% Tween-20) at room temperature for 2 h. Subsequently, the membranes were hybridized at 4°C overnight in TBST with the primary antibodies: Anti-rPARP-1 (1:5,000; cat. no. 1051-1; Epitomics, Burlingame, CA, USA), and anti-rCaspase-3 (1:1,000; cat. no. 9661; Cell Signaling Technology, Boston, MA, USA) overnight at 4°C. Following incubation with a horseradish peroxidase-conjugated anti-rabbit immunoglobulin antibody, proteins were visualized with an enhanced chemiluminescence kit (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, MD, USA). For normalization, blots were probed with β-actin, a housekeeping antibody (anti-β-actin; 1:1,000; Beijing Zhongshan Golden Bridge Biotechnology Co., Ltd.). The signal was detected by exposing X-ray films to the processed blots and analyzed by laser scanning densitometry (Personal Densitometer; GE Healthcare, Piscataway, NJ, USA).

Paraffin-embedded sections (n=6, 4 µm thick) were deparaffinized and rehydrated. Antigen retrieval was achieved by boiling in 10 mmol/l sodium citrate buffer for 10 min and then steadily cooling to room temperature. Subsequently, the sections were blocked using 3.0% H₂O₂ in methanol for 15 min in order to inhibit endogenous peroxidase activity. Following washing in PBS, the sections were incubated overnight at 4°C with monoclonal rabbit anti-PARP-1 primary antibody (cat. no. 1051-1) at a dilution of 1:40 (Epitomics). The sections were incubated with peroxidase-conjugated goat anti-rabbit IgG (Beijing Zhongshan Goldenbridge Biotechnology Co., Ltd.) at the same dilution of 1:500 for 2 h at 37°C. The sections were washed in PBS, developed in prepared DAB chromogen solution, lightly counterstained with hematoxylin, dehydrated and mounted.

All data are presented as the mean ± standard deviation. Data were analyzed using one-way analysis of variance conducted between groups with post hoc analysis through Bonferroni multiple comparisons test. Student’s t-test was used to compare data between two groups. Statistical analyses were conducted using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA). P<0.05 was considered to indicate a statistically significant difference.

The histological alterations of RIR injury occurred after 1 week of reperfusion (21). The thinning of retinas and cell loss in the GCL demonstrated the successful induction of this model. Abdominal administration of HRS effectively reversed the morphological alterations, including thinning of retinas and cell loss in the GCL (Fig. 1). By contrast, the retinal thickness of the RIR injury with NS group was 73.39±6.07 µm, which was markedly thinner than the retinal thickness in the NS-treated control group (118.85±4.31 µm; P<0.001). The retinal thickness in the HRS-treated control group was 117.71±4.95 µm, which demonstrated that hydrogen did not alter the thickness of the retina in normal rats. However, following HRS administration the thinning of retina following RIR injury partially recovered to 97.72±8.25 µm, which was significantly different to the NS-treated control (P<0.001).

Furthermore, the number of cells in the GCL in the IR injury with NS group decreased to 21.2±1.9, compared with 31.9±2.2 in the NS-treated control (P<0.001). No significant difference was identified between the HRS-treated (31.6±2.4) or NS-treated controls. By contrast, the number of cells in the IR injury with HRS group was 28.3±2.4 (P<0.001).

The NS-treated control and HRS-treated control had few TUNEL positive cells, which illustrated that HRS did not induce cell death. By contrast, retinas with RIR injury experienced severe cell death in the GCL 7 days after reperfusion, which was in accordance with previous studies (22,23). Treatment with HRS significantly decreased the number of TUNEL-positive cells in the GCL of RIR rats on the seventh day after reperfusion (P<0.001), suggesting that HRS may have a protective effect on cell apoptosis (Fig. 2).

The DNA oxidative breaks in RIR injury were detected through immunofluorescent staining of 8-OHdG, a DNA oxidative product (Fig. 3). There was moderate 8-OHdG immunoreactivity in the photoreceptor cell layer compared with faint immunostaining in the GCL in the untreated control. Following administration of HRS, 8-OHdG immunoreactivity partially decreased. In addition, IR injury increased the strength of 8-OHdG immunoreactivity in the GCL as well as in photoreceptor cell layers, suggesting severe DNA oxidative stress due to ROS. By contrast, the immunoreactivity of 8-OHdG decreased significantly in the GCL immediately following HRS intervention in IR injury, which may confirm the reductive potential of hydrogen to DNA oxidative stress.

Excessive DNA breaks may induce overactivation of PARP-1 and cell death. In addition, as the substrate of caspase-3, activated PARP-1 can be cleaved into 29 and 84 kDa fragments. The primary anti-capase-3 antibody could react with the whole molecule and the 19 kDa fragment of caspase-3, while anti-PARP-1 antibody recognized full length PARP-1 and its 29 kDa fragment.

Cleavage of PARP-1 was significantly elevated in RIR retina 24 h and 7 days after reperfusion (Fig. 4). Compared with the relatively low immunofluorescent staining of PARP-1 at 116 kDa in the untreated control and hydrogen control, the RIR model demonstrated strong immunoreactivity of the 29 kDa C-terminal catalytic domain. At the same time, the pro-form of PARP-1 decreased 24 h after RIR injury, but increased and surpassed the control 7 days after RIR injury, while the 29 kDa cleaved fragment remained visible. Following peritoneal administration of HRS, cleaved PARP-1 was reduced 24 h and 7 days after reperfusion, demonstrated as decreased immunoreactivity of 29 kDa PARP-1 fragments.

The expression of capase-3 was detected in RIR injury by immunostaining. In the untreated control retina and hydrogen control retina, there was almost no caspase-3 immunoreactivity. However, cleaved caspase-3 was detected between 24 h and 7 days after reperfusion. However, hydrogen intervention inhibited caspase-3 cleavage, showing the loss of cleaved fragment immunoreactivity 24 h and 7 days after reperfusion.

In normal adult retina, PARP-1 was mildly detected in the cytoplasm of the GCL and nerve fiber layer (Fig. 5A). There were markedly fewer PARP-1 positive cells in the IR injury groups at 7 days. However, the immunoreactivity of PARP-1 was concentrated predominantly in the nuclei of the GCL in IR retina (Fig. 5C). This altered location of PARP-1 from the cytoplasm to the nuclei was previously observed by Huang et al (24) in the same rodent experiment model. Since the cleaved 29 kDa fragment of PARP-1 can translocate to the nuclei and bind to the oxidized DNA breaks, the translocation of PARP-1 immunoreactivity illustrated the sustained existence of DNA breaks in RIR injury. By contrast, after 7 days of intraperitoneal administration of HRS in IR rats, there was apparent immunoreactivity of PARP-1 not only in the nuclei but also in the cytoplasm of the GCL as well as in the nerve fiber layer (Fig. 5D).