Hydrogen prevents corneal endothelial damage in phacoemulsification cataract surgery

In phacoemulsification, ultrasound induces hydroxyl radical (·OH) formation, damaging corneal endothelium. Whether H2 can prevent such oxidative damage in phacoemulsification was examined by in vitro and in vivo studies. H2 was dissolved in a commercial irrigating solution. The effects of H2 against ·OH generation were first confirmed in vitro by electron-spin resonance (ESR) and hydroxyphenyl fluorescein (HPF). ESR showed a significantly decreased signal magnitude, and fluorescence intensity by oxidized HPF was significantly less in the H2-dissolved solution. The effects of H2 in phacoemulsification were evaluated in rabbits, comparing H2-dissolved and control solutions. Five hours after the procedure, the whole cornea was excised and subjected to image analysis for corneal edema, real-time semiquantitative PCR (qPCR) for heme oxygenase (HO)-1, catalase (CAT), superoxide dismutase 1 (SOD1), and SOD2 mRNA, and immunohistochemistry. Corneal edema was significantly less and the increases in anti-oxidative HO-1, CAT and SOD2 mRNA expressions were significantly suppressed in the H2 group. In addition, corneal endothelial cell expressions of two oxidative stress markers, 4-HNE and 8-OHdG, were significantly lower in the H2 group. In conclusion, H2 dissolved in the ocular irrigating solution protected corneal endothelial cells from phacoemulsification-induced oxidative stress and damage.

localized high pressures of > 600 atmospheres and temperature elevations of > 5000 K. The energy created extends to neighboring water molecules, causing direct disintegration. This phenomenon (H 2 O→ ·OH + ·H) is called sonolysis, and the ·OH, i.e. hydroxyl radical, is the most reactive of the various reactive oxygen species (ROS), including superoxide anion, singlet-dioxygen, and hydrogen peroxide. The production of ·OH by an ophthalmic phacoemulsificator was first reported by Cameron et al. 10 . They detected ·OH using electron spin resonance (ESR) in a closed circuit and demonstrated that ·OH production was proportional to the US oscillation time. We then also demonstrated ·OH production by ESR in the anterior chamber of a model eye under clinical conditions of phacoemulsification 11 . In addition, we proved that free radicals produced by phacoemulsification were the cause of corneal endothelial damage using 8-hydroxy-2-deoxyguanosine (8-OHdG) as an oxidative stress marker in animal eyes 12 .
To maintain the anterior chamber of the eye and to protect the corneal endothelium from various forms of surgical damage during phacoemulsification, an ophthalmic viscosurgical device (OVD) is injected into the anterior chamber before phacoemulsification as a standard procedure. The major ingredient of an OVD is sodium hyaluronate (HA), which is a known free radical scavenger. The protective properties of HA against oxidative stress in the corneal endothelium have been reported 13 , and we also demonstrated the scavenging effect of OVD against ·OH in a model eye study 11,14 . The effect of the material, however, depends on its retention in the anterior chamber during phacoemulsification 11,14 , because the material can be aspirated and disappear in the process of surgery. A new method of constant protection from oxidative insults during surgery has been awaited.
One promising candidate for this purpose is the hydrogen molecule, H 2 . In 2007, we reported that H 2 selectively reduced cytotoxic ROS, ·OH, in particular, in vitro and exerted a therapeutic antioxidant activity in an ischemia-reperfusion injury model 15 . Since then, the effect of H 2 against oxidative stress-induced injury in several organs, including the brain, heart, lung, intestine, or bone marrow, has been reported [16][17][18][19] . In the field of ophthalmology, we reported that H 2 was effective to prevent ischemia reperfusion injury in the retina in a retinal artery occlusion model 20 . Considering the effect of H 2 against ·OH, H 2 dissolved in ocular irrigating solutions should work as a free radical scavenger in the anterior chamber. In this study, we examined the effect of H 2 in phacoemulsification. We have demonstrated that US in phacoemulsification induces ·OH, which damages corneal endothelium 11,12 . If H 2 dissolved in ocular irrigating solutions could prevent oxidative damage to the corneal endothelium, it would have great clinical usefulness.

H 2 concentration in the solution.
To prepare the H 2 -dissolved ocular irrigating solution, bags containing the solution were placed in H 2 gas. After 24-hour and 1-week exposure of the solution to 100% H 2 gas in an acrylic chamber (Fig. 1a), the dissolved concentration of H 2 in the solution was 61.9% and 61.8%, respectively, indicating that H 2 gas easily penetrated through a plastic bag, and that the solution was almost equilibrated with H 2 gas in the chamber within 24 hours. Thus, the H 2 -dissolved irrigating solution after 24-hour exposure was used for the following experiments. The H 2 concentration in the anterior chamber of the porcine eye (Fig. 1b) under continuous irrigation (10 ml/min) with the H 2 -dissolved solution for 30 minutes showed little change, from 56.5% to 53.7% (Fig. 1c). To confirm O 2 supply to the tissue, the O 2 concentration was also measured. At 2 minutes of irrigation with the H 2 -dissolved solution, the O 2 concentration in the anterior chamber was 5.4% ± 0.6%.

Detection of ·OH by ESR and by HPF.
To demonstrate the reduction of ·OH with dissolved H 2 , ocular irrigating solutions containing DMPO were treated with the US device. Their ESR spin adducts showed the characteristic quartet signal pattern of DMPO-OH. The signal magnitude in the H 2 -dissolved solution was apparently suppressed (Fig. 2a). When the intensities of the signals were calculated by image analysis after standardization using the amplitudes of the Mn signal, the difference was significant ( Fig. 2b; control 100 ± 30.8, H 2 group 60.5 ± 7.8, p < 0.05). Similarly, ocular irrigating solutions containing HPF were treated with the US device. Their fluorescence intensity by oxidized HPF was significantly smaller in the H 2 -dissolved solution than in the control ( Fig. 2c; Control 3588 ± 210.0, H 2 group 1032.5 ± 324.7, p < 0.005). Figure 3a (control) and b (H 2 group) shows the US probe position in the anterior chamber of the rabbit when US oscillation was performed. The representative photographs of the anterior segment at 5 hours after US exposure are shown in Fig. 3c (control) and d (H 2 group). The opaque, i.e. edematous lesion, was less apparent in the H 2 group eye than in the control eye. To compare the amplitude of the edema quantitatively, the intensity of the opaque lesion was analyzed using ImageJ software. Representative images of the excised corneas in the control and the H 2 groups are shown in Fig. 3e,f, respectively. The average index of the intensity was significantly smaller in the H 2 group (5.6 ± 8) than in the control (47.8 ± 15.2) ( Fig. 3g; p < 0.005).

Discussion
Corneal endothelial damage induces corneal edema, and when the damage is excessive, it results in irreversible bullous keratopathy (BK). Because the corneal endothelial cells lack the ability to regenerate, transplantation of corneal tissue is the only cure currently available. A recent national survey on BK in Japan showed that cataract surgery was the most common cause of penetrating keratoplasty (24.2%) 21 , and a more recent study in the UK showed that, among the indications for endothelial keratoplasty, cataract surgery ranked second following Fuchs' endothelial dystrophy 22 . Furthermore, corneal endothelial damage in a limited area, if it does not result in irreversible BK, is one of the most frequent complications after phacoemulsification and can cause temporary visual dysfunction in many post-phacoemulsification patients. Even though the safety of phacoemulsification has been dramatically improved due to progress in the apparatus and development of surgical devices including OVDs, the prevention of corneal endothelial damage in phacoemulsification is still an important issue for cataract surgeons.
The effect of H 2 as a free radical scavenger has been vigorously investigated in various conditions in vitro or in vivo 23 , since our report first described its effect 15 . We first showed that H 2 dose-dependently reduces ·OH

Figure 2. Detection of ·OH by ESR and by HPF. (a)
Representative signals in the control and H 2 groups on electron-spin resonance (ESR). Mn denotes the signal of manganese, which is used as a marker. The quartet signal suggesting DMPO-OH is obtained in both groups; however, the signal amplitude in the H 2 group is apparently smaller than that of the control. (b) The intensities of the signals were calculated through image analysis after standardization using the amplitudes of the Mn signal. The signal intensity in the H 2 group is significantly smaller than that of the control (each group n = 3, * P < 0.05 by the unpaired t-test). (c) HPF relative fluorescence units (RFU). RFU of the H 2 group (1032.5 ± 324.7) is decreased to almost 30% of the control (3588 ± 210.0, * p < 0.05 by the unpaired t-test).
in vitro, whereas H 2 is too weak to reduce physiologically important ROS such as NO· and superoxide. H 2 , the smallest molecule in the universe, has the unique ability of rapidly diffusing across membranes; it can react with cytotoxic ·OH in all organelles, including mitochondria and the nucleus, and thus effectively protects cells against oxidative damage. Indeed, H 2 prevented a decrease in the cellular levels of ATP synthesized in mitochondria 15 .
In phacoemulsification, ·OH production by US has been demonstrated 10,11 , and corneal endothelial damage by oxidative insults has also been demonstrated 12 . Considering the effect of H 2 and the consequence of sonolysis in phacoemulsification, it seemed reasonable and worthwhile to investigate the usefulness of H 2 dissolved in the irrigating solution, because, by this method, a potent free radical scavenger can be continuously provided during the surgery.
In the current study, the H 2 concentration in the irrigating solution was examined first. H 2 was dissolved into solution by placing the bag into a chamber of 100% H 2 . H 2 spontaneously penetrates through the plastic film of the bag, even without positive pressure in the chamber. This was possible because the solution, Ope Guard ® neo  kit, is a soft plastic product, which is why this solution was used. The H 2 concentrations in the 24-hour and 1-week exposure solutions were almost the same, indicating H 2 was almost saturated in the solution by 24-hour exposure with this method. Although this method has the advantage of eliminating concern about contamination, the H 2 concentration in the bag decreased gradually after it was retrieved from the chamber, because H 2 can escape from the bag. To confirm its validity in clinical use, H 2 concentration was measured under clinical conditions. At 30 minutes of continuous irrigation, the H 2 concentration in the anterior chamber was still 53.7%, suggesting that sufficient H 2 concentration can be maintained during a standard phacoemulsification time (Fig. 1). Furthermore, the O 2 concentration in the anterior chamber (5.4%) was confirmed to be within a safe range considering the short surgery time and the reported O 2 concentration in the aqueous humor in the human eye 24 .
The effects of H 2 in ·OH production caused by US were then examined in in vitro experiments using ESR and HPF analysis. US oscillation in the tube was performed as in our previous report 11 . Results of both analyses clearly demonstrated that H 2 suppressed ·OH production (Fig. 2). Next, phacoemulsification simulation was performed in animal eyes to observe the effect of H 2 in an in vivo model. In the procedure, the lens was not touched to avoid the effect of lens fragments on corneal endothelial damage. The difference between the H 2 group and the control group was obvious. The area of the edematous lesion of the cornea evaluated by image analysis was significantly smaller in the H 2 group (Fig. 3), suggesting that H 2 protected the corneal endothelium from oxidative insults. To observe the physiological reaction to free radicals in the corneal endothelial cells, the expression of HO-1 mRNA was then examined in the cells after US exposure. HO-1 removes the pro-oxidant molecule heme and generates free radical scavengers, biliverdin and bilirubin; therefore, HO-1 is usually regarded as an antioxidant-inducible cellular defense protein, and an increase of HO-1 mRNA indicates that oxidative stress occurred in the cells 25 . The mRNA expressions of other anti-oxidative enzymes, including CAT, SOD1, and SOD2 were also examined, and all of them were suppressed in the H 2 group, with significant changes in CAT and SOD2. Suppression of US exposure-induced HO-1, CAT, and SOD2 mRNA expressions in the H 2 group (Fig. 4a,b,d) indicates that the corneal endothelial cells were protected by H 2 from oxidative insults in phacoemulsification. Lastly, immunohistochemistry showed that the number of 4-HNE and 8-OHDG-positive cells was significantly smaller in the H 2 group (Fig. 5), proving histologically that the oxidative damage was truly suppressed by H 2 .
All evidence presented in the current study strongly supports the usefulness of H 2 in phacoemulsification. Among the various H 2 applications in medicine reported so far, H 2 dissolved in the ocular irrigating solution may be one of the most reasonable and promising uses of H 2 . In this system, H 2 is continuously provided to the anterior chamber of the eye where ·OH is produced by US oscillation without any additional intervention such as injection or inhalation. The effect of H 2 against ROS has been reported in various ways in previous studies 23 . Among these, although the effect in an ischemia-reperfusion injury mode has been regarded as the representative study since our first report 15,20,26,27 , we have also reported the effect of H 2 in an irradiation-induced lung damage model 19 . One of the harmful effects of ionizing radiation, i.e. the indirect action, occurs when ionizing radiation interacts with water molecules in cells, which leads to production of ROS including ·OH. The ROS reacts rapidly with cellular macromolecules to damage DNA, lipids, and proteins and exert strong cytotoxic effects. It has been estimated that ·OH causes 60-70% of ionizing radiation-induced cell damage 28 . The sonolysis-induced ·OH in phacoemulsification is another mode of oxidative insult to the cells, and it seemed similar, at least partially, to irradiation-induced damage.
The standard composition of the ocular irrigating solution includes glutathione, which is a known anti-oxidative agent 29,30 . Although the solution, Ope Guard ® neo kit, used in the current study contains glutathione, production of ·OH shown by ESR and HPF was evident. However, the results of in vivo experiments clearly showed more corneal endothelial damage in the control than in the H 2 group, suggesting that glutathione alone is not enough to protect cells from oxidative stresses caused by US oscillation in this model. In addition, several reports have shown the effect of ascorbic acid as an anti-oxidative agent in an in vivo phacoemulsification model 31,32 and in an in vitro model 33 , but no clinical trial of ascorbic acid use in phacoemulsification has yet been done. This may be because ascorbic acid can be easily oxidized to dehydroascorbic acid, which can have an effect on glutathione 33 . In this regard, there is no concern about oxidative adducts with H 2 use. Moreover, because there is literally no barrier to H 2 , it can function either outside or inside the cell membrane by rapid diffusion.
In conclusion, H 2 dissolved in the ocular irrigating solution effectively protected corneal endothelial cells from oxidative stress and damage caused by phacoemulsification. H 2 dissolved in the irrigating solution can have a dual action: as a direct scavenger against ·OH produced by sonolysis and as an anti-oxidative agent in the cell. In the current study, the former was clearly proven by the ESR and HPF analyses. In in vivo experiments, the favorable effect of H 2 was evident, but to prove the anti-oxidative function of H 2 in cells clearly, further investigation is needed.

Animals.
Nine-week-old, male, Japanese white rabbits, weighing 2.5 to 3.0 kg, were purchased from Tokyo Laboratory Animals Science Co. Ltd. (Tokyo, Japan). The animals were kept individually under standardized laboratory conditions and given tap water and food ad libitum. All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Porcine eyes that were used for H 2 concentration measurements were obtained from a local abattoir.

Preparation of H 2 -dissolved irrigating solution.
The ocular irrigating solution is used to maintain the anatomic and physiologic integrity of intraocular tissues in ocular surgeries including phacoemulsification. The standard composition of commercially available irrigating solutions includes glutathione, glucose, and sodium bicarbonate, which has been proven to be effective for maintenance of normal corneal endothelial functions 29 . In the current study, Ope Guard ® neo kit (Senju Pharmaceutical, Osaka, Japan), which contains the standard ingredients, was used. Ope Guard ® neo kit is a soft plastic bag product and one of the most popular irrigating solutions in Japan. To dissolve H 2 in the solution, the bag was placed in an acrylic vacuum chamber (SNS-type, Sanplatec, Osaka, Japan), of which the air was replaced by 100% H 2 gas (Fig. 1a). Because of its molecular size, H 2 penetrates spontaneously through the plastic film of the bag even without positive pressure in the chamber. The bag was retrieved from the chamber after 24-hour or 1-week exposure, and the dissolved H 2 concentration in the solution was immediately measured using a needle-type H 2 sensor (Unisense, Aarhus N, Denmark). Then, to assess H 2 concentration changes in the anterior chamber with irrigation of H 2 -dissolved solution, enucleated porcine eyes were used. The H 2 sensor was inserted through the incision of an enucleated porcine eye, and H 2 concentration was measured every 5 minutes for 30 minutes under continuous irrigation of the solution at 10 ml/min (Fig. 1b). Furthermore, to confirm O 2 supply to the tissue, O 2 concentration in the anterior chamber of an enucleated porcine eye at 2 minutes after the irrigation started was also measured using a needle-type O 2 sensor (Unisense).

Detection of ·OH by ESR.
The ·OH detection by ESR was performed similarly to our previous paper 11 . For spin trapping, 10% aqueous solution of 5, 5′ -dimethyl-1-pyrroline N-oxide (DMPO; Labotec, Tokyo, Japan) was used. DMPO was added to the Ope Guard ® neo kit solution at a final concentration of 1%. In 50-ml plastic test tubes, 10 ml of 1% DMPO/H 2 -dissolved solution (H 2 group) or normal solution (control) was prepared. The US probe of a commercially available phacoemulsification device (Stellaris ® , Bausch & Lomb, Rochester, NY) was placed in the center of the tube, and US was produced at a power level of 30% for 10 seconds without irrigation and aspiration. Immediately after US, 300 μ l of the solution were transferred to a flat quartz ESR cuvette. The cuvette was then placed in an ESR spectrometer (model JES-RE3X; JEOL, Tokyo, Japan), and the signals of the spin adducts, DMPO-OH, were measured by double integration wave height using a computer software program (ES-IPEITS data system version 7.0; JEOL). All measurements were performed 3 times.

Detection of ·OH by HPF.
HPF is a novel reagent that directly detects certain highly reactive oxygen species (hROS) 34 . Although HPF itself has little fluorescence, it selectively and dose-dependently reacts with hROS, such as ·OH and peroxynitrite, and shows strong fluorescence. HPF (Goryo Chemical, Hokkaido, Japan) was added to H 2 -dissolved solution (H 2 group) or normal solution (control) at a final concentration of 5 μ M. As in the ESR experiment, in 10 ml of the solution in the 50-ml plastic test tubes, US oscillation without irrigation and aspiration was performed for 10 seconds at 30% power. Immediately after US, fluorescence of the samples was measured by a plate reader (Wallac 1420 ARVO; PerkinElmer, Waltham, MA) with excitation at 485 nm and emission at 535 nm. All measurements were performed 4 times.
Scientific RepoRts | 6:31190 | DOI: 10.1038/srep31190 Effects of H 2 in phacoemulsification in vivo. Rabbits were anesthetized via intramuscular injection of ketamine (30 mg/kg) and xylazine (4 mg/kg), and topical oxybuprocaine hydrochloride (Santen Pharmaceutical, Osaka, Japan) was used for local anesthesia. The US probe (Stellaris ® , Bausch & Lomb) was introduced into the center of the anterior chamber through the incision, and then continuous US oscillation with irrigation by H 2 -dissolved solution (H 2 group) or normal solution (control) was performed for 90 seconds at 30% power of the US setting of the device (vacuum pressure: 175 mmHg; bottle height: 85 cm). To avoid the effect of probe manipulation, the phacoemulsification procedure was performed in a completely similar fashion in both groups. The US probe was moved slowly on the iris plane roundly without touching the lens. At 5 hours after the procedure, the rabbits were euthanized by overdose injection of ketamine. After the anterior segments were photographed using the ophthalmologic surgical microscope, the whole cornea was excised and submitted to the following experiments.
Image analysis of corneal edema. The excised corneas (control n = 4, H 2 group n = 3) were photographed and analyzed using ImageJ software (Version 1.44, NIH, Bethesda, MD) as described previously 35,36 . Each image was captured using the same camera settings for gain and time, and pixel intensity was analyzed for 25,000 pixels (500 × 500 pixels) at the center of the cornea. Background intensity was calculated from the untreated cornea and subtracted from each image.
Semiquantitative PCR for HO-1, CAT, SOD1, and SOD2 mRNA. HO-1 has been known to play an important role in cell membrane protection against various oxidative stresses 25 . CAT, SOD1, and SOD2 are also crucial for breaking down the harmful end products of oxidative phosphorylation 37 . The effect of H 2 on their mRNA expression in the corneal endothelium was evaluated. At 5 hours after US exposure, the cornea was excised, and the endothelial cells were immediately scraped using a blunt scalpel. Total RNA was extracted from corneal endothelial cells and subjected to reverse transcription, with portions of the resulting cDNA then subjected to real-time semi-quantitative PCR (qPCR) analysis with a 7500 Fast Real-Time PCR System (Life Technologies, Tokyo, Japan). The Fast Real-Time PCR System was used with the following primers: for HO-1, 5′ -CAGGTGACTGCCGAGGGTTTTA-3′ (forward) and 5′ -GGAAGTAGAGCGGGGCGTAG-3′ (reverse) 38 ; for CAT, 5′ -CCCAATAGGAGACAAACTGA-3′ (forward) and 5′ -ACTCTCTCCGGAATTCTCTC-3′ (reverse) were designed using DNASIS Pro software (Hitachi Software Engineering Co., Ltd., Tokyo, Japan); for SOD1, 5′ -GACGCATAACAGGACTGACCG-3′ (forward) and 5′ -AACACATCAGCGACACCATTG-3′ (reverse) 39 , for SOD2, 5′ -TGACGGCTGTGTCTGTTGGT-3′ (forward), and 5′ -GCAGGTAGTAAGCGTGTTCCC-3′ (reverse) 39 , for rabbit glyceraldehyde -3-phosphate dehydrogenase (GAPDH) gene, 5′ -GCCGCTTCTT CTCGTGCAG-3′ (forward) and 5′ -ATGGATCATTGATGGCGACAACAT-3′ (reverse) (accession no. L23961) 40 . The cycling protocol entailed incubation at 90 °C for 10 s followed by 40 cycles of 95 °C for 5 s and 60 °C for 34 s. Relative gene expression was calculated using the standard curve method. The HO-1 mRNA levels were normalized to those of the housekeeping GAPDH gene. The HO-1/GAPDH ratio in untreated corneal endothelial cells was defined as 1.0. To assess the average value of the HO-1/GAPDH ratio, measurements were performed 5 times. Each sample was run in duplicate, and each real-time PCR was repeated three times (control n = 7, H 2 group n = 6).
Immunohistochemistry. The excised corneas were fixed in Bouin's solution (Wako Pure Chemical Industries, Osaka, Japan) for 1 hour at 4 °C, washed 3 times for 10 minutes each in PBS with 1% Triton X-100 (Bio-Rad Laboratories, CA) (PBST), and treated with acetone for 3 minutes at − 20 °C. They were further washed 3 times for 10 minutes with PBST and incubated in 10% horse serum diluted in PBST to block nonspecific staining. The corneas were then incubated in a 1:30 dilution of a monoclonal antibody against 8-OHdG (control n = 5, H 2 group n = 6) or 4-hydroxy-2-nonenal (4-HNE) (control n = 3, H 2 group n = 5) overnight at 4 °C (18). Both antibodies were purchased from the Japan Institute for the Control of Aging (Shizuoka, Japan). The corneas were washed with PBST 3 times for 10 minutes and incubated with horse biotinylated anti-mouse IgG (Vector Laboratories, Burlingame, CA) for 1 hour at room temperature. Tissues were then washed with PBST 3 times for 10 minutes and incubated with ABC reagent (Vector Laboratories) for 1 hour at room temperature. After being washed with PBST 3 times for 10 minutes, corneas were stained with DAB solution (Vector Laboratories) for 10 minutes at room temperature. Finally, they were washed with distilled water for 5 minutes each and mounted on slides.

Statistics.
Morphometric data from different regions in each eye were averaged to provide one value per eye. The mean and SD for these measurements were calculated for each group, and comparisons between groups were made using the unpaired t-test (Stat Flex ver. 6, Artec, Osaka, Japan). A p value of < 0.05 was considered significant.