Drinking hydrogen water improves photoreceptor structure and function in retinal degeneration 6 mice

Retinitis pigmentosa (RP) is a genetically heterogeneous group of inherited retinal disorders involving the progressive dysfunction of photoreceptors and the retinal pigment epithelium, for which there is currently no treatment. The rd6 mouse is a natural model of autosomal recessive retinal degeneration. Given the known contributions of oxidative stress caused by reactive oxygen species (ROS) and selective inhibition of potent ROS peroxynitrite and OH·by H2 gas we have previously demonstrated, we hypothesized that ingestion of H2 water may delay the progression of photoreceptor death in rd6 mice. H2 mice showed significantly higher retinal thickness as compared to controls on optical coherence tomography. Histopathological and morphometric analyses revealed higher thickness of the outer nuclear layer for H2 mice than controls, as well as higher counts of opsin red/green-positive cells. RNA sequencing (RNA-seq) analysis of differentially expressed genes in the H2 group versus control group revealed 1996 genes with significantly different expressions. Gene and pathway ontology analysis showed substantial upregulation of genes responsible for phototransduction in H2 mice. Our results show that drinking water high in H2 (1.2–1.6 ppm) had neuroprotective effects and inhibited photoreceptor death in mice, and suggest the potential of H2 for the treatment of RP.


Results
Preservation of the H 2 concentration in H 2 water. To ensure that mice drank water with a stable, high concentration of H 2 , we developed unique water drinking valves designed to preserve H 2 (Fig. 1a). The average amount of water consumed per mouse was 3.42 ± 0.14 ml/day. The hydrogen concentration before drinking was 96.84 ± 1.03%, and was maintained at 70.68 ± 0.31% in the first week after drinking (C57BL/6J mice, n = 4) (Fig. 1b).
Hydrogen prevents outer retina thinning in rd6 mice. In rd6 mice, photoreceptor degeneration occurs from 3-4 weeks of age, and progressive loss of the photoreceptor outer segments continues over approximately 16 months 31 . To examine whether H 2 water prevents thinning of the outer retina in rd6 mice, we measured outer retinal thickness using optical coherence tomography (OCT), a non-invasive imaging modality that produces cross-sectional reflectance images of the retina. Figure 2a shows representative images for the control and H 2 groups. Outer retinal thicknesses (in pixels) of the control group (n = 8) and H 2 group (n = 10),   11.14 ± 1.05, p = 0.00005) (Fig. 2b). OCT imaging showed that the outer retina was significantly thicker in the H 2 group than in the control group from 21 to 47 weeks postnatally. These results suggest that drinking H 2 water can prevent outer retinal thinning.
Hydrogen protects rod function in rd6 mice. To determine whether preservation of the outer retina results in improved retinal function in the H 2 group, we performed scotopic ERG to measure rod response and mixed rod-cone response elicited with 0.02 cd·s/m 2 and 2 cd·s/m 2 stimuli, respectively (H 2 group: n = 10; control group: n = 8). Figure 3a shows representative ERG recordings of the b-wave amplitudes at 0.02 cd·s/m 2 and 2 cd·s/m 2 . The rod response represented by b-waves at 0.02 cd·s/m 2 was significantly preserved in the H 2 group beginning at 10 weeks postnatal through 42 weeks ( Fig. 3b). On the other hand, the mixed rod-cone response reduction in rd6 mice was relatively moderate compared to the rod response, and a significantly higher amplitude of the mixed rodcone response was observed only at 27 weeks old in the H 2 group (p = 0.006, Fig. 3b). These results suggest that drinking H 2 water can rescue rod function in rd6 mice.
Effect of hydrogen on histopathological changes, cell density in the outer nuclear layer, opsin red/green positivity, and opsin blue positivity in retinal cross sections. To evaluate the protective effect of hydrogen, we then examined histopathological and morphometric changes in 49-week-old rd6 mice. Figure 4a shows images of representative slices from the control group (n = 4) and H 2 group (n = 4). The photoreceptor inner and outer segments thickness of the control group was 8.8 ± 2 μm, and that of the H 2 group was 17.2 ± 1.8 μm (p = 0.00039; Fig. 4b). The outer nuclear layer thickness was 24.9 ± 1.9 μm in the control group, and 34.6 ± 2.2 μm in the H 2 group (p = 0.00027, Fig. 4c). The number of cells per slide in the outer nuclear layer was 275.8 ± 20 in the control group, and 316.8 ± 20.6 in the H 2 group (p = 0.015; Fig. 4d). The photographs in Fig. 5a,b show representative retinal cross sections with cells stained red for rhodopsin and green for opsin red/green or opsin blue. Nuclei were stained with DAPI (blue). The number of cells positive for both rhodopsin and opsin red/green (yellow in the merged image) per vertical section through the retina was 216.7 ± 26.4 in the control group, and 337.3 ± 60.6 in the H 2 group (p = 0.037; Fig. 5a). The number of cells positive for both rhodopsin and opsin blue (yellow in the merged image) per vertical section through the retina The outer retina thickness with H 2 water (n = 10) was significantly greater than that without H 2 water (n = 8) (p < 0.001 and 0.05). Bars depict mean ± standard deviation (SD). *p < 0.05, ***p < 0.001.  Fig. 5b). These results suggest that drinking H 2 water can protect photoreceptor cells and opsin red/green-positive cells in rd6 mice.

Hydrogen induces high expression of genes involved in phototransduction.
To characterize the effects of H 2 water on gene expression in rd6 mice, we performed whole transcriptome analysis of the neural retina with/without H 2 water. RNA-seq analysis of differentially expressed genes (DEG) in the H 2 group (n = 3) vs control group (n = 4) revealed 1996 genes with significantly different expression (upregulation of 856, downregulation of 1140 genes), as indicated in the heatmap (Fig. 6a). To identify signature trends for upregulation or downregulation of downstream pathways, we performed gene and pathway ontology analysis and showed the four most upregulated pathways and the four most downregulated pathways (Fig. 6b). Top molecular pathways differentially regulated following hydrogen drinking included transcriptional changes in approximately 18 genes involved in phototransduction pathways (Fig. 6c). The diagram produced by Ingenuity Pathway Analysis (IPA) illustrates phototransduction in rod cells and cone cells. As illustrated, the majority of genes included in phototransduction were upregulated in the H 2 group. We also examined the superpathway of cholesterol biosynthesis, another upregulated pathway in the H 2 group. Genes in this superpathway including Idl1, Acta2, Cyp51A1, and Hmgcs1 were slightly elevated (see Supplementary Fig. S1 online). Gene ontology groups such as inflammatory response (GO:0006954) and response to oxidative stress (GO:0006979), which were expected to be different, were not significantly changed as a system ( Supplementary Fig. S1).
GFAP, monocytes/macrophages-2 (MOMA-2) positivity in retinal cross-sections and ionized calcium-binding adapter molecule 1 (Iba-1). We also carried out immunohistochemical analyses of retinal cross-sections to evaluate the effect of H 2 water on retinal inflammatory response in rd6 mice.

Discussion
This study shows that drinking H 2 water delayed retinal degeneration in rd6 mice. H 2 water inhibited photoreceptor death. We found that a high concentration (1.2-1.6 ppm) of H 2 in drinking water led to neuroprotective effects. H 2 water also increased expression of genes of phototransduction in photoreceptors. Thus, our study may pave the way toward a new neuroprotective strategy using H 2 water in RP patients. In rd6 mice, the photoreceptor cell outer segments are reduced slightly in length, with a decrease in the number of photoreceptor cells 31 . As shown in Fig. 2, outer retina thickness decreased in the control group, but this decrease was significantly suppressed beginning at postnatal 21 weeks in the hydrogen drinking group. Similarly, the same result was obtained with histopathology in the analysis of outer retina thickness (Fig. 4), and a significant effect of protecting photoreceptor cells was observed in terms of photoreceptor thickness, outer retina thickness, and the number of photoreceptor cells.
We found a significant difference in rod response of ERG (0.02 cd·s/m 2 ; Fig. 3). Previous studies have shown that the decreased ERG amplitude in rd6 mice is detected beginning at P25 36 to 70 weeks of age 31 , and the rod response reduces earlier than the cone response 36 . Our data showed the rod function could be rescued by H 2 water, whereas a slight effect was observed for the mixed rod-cone function. Longer observation may be required to determine the effect of the H 2 water for mixed rod-cone function.
We investigated two cone opsins, long wavelength-sensitive red and green opsin, and short wavelengthsensitive blue opsin. The number of red and green opsin-positive cells in the H 2 group was significantly higher than that in the control group (Fig. 5a). The number of blue opsin-positive cells in the H 2 group tended to be higher than that in the control group (Fig. 5b). In RP, not only rod cells but also cone cells disappear, and drinking H 2 water was effective in suppressing the decrease in cone cells in rd6 mice.
The rd6 mouse has a mutated MFRP and 4-bp deletion in a splice donor sequence, resulting in exon 4 being skipped and a truncated protein 25 . Although MFRP mutations are linked to photoreceptor cell degeneration, MFRP protein function is not completely understood 32 . In this RNA-seq experiment, expression of the phototransduction gene cluster in the hydrogen group was increased (Fig. 6). When exposed to light, rhodopsin is activated, and phosphodiesterase is activated via transducin, thereby degrading the second messenger cyclic guanosine monophosphate. As a result, the cyclic nucleotide-gated channel in the plasma membrane closes, Photoreceptor inner and outer segments thickness with/without H 2 water. Thickness was significantly greater with H 2 water than without H 2 water (p < 0.01). (c) Outer nuclear layer (ONL) thickness with/without H 2 water. Thickness was significantly greater with H 2 water than without H 2 water (p < 0.01). (d) Outer nuclear layer cells/ slide with/without H 2 water. Thickness was significantly greater with H 2 water than without H 2 water (p < 0.05). Bars depict mean ± standard deviation (SD). Scale bar 50 μm. *p < 0.05, **p < 0.01. www.nature.com/scientificreports/ and inward current stops flowing, resulting in a decrease in the membrane potential. Phototransduction in photoreceptor cells is illustrated in Fig. 6c. Expression of genes related to phototransduction such as rhodopsin, transducin, phosphodiesterase, and cyclic nucleotide-gated channel was elevated. Drinking H 2 water contributed not only to neuroprotection of photoreceptor cells but also to improvement in photoreceptor function. We examined the superpathway of cholesterol biosynthesis, which was an upregulated pathway in the H 2 group. Idl1, Acta2, Cyp51A1, and Hmgcs1 were slightly elevated ( Supplementary Fig. S1). Our gene expression analysis also revealed that the pathway of estrogen receptor signaling was slightly downregulated in the H 2 group. Although an influence of sex on phenotype was not observed in rd6 mice, the possibility of different responses to H 2 treatment cannot be ruled out, due to differences in hormonal profiles and inflammatory responses. As our study used only male mice, further research is needed to clarify potential sex differences in H 2 effects.

Scientific Reports
The inflammatory response and response to oxidative stress were expected to be affected, but the genes involved in these responses were not significantly changed. For the RNA-seq data of this experiment, we analyzed the whole neural retina, but if we had performed single-cell RNA-seq with photoreceptor cells, different results may have been obtained. In addition, the analysis was performed at 49 weeks of age, but because photoreceptor degeneration has progressed considerably, performing the analysis at an early stage when photoreceptor cells remain may be important.
In this experiment, as shown in Fig. 7, we found no difference in the average expression intensity of GFAPand MOMA-2-positive cells. GFAP is an intermediate filament protein and is a marker of Müller glial cells, but its expression is increased by inflammation. However, unlike during acute inflammation, inflammation is minimal in RP, and a change in GFAP expression is unlikely. MOMA-2 is a monocyte/macrophage-specific protein. MOMA-2 positive cells are found in the subretinal space of rd6 mice. In our experiment, we found no significant difference in the number of MOMA-2 positive cells between the H 2 group and the control group (Fig. 7b).
Microglia are involved in the progression of RP and pathologically accumulate in the outer layer of the retina in RP mice 37,38 . Mechanistically, microglia have been shown to play a key role in photoreceptor degeneration in RP 39 . Iba-1 is a microglia/macrophage-specific calcium-binding protein with actin-bundling activity, and shows www.nature.com/scientificreports/ increased expression during neuroinflammation. We assessed the microglial infiltration into the outer layer of the retina by counting Iba1+ cells, but found no significant change in the H 2 group. Further evaluation of the effect of hydrogen on microglial inflammatory response in rd6 mice will require specific evaluation of inflammatory gene expression in microglia. As an animal model with slow retinal degeneration, rd6 is suitable for observing the effects of long-term interventions 31 . Hadziahmetovic et al. reported that 5 months of treatment with the oral iron chelator deferiprone (DFP) prevented thinning of the outer retina in rd6 mice 40 . As DFP appears effective in the reversal of oxidative stress-related tissue damage, the mechanism by which DFP delays the progression of RP might be similar to that of hydrogen. The present study evaluated the therapeutic effects of antioxidant therapy on RP progression from multiple perspectives, including not only morphological changes, but also by ERG and RNA-seq.
Gene and cell therapy has also been investigated in rd6 mice, and intravitreal injection of genetically engineered bone marrow-derived mesenchymal stromal cells (MSCs) deigned to overexpress brain-derived neurotrophic factor (BDNF) resulted in rescue from the chronic degenerative process of slow retinal degeneration in recipient rd6 mice 41 . These findings suggested that anti-apoptotic signaling induced by MSC-BDNF rescued retinal cells. Our RNA-seq data showed upregulation of anti-apoptotic factor Bcl2 in the H 2 group (Supplementary Fig. S1), suggesting that hydrogen therapy may also be involved in the inhibition of apoptosis in rd6.
In P23H-1 and Royal College of Surgeons (RCS) rats as other animal models, administration of basic fibroblast growth factor (FGF2) and minocycline has been shown to increase photoreceptor survival. Minocycline reduced microglial activation and migration, and the combination of FGF2 and minocycline exhibited greater neuroprotective effects than the effects of either agent alone 42 . The therapeutic effects of combined administration of hydrogen and treatments with different mechanisms of action clearly merits further research.
Currently, H 2 can be administered via multiple routes. In clinical applications, the common routes of H 2 administration include H 2 gas, drinking H 2 -rich water, injection of H 2 -rich saline, bathing in H 2 water, H 2 intake of a solid carrier (coral calcium hydride), and ocular instillation of H 2 -rich saline. Previously, we reported that ocular instillation of H 2 -rich saline is a useful therapy for retinal artery occlusion 12 . Ocular instillation of H 2 -rich www.nature.com/scientificreports/ saline is effective for sudden onset acute diseases such as retinal artery occlusion. However, drinking water is more effective for chronic diseases such as RP because eye drops cannot be applied all the time. In this experiment, we planned our study using drinking water. Shimouchi et al. reported that after the intake of 500 ml of H 2 water, the concentration of H 2 in the breath increases to the level of 36 ppm after 10 min and gradually decreases to the baseline level of 7 ppm at 60 min 43 . Sano et al. reported that 60 min after a single dose of hydrogen, the blood hydrogen concentration is higher than the steady state 44 . We consider that a small amount of drinking water can sufficiently supply hydrogen to the retina. To further improve the effect, the combined use of inhalation of hydrogen during sleep may also prove effective. H 2 has been reported as a novel potential therapeutic strategy for the prevention and treatment of chronic neurological diseases, including AD 45,46 , cognitive dysfunction 47 , mood disorders 48,49 , and PD 50 . We hope that H 2 will play a similar role for RP.

Methods
Animals. Male C57BL/6J mice from Charles River Laboratories Japan (Tokyo, Japan) were used to examine changes in hydrogen concentration during drinking of H 2 water (Fig. 1). Male rd6 mice from The Jackson Laboratory (Bar Harbor, ME) were used for the other experiments. From postnatal 4 weeks, rd6 mice started to drink either regular water or H 2 water. Mice were housed individually in 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. The studies were approved by the Animal Care and Use Committee of Nippon Medical School (approval number; H28-049, 2021-021). All experiments were performed in accordance with the ARRIVE guidelines.

Preservation of the H 2 concentration in H 2 water.
Four male 11-week-old C57BL/6J mice were allowed to drink H 2 water (1.2-1.6 ppm of hydrogen) in an aluminum pack (Merodian Co., Osaka, Japan) for 1 week. Hydrogen leaks rapidly from the H 2 water in regular drinking bottles or valves, so to ensure that mice in this experiment drank a stable, high concentration of H 2 water, we developed unique water drinking valves, which were designed to completely match the water outlet of the H 2 water in an aluminum pack and prevent gas from leaking. In addition, a backflow prevention valve was incorporated into the water drinking valve to prevent air from entering the package. The H 2 concentration was measured before drinking and 1 week after drinking using a needle-type H 2 sensor (Unisense, Aarhus N, Denmark).
OCT imaging. Mice were anesthetized, and pupils were dilated. Mice were placed on the rodent alignment stage. An ophthalmic viscosurgical device was applied with cover glass. OCT images were acquired using a Cirrus HD-OCT Model 4000 (Carl Zeiss, Oberkochen, Germany). A specific adaptor including a 90D lens was placed on the objective lens of the Multiline OCT to focus on the mouse retina. The OCT image resolution was 500 pixels (height) × 750 pixels (width). All images were location matched by scanning vertically through the center of the optic nerve head. The average thickness of the outer retina (between the outer plexiform layer and the RPE) was measured at 200 pixels from the optic nerve head using Adobe Photoshop (Adobe Inc., San Jose, CA). In this study, the maximum number of B-scans set by the manufacturer (20 times) was used for averaging. Experimental (n = 10) and control (n = 8) eyes from each mouse were compared at postnatal 6, 11, 21, 32, 39, and 47 weeks. ERGs. After overnight dark adaptation, mice were anesthetized with an intraperitoneal injection of normal saline solution containing ketamine (80 mg/kg) and xylazine (10 mg/kg). ERGs were recorded using a synchronized trigger and summing amplifier (Primus; Mayo, Nagoya, Japan) with a stimulation device (LS-W; Mayo), as described in our previous report 51,52 . After pupil dilation (0.5% tropicamide and 0.5% phenylephrine ophthalmic solution; Santen Pharmaceutical Co., Osaka, Japan), scotopic responses were examined. ERG responses were measured according to the International Society for Clinical Electrophysiology of Vision guidelines. Scotopic-adapted standard white flash stimuli were set at 0.02 cd·s/m 2 and 2 cd·s/m 2 . At least three ERG readings were collected from each eye. Experimental (n = 10) and control (n = 8) eyes from each mouse were compared at postnatal 5, 10, 16, 20, 27, and 42 weeks.
Histology and thickness of the outer retina. At

RNA-seq and DEG analysis.
Total RNA was extracted from each sample (H 2 group: n = 3; control group: n = 4) of neural retina, treated with DNase 1, and purified using a RNeasy Mini Kit according to the manufacturer's instructions (Qiagen, Valencia, CA). Libraries were sequenced (150 bp × 2 paired-end) on a Novaseq 6000 (Illumina, Inc. San Diego, CA) with a depth of > 40 million reads. Library preparation and sequencing procedures were performed by Rhelixa (Tokyo, Japan), a company specializing in life sciences. Data quality of raw RNA-seq reads in FASTQ files was assessed using FastQC (ver. 0.11.7) to identify potential sequencing cycles with low average quality and base distribution bias. Reads were processed with Trimmomatic (version 0.38), allowing spliced read alignment to the mouse reference genome (GRCm38: mm10) using HISTAT2 (ver. 2.1.0). Fragments per kilobase of exon per million reads mapped (FPKM), FPKM-upper quartile (UQ), and transcripts per million (TPM) data were calculated using featureCounts (version 1.6.3) from the mapped reads. FPKM values were analyzed using iDEP, an integrated web application for RNA-seq data analysis 56 . DEG between H 2 -treated and control groups were identified using the two-tailed permutation FDR-based Student's t test (FDR < 0.15). We then performed a pathway analysis based on the identified genes and generated images using QIAGEN IPA (Ingenuity ® Systems, www. ingen uity. com).
Statistics. All comparisons between the control group and H 2 group were done with the paired t-test. The mean and standard deviation for these measurements were calculated for each group. Values of p < 0.05 were considered statistically significant.
Ethics approval. All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The studies were approved by the Animal Care and Use Committee of Nippon Medical School (approval number; H28-049, 2021-021).

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.