Dulbecco’s modified Eagle’s medium was purchased from Nissui Pharmaceutical Co., Ltd. (Tokyo, Japan). All solutions were prepared using ultrapure water from a Milli-Q® Synthesis system (Millipore, Tokyo, Japan). Individual gas cylinders each containing O₂ (purity, >99.7%; quality, N₂ < 1000 ppm, H₂O < 10 ppm), N₂ (purity, >99.999%; quality, O₂ < 2 ppm, CO < 1 ppm, CO₂ < 1 ppm, CH₄ < 1 ppm, H₂O < 5 ppm), and H₂ (purity, >99.998%; quality, O₂ < 1 ppm, N₂ < 10 ppm, CO < 1 ppm, CO₂ < 1 ppm, CH₄ < 1 ppm, H₂O < 5 ppm) were obtained from Fukuoka Sanso Co., Ltd. (Fukuoka, Japan). Trolox is an antioxidant and commonly used as the positive control for ROS scavenging activity measurements (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan). Other reagents not detailed herein were obtained from Wako Pure Chemical Inc. (Tokyo, Japan).

Nihon Trim Co. Ltd. has the highest share in the market for alkaline water producers. Thus, we used a water flow-type apparatus (Trim Ion Hyper, Nihon Trim Co. Ltd., Osaka, Japan) as the test apparatus. The test apparatus has the same structure to the previously reported apparatus [3]. Briefly, the test apparatus contains two units, a micro-carbon cartridge unit and an electrolysis unit. Tap water flows into the cartridge unit and passes through the nonwoven-fabric filter, activated charcoal powders and cationic ion-exchange material to remove most of the impurities. Purified tap water then flows into the electrolysis unit, and the water is electrolyzed while passing through the gaps between the electrodes. Electrolyzed tap water near the cathode typically exhibits a high pH, a low amount of dissolved oxygen, a high negative redox potential and a high concentration of dissolved hydrogen (0.4–0.9 ppm). Water produced in this manner, with the above characteristics, is designated as ERW. The apparatus is designed to produce five types of water; four types of ERW: levels 1 (ERW_LV1), 2 (ERW_LV2), 3 (ERW_LV3) and 4 (ERW_LV4) electrolyzed with a constant electric current for each level (0.8 to 4.2 A) at a maximum of 50 volts while passing through the gaps between the electrodes equipped in the test apparatus and one type of filtered water without electrolysis (FW). We also prepared two types of water as controls: MilliQ (MQ) water and filtered water, both are saturated with mixed gas. The mixed gas consisted of 80% H₂, 15% O₂, and 5% N₂, and this gas was bubbled through the water to achieve the same dissolved oxygen (DO) and dissolved H₂ (DH) concentration to those of ERW_LV4. The pH was adjusted to match ERW_LV4 using sodium hydrate.

Dissolved H₂ concentrations were measured using a DM-10B2 meter (ABLE Corporation, Tokyo, Japan). Degassed ultrapure water was saturated with H₂ gas by bubbling for 30 min. This solution was defined as standard H₂-saturated (H₂Sat) water. Dissolved H₂ concentrations in sample solutions were calculated using H₂Sat as the standard value.

Dissolved O₂ was measured using an YSI-3000 dissolved O₂ meter (YSI/Nanotech Inc. Kanagawa, Japan). The oxygen concentration was determined by setting an air-saturated MQ water, which contains 7.87 ppm of oxygen at 1 atm and 27°C, as 100% [19]. We also measured the oxidation-reduction potential (ORP) (F-52 with electrode, 9300-10D, HORIBA, Ltd., Tokyo, Japan), electrical conductivity (EC) (CM-14P, DDK-TOA, Co., Tokyo, Japan) and pH (F-52 with electrode, 9618, HORIBA, Ltd., Tokyo, Japan).

The ORAC assay measures the change of probe fluorescence in response to the scavenging of peroxyl radicals generated by 2,2ʹ-azobis (2-methylpropionamidine) dihydrochloride (AAPH) by antioxidants that prevent the degradation of the fluorescein probe. The analysis of the ORAC assay was carried out following the method by Huang et al. [20]. Briefly, 20 μL of 15 μM fluorescein dissolved in a 100 mM phosphate buffer (pH 7.0) solution was added to the designated wells of a 96-well black plate, followed by the addition of 25 μL of blank, standard (Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) 5–40 μM), or sample to the wells. Then, 15 μL of freshly prepared 30 mM 2,2′-azobis (2-methylpropionamidine) dihydrochloride solution was added to all wells. Fluorescence was monitored using 485 nm of excitation and 528 nm of emission at 10-min intervals for 180 min to generate fluorescence intensity versus time plots for samples, blank and Trolox. Using the plots, the net area under the curve for the sample was calculated by subtracting the integrated values of the area under the curve for the blank from that of the sample and Trolox. The value for individual samples was then obtained by subtracting the net Trolox value from that of the samples.

An adopted method of the 2,2-diphenyl-1-picrylhydrazyl (DPPH) assay by Brand-Williams et al. was used for measuring total antioxidative capacity [21]. The method measures the absorbency reduction of stable DPPH nitrogen radicals by the action of antioxidants. Trolox is commonly used to determine the equivalent value of an antioxidant [22,23]. Briefly, 500 μM of DPPH was prepared in ethanol or acetonitrile as the stock solution. Reaction cocktails consisted of 50 μL of either blank, Trolox (positive control) or samples, 10 μL of acetate butter (pH 5.5) and 40 μL of DPPH were mixed, and kept in the dark for 30 min. The absorbance values were measured at 515 nm using a microplate reader. A standard curve was prepared using 50–250 μM Trolox.

HT1080 cells (CCL-121; American Type Culture Collection, Manassas, VA, USA), a human fibrosarcoma cell line, were cultured with medium supplemented with 10% (v/v) fetal bovine serum (Lot No. 12D168, Sigma-Aldrich Japan Co. LLC, Tokyo, Japan). The cell number was measured using a cell counter (Blood cell counting device F-520, Sysmex Co., Hyogo, Japan). HT1080 cells at a density of 0.5 × 10⁵ cells in 100 μL DMEM were used to inoculate wells of 96-well plates and cultured for 24 h in a CO₂ gas incubator. Then, the cells were placed in a medium containing various sample waters with or without hydrogen peroxide (H₂O₂). Briefly, cells were first treated with 60 μM H₂O₂ for 30 min and replaced with a medium made from 14 types of water, followed by 120 min of incubation. Sample media were prepared immediately before use. This was done by mixing 5× concentrated DMEM medium with sample water at a ratio of 1:4. Cells were then incubated with 2 μM 3′-O-acetyl-6′-O-pentafluorobenzenesulfonyl-2′,7′-difluorofluorescein (BES-H₂O₂-AC) [24] and 1 μM Hoechst 33342 for 15 min. The intracellular fluorescence intensity generated by the sample waters was probed with BES-H₂O₂-AC designed to detect specifically intracellular H₂O₂ and the fluorescence levels were detected using an IN Cell Analyzer 1000 (GE Healthcare, Tokyo, Japan) with an excitation filter at 480 nm and an emission filter at 535 nm. Data derived from 1350 cells were statistically analyzed. The IN Cell Analyzer 1000 was able to rapidly and quantitatively determine the fluorescence intensity of BES-H₂O₂-AC in each well to provide statistically reliable data derived from several thousand to several million cells. In this study, we used a specific lot of fetal bovine serum that we empirically determined as the best-suited serum that ensured high sensitivity and reproducibility for the detection of intracellular ROS scavenging activity in HT1080 cells. We provide more detail in the Supporting Information. We also performed immuno spin trapping assay to evaluate the antioxidative activity of ERW. Immuno spin trapping assays with 5,5-dimethyl-1-pyrroline N-oxide (DMPO; Labotec Co., Tokyo, Japan) were performed by one of the following three procedures: 1. Cells were treated with 80 μM H₂O₂ for 30 min, which was replaced with water sample medium with DMPO for 30 min. 2. Cells were treated with 80 μM H₂O₂, water sample medium and DMPO at the same time for 90 min. 3. Cells were treated with 80 μM H₂O₂ or 100 μM menadione for 30 min, which was replaced with water sample medium followed by 90 min of incubation. In all of these three alternatives, cells were then incubated with 40 mM DMPO for 30 min. Sample media were prepared immediately before use. This was done by mixing 5× concentrated DMEM medium with sample water at a ratio of 1:4. Then, cells were fixed with 4% paraformaldehyde/phosphate-buffered saline (PBS) on ice. Anti-DMPO nitrone adduct antibody (ab110430; Abcam, Cambridge, UK) was used at 1:1000 dilution with PBS-T/1% bovine serum albumin after treatment with PBS/0.25% Triton X for 10 min. After that, cells were incubated with fluorescein-labeled avidin (1:400 dilution; Fluorescein Avidin DCS, cell sorting grade; Vector Laboratories Ltd., Burlingame, CA, USA) and 1 μM Hoechst 33342 for 60 min. The intracellular fluorescence intensity was detected using an IN Cell Analyzer 1000 (GE Healthcare, Tokyo, Japan) with an excitation filter at 480 nm and an emission filter at 535 nm. The IN Cell Analyzer 1000 can rapidly and quantitatively determine the fluorescence intensity in each well to provide statistically reliable data derived from several thousand to several million cells. We established this protocol to detect weak intracellular oxidation. Therefore, hydrogen peroxide treatment is used to induce oxidation and/or injury to the intracellular components of cells. The probe is expected to detect the production of intracellular ROS resulting from such injury of organelles, from oxidized cellular molecules, from the process of recovery of oxidized molecules, and from the retention of intracellular hydrogen peroxide. There are additional information for redetection of intracellular scavenging activity of ERW (S1 Supporting Information)

Ten milliliters of water samples (MQ, FW, LV1, LV2, LV3, LV4) were cleaned by filtration and analyzed semi-quantitatively for elements present in the samples using ICP-MS (Agilent 7500c, Agilent Technologies Co. Ltd., Santa Clara, CA, USA) at the Center of Advanced Instrumental Analysis of Kyushu University.

The structure of the surface and cross-section of the electrode plate were observed under a field-emission scanning electron microscope (JEOL-6390, JEOL Ltd. Tokyo, Japan). Photographs were taken at 10 kV with magnifications of 10,000 or 30,000.

Microsoft Excel 2016 was used to calculate the means, standard deviations or coefficient of determinations in triplicate experiments. One- or Two-way ANOVA (STATMATE III for Windows) was used to determine if there were statistically significant differences.

Parameters such as pH, EC, ORP, DO and DH were measured for several types of water, including MQ, tap water, FW and ERWs (ERW_LV1–LV4). FW is filtered water by passing tap water through the micro-carbon cartridge without electrolysis. ERW_LV1 is ERW generated by electrolyzing filtered water at level 1 with constant electric current at 50 V upper limit voltage and a flow rate of 1.8–2.0 l/min. Likewise, other ERWs were produced using identical conditions, except selection of the LV2, LV3 or LV4 switch. As shown in Table 1, pH, ORP negativities and DH values increased and were dependent on the electrolysis strength, whereas EC and DO remained essentially constant. Such fundamental characteristics of ERWs produced by the test apparatus are in agreement with previously published data, demonstrating the reproducibility of the apparatus [3,9]. Using these waters as starting materials, we examined whether the ERWs have ROS scavenging activity in HT1080 cells after H₂O₂ stimulation.

As shown in Fig 1, the Bes-H₂O₂-AC probe detected basal intracellular fluorescence (FL) in HT1080 cells treated with all sample waters, including MQ, FW and ERW_LV1‒ERW_LV4 (Fig 1, left). The FL intensity was found to increase only when treated with 60 μM of hydrogen peroxide for 30 min (Fig 1, right). ERW_LV1-treated cells did not show a difference in this regard from those treated with FW; however, intracellular FL intensities in the cells treated with ERW_LV2‒ERW_LV4 were found to decrease significantly and in an electrolysis intensity-dependent manner (Fig 1, right). The sample waters have different pH, as shown in Table 1; however, we confirmed that the pH levels of all of the prepared media containing the sample waters had been adjusted using medium buffer to the range of 7.39 to 7.49 (Sl Table). We used 5 mM of N-acetyl cysteine (NAC) as a positive control, which showed a significant reduction of H₂O₂ induced intracellular ROS levels. Thus, confirming that the assay system was working properly. NAC was used here because it acts as a direct ROS scavenger for hydrogen peroxide, the superoxide radical and hydroxyl radical, and as an indirect ROS scavenger by functioning as a precursor of reduced glutathione. Therefore, electrolysis levels are closely correlated with dissolved hydrogen levels and ROS scavenging ability. To deepen our understanding, we prepared two types of water as controls: (i) MQ water and (ii) FW, in which both are saturated with mixed gas (MQ+Mixed gas, FW+Mixed gas, parameters shown in Table 2)

The control waters and plain MQ, FW and ERW_LV4 were assessed for their intracellular ROS scavenging activity. All five types of water showed basal FL levels (Fig 2, left). In contrast, the cells cultured with plain MQ and FW-based medium treated with 60 μM H₂O₂ for 30 min showed approximately 750 cps of FL intensity, referring to them as absolute controls, whereas these two types of water saturated with mixed gas significantly reduced the intracellular FL intensity when compared with that of the absolute controls (Fig 2, right). Moreover, the cells cultured with ERW_LV4 based medium significantly reduced the intracellular FL intensity when compared with that of the absolute controls and mixed gas saturated waters. NAC (5 mM, 30 min) as a positive control exerted an expected antioxidative effect (Fig 2, right).

To confirm whether intracellular ROS scavenging activity is conferred by H₂ gas, we treated the water with degassing and autoclaving steps, because such treatments are thought to eliminate and purge gases, respectively [5,25]. Intracellular ROS scavenging activity in the cells treated with the degassed ERW_LV4-based medium still showed significant ROS reduction; although the level was slightly weaker than that of the sample without degas treatment (Fig 3A). This is important because the ROS scavenging activity remains in ERW_LV4 after degassing treatment, thus indicating the existence of non-gaseous ROS scavenging factors in addition to gaseous factors (e.g., H₂). We then examined the effect of an autoclaving step for ROS scavenging ability. The result demonstrated that the cells cultured with the autoclaved ERW_LV4-based medium completely lost intracellular ROS scavenging activity, whereas a consistent ROS scavenging effect was obtained with non-autoclaved ERW_LV4-based medium (Fig 3B). These results suggest that the prospective antioxidative factor has heat and pressure labile characteristics.

To elucidate the nature of the prospective antioxidative factors contained in ERW_LV4, we first prepared several types of water simulating the ERW_LV4 characteristics (Table 2) by adjusting the pH, DO and DH of MQ and FW by bubbling the mixed gas with N₂, O₂ and H₂. These sample water-based culture media were used to culture HT1080 cells after either degassing or autoclave treatment, and the intracellular ROS scavenging activity was measured using BES-H₂O₂-AC after the addition of H₂O₂. Degassed and H₂O₂-stimulated sample waters (MQ, FW) could not reduce intracellular ROS, but ERW_LV4 and NAC reduced ROS significantly (Fig 3A). In contrast, the same experiments using the autoclaved waters showed that ERW_LV4, but not NAC, lost its ROS scavenging activity (Fig 3B). Therefore, the results indicate that the antioxidative factors in ERW_LV4 are produced during electrolysis.

We also performed immuno spin trapping assays of ERWs to confirm the antioxidative activity of ERW. We firstly tested effects of treatment with the identical BES-H₂O₂-AC protocol but could not detect a difference of FL intensity in cells treated with H₂O₂ at concentrations of between 0 and 60 μM (data not shown). Then, we changed the treatment order as follows: Cells were treated with 80 μM H₂O₂ for 30 min, which was then replaced with water sample medium with DMPO for 30 min (Fig 4A). As a result, we detected a significant decrease in FL intensity in LV4 and NAC sample water, but not in mixed gas FW. Therefore, we next changed the treatment order as follows: treatment with 40 mM DMPO, sample water and 80 μM H₂O₂ at the same time for 90 min (Fig 4B). As a result, we detected significant differences between FW and FW+mixed gas, FW and LV4, and FW+mixed gas and LV4. These results suggest that DMPO requires a certain volume of hydrogen peroxide and a supply of oxygen radicals nearby to produce DMPO adducts. Therefore, to establish similar conditions in our study using a BES-H₂O₂ probe, we additionally performed experiments using menadione instead of hydrogen peroxide.

As shown in Fig 5, the FL intensity was increased upon treatment with 80 μM menadione for 45 min. The CL intensity of cells which cultured with LV4-contained medium significantly decreased and FW+Mixed gas had a tendency to reduce the intracellular FL intensity when compared with that of the absolute controls. NAC was used as a positive control. In addition, LV4 and FW seemed to lose ROS scavenging activity upon degas treatment; moreover, LV4 degassed water showed a tendency for persistence in its activity.

In parallel with this experiment, sample waters were tested to determine whether the prospective antioxidative factors exert direct ROS scavenging activity for the DPPH radical and hydrogen atom transfer-based ORAC assays. Trolox was used as a positive control as well as a system control, and was confirmed to reduce absorbency in a concentration-dependent manner where 40 μM and 80 μM Trolox significantly reduced the absorbency at 520 nm compared with MQ in the DPPH assay (Fig 6A). However, FW, H₂-dissolved water (plus the gas mix) of MQ and FW and H₂-saturated water (H₂Sat) as well as ERW_LV1 and ERW_LV4 did not show DPPH radical scavenging activity (Fig 6A). Similarly, 10 μM of Trolox showed ROS scavenging activity, whereas FW, +H₂ gas of MQ and FW, H₂Sat, ERW_LV1 and ERW_LV4 did not show 2,2′-Azobis 2-amidinopropane dihydrochloride-derived ROS-scavenging activity in the ORAC assay (Fig 6B). Thus, the DPPH and ORAC assays indicated that the prospective antioxidative factors did not involve stable radical reduction and hydrogen-atom-donation reactions. However, complete exclusion of weak antioxidative activities by the factors cannot be ruled out completely, because the detection limit of DPPH and ORAC is ~10 μM and 1 μM Trolox equivalent, respectively. We additionally measured the ROS-scavenging activity of dissolved hydrogen using a chemiluminescence assay because this assay has a more sensitive ROS response, which can be seen in the supporting information (S2 Supporting Information and S1 Fig).

Based on the present results, the prospective antioxidative factors are considered to be dissolved H₂ and non-gaseous materials with heat and pressure lability; both species are produced in an electrolysis strength-dependent fashion. To further specify the latter prospective factors, we analyzed elements contained in various types of water using semi-quantitative ICP-MS analysis (S2–S4 Figs). We searched for an element from 62 elements that increased in amount in an electrolysis-dependent manner, yet existed at low levels in the absence of electrolysis. Most elements were found to show minimal change and exist at low levels. However, Pt was found to increase in an electrolysis-dependent manner. This tentative conclusion is probable because we reported previously that ERW contains platinum nanoparticles (Pt NPs), as visualized by transmission electron microscopy [26]. It has been suggested that the elevated Pt NPs in ERW originate from the platinum electrode equipped in the ERW producing apparatus [26]. Further support for the presence of Pt NPs comes from a report that Pt NPs form aggregates upon autoclave treatment and such aggregates no longer show ROS scavenging activity [27]. Our data and previous reports suggested Pt NPs exist in ERW, but the process of formation of Pt NPs has not been considered in depth. It has been reported that ERW could contain metal and/or nonmetal nanoparticles eluted from electrode ions during electrolysis only when electrodes are constructed with a metal with a high ionization tendency, which excludes platinum as a candidate metal [28]. However, an alternative view suggests that the surface structure of metal-coated plates could be formed as a layer of very tiny globular structure [29,30]. The presently used apparatus contains platinum-coated plate electrodes and the surface structure appears to be a sheet of aggregated nanoparticles (Fig 7). Although SEM magnification could not clearly reveal nanometer-sized globular structures, it is assumed that they are nanometer-sized particle-like structures, as suggested previously [29,30]. These reports lead us to consider that electrolysis increases the number of Pt NPs in ERW due to physical force-detachment during electrolysis, and the produced Pt NPs are a potential active factor of the ROS-scavenging activity of ERW. However, a drawback to this hypothesis is that the amount of Pt NPs present in ERW is very low, with values between 0 and 0.7 ppb [7]. The Pt NPs concentration of ERW produced by the currently used apparatus is 0.2 ppb, as determined by quantitative analysis of ICP-MS data (data not shown). The molar concentration is calculated to be 20 pM based on our previous study [27]. Therefore, such a Pt NPs concentration appears to be too low to exert direct intracellular ROS scavenging activity, as observed in Fig 2. In an attempt to link the intracellular ROS scavenging activity and the trace amount of Pt NPs in ERW, we propose an indirect ROS-scavenging activity mechanism, in which the Pt NPs work as a ligand for activation of the intracellular antioxidant systems. This mechanism holds true only when an additional assumption is fulfilled; the existence of high affinity nanoparticle-receptors that have affinity comparable to low molecular weight hormone receptors present in cells [31].