All animal experimental protocols were approved by the ethics committee of Changhai Hospital affiliated to the Second Military Medical University (Shanghai, China) and were conducted in accordance with their guidelines.

A total of 24 male Sprague-Dawley rats (age, 10 weeks; weight, 250–280 g) were purchased from the Shanghai Laboratory Animal Center of the Chinese Academy of Sciences (Shanghai, China). Rats were housed at 24±2°C under a 12-h light/dark cycle, and fed ad libitum. All rats received humane care according to the Guide for the Care and Use of Laboratory Animals (16). The animals were randomly divided into three groups: i) The control group (n=8), which was fed a normal diet (15% kcal from fat), and normal saline (5 ml/kg) was administered intraperitoneally twice daily; ii) the model group (n=8), which was fed a high-sugar and high-fat diet, and normal saline (5 ml/kg) was administered intraperitoneally twice daily; and iii) the H₂-rich saline treatment group (n=8), which was fed a high-sugar and high-fat diet, and H₂-rich saline (5 ml/kg) was administered intraperitoneally twice daily. The high-fat diet was prepared according to Li et al (17) and contained 2% cholesterol, 7% lard, 8.3% yolk, 16.7% sucrose and 66% standard diet, which provided 4.66 kcal/g with an energy composition of 31.59% from fat, 51.73% from carbohydrate and 16.68% from protein.

Lipotoxicity was induced by a high-fat diet, and streptozotocin (STZ; Sigma-Aldrich; Merck Millipore, Darmstadt, Germany) was injected to cause glucotoxicity, which subsequently led to islet β-cell failure and apoptosis (18). Briefly, a single injection of STZ (25 mg/kg) was administered via the tail vein in 0.1 mol/l citrate buffer (pH 4.2) followed by a continuous high-fat and high-sugar diet for 8 weeks. Subsequently, an oral glucose tolerance test was performed as described previously (19). Rats from the control and treated groups received 2 g/kg glucose orally; a 2 h blood glucose result measuring >1.2 mmol/l or a random blood glucose result measuring >16.8 mmol/l in rats was considered successful model establishment.

H₂-rich saline was prepared as previously described (16). Briefly, H₂ gas was dissolved in normal saline for 2 h under 0.4 MPa pressure to saturation. The concentration was measured with gas chromatography to ensure the hydrogen level was >0.6 mmol. H₂-rich saline was prepared each week and stored at 4°C in aluminum bags until ready for use.

Blood was collected from the tail vein at 8 weeks, and serum alanine transferase (ALT), total bilirubin (TBIL), total cholesterol (TC), triglycerides (TG), fasting blood glucose (FBG) and fasting insulin (FINS) levels were determined using a biochemistry analyzer. In addition, the insulin sensitivity index (ISI) and homeostasis model assessment-insulin resistance (HOMA-IR) were calculated as follows: HOMA-IR = (fasting blood glucose × fasting insulin) / 22.5; ISI = 1/(fasting blood glucose × fasting insulin). Insulin tolerance tests were conducted on the three rat groups as described previously (20).

After 8 weeks, the rats were anesthetized using by an intraperitoneal injection of 10% aqueous solution (0.3 ml/100 g) of chloral hydrate (Huai'an Xingzhi Biological Technology Co., Ltd., Huaian, China) and their livers were harvested, sectioned, fixed in 4% paraformaldehyde and embedded in paraffin. The paraffin blocks were then sliced into 4 µm sections and deparaffinized according to Shi et al (21).

Sections were stained with H&E and images were captured using a light microscope. A pathologist that was blinded to the animal groups evaluated the slides and scored each liver tissue specimen for steatosis, inflammation and fibrosis based on the criteria proposed by previous studies (22–24). Briefly, for steatosis: Score 0, none present; score 1, steatosis <33% of the parenchyma; score 2, steatosis 34–66% of the parenchyma; score 3, steatosis >67% of the parenchyma. For inflammation: Score 0, no foci of inflammation; score 1, <1 foci per two 200x fields; score 2, one foci per two 200x fields to one foci per one 200x field; score 3, one to two foci per one 200x field; score 4, >2 foci per one 200x field. For fibrosis: Score 1, zone-3 perisinusoidal fibrosis; score 2, zone-3 perisinusoidal fibrosis with portal fibrosis; score 3, zone-3 perisinusoidal fibrosis and portal fibrosis with bridging fibrosis; and score 4, cirrhosis. The total score (steatosis + inflammation + fibrosis) of each rat was calculated and the average score for each group was determined.

TUNEL staining was performed using an In Situ Cell Death Detection kit (Roche Diagnostics, Basel, Switzerland). Liver sections were heated to 60°C for dewaxing and were then rehydrated according to standard protocols and as previously reported (21). After cooling to room temperature, the sections were incubated with 20 µg/ml proteinase K solution for 20 min followed by incubation in TUNEL reaction mixture for 1 h at 37°C. Converter-alkaline phosphatase antibody was then added to the samples for 1 h at 37°C. Subsequently, sections were washed with PBS three times for 5 min, followed by color development in the dark with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate. Ten fields were randomly chosen (x200). The number of apoptotic hepatocytes was counted in each field and an average was calculated.

Rat livers were collected and washed in normal saline, and were then homogenized immediately on ice in 1 ml normal saline at 2–8°C. The homogenates were centrifuged at 3,000 × g at 4°C for 15 min. The expression levels of TNF-α (catalog no. H052) and IL-1β (catalog no. H002) were measured using commercial ELISA kits (Nanjing Jianchen Bioengineering Institute, Nanjing, China) according to the manufacturer's protocol.

Similar methods were used to determine CASP-3 activity as previously described (21). Livers were harvested at 8 weeks, and CASP-3 activity was determined with the CASP-3/CPP32 Fluorometric Assay kit (BioVision, Inc., Milpitas, CA, USA). Briefly, hepatocyte lysates were incubated with 10 µg/ml proteinase K for 30 min at 37°C in a 96-well plate. The peptide substrate DEVD-AFC (5 µl) was then added to initiate the reaction. After incubation in the dark at 37°C, a fluorometer was used to read the plate using a 400-nm excitation filter and a 505-nm emission filter. Fold increase in CASP-3 activity was calculated by comparing with the level of the control.

Similar methods were employed as previously reported (25). Liver sections were deparaffinized in xylene, rehydrated with ethanol and pretreated with 10 µg/ml proteinase K for 30 min at 37°C. Subsequently, the sections were incubated in 10% bovine serum albumin (Nanjing Jianchen Bioengineering Institute) for 20 min. After overnight incubation at room temperature with anti-8-OHdG (1:200; catalog no. ZY-1131R) and anti-3-NT antibodies (1:200; catalog no. 28-60252P) from Heyzer Ye Biological Technology Co., Ltd. (Shanghai, China). An alkaline phosphatase-conjugated secondary antibody (1:500; catalog no. E030210; Shanghai Yanhua Bio-Tech Co., Ltd., Shanghai, China) were added for 1 h at 37°C, and the sections were incubated with diaminobenzidine. The number of 8-OHdG- and 3-NT-positive cells were counted under a light microscope, and integral optical densities were calculated with Image-Pro Plus software version 6 (Media Cybernetics, Inc., Rockville, MD, USA).

The concentrations of 3-NT and 8-OHdG in the liver were determined using commercial ELISA kits [catalog nos. JK-(a)-5053 and JK-(a)-1571; Shanghai Jinkang Medicine Technology Co., Ltd., Shanghai, China], according to the manufacturer's protocol. Liver tissues were homogenized in 2 ml 10 mM phosphate-buffered saline (pH 7.4). After centrifugation at 10,000 × g for 30 min at 37°C, the levels of 3-NT and 8-OHdG in the supernatant were measured using the corresponding kits.

The protein expression levels of PPARα and PPARγ in the liver were examined using immunohistochemistry. The transcription of PPARα and PPARγ were determined by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). Total RNA was extracted using TRIzol^® reagent according to the manufacturer's protocol (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The RT reaction was performed using the Superscript II Reverse Transcriptase kit (Invitrogen; Thermo Fisher Scientific, Inc.) in a 20 µl volume with 1 µg total RNA; RT was conducted at 16°C for 30 min, 42°C for 42 min and 85°C for 5 min. The RT product (1 µl cDNA, corresponding to 6.25 ng RNA) was used to conduct subsequent qPCR analyses in an ABI Prism 7300 Sequence Detection system (Applied Biosystems; Thermo Fisher Scientific, Inc.), with the SYBR^® Green PCR kit (Applied Biosystems; Thermo Fisher Scientific, Inc.). Each experiment was performed in duplicate for each gene under the following cycling conditions: Initial template denaturation at 95°C for 5 min, followed by 40 cycles at 95°C for 10 sec, 60°C for 20 sec, 72°C for 20 sec and 78°C for 20 sec, and a final 10 min extension step at 72°C. The primer sequences were as follows: PPARα, forward 5′-GGTCTTAACCGGCCC-3′, reverse 5′-AAACGCAACGTAGAG-3′; PPARγ, forward 5′-CTGTTTTATGCTGTTATGGGTGAAA-3′, and reverse 5′-GCACCATGCTCTGGGTCAA-3′; and GAPDH, forward 5′-TGGGTGTGAACCACGAGAA-3′, and reverse 5′-GGCATGGACTGTGGTCATGA-3′. The ΔΔCq method was used to to normalize mRNA expression levels to those of GAPDH (26).

All data are presented as the mean ± standard deviation (n=8) and were analyzed with SPSS 9.1 software (SPSS, Inc., Chicago, IL, USA). Differences between groups were compared with one-way analysis of variance followed by the Student-Newman-Keuls post hoc test. Glucose tolerance and insulin tolerance test results were analyzed by mixed model for repeatedly measured data for multiple comparisons. P<0.05 was considered to indicate a statistically significant difference.

H&E staining was used to observe liver morphology by microscopy (Fig. 1). As presented in Fig. 1A, rats in the control group demonstrated a normal architecture with hepatocytes arranged in plates aligned to sinusoids converging to centrilobular veins. Rats in the model group, which were fed a high-sugar and high-fat diet, presented with moderate to severe steatosis, ballooning degeneration, piecemeal necrosis and inflammatory cell infiltration. In addition, early fibrosis was detected. In the H₂ group, a relatively normal histologic architecture was observed, with little steatosis, ballooning degeneration, necrosis and inflammatory cell infiltration. The average histopathological scores are presented in Fig. 1C. Scores in the model group were significantly higher than those in the control and H₂ groups (P<0.05 model group vs. control and H₂ groups).

For analysis of apoptosis, TUNEL staining results are presented in Fig. 1B. At a magnification of ×200, the nuclei of hepatocytes were clearly stained. Some occasional TUNEL-positive cells were detected in the control group. Significantly more apoptotic cells were present in the model group, in which the apoptotic cells were significantly shrunken and the nuclei were strongly stained. In the H₂ group, the number of apoptotic cells was significantly smaller than that in the model group (Fig. 1D; P<0.05 model group vs. control and H₂ groups).

To assess lipid metabolism and liver function, serum levels of ALT, TBIL, TC and TG were measured at 8 weeks. ALT and TBIL levels in the model group were slightly higher compared with the control and H₂ groups (Fig. 2A and B); however, the differences were not statistically significant (P=0.08). The TC and TG levels in the model group were significantly higher compared with those in the control and H₂ groups (Fig. 2C and D; P<0.05 model group vs. control and H₂ groups).

At 8 weeks, the serum FBG and FINS levels were also examined to assess the function of pancreas islets (Fig. 2E and F). For FBG, the differences between the model group and the control or H₂ groups were not statistically significant; for FINS, the difference was statistically significant (P<0.05 model group vs. control and H₂ groups).

Finally, the ISI and HOMA-IR were calculated to assess insulin resistance, and it was revealed that in the H₂ group, when compared to the model group, the ISI was significantly elevated and the HOMA-IR was significantly reduced (P<0.05 H₂ vs. model group).

At 8 weeks, the insulin tolerance test and glucose tolerance test were performed. The insulin tolerance test was performed to assess pituitary function and adrenal function, and as presented in Fig. 2I, following injection of insulin serum glucose levels gradually decreased over time. The rate of decrease in the model group was significantly lower than that in the other two groups. At 60 and 90 min after the injection, the difference was statistically significant (P<0.05 model group vs. control and H₂ groups, P<0.01 model group vs. control and H₂ groups, respectively). The glucose tolerance test was performed to diagnose diabetes, insulin resistance and impaired beta cell function. As presented in Fig. 2J, following glucose intake blood glucose gradually increased, reached a peak at 30 min and subsequently decreased (*P<0.05 model vs. control and H₂ groups).

3-NT, a biochemical marker of peroxynitrite, and 8-OHdG, a product of free radical oxidative damage to DNA, have been reported to be ameliorated by H₂ (25). The number of 3-NT- and 8-OHdG-positive cells were detected and are presented in Fig. 3A-D. The model group had more 3-NT-positive cells (Fig. 3A and C) and more 8-OHdG-positive cells (Fig. 3B and D) compared with the control and H₂ groups (P<0.05 model group vs. control and H₂ groups).

The levels of the inflammatory cytokines TNF-α (Fig. 3E) and IL-1β (Fig. 3F) in the model group were significantly greater compared with the control and H₂ groups. CASP-3 activity in the model group was significantly greater compared with the control and H₂ groups (Fig. 3G; P<0.05 model group vs. control and H₂ groups).

PPARα and PPARγ protein expression was detected using immunohistochemistry (Fig. 4A-D). As shown in Fig 4A and C, PPARα protein expression was enhanced in the H₂ group compared with the control and model groups. As shown in Fig. 4B and D, PPARγ protein expression was also increased in the H₂ group compared with the control and model groups. These differences were statistically significant. The immunohistochemical results were confirmed by RT-qPCR (Fig. 4E and F; P<0.05 model vs. H₂ group).