Hydrogen gas alleviates blood-brain barrier impairment and cognitive dysfunction of septic mice in an Nrf2-dependent pathway
Abstract
Sepsis-associated encephalopathy (SAE) is a cognitive impairment caused by sepsis and is related to increased morbidity and mortality. Damage to the blood–brain barrier (BBB) has been proved to be one of the important causes of SAE. Molecular hydrogen (H2) is a promising method for the treatment of SAE, yet the underlying mechanism is not clear. This study was designed to demonstrate whether H2 can alleviate SAE by protecting the BBB, and whether it is protected by Nuclear factor erythroid-2-related factor 2 (Nrf2) and its downstream sig- naling pathways. Either a sham or a cecal ligation and puncture (CLP) procedure was applied to female wild-type (WT) and Nrf2-knock-out (Nrf2-/-) C57BL/6J mice. H2 (2%) was given for 60 min starting at 1 h and 6 h after the sham or CLP procedure. In addition, bEnd.3 cells cultured with medium which contained LPS, Saline, DMSO or ML385 (a Nrf2 inhibitor) were also used in the research. The 7-day survival rates were recorded. The Morris water maze was used to determine cognitive function. Pro-inflammatory and anti-inflammatory cytokines [tumor necrosis factor-alpha (TNF-α), interleukin 6 (IL-6), HMGB1, and IL-10), antioxidant enzymes, and oxidation products [superoxide dismutase (SOD), chloramphenicol acetyltransferase (CAT), malondialdehyde (MDA), and (8-iso-PGF2α)] were determined by enzyme-linked immunosorbent assay (ELISA). Brain water content, Dextran tracer, and Evans blue extravasation were used to detect the damage of the BBB. Western blot analysis was used to detect β-catenin, phosphorylated β-catenin, adhesion-linked protein VE-cadherin, and associated tight junction protein ZO-1. We found that H2 can improve survival in septic mice, decrease escape latency and platform crossing times, decrease pro-inflammatory cytokines and oxidative product levels in the mouse cortex, and increase the expression of anti-inflammatory factors in WT, but not Nrf2-/-, mice. Moreover, H2 can also decrease brain water content, extravascular dextran, extravascular Evans blue dye, and β-catenin level, and increase ZO-1 and VE-cadherin expressions in WT mice, but not in Nrf2-/- mice. Our result shows that H2 can protect the BBB by decreasing its permeability, thereby reducing SAE and improving cognitive function, which is mediated through Nrf2 and its downstream signaling pathways.
1. Introduction
As a serious systemic inflammatory response syndrome, sepsis mainly includes excessive imbalance of inflammatory molecules, in- creased accumulation of venous leukocytes, microthrombus formation, and microvascular vasoconstriction, which in turn causes multiple organ failure [1]. Because of its high morbidity and mortality, sepsis has become the tenth main cause leading to death in the US intensive care unit [2]. Many septic patients exhibit cognitive impairment known as sepsis-associated encephalopathy (SAE) [3]. The incidence of SAE in patients with sepsis is as high as 70%, and it is an important cause of acute and long-term death in patients with sepsis [4].
The pathophysiological mechanism of SAE is very complicated, and some studies have shown that it is related to brain damage, among which blood–brain barrier (BBB) injury is one of the main reasons [3–5]. The BBB contains vascular endothelial cells showing specialized tight junctions and quite rare transcellular vesicular transport (trans- cytosis) [6]. It mainly relies on the tight junction complex between endothelial cells to achieve its barrier function. Its existence and in- tegrity maintenance are important for environmental stability in the brain tissue and the normal function of the central nervous system [7,8]. BBB destruction will cause vasogenic/cell-derived cerebral edema, tissue metabolic imbalance, excitotoxicity, infiltration of per- ipheral macrophages and lymphocytes, promotion of inflammatory re- sponse, and inhibition of nerve repair, thereby affecting the function of the corresponding brain parts and causing long-term cognitive im- pairment [9,10]. Specifically, SAE can profoundly affect the function of the central nervous system through an impaired BBB, including cogni- tive impairment and delirium [8,11,12].
Molecular hydrogen (H2), a selective antioxidant, can be effica- ciously administered to treat > 70 diseases [13]. Our previous research studies have shown that inhaling hydrogen or intaking hydrogen-rich water is efficacious in preventing organ damage (including liver, in- testine, lung, and brain) caused by sepsis [14–17]. We also found that in rodent models, 2% H2 inhalation can alleviate cognitive dysfunction caused by sepsis [17,18]. However, the specific mechanism of H2 pro- tection is still unknown. We discovered that 2% H2 inhalation treatment has an obvious protective effect on BBB destruction caused by sepsis and corresponding brain injury [17]. Therefore, BBB protection may be an important mechanism for H2 treatment of SAE.
Nuclear factor erythroid-2-related factor 2 (Nrf2), one kind of leu- cine zipper (bZIP) transcription factor, modulates the expression of antioxidant proteins, protects cells from oxygen injury, and is widely recognized as a multi-organ protection agent [19]. Our previous studies have found that Nrf2 and its downstream signaling pathway have a meaningful influence on sepsis resistance with hydrogen and partici- pate in and regulate the protective effects of hydrogen on brain, lung, liver, and bowel injury caused by sepsis [14,16,20,21]. Moreover, we confirmed in vitro that hydrogen can upregulate the expression of cadherin (VE-cadherin, an adhesion-linked protein), which has a good protective effect on vascular endothelial barrier damage caused by sepsis. In view of our previous researches, the target of the current study was to explore the function and underlying mechanisms of Nrf2 in the positive effect of H2 in the area of SAE caused by sepsis in wild-type (WT) and Nrf2 knock-out (Nrf2-/-) mice.
2. Materials and methods
2.1. Animals
We purchased female WT and Nrf2-/- C57BL/6J mice (both 8 weeks old and 20–25 g in weight) from the Better Biotechnology Company (Nanjing, China). We fed all animals standard food and sufficient water in cages (five mice per cage) under a controllable circumstance (tem- perature 20 °C–23 °C, humidity 55%–65%, and a 12-hourlight-dark cycle). The animal program was authorized by the Institutional Animal Care and Use Committee of Tianjin Medical University General Hospital (Tianjin, China). During this experiment, humane endpoints were used according to Tianjin Medical University General Hospital standard op- erating protocol. If a mouse was thought to be infected at the surgical site, as evidenced by wound splitting or weight loss (> 20%), cachexia, or difficultly eating, drinking, or walking, the mouse was euthanized by inhalation of carbon dioxide.
2.2. Cecal ligation and puncture (CLP) model
Sepsis was established by CLP as depicted in a previous study [17]. After a week of laboratory environment acclimatization, the mice were induced to an anesthetic state with 2% pentobarbital sodium solution (50 mg/kg) and sterilized in the prone position. Ophthalmic scissors were used to open the skin. Then, an incision (1 cm) in the midline of the abdomen was made to expose the cecum. The cecum was fastened with surgical sutures 1/4 distance to the end below the ileocecal flap and then punctured with a 20G sterilized needle. The intestinal contents were pushed out for approximately 0.3 mL; then, the cecum was re- turned to the abdominal cavity and the peritoneum and skin were closed by stitching. The sham-operated mice received just laparotomies without ligation and puncture. Some lidocaine cream (Cat#H20063466, Ziguang Beijing) was applied after the CLP or sham procedure to alleviate the ache. Moreover, a dose of antibiotic (Primaxin, 0.5 mg/mouse in 200 μL sterile saline) was subcutaneously injected at the end of surgery.
2.3. bEnd.3 cell line culture
Mouse brain microvascular endothelial cells (bEnd.3) were pur- chased from American Type Culture Collection (ATCC) and were cul- tured in Dulbecco’s modified Eagle’s medium (DMEM, Cat#D5796, Sigma-Aldrich, USA) with 10% fetal bovine serum (FBS, Cat#F8687, Sigma-Aldrich, USA) and 1% penicillin/streptomycin solution (Cat#V900929, Sigma-Aldrich, USA). The cells were grown in a humid atmosphere of 5% CO2 at 37 °C. When the cells reached 80%, sub- culture was conducted and the second and third generations were used in the test.
Fig. 1. Experimental design. (A) WT Female C57BL/6J mice (aging 6–8 weeks, weighing 20–25 g) were subjected to sham or CLP opera- tion. H2 or fresh air was inhaled for 60 min starting from 1 and 6 h after CLP or sham sur- gery, respectively. The brain tissue of different groups were obtained for tests 24 h after the sham or CLP procedure. The Morris water maze task was conducted from the 4th to 7th days after the sham or CLP procedure. Different groups of brain tissues were applied for all the examinations, except for 7-day survival rates, as described in Materials and methods. (B) The mouse bEnd.3 cells were incubated with Control + DMSO or ML385 medium, Control + H2 + DMSO or ML385 medium, LPS + DMSO or ML385 medium, and LPS + H2 + DMSO or ML385 medium. The cells and culture medium supernatant were collected for test on 24 h after incubation. CLP, cecal li- gation and puncture; LPS, lipopolysaccharide; ML385, a Nrf2 inhibitor.
2.4. Experimental design
Fig. 1A shows the experimental plan and process of animal research. WT and Nrf2-/- animals were randomly assigned into 4 × 2 groups (four groups for WT mice and four groups for Nrf2-/- mice) as follows: sham group, sham + H2 group, CLP group, and CLP + H2 group. The SAE model was established by the cecal ligation and puncture (CLP) pro- cedure in the CLP and CLP + H2 groups. The sham groups received the identical treatment with the exception of CLP. H2 was administered for 60 min to mice in the sham + H2 and CLP + H2 groups starting at 1 h and 6 h post-surgery. Mice in the sham and CLP groups just inhaled air. At 1 h and 6 h after the sham or CLP procedure, H2 concentration in the mouse brain of different groups (n = 4/group) was detected at 0, 10, 20, 30, 45, 60, 75, 90, 105, and 120 min after the start of inhaling and at 5, 15, 30, 45, and 60 min after the end of inhaling H2. On the time point of 24 h after the sham or CLP procedure, survival rates in each group (n = 20/group) were calculated for 7 days, and other mice were immolated and perfused for brain tissue (cortex) harvest. Moreover, different groups of mice (n = 10) given the Morris water maze (MWM) test from day 4 to day 10 after the sham or CLP operation. The cortex from four mice from each group was applied for brain water content determination and for Evans blue extravasation. The cortex from six mice from each group was used for phosphorylated (p)-β-catenin, β- catenin, VE-cadherin, and ZO-1 protein detection. A brain slice from three mice of each group was used for dextran quantitative analysis by immunofluorescence. The cortex from four mice per group was used for antioxidant enzyme [chloramphenicol acetyltransferase (CAT) and su- peroxide dismutase (SOD)], oxidative product [8-iso-PGF2α and mal-
ondialdehyde (MDA)], and inflammatory cytokine [tumor necrosis factor-alpha (TNF-α,) interleukin-6 (IL-6), IL-10, and HMGB1] de- termination by enzyme-linked immunosorbent assay (ELISA).
2.5. Hydrogen gas treatment
In light of our previous researches [16], the mice in the H2 treat- ment groups (sham + H2 group and CLP + H2 group) were put in a plastic square device with an entrance and an exit. H2 was given by a TF-1 gas flow meter (Yutaka Engineering Corp., Tokyo, Japan) and was mixed with air at a rate of 4 l/min. The H2 concentration in the con- tainer was constantly monitored by a gas sensor (HY-ALERTA Handheld Detector Model 500; H2 Scan, Valencia, CA) and was held at 2% from the beginning of the administration to the end. Carbon dioxide was taken away with Baralyme.
2.6. Preparation of hydrogen-rich medium
Hydrogen-rich medium was prepared as previously described [22]. Briefly, H2 (1 l/min) mixed with air (1 l/min) was dissolved in low glucose (5.6 mM of glucose) Dulbecco’s modified Eagle’s medium for 4 h in order to reach supersaturation (0.6 mM of hydrogen) under the pressure conditions of 0.4 MPa. H2 was produced by a TF-1 gas flow meter (Yutaka Engineering Corp., Tokyo, Japan). A special sealed alu- minum bag (no dead volume was left) was used to store the saturated HM under the atmospheric conditions at 4˚C. In order to ensure the saturated concentration of hydrogen, the medium was prepared freshly every week.
2.7. Survival rates
The survival rates of different groups were estimated within 7 days at the end of the operative treatment as described previously [15]. All experiments were performed twice.
2.8. Detection of H2 concentration
H2 concentration in the brain of mice was detected by a needle-type hydrogen sensor (Unisense A/S, Aarhus, Denmark) based on our pre- vious method [17]. In brief, the sensor was planted straight into the cerebrum of mice for the measurement of H2 concentration at 0, 10, 20, 30, 45, and 60 min after the start of inhaling H2 and at 5, 15, 30, 45, and 60 min after stopping H2 administration.
2.9. Morris water maze (MWM)
The MWM tests followed Tao, et al.’s study [23]. In brief, the mice in different groups were examined in the MWM via four trails per day for 7 (day 4 to day 10) days. Escape latency was recorded on each day. On the last day (day 10), the platform was removed, and the platform crossing times were measured. After each test, every animal was dried in a container under a heat lamp for 5 min and then placed in its home cage.
2.10. Detection of antioxidant enzymes activities
The cortex tissues were acquired to detect the activities of the CAT (Cat#ab238537, Abcam, Britain) and the SOD (Cat#7500-100-K, R&D Systems, Inc, USA) by commercial kits based on the instructions of the manufacturers.
2.11. Detection of oxidative products
The MDA (Cat#ab118970, Abcam, Britain) and 8-iso-PGF2α (Cat#ab133025, Abcam, Britain), both in the cortex, were determined by commercial kits in accordance with the instructions through a micro
plate reader (Cat#94089, Molecular Device, Sunnyvale, USA).
2.12. Determination of inflammatory cytokines
Commercial kits were applied for estimating the expressions of HMGB1 (Cat#BOS-14703, BOSK, Wuhan, China), IL-6 (Cat#RAB0311,
Sigma-Aldrich, USA), TNF-α (Cat#RAB0479, Sigma-Aldrich, USA), and
IL-10 (Cat#RAB0247, Sigma-Aldrich, USA) in the cortex by a micro- plate reader following the instructions with minor modifications at 24 h after CLP and sham procedures [15,19].
2.13. Dextran imaging studies to detect Blood-Brain barrier (BBB) permeability
BBB permeability was estimated by injecting dextran as mentioned in Yang, et al. [24]. In detail, at 24 h after the CLP or sham procedure, animals were induced to an anesthetic state by injection of pento- barbital sodium solution (50 mg/kg) into the spinal canal at a 2% concentration. Next, we injected 150 µL 10-kDa dextran tetra- methylrhodamine lysine fixable (4 mg/mL, Cat#D3312, Invitrogen, USA) into the tail vein of each mouse. Thirty minutes after injecting the 10-kDa dextran, mice were injected with a large amount of phosphate- buffered saline (PBS) and 4% paraformaldehyde from the ventricle and euthanized. Next, brain tissues were taken and fixated by 4% paraf- ormaldehyde (PFA) for 24 h. Then the brains were dehydrated with 20% and 30% sucrose PBS solution one day each. After dehydration, the brain tissues were frozen in TissueTek OCT (Sakura) at −80 °C. immunofluorescence staining was used to discover the infiltration ca- pacity of dextran in the frozen sections (10 µm thickness per slide) cut from mouse brain tissues. Tissue slides were immersed in 4% PFA for 30 min at stated temperature (25 °C), washed 3 times in PBS (5 min/ time), and covered by 0.5% Triton X-100 lasting 0.5 h. The slides were then shaken in PBS three times, blocked with 2% albumin in bovine serum, and incubated with isolectin B4 (1:200, Cat#I21411, Molecular Probes, San Francisco, CA, USA) for staining blood vessels. Dextran has the fluorescence for immunofluorescence imaging. The slides were detected using cellSens standard under a 40-fold objective lens of the fluorescence microscope, and photographs of the sections were taken. We analyzed the results via Image-pro Plus to measure the level of the dextran tracer-positive area.
2.14. Evans blue extravasation to detect BBB permeability
Evans blue (EB) dye was applied to assess BBB permeability as just described [17]. At 24 h after the procedure, animals were anesthetized and administered with EB (2% in saline, Cat#E2129, Sigma, USA), 3 mL/kg via the vessel located at the distal end of the tail, and it was allowed to spread for 2 h. Then, all the mice were injected with a large amount of saline from the ventricle and euthanized. The cortex of every sample was taken to record their weight, respectively; homogenized in formamide (1 mL); and incubated at 37 °C for 48 h. After centrifuga- tion, the optical density of the supernatant was calculated at OD 625 nm by a microplate reader. On the basis of a linear standard curve, a large quantity of EB (μg/g wet weight) was quantified and showed as relative amount.
2.15. Brain water content
The mice were induced to an anesthetic state and euthanized to retrieve the whole brain at 24 h after sham or CLP treatment. The brain water content was evaluated by a dry-wet method as we described before [17]. The brain tissues were instantly weighed as the wet weight and then dried in the oven at 100 °C for 24 h to acquire the dry weight. The brain water content (WC) was determined with following the for- mula: WC = [wet weight/dry weight] × 100%.
2.16. Western blot analysis
Western blot analysis was applied to confirm the expressions of phosphorylated (p)-β-catenin, β-catenin, VE-cadherin, and ZO-1 in mice cortex or bEnd.3 cells. The samples were separated from sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) by 4% to 20% double TIS polyacrylamide gel (Cat#M00655, Gen Script Biotech Corp., Nanjing, China) or standard XT 4% to 12% double TIS gel (Bio-Rad, USA), transmitted to a nitrocellulose membrane (Bio-Rad, USA), and then incubated in 5% skimmed milk powder and detected with the main antibodies to the following proteins: phosphorylated (p)- β-catenin (Ser33/37/Thr47, 1:1000, Cat#9561, Cell signaling, USA), β-catenin (1:1000, Cat#9562, Cell signaling, USA), VE-cadherin (1:1000,
Cat#ab205336, Abcam, Britain), ZO-1 (1:1000, Cat#ab216880, Abcam, Britain), and β-actin (1:2000, Cat#A5060, Sigma, USA). After rinsing at 4 °C overnight, goat anti-mouse antibody (1:4000, Cat#31430, Invitrogen, USA) or goat anti-rabbit antibody (1:4000, Cat#31466, Invitrogen, USA), was added and incubated at 37 °C for 1 h. The nitrocellulose membrane was washed in tris-buffered saline Tween (TBST) and dripped with enhanced chemiluminescence (ECL) liquid (Cat#34577, Invitrogen, USA), then scanned and photographed by the Bio-Rad image analysis system. The expression level of the target protein is reflected by the ratio of the integral optical density of the target line compared with the β-actin version. 100% of changes refers to control or sevoflurane levels for the purpose of comparison to ex- perimental states.
2.17. Statistical analysis
The survival rates were showed as percentage (%), the escape la- tency of MWM in different groups were shown as means ± standard deviation (SD), the numbers of the platform crossing time of MWM were presented as median with interquartile range and another aspect showed as means ± standard deviation (SD). There were 20 samples for each group in survival rates and 10 per group in behavioral researches, 4 per group in H2 concentration examination, brain water content de- tection, Evans Blue extravasation and ELISA studies, 6 per group in Western blot tests, 3 per group in Dextran imaging study. Long-rank (Mantel-Cox) Test was performed for analysis of the differentiation of survival rates among groups. Several totally same Two-way ANOVA was applied to determine the interplay between time and group factors (according to escape latency) between two groups in the MWM, Post- hoc Bonferroni test was applied in aims of contrasting the differentia- tion in escape latency between the two groups in every day of the MWM and the Mann-Whitney test was applied to estimate the differentiation of all groups on platform crossing times. All the data for the variables of survival rates and MWM (escape latency and platform crossing time) studies were complete in the process of the data analysis. Finally, the unpaired t-test (if the values were normal distribution) or Mann- Whitney test (if values were non-normal distribution) were employed for analysis of the difference between two groups in other biochemistry data. P < 0.05 was recognized to be statistically significant, and the significance testing was two-tailed. Statistical analysis was performed by the GraphPad Prism software (version 6.0) and SPSS statistic soft- ware (version 21.0).
3. Results
3.1. H2 gas concentration at different time points in the brain of WT and Nrf2-/- mice
At 1 h and 6 h after the sham or CLP surgery, the H2 concentration in the brain of WT and Nrf2-/- mice was determined. As shown in Fig. 2, brain H2 concentration from the sham + H2 group and the CLP + H2 group increased with time from 0 min to 60 min after inhalation of H2 in WT and Nrf2-/- mice (P < 0.05 vs. sham in WT mice at 10, 20, 30, 45, 60 min; P < 0.05 vs. sham in Nrf2-/- mice at 10, 20, 30, 45, 60 min). H2 concentration reached its peak at 45 min, and the highest level was held from 45 to 60 min after the start of H2 administration in the brain of WT and Nrf2-/- mice. After stopping the inhalation of H2, its concentration in the mouse brain from the sham + H2 group and the CLP + H2 group decreased quickly to the baseline (P < 0.05 vs. sham in WT mice at 65 and 75 min; P < 0.05 vs. sham in Nrf2-/- mice at 65 and 75 min). Nevertheless, no significant difference was obtained be- tween the CLP + H2 and the sham + H2 groups (P < 0.05 vs. sham + H2 in WT mice at all mark points; P > 0.05 vs. sham + H2 group at all time points in Nrf2-/- mice).
3.2. Survival rates of septic mice were increased by 2% H2 inhalation in WT mice, but not Nrf2-/- mice
Compared with the sham group, the 7-day survival rates of the CLP group in WT and Nrf2-/- mice decreased markedly (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). After 2% H2 inhala- tion, survival rates increased in the CLP + H2 group in WT mice (P < 0.05 vs. CLP in WT mice). But, this indicator of CLP and CLP + H2 group in Nrf2-/- mice did not differ from each other (P < 0.05 vs. CLP in Nrf2-/- mice) (Fig. 3).
3.3. Cognitive impairment induced by sepsis was mitigated by 2% H2 inhalation in WT mice, but not in Nrf2-/- mice
To assess the therapeutic effect of 2% H2 inhalation, we conducted a SAE mouse model induced by CLP and detected spatial learning and memory function by Morris water maze from day 4 to day 10 after CLP or sham operation. As demonstrated in Fig. 4, the escape latency of mice in the CLP group of WT and Nrf2-/- mice were obviously longer than that in sham group (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). H2 (2%) inhalation decreased escape latency in the CLP + H2 group compared with the CLP group just in WT mice (P < 0.05 vs. CLP in WT mice), absent in Nrf2-/- mice (P > 0.05 vs. CLP in Nrf2-/- mice). For the water maze trial, the times of target platform crossing decreased significantly in the CLP group compared with the sham group in both WT and Nrf2-/- mice (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). H2 treatment added to the times of platform crossing in the CLP + H2 group compared with the CLP group only in WT mice, not in Nrf2-/- mice (P < 0.05 vs. CLP in WT mice; P > 0.05 vs. CLP in Nrf2-/- mice).
3.4. The imbalance of the oxidative stress and inflammation condition caused by sepsis were reversed by 2% H2 inhalation in WT, but not in Nrf2-/- mice
The efficiency of antioxidant enzymes SOD and CAT decreased markedly, whereas the oxidative products MDA and 8-iso-PGF2α were improved dramatically after the CLP procedure in the WT and Nrf2-/- groups (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). H2 gas inhalation was able to enhance the expressions of SOD and CAT, yet lower the MDA and 8-iso-PGF2α production in WT septic mice (P < 0.05 vs. CLP in WT mice), but not in Nrf2-/- septic mice (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 5A-D).
Compared with the sham group, the level of TNF-α, IL-6, HMGB1, and IL-10 were upregulated significantly in the CLP group in both WT and Nrf2-/- mice (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). In WT mice, after inhalation of H2, the “early” pro-in- flammatory cytokines, such as TNF-α and IL-6, and the “late” pro-in- flammatory cytokine HMGB1 were decreased drastically in the CLP + H2 group compared with the CLP group; yet, the expression of anti-inflammatory cytokine IL-10 was enhanced compared with the CLP procedure alone (P < 0.05 vs. CLP in WT mice). Nevertheless, in Nrf2- deficient mice, there were no obvious differences between the CLP + H2 and CLP groups (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 5E- H).
4. The increased permeability of the BBB caused by sepsis could be prevented in WT mice, but not in Nrf2-/- mice.
First, we measured the brain water content of mice to reflect brain edema. As demonstrated in Fig. 6D, in WT and Nrf2-/- mice, the brain water content of the CLP group was strikingly enhanced compared with the sham group (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). After H2 treatment, the brain water content of WT mice in the CLP + H2 group was radically decreased (P < 0.05 vs. CLP in WT mice), but there was no evident difference in Nrf2-/- mice in the CLP + H2 and CLP groups (P > 0.05 vs. CLP in Nrf2-/- mice). These data indicate that H2 treatment can alleviate brain edema caused by the CLP procedure in WT, but not in Nrf2-/- mice.
Then, we used immunohistochemical staining of blood vessels (green) and 10 kDa dextran (red) to measure the level of extravascular dextran in the brain of mice to assess the permeability of the BBB. Compared with the sham group, CLP treatment enhanced the level of extravascular dextran in the brain tissue of WT and Nrf2-/- mice, but only in WT mice can H2 decrease extravascular dextran levels in brain tissue as a result of CLP treatment (Fig. 6A). Quantification of im- munohistochemistry images indicated that the levels of extravascular dextran in the brain tissue of the CLP group mice in WT and Nrf2-/- mice were surprisingly higher than those of the sham group (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). In WT mice, compared with the CLP group, the level of extravascular dextran in the brain tissue of the CLP + H2 mice was dramatically decreased (P < 0.05 vs. CLP in WT mice), but in Nrf2-/- mice, there were no obvious differences between the CLP + H2 and CLP groups (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 6B). We also tested the permeability of the BBB by the Evans blue extravasation. After quantifying it, it can be seen that the Evans blue dye of the CLP group of WT and Nrf2-/- mice leaked more than the sham group (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice). Inhalation of H2 decreased the leakage of Evans blue dye in WT septic mice (P < 0.05 vs. CLP in WT mice), but not in Nrf2-/- septic mice (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 6C).
4.1. Levels of BBB-associated proteins changed differently after 2% H2 inhalation or H2-riched medium treatment in septic mice and cell models
Previous studies have indicated that the permeability of the BBB is related to changes in cell junction proteins in endothelial cells [17,18]. Therefore, we compared the changes of β-catenin, phosphorylated (p)-β-catenin, adhesion-linked protein VE-cadherin, and related tight
junction protein ZO-1 in the cerebral cortex of mice or in bEnd.3 cells in each group. In mice research, quantitative western blot analysis in- dicated that CLP enhanced the expression of β-catenin, and CLP also enhanced the expression of P-β-catenin in the cerebral cortex of mice (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/- mice), but H2 lowered β-catenin and P-β-catenin in the cerebral cortex of only WT septic mice (P < 0.05 vs. CLP in WT mice), and had no significant effect in Nrf2-/- septic mice (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 7A, B). Compared with the sham group, VE-cadherin and ZO-1 were dras- tically decreased in the cerebral cortex of WT CLP and Nrf2-/- CLP groups (P < 0.05 vs. sham in WT mice; P < 0.05 vs. sham in Nrf2-/-mice). After H2 treatment, VE-cadherin and ZO-1 were markedly en- hanced in the cerebral cortex of WT mice in the CLP + H2 group (P < 0.05 vs. CLP in WT mice), but in Nrf2-/- mice in the CLP + H2 group, there was no noticeable difference in VE-cadherin and ZO-1 of the mouse cerebral cortex compared with its CLP group (P > 0.05 vs. CLP in Nrf2-/- mice) (Fig. 7C-F).
In bEnd.3 cells research, compared with control + DMSO group, the ratio of p-β-catenin / β-catenin was increased significantly in LPS + DMSO group (P < 0.05 vs. Control + DMSO), hydrogen-rich medium treatment could decrease the p-β-catenin / β-catenin ratio obviously in LPS + H2 + DMSO group compared with LPS + DMSO group (P < 0.05 vs. LPS + DMSO). Moreover, compared with Control + ML385 group, LPS treatment could reduce the p-β-catenin/ β-catenin ratio markedly (P < 0.05 vs. Control + ML385), but there were no big difference between LPS + H2 + ML385 group and LPS + ML385 group (P > 0.05 vs. LPS + ML385) (Fig. 8A-B). Compared with Control + DMSO group, LPS treatment could decrease the levels of VE-cadherin and ZO-1 in LPS + DMSO group (P < 0.05 vs. Control + DMSO). Moreover, hydrogen-rich medium treatment could increase the levels of VE-cadherin and ZO-1 compared with LPS + DMSO group (P < 0.05 vs. LPS + DMSO). Compared with Control + ML385 group, there were less VE-cadherin and ZO-1 ex- pressions in LPS + ML385 group (P < 0.05 vs. Control + ML385). However, there were no statistical difference in VE-cadherin and ZO-1 expressions between LPS + H2 + ML385 and LPS + ML385 group (P > 0.05 vs. LPS + ML385) (Fig. 8C-F).
5. Discussion
Sepsis is a severe systemic inflammatory response syndrome in- duced by a variety of infectious factors [1]. It is believed that sepsis may lead to impaired BBB, which in turn causes brain damage, manifested as cognitive impairment and behavioral defects, namely SAE [4,25]. Our previous researches have shown that H2 could predominantly prevent sepsis, and SAE and can alleviate cognitive impairment caused by sepsis [17,18]. The CLP model is a widely recognized and clinically relevant model for sepsis research [26]. In this research, we not only showed that H2 has a meaningful protective effect on severe sepsis and SAE- induced BBB disruption, but also showed the important function of Nrf2 and its downstream molecules in these effects.
In this research, for the purpose of assessing the overall effect of H2 treatment in septic mice, we administered inhaled 2% H2 for 60 min at 1 h and 6 h after the procedure in the sham group and the CLP group, and recorded the 7-day survival rates of mice. Inhalation of hydrogen markedly improved the 7-day survival rates of mice in the CLP + H2 group compared with the CLP group in WT mice, but not in Nrf2-/- mice. After mouse inhalation of H2 for 60 min at 1 h and 6 h after surgery, we also measured the H2 concentration and changes in mouse brain tissue. The results showed that the H2 concentration in the brain tissue increased with increasing inhalation duration and peaked at 45 min, indicating that hydrogen can enter the brain.
Excessive release of inflammatory factors and oxidative stress are important factors causing brain damage [27]. To evaluate the effect of H2 on SAE caused by severe sepsis, we tested the expressions of in- flammatory factors, involving the pro-inflammatory factor TNF- ɑ, IL-6 (two “early” pro-inflammatory factors in the pathogenesis of sepsis) and HMGB1 (a “late” pro-inflammatory factor), and the anti-inflammatory factor IL-10. At the same time, we examined the expressions of some oxidative stress factors, such as oxidation products (MDA, 8-iso-PGF2ɑ) and antioxidant enzymes SOD and CAT. Oxidative stress is caused by insufficient activity of the endogenous antioxidant defense system against reactive oxygen species (ROS) [28], namely the inflammatory response and oxidative stress imbalance. Excessive release of ROS ac- tivates lipid peroxidation, leading to rupture of cell membranes and mitochondrial membranes, and eventually leading to apoptosis and necrosis [29]; whereas decreased levels of antioxidants and antioxidant enzymes can also lead to excessive ROS production [30]. The con- sequences indicated that inhalation of H2 decreased the expressions of TNF-ɑ, IL-6, HMGB1, MDA, and 8-iso-PGF2ɑ in the cortex and hippo- campus of mice compared with CLP in WT mice, but increased IL-10, SOD, and CAT expression. However, in Nrf2-/- mice, these indicators did not differ between the CLP and CLP + H2 groups. It was demon- strated that H2 can reverse the imbalance of inflammatory state and oxidative stress in WT septic mice, but there was no similar effect in Nrf2-/- mice.
Studies have shown that BBB damage is one of the main causes of SAE [5]. BBB destruction can cause vasogenic/cell-derived brain edema, tissue metabolic imbalance, excitotoxicity, infiltration of sur- rounding macrophages and lymphocytes, promotion of inflammatory response, and inhibition of nerve repair, thereby affecting the function of the corresponding brain, causing long-term cognitive dysfunction [8,25]. For cognitive assessment, inhaled hydrogen reduced escape la- tency in WT septic mice and increased the times of platform crossing; but, H2 did not correspondingly reduce escape latency and increase platform crossing times in Nrf2-/- septic mice. In addition, we measured the brain water content to reflect the brain damage and we also em- ployed the dextran tracer injection and Evans blue extravasation to assess the effects of sepsis on BBB formation and function. Brain water content, extravascular dextran, and extravascular Evans blue dye were positively correlated with BBB damage, as these three indicators were higher than those of the CLP group in both WT and Nrf2-/- mice. This is consistent with the results of inflammatory factors and oxidative stress indicators. H2 gas therapy can reduce the three upregulating indexes of brain tissue in WT mice due to CLP treatment, but cannot reduce same indicators in Nrf2-/- mice. It was shown that in WT mice, H2 can alle- viate SAE by protecting the BBB, but not in Nrf2-/- mice.
We have conducted further research on how H2 protects the BBB. The BBB is formed by the brain microvascular endothelial cells, peri- vascular astrocytes, pericytes and neurons, which comprise a neuro- vascular unit that provides selective brain permeability to various substances and creates a tightly regulated microenvironment [31]. The permeability of BBB is associated with changes in cell junction proteins, a tight junction of proteins between endothelial cells [25]. We found that the expressions of the adherens junction (AJ) protein VE-cadherin and tight junction (TJ) protein ZO-1 in the cortex of WT and Nrf2-de- ficient mice decreased after the CLP operation. Compared with the CLP group, H2 gas inhalation could increase both VE-cadherin and ZO-1 levels in the brain of WT mice, but not Nrf2-/- mice. In bEnd.3 cells (kinds of brain microvascular endothelial cells) experiment, the ex- pressions of VE-cadherin and ZO-1 were decreased with LPS treatment in FBS-free DMED medium which added DMSO or ML385 (Nrf2 in- hibitor). Compared with LPS group, the levels of VE-cadherin and ZO-1 increased in LPS + H2 + DMSO group, but not in LPS + H2 + ML385 group in bEnd.3 cells. β-catenin is a multifunctional protein which is found in cell membranes, cytosol, and nucleus, with different roles in different locations [32]. Evidences had shown that β-catenin may be a key regulator for maintaining the BBB formation and function [33,34]. In certain conditions, β-catenin can be degraded through phosphor- ylating at the Ser33/37/Thr41site by GSK-3β. The β-catenin degrada- tion may break the junctional stability of cerebrovascular endothelial cells and injure the BBB [35,36]. Interestingly, in WT mice, CLP treatment enhanced the level of p-β-catenin but decreased the level of β-catenin; 2% H2 gas inhalation could decrease the p-β-catenin level but increase the β-catenin level. However, there was no significant difference between the CLP and CLP + H2 group in Nrf2-/- mice. In addition, the bEnd.3 cells experiment gave the same results as what the animal research did. The current study showed that the levels of p-β- catenin at the Ser33/37/Thr41 site increased in the cortex of CLP mice. Since the p-β-catenin (Ser33/37/Thr41) is expressed only in cytosol, it suggests that sepsis may specifically decrease the cytosol β-catenin level. Further studies should be conducted on the expressions of β-ca- tenin and p-β-catenin in three different locations. In summary, these data suggest that CLP can damage tight junction proteins and adhesion- associated proteins, resulting in increased permeability of the BBB, causing its destruction, and leading to cognitive impairment; whereas, H2 can reverse these damages in WT mice, but not in Nrf2-deficient mice.
Among our results, there is a common conclusion that H2 did not have any protective effects on the septic brain of in Nrf2-deficient mice or on the LPS treated bEnd.3 cells cultured with medium contained Nrf2 inhibitor (ML385). Nrf2 is a leucine zipper (bZIP) transcription factor that regulates the level of antioxidant proteins and defends cells from oxidative damage; as such Nrf2 is widely recognized as a multi-organ protection agent [37]. Under oxidative stress or other stimuli, Nrf2 can be triggered and transferred from the cytoplasm to the nucleus, and binds to the ARE gene to control the activation and expression of an- tioxidant including SOD and CAT, and Phase II genes involving heme oxygenase-1 (HO-1), glutathione S-transferases (GSTs), and NAD(P)H quinine oxidoreductase [15,17]. Our previous studies have found that Nrf2 and its downstream signaling pathway have a meaningful influ- ence on sepsis resistance with hydrogen, and participate in and regulate the protective effects of hydrogen on brain damage induced by sepsis [15,19,38]. In this study, it was demonstrated that there was no sig- nificant brain protective effect on Nrf2-deficient septic mice with H2 treatment. It also proves that Nrf2 and its downstream signaling path- ways participate in and regulate the positive effect of hydrogen on sepsis and SAE. The limitation is that the current experiment did not discuss the downstream signaling pathway of Nrf2, and subsequent studies should be conducted.
6. Conclusion
In summary, we found that inhalation of 2% H2 can protect the BBB by decreasing its permeability, thereby reducing SAE and improving cognitive function, which is mediated through Nrf2 and its downstream signaling pathways. Therefore, we speculate that molecular hydrogen may be a promising method to alleviate SAE.