Hyperoxygenation revitalizes Alzheimer’s disease pathology through the upregulation of neurotrophic factors

Abstract Alzheimer’s disease (AD) is a neurodegenerative disease characterized by Aβ‐induced pathology and progressive cognitive decline. The incidence of AD is growing globally, yet a prompt and effective remedy is not available. Aging is the greatest risk factor for AD. Brain aging proceeds with reduced vascularization, which can cause low oxygen (O2) availability. Accordingly, the question may be raised whether O2 availability in the brain affects AD pathology. We found that Tg‐APP/PS1 mice treated with 100% O2 at increased atmospheric pressure in a chamber exhibited markedly reduced Aβ accumulation and hippocampal neuritic atrophy, increased hippocampal neurogenesis, and profoundly improved the cognitive deficits on the multiple behavioral test paradigms. Hyperoxygenation treatment increased the expression of BDNF, NT3, and NT4/5 through the upregulation of MeCP2/p‐CREB activity in HT22 cells in vitro and in the hippocampus of mice. In contrast, siRNA‐mediated inhibition of MeCP2 or TrkB neurotrophin receptors in the hippocampal subregion, which suppresses neurotrophin expression and neurotrophin action, respectively, blocked the therapeutic effects of hyperoxygenation on the cognitive impairments of Tg‐APP/PS1 mice. Our results highlight the importance of the O2‐related mechanisms in AD pathology, which can be revitalized by hyperoxygenation treatment, and the therapeutic potential of hyperoxygenation for AD.

Animals exposed to hypobaric hypoxia show dendritic atrophy in the hippocampus and cognitive impairment (Titus et al., 2007). Concerning that the incidence rate of AD increases with age (Wyss-Coray, 2016), the question may be raised whether reduced O 2 availability in aging brains triggers or aggravates AD pathology. Considering the close relationship between hypoxic conditions and aging-related changes in the brain, hyperoxygenation therapy might be considered to antagonize hypoxic states and aging-related changes in the brain of AD patients. However, hyperoxygenation treatment has rarely been studied in research on aging and AD, partly due to the concern that hyperoxygenation may result in harmful oxidative stress (Oter et al., 2005).
Nonetheless, recent studies have reported that hyperoxygenation with 100% O 2 at 2-3 atmospheres absolute (ATA) is beneficial for treating various brain disorders (Yan, Liang, & Cheng, 2015). Hyperoxygenation treatment improved cognitive sequelae in patients with carbon monoxide poisoning (Weaver et al., 2002) and ameliorated traumatic brain injury (Huang & Obenaus, 2011) and ischemic brain injury (Baynosa et al., 2013). Furthermore, hyperoxygenation treatment (100% O 2 , 2.5 ATA), with 30 intermittent exposures, improved cognitive function in elderly individuals with cognitive deficits (Jacobs, Winter, Alvis, & Small, 1969). Hyperoxygenation treatment (100% O 2 , 2.0 ATA) of poststroke patients for daily 90 min for 40-60 days improved memory impairments on the tasks in the immediate and delayed verbal and nonverbal recall conditions (Boussi-Gross et al., 2015). Thus, the results of those studies support that hyperoxygenation could produce certain therapeutic effects on neuronal function. However, the mechanism afforded by hyperoxygenation of the brain is poorly understood.
In the present study, we investigated whether hyperoxygenation treatment changes the Aβ-induced pathology and cognitive impairment seen in Tg-APP/PS1 mice. Our analyses demonstrated that hyperoxygenation treatment improved the Aβ pathology and cognitive deficits of Tg-APP/PS1 mice through the induction of MeCP2mediated neurotrophin expression.

| Hyperoxygenation suppressed the expression of hypoxia-related markers in hippocampal neurons
HT22 hippocampal cells treated with Aβ42 showed increased expression of hypoxia-related genes, including hypoxia-inducible factor-1α (Hif-1α), Vegf-a, Hmox1, and Pdk1. In contrast, treatment of HT22 cells with perfluorodecalin (PFD), a synthetic biomaterial that noncovalently dissolves large amounts of molecular oxygen (O 2 ) (Lowe, Davey, & Power, 1998), reversed the Aβ42-induced increase in these markers (Supporting information Figure S1a-d). We examined whether similar changes could occur in the AD-like brain. Tg-APP/PS1 mice, which express high levels of Aβ in the brain at 8 months of age (Kim et al., 2012), showed increased expression of hypoxia-related markers, including Hif-1α, Hmox1, and Pdk2, in the hippocampus, whereas Tg-APP/PS1 mice treated with hyperoxygenation (HO 2 ; 100% O 2 , 2 ATA) for 1 hr daily for 28 days had reduced expression of these factors (Supporting information Figure S1e-f).

| Hyperoxygenation upregulated the expression of neurotrophins in the hippocampus
HT22 cells treated with Aβ42 showed reduced expression of Bdnf, Nt3, and Nt4/5, and Trkb, whereas PFD treatment reversed the decrease in the expression of these factors ( Figure 1a). The expression of Bdnf, Nt3, and Nt4/5 was also increased in the hippocampus of wild-type mice that were exposed to HO 2 (100% O 2 , 2 ATA) for 1 hr daily for more than 7 days (Figure 1b), suggesting that the expression of neurotrophic factors is regulated by hyperoxygenation in vitro and in vivo.
Tg-APP/PS1 mice had reduced levels of the neurogenesis markers DCX and Ki-67 in the dentate gyrus, whereas HO 2 treatment reversed the reduction of those markers (Supporting information Figure S2f-i).

| HO 2 treatment partially suppressed the ROS levels accumulated in the brain of Tg-APP/PS1 mice
Oxidative stress is elevated by amyloidopathy in AD (Dumont & Beal, 2001). We examined whether or not HO 2 treatment in mice increased ROS levels in the brain. Analyses of ROS levels by staining with the superoxide sensitive dye dihydroethidium (DHE), immunological staining for 4-hydroxynonenal (HNE), a lipid peroxidation marker, or the biochemical assessment of the level of malondialdehyde (MDA), another lipid peroxidation marker, indicated that Tg-APP/PS1 mice had increased ROS levels in the hippocampus, whereas those increases were significantly suppressed after HO 2 treatment (Supporting information Figure S3a-h). The expression levels of Cox1, eNos, nNos, Ho-1, and the NADPH oxidase subunits in the hippocampus tended to increase in Tg-APP/PS1 mice, whereas HO 2 treatment suppressed those increases. Among antioxidant genes, the level of Prx3 increased after HO 2 treatment (Supporting information Figure S3i-k). Overall, these results suggest that hyperoxygenation partially reduced, rather than increased, oxidative stress levels in the hippocampus of Tg-APP/PS1 mice.

| HO 2 treatment improved the cognitive deficits of Tg-APP/PS1 mice
Next, we examined whether HO 2 treatment changed behaviors. To address this, Tg-APP/PS1 mice treated with HO 2 (100% O 2 , 2 ATA), starting from 7 months of age, for 1 hr daily for 2 weeks were placed in the indicated behavioral tests while continuing HO 2 treatment on the given schedule ( Figure 5a). Tg-APP/PS1 mice at 7.5 months of age display severe cognitive deficits (Kim et al., 2012).
During the training phase of the water maze test, Tg-APP/PS1 mice given HO 2 exhibited a markedly reduced latency to find the hidden platform compared to Tg-APP/PS1 control mice (Figure 5b) However, Tg-APP/PS1 mice exposed to moderate hyperoxygenation (mHO 2 ; 42% O 2 , 2 ATA), starting from 7 months of age, for 1 hr daily for 2 weeks or more, in a manner similar to the treatment with 100% O 2 at 2 ATA (Supporting information Figure S6a), showed the same cognitive deficits in the water maze test (Supporting information Figure S6b-f) and novel object recognition test (Supporting information Figure S6g-i), as the Tg-APP/PS1 control mice. In the passive avoidance test, Tg-APP/PS1 mice given mHO 2 treatment exhibited an increased latency to enter the dark chamber 24 hr after foot shock and a tendency toward increased latency 72 and 120 hr after shock (Supporting information Figure S6j). Mice were treated with HO 2 (100% O 2 , 2 ATA) for 1 hr daily for 1, 7, or 14 days (f). MeCP2 levels (g) and ChIP assay data showing MeCP2 (h) and p-CREB (i) binding levels to the promoter of Bdnf1, Bdnf3, Bdnf4, Nt3, Nt4/5 in the hippocampus of mice treated with HO 2 for indicated days. Data are presented as mean ± SEM. *p < 0.05; **p < 0.01 (one-way ANOVA followed by Newman-Keuls post hoc test and two-way ANOVA followed by Bonferroni post hoc test)

|
independently inhibition of Mecp2, within the hippocampus could block the behavioral effects of HO 2 .
Tg-APP/PS1 mice treated with HO 2 (100% O 2 , 2 ATA) from 7 months of age for 14 days, followed by injection of siRNA-MeCP2 or siRNA-TrkB into the CA3 subregion (Figure 6f), did not distinguish F I G U R E 5 HO 2 improved the cognitive deficits of Tg-APP/PS1 mice. (a) Experimental design. Mice were treated with HO 2 (100% O 2 , 2 ATA) from 7 months of age for 1 hr daily for 28 days. Behavioral tests were performed in this order: water maze test (WM), novel object recognition test (NOR), and passive avoidance test (PA). The color codes in the legends are applicable to all parts of Figure 5. (e and f) The latency to finding the platform (e) and swim speed (f) during the visual platform trial of WT-CON, WT-HO 2 , Tg-CON, and Tg-HO 2 . (g-i) The effects of HO 2 on novel object recognition memory. Time spent exploring between the two identical objects during the familiarization (g), and between a new and an old object 2 hr after familiarization (h, NOR-2h) and 24 hr after familiarization (i, NOR-24h). (j and k) The effects of HO 2 on fear memory in the PA. The latency to entering the dark chamber at the preshock, and 24 and 72 hr after shock (j), and the freezing time 24 hr after shock (k). Data are presented as mean ± SEM. WT-CON, n = 7-18; WT-HO 2 , n = 8-18; Tg-CON, n = 7-12; Tg-HO 2 , n = 8-14 per each group. *p < 0.05; **p < 0.01, difference between indicated groups; # p < 0.05; ## p < 0.01, difference between Tg-CON and Tg-HO 2 (two-way ANOVA followed by Bonferroni post hoc test and two-way repeated-measures ANOVA followed by Bonferroni post hoc test) between novel and familiar objects in the novel object recognition (NOR) test (Figure 6g,h). In the subsequent novel location recognition (NLR) test, Tg-APP/PS1 mice injected with siRNA-TrkB or siRNA-MeCP2 also did not preferentially explore the displaced object relative to the non-moved object, despite the fact that they received HO 2 treatment (Figure 6i). The total interaction time with the objects in these tests was comparable among the test groups.

| DISCUSSION
We demonstrated that HO 2 treatment suppressed the Aβ accumulation ( Figure 2) and hippocampal neuritic atrophy (Supporting information Figure S2) and improved the cognitive deficits in Tg-APP/PS1 mice (Figures 5 and 6). These results support the notion that O 2 availability is critical for AD-like pathology. A postmortem study indicated that 77% of vascular dementia patients had AD pathology (Barker et al., 2002). Higher cerebral blood flow velocity, measured by a transcranial Doppler flowmetry, was correlated with a lower prevalence of cognitive decline and dementia (Ruitenberg et al., 2005). Patients diagnosed with amnestic mild cognitive impairment (MCI) aged 65 years and above advance to AD at a rate of 15% per year (Davatzikos, Bhatt, Shaw, Batmanghelich, & Trojanowski, 2011).
A recent study reported that subjects with MCI showed global brain hypoperfusion and low oxygen metabolism in the brain (Liu et al., 2014). Together, these studies raise the possibility that the cellular and molecular mechanisms, which might be hypoxia-related mechanisms as inferred from the results reversed by hyperoxygenation, are involved in AD-related pathology. Further studies to elaborate the physiological significance of this possibility are highly encouraged. with HO 2 (100% O 2 , 2 ATA) from 7 months of age for 1 hr daily for 14 days. The siRNA-MeCP2 or siRNA-TrkB was injected into the CA3 subregion of the hippocampus on day 15, and behavioral tests were performed in all mice on day 17. WT-CON, wild-type control; Tg-CON, Tg-APP/PS1 mice; Tg-siCON, Tg-siMeCP2; and Tg-siTrkB, Tg-APP/PS1 mice treated with HO 2 followed by injection with siRNA-control, siRNA-MeCP2, or siRNA-TrkB, respectively. (g-i) Time spent exploring the two identical objects during the familiarization (g), between a new and an old object 2 hr after familiarization (h, NOR), and between a displaced object and an old object from the previous test (i, NLR). The black circle is a spatial marker posted outside the open field. Blue shaded, mouse groups treated with HO 2 . Data are presented as mean ± SEM. WT-CON +siCON, n = 13; Tg-CON +siCON, n = 8; Tg-HO 2 + siCON, n = 7; Tg-HO 2 + siTrkB, n = 7; Tg-HO 2 + siMeCP2, n = 7 per each group. *p < 0.05; **p < 0.01 (Student's t test and two-way ANOVA followed by Bonferroni post hoc test) did not (Supporting information Figure S6). HO 2 treatment for more than 7 days in wild-type mice was required for the increase in MeCP2 expression (Figure 4f-i). These results suggest that the repeated and sufficient HO 2 stimulation is required for the therapeutic effects of HO 2 . In normal physiological conditions, normal O 2 concentration in arterial blood in humans is 9.5%, whereas the partial O 2 concentration in the brain drops to 3.4% (McKeown, 2014).
HO 2 treatment with 100% O 2 at 2.0 ATA increases the O 2 tension in the brain tissues by 7-10-fold compared to the pO 2 normally achieved with room air at 1.0 ATA (Demchenko et al., 2005). In the presence of high O 2 , the sulfur-containing amino acids cysteine and methionine can be reversibly oxidized (Sanchez, Riddle, Woo, & Momand, 2008). Cysteine and methionine oxidations are known to play a role in antioxidant defense, redox sensing, and regulation, and changes in protein activity (Kim et al., 2014). Therefore, it is possible that HO 2 -initiated oxygenation of unidentified proteins might play a role in HO 2 -induced changes.
Hyperoxygenation is known to cause harmful oxidative stress (Oter et al., 2005). However, it was also suggested that certain therapeutic effects of hyperoxygenation act through oxidative stress brought about by hyperoxia (Thom, 2009). A recent study with a rat model of stroke reported the protective effect of repeated HO 2 treatment (100% O 2, 1.3 ATA, 45 min/once, 40 sessions) was blocked by inhibition of ROS (Hu et al., 2014). It remains to be explored whether or not the initial and/or mildly increased oxidative stress during each HO 2 exposure is required for the ultimately therapeutic effects of HO 2 . The ROS levels, as measured by DHE staining, anti-HNE staining, and MDA assay, in the brain of Tg-APP/PS1 mice treated with hyperoxygenation tended to be higher than those of age-matched wild-type control (Supporting information Figure S3).
Thus, our results indicated that the repeated HO 2 treatment conditions partially suppressed the oxidative stress in the brain of Tg-APP/PS1 mice (Supporting information Figure S3).
The results of present study support that HO 2 treatment procedures can be developed for people with AD. However, HO 2 treatment at certain conditions cloud produce some complications such as middle ear, lung, or sinus barotrauma when hyperbaric pressure is used, and neurological, retinal, and pulmonary oxygen toxicity caused by high oxygen level (Grim, Gottlieb, Boddie, & Batson, 1990;Leung & Lam, 2018). Pulmonary toxicity occurs when exposed to HO 2 at over 2.5 ATA for 6 hr or 1.5 ATA for over 12 hr at one session (Clark et al., 1999). HO 2 treatment at 2.0 ATA or higher conditions can develop a risk of seizure (Jain, Torbati, Tao, & Ni, 1999), although its incidence is low (Yildiz, Aktas, Cimsit, Ay, & Toğrol, 2004). Therefore, possible side effects and complications of HO 2 treatment should be studied in conjunction with specific conditions of AD or other disease states. It also requires to investigate possible side effects of prolonged treatments with HO 2 .
Neurotrophic factors play a role in synaptic plasticity, neuronal circuit activity, and circuit formation (Vicario-Abejón et al., 2002) and have neuroprotective roles also in hypoxic-ischemic brain injury (Huang et al., 2017). BDNF expression can be regulated by multiple factors, including CREB (Koo et al., 2015) and MeCP2 (Chang et al., 2006). MeCP2 overexpression in mouse cortical neurons increased the level of BDNF (Klein et al., 2007), and exogenous BDNF revived defective synaptic transmission in MeCP2 null mice (Kline, Ogier, Kunze, & Katz, 2010). In our study, HO 2 treatment increased the expression of BDNF, NT-3, and NT-4/5 through the upregulation of MeCP2/p-CREB activity (Figures 2 and 3), whereas siRNA-mediated knockdown of MeCP2 or TrkB in the hippocampus blocked the therapeutic effects of HO 2 on the cognitive deficits of Tg-APP/PS1 mice ( Figure 6). Thus, our results suggest that HO 2 -induced MeCP2 upregulation triggers not only BDNF, but also NT-3, and NT-4/5 in the hippocampus. Moreover, increased these neurotrophic factors appear to act through TrkB receptors in the hippocampus. Our results highlight the importance of the mechanisms revived by hyperoxygenation in AD pathology, which have not been carefully explored in AD research.

| EXPERIMENTAL PROCEDURES
The detailed procedures and information are described in the Supporting information Appendix S1 and Table S1.

| Animals
Tg-APPswe/PS1dE9 (Tg-APP/PS1) mice, which show plaque deposition from 6.5 months of age and severe cognitive deficits at 7-7.5 months of age (Kim et al., 2012), were used. All animals were handled in accordance with the animal care guidelines of the Ewha Womans University (IACUC 16-019).

| Hyperoxygenation treatment in mice
Mice were treated with hyperoxygenation (HO 2 ) using a hyperbaric chamber as described previously (Kim et al., 2014). Mice were exposed to 100% oxygen daily for 60 min at 2.0 ATA in a hyperbaric chamber for the indicated number of days. For moderate hyperoxygenation (mHO 2 ), mice were treated as above, but with atmospheric air.
Immunoprecipitated DNA was used for quantitative real-time PCR, as described above.

| Behavioral tests
Behavioral tests were carried out as described previously (Choi et al., 2015). Behavioral performance was recorded with a video-tracking system (SMART; Panlab S.I., Barcelona, Spain) and a webcam recording system (HD Webcam C210, Logitech, Newark, CA, USA).

| Water maze test
The water maze test was performed as described previously (Kim et al, 2012). The training was performed in a circular tank pool filled with opaque water (made using Sargent ® White Art Tempera Paint) twice a day for 5 days. On day 6, mice were given the probe trial test. At the end of the experiment, mice were placed on the visible platform test to control for possible locomotor and visual deficits.

| Novel object recognition tests
The standard and modified novel objective recognition tests were performed as described previously . A subject mouse was presented to two identical objects (object A) for 10 min. Two hours after the familiarization sessions, one familiar object was replaced with a new object (object B). The time spent exploring each object was recorded for 10 min. Twenty-four hours, the subject mouse was presented to a familiar object (object A) and a third, new object (object C), and the time spent with each object was recorded for 10 min and analyzed in a blind manner with two researchers.

| Passive avoidance test
The passive avoidance test has been described previously (Kim et al, 2012). The test apparatus consisted of a lighted chamber (1,500 lx) and a dark chamber equipped with a metal grid floor. On the first day, subject mice were individually placed in the lighted chamber with the door opened and allowed to explore freely the equipment for 5 min. On the second day, mice were given with two foot shocks. On the test day, the latency to entering the dark chamber was recorded. The total freezing time during the testing period was manually analyzed.

| Statistical analysis
Two-sample comparisons were carried out using Student's t test, whereas multiple comparisons were performed using one-way ANOVA followed by the Newman-Keuls post hoc test or two-way ANOVA or two-way repeated-measures ANOVA followed by the Bonferroni post hoc test. All data are presented as mean ± SEM, and statistical significance was accepted at the 5% level.