Abstract
INTRODUCTION
Alzheimer’s disease (AD) is a devastating neurodegenerative disease characterized by progressive dementia with onset in middle age or late life. The primary pathological characteristics of AD are neurofibrillary tangles resulting from tau protein hyperphosphorylation and senile plaques formed from extracellular amyloid-β (Aβ) deposits [1–3]. Though the incidence of AD is rising steadily, particularly in countries with aging populations, an effective therapy has yet to be developed [4, 5].
The underlying mechanism driving AD pathology remains unclear [5]. The oxidative stress hypothesis posits that free radical damage due to excess reactive oxygen species (ROS) or reactive nitrogen species production may be the etiological basis of AD [6]. However, the fact that not only hyperoxidation, but also dysoxidation, is disruptive to the cellular redox state has been largely neglected by researchers. Watson [7] hypothesized that both diabetes and AD might be due to cellular oxidant production failure and an insufficient oxidative redox potential of the endoplasmic reticulum resulting in a failure to oxidize the free sulphydryl groups of cysteine residues into disulfide bonds, which are essential for the physiological activity of proteins.
Ample evidence supports the notion that hypoxia may contribute to impaired cognitive function and increase the risk of dementia [8, 9]. Neural activity relies greatly on aerobic metabolism and the brain consumes a large amount of oxygen in proportion to its volume to meet its high-energy needs. Consequently, brain neurons are highly vulnerable to hypoxia-induced damage. It has been reported that patients with AD have reduced cerebral blood flow due to a decreased vascular density [10]. Moreover, cerebral blood supply and the ability to recover from ischemic brain damage decreases with age [10, 11], resulting in a higher risk of neurological disease in the elderly. Because increased oxygen inhalation can raise the arterial partial pressure of oxygen and oxygen saturation, we are interested in exploring whether supplemental oxygen therapy could relieve AD symptoms.
Several AD-associated changes in the lens may underlie AD-associated visual symptoms [12]; the observation of such changes supports the hypothesis that AD pathology in the brain and lens are linked. The lens contains highly specialized antioxidant defense and repair systems that function to minimize oxidative damage. Compromise of these symptoms may leave patients’ eyes vulnerable to oxidative stress with aging or in neurodegenerative diseases [12]. AD impacts visual function early in the progression of the disease, even prior to detectable memory impairments [13]. McKee et al. [14] observed Aβ protein and neurofibrillary tangles in the visual association areas of 52% and 100% of brains from cognitively intact and mildly impaired elderly people, respectively, indicating that AD disease processes may develop in visual association area before developing in memory processing regions. To explore the effect of oxygen treatment on AD in this study, we examined the differential expression of proteins in lenses, which may be more sensitive than that in brain tissues, of AD model mice given a supplemental oxygen treatment.
Firstly, transgenic AD model mice were subjected to a behavioral Morris water maze (MWM) experiment. Secondly, we employed quantitative proteomics to examine AD-related changes in protein expression levels in lenses from the same model mice. Relative changes in protein expression were assessed by isobaric tag for relative and absolute quantitation (iTRAQ) followed by tandem mass spectrometric (LC-MS/MS) analysis, the standards of the present study to judge the significance of change after oxygen treatment were two-fold.
MATERIALS AND METHODS
Animals
The triple-transgenic mouse model of AD (3xTg-AD) was employed. Twenty-one 3xTg-AD mice [B6;129-Psen1tm1Mpm Tg(APPSwe,tauP301L)1Lfa/Mmjax] were purchased from The Jackson Laboratory and divided into an oxygen-treated (OT) group (n = 12), given supplemental oxygen in their cages (40% for 20 min/day for 2 months), and a control (Ctrl) group (n = 9), which did not receive the oxygen treatment. The concentration of oxygen in the bottom of the OT group cages [measured with an injection oxygen analyzer (OX-100A, Huirui, China)] increased from 21% pretreatment to 32% posttreatment. The treatments were continued through behavioral testing. Mice were 6 months of age when the treatment protocol began. Both groups were housed in a temperature-regulated environment with free access to standard food and water. All procedures were carried out according to the Institutional Animal Care and Use Committee Guide.
Morris water maze
MWM training and testing was performed over 6 dand consisted of place navigation trial blocks on days 1–5 and a spatial probe test on day 6. The MWM apparatus (TMETech, china) was a 100-cm-diameter, 45-cm-high circular water pool with a 7-cm-diameter, 15-cm-high circular platform in the center of the first quadrant. The platform was kept 1 cm under the water surface to be invisible to the mice in thepool.
Each place navigation trial block included four trials, with the mouse being placed in a different starting quadrant in each trial. The mice were supposed to escape to the platform within 120 s. If a mouse did not reach the platform within 120 s, it was guided to the platform, where it was left to stand for 10 s before being placed back in its home cage. The escape latencies and swim paths were recorded by the video path analysis software.
On the spatial probe test day, the platform was removed from the maze. Mice were placed in the second and fourth quadrants, both of which were adjacent to the platform quadrant, and allowed to swim freely for 120 s. The number of platform-site crossovers made by each mouse was recorded.
Protein sample preparation
Mice were anesthetized and sacrificed 1 week after MWM testing. Their intact eyes were removed from the orbits, and the corneas were peeled back with tweezers to expose the lens. The lenses were then removed and placed in tubes on ice. The lens tissues from each group were homogenized in 1 mL ice cold lysis buffer containing 10 μL phosphatase inhibitor, 1 μL protease inhibitor, and 10 μL pmsf (keyGEN, china) first with a plastic homogenizer and then with an ultrasonic crusher (branson ultrasonics, USA). Fractionation was then carried out at 4°C, except where otherwise specified. Homogenates were centrifuged at 1000×g for 5 min. The extracted phosphorylated proteins were dissolved in the supernatent and then quantified by the Bradford method with a BCA Protein Quantification Kit (Beyotime, China). The proteins were digested by the filter-aided sample preparation method (FASP). Each digested protein sample was diluted in 200 μL of pH 8.5 urea and then centrifuged twice at 12,000 rpm for 30 min. Then, 10 μL of 100 mM dithiothreitol was added to the filter and shook for 1 h at room temperature. A 100- μL aliquot of 50 mM iodoacetamide (in pH 8.5 urea solution) was added and the sample was incubated in darkness for 20 min. The sample was then centrifuged at 12,000 rpm for 30 min and washed twice with 150 μL of 50 mM NH4HCO3, and washed twice with 100 μL of pH 8.0 urea solution. The washed proteins were digested overnight in 250 μL dissolution buffer with trypsin (Amersco) at an enzyme-to-protein ratio of 3:100 at 37°C. The released peptides were collected by centrifugation and washed with dissolution buffer.
iTRAQ labeling
We chose the iTRAQ protein quantitation method because of its prior utility in proteomics AD research [15]. Wu et al. [16] found that it was more sensitive than fluorescence difference gel electrophoresis and isotope-coded affinity tag-MS/MS. To normalize for total protein content, a volume corresponding to 30 μg of protein from each group sample was labeled with iTRAQ reagents (AB SCIEX, USA) according to the manufacturer’s protocol. The Ctrl and OT group samples were labeled with reagent 118 and 119, respectively. The two samples were combined into a mixed sample, which was then divided equally between two tubes and dried under avacuum.
LC-MS/MS analysis
Samples were loaded into a ChromXP C18 (3 μm, 120 Å) nanoLC trap column. Online chromatography separation was performed in a EksigentnanoLC-Ultra™ 2D System (AB SCIEX, Concord, ON). The trapping and desalting procedure were carried out at 2 μL/min for 10 min in 100% solvent A. The running solvents were water/acetonitrile/formic acid mixtures (A, 98/2/0.1%; B, 2/98/0.1%). Then, a 5–40% acetonitrile (0.1% formic acid) elution gradient over 70 min was applied on an analytical column (75 μm×15 cm C18, 3 μm 120 Å, ChromXPEksigent). LC MS/MS analysis was performed with a TripleTOF 5600 System (AB SCIEX,Concord, ON) fitted with a NanosprayIII source (AB SCIEX, Concord, ON). Data were acquired with a 2.4-kV ion spray voltage, 30-psi curtain gas, 5-psi nebulizer gas, and an interface heater temperature of 150 °C. The scan scope for TOF-MS is 350–1500 and for MS/MS is from 400 to 1250. For IDA, survey scans were acquired in 300 ms and thirty 80-ms product ion scans were run.
Protein identification and bioinformatics analysis
The MS/MS data were analyzed for protein identification with ProteinPilot Software v.4.5 (AB Sciex Inc., USA). The local false discovery rate was estimated to be 1.0% by the integrated PSPEP tool in the ProteinPilot Software after searching against the mouse proteome set database, which contains 16,615 protein sequences downloaded from http://www.uniprot.org/uniprot/. Our search parameters included no digestion, phosphorylation emphasis and biological modifications, thorough searching mode and a minimum protein threshold of 95% confidence (unused protein score > 1.3). The subcellular location, function, and biological process of the identified differential proteins were elucidated by gene ontology (GO) component, functional, and process terms, respectively. We performed GO categorization and statistical overrepresentation tests using the PANTHER database.
Data analysis
Group mean data [reported with standard deviations (SDs)] were compared with Prism5 statistical software (GraphPad Software, Inc., USA). The statistical tests were based on two-tailed probabilities and p-values <0.05 were considered to be statistically significant. Additionally, we compared classifications of up- and downregulated proteins to a reference list in the PANTHER database, and determined whether any classification categories of biological processes, cellular components, and molecular functions were over- or under-represented. That is, the numbers of proteins classified in each category were compared with the expected number for that category based on the PANTHER classification reference list [17].
RESULTS
MWM behavior
Both the OT and the Ctrl groups exhibited progressively shorter escape latencies with further training (Fig. 1A). The daily average escape latencies of the OT group were shorter than those of the Ctrl group on days 2, 3, 4, and 5 (p < 0.05), indicating that the OT group had better learning and memory performance than the Ctrl group. In the spatial memory probe test, the OT group exhibited slightly more platform-location crossings, but the difference did not reach statistical significance (Fig. 1B). Nevertheless, compared with the Ctrl group, the swim paths of the OT group were shorter, and OT mice (Fig. 1C, D) were less inclined to swim along the pool wall (Fig. 1E, F).
Identification of differentially expressed proteins by iTRAQ proteomic analysis
Using iTRAQ coupled with LC TRIPLE-TOF, we identified a total of 4,032 proteins from 377,038 peptides in the eye lenses of both groups. Among them, the expression levels of 205 proteins differed significantly between the OT and Ctrl groups (≥2-fold change in expression; p < 0.05), including 90 that were upregulated (Table 1) and 115 that were downregulated (Table 2) in the OT group. Among the differentially expressed proteins, 30 were verified to be closely related to AD, including ATCAY, NDRG3, APMAP, CPLX1, STX5, and ABCG2, which were upregulated (Table 1), and HMGB1, TS101, PINT, LMNB1, and ALDOC, which were downregulated (Table 2). The full names of the proteins are listed in Tables 1 and 2.
GO categorization of differentially expressed proteins
Of 205 proteins uploaded into PANTHER, 202 were mapped successfully. Following GO analysis, they were divided into 13 biological process categories (Fig. 2), including metabolic process (40%), cellular process (20%), biological regulation (8%), developmental process (6%), and apoptosis process (1%). In terms of cellular component, they were classified into seven categories, namely cell part (42%), organelle (29%), macromolecular complex (14%), and membrane (11%). Regarding molecular function, there were nine categories, including catalytic activity (42%), binding (28%), structural molecule activity (14%), and antioxidant activity (1%). A number of the protein classes identified (e.g., oxidoreductase, phosphatase, and calcium binding proteins) are critical components of pathways such as the apoptosis signaling pathway, the AD amyloid secretase pathway, the AD-presenilin pathway, the oxidative stress response, and the oxytocin receptor mediated signaling pathway.
Statistical overrepresentation
The number of categories in which downregulated proteins were over- or under-represented was greater (p < 0.05) than the number for upregulated proteins that were over- or under-represented (Table 3). The Ctrl group had more categories with unexpected numbers of relatively abundant proteins than the OT group, indicating that proteins in lenses of 3xTg-AD mice may differ substantially from that in normal mice, and that oxygen therapy has a normalizing effect on the protein content of lenses.
DISCUSSION
In the present study, we found that AD model mice given a daily oxygen treatment performed better in the MWM than AD model mice not given the treatment. This finding supports the notion that hypoxia may worsen functioning in AD and, further, supports our hypothesis that enhanced oxygen exposure may help to relieve AD symptoms. Using iTRAQ and LC-MS/MS based proteomic technology, we then identified and quantified proteins that were differentially expressed in the lenses of the OT versus Ctrl groups. A total of 205 proteins were found to be differentially expressed in relation to the treatment, including 90 that were upregulated and 115 that were downregulated. Among the differentially expressed proteins, 30 were closely related to AD.
Aging may reduce microvascular plasticity and contribute to a decline in cerebral blood flow, which reduces metabolic support for neural signaling, especially when levels of neuronal activity are high [18, 19]. Thus, compromised cerebral oxygen availability has been speculated to be a risk factor for a range of neurodegenerative diseases and oxygen treatment has been suggested as an adjunct therapy for AD. Hypoxia produces weariness and dizziness, and if sustained can lead to neuronal injury or even brain death. Hypoxic ischemic encephalopathy can disrupt cerebral blood flow and oxygen delivery to the brain [20]. Patients with ischemic stroke are at an elevated risk of developing incident dementia, particularly when stroke is associated with illnesses thought to cause cerebral hypoxia or ischemia [21]. Quantitative changes in both β-secretase and amyloid precursor protein genes, which are considered to be associated with AD neuropathology, were examined in rats following global brain ischemia [22]. Oxygen intake increases during physical exercise and exercise training has been used to alleviate AD symptoms [23]. Furthermore, Fang et al. [1] found that compared to a normoxia (20% O2) control condition, hypoxia (1% O2) promoted phosphorylation of tau protein, a major hallmark of AD, in culture primary hippocampal neurons.
Developmentally, the eyes are sensory extensions of the brain and anatomical changes in the eyes have been detected before signs of cognitive impairment and memory loss are apparent [12]. Amyloid-β protein precursor (AβPP) and Aβ peptide cleavage products have been found in lenses at levels comparable to those in the brain of humans with and without AD [24]. Increased levels of AβPP and Aβ have also been observed in lens and lens-cell cultures in the presence of cataractogenic agents [25]. Equatorial supranuclear cataracts (likely due to Aβ aggregation) were found consistently in lenses from AD patients (n = 9) but not in controls without the disorder (n = 8) [24]. Such observations support Gardner’s [26] assertion that Aβ accumulation processes that are going on in the brain are also going on in the eye and further suggest that it is worthwhile to examine the lenses of AD model mice.
The 3xTg-AD mice used in our study harbor PS1M146V, APPSwe, and tauP301L transgenes [27]. They, characteristically, exhibit deficits in synaptic plasticity, develop extracellular Aβ deposits, and form tangles. Aβ plaques formed in the brains and eyes of transgenic mice can impair vision and brain function. Aβ plaques were observed recently in the retinas of 3xTg-AD mice (Z. Tan, unpublished observation), supporting the notion that these mice can serve as models of AD-associated eye anomalies.
Several of the proteins that we found to be upregulated in the OT group are noteworthy. One of them, CPLX1, was reported to be reduced in the inferior temporal cortex of AD brains [28]. Abnormal expression of CPLX1 is seen in several neurodegenerative diseases associated with disturbed social behavior and CPLX1 knockout mice show pronounced social behavioral deficits [29]. Hence, we speculate that increased CPLX1 expression after oxygen treatment may help to normalize the behavior of AD mice. Secondly, ACOX1, a key enzyme involved in peroxysomal β-oxidation, was also upregulated in the OT group. ACOX1 activity decreases with aging, which alters the fatty acid composition of the brain and contributes to the progression of brain aging and certain neurodegenerative diseases, including AD and Parkinson’s disease [30]. Interestingly, siRNA knockdown of ACOX1 expression results in an elevation of oxidative stress, which has been strongly implicated in the pathogenesis of AD [31]. Thirdly, APMAP, a recently discovered endogenous suppressor of Aβ generation through its interaction with AβPP and γ-secretase that may help delimit AD progression [32] was upregulated in the OT group. Fourthly, the OT upregulated protein ABCG2 has also been reported to be involved in Aβ clearance [33]. Finally, the lysosomal endopeptidase TPP1 was upregulated in the OT group. Loss-of-function TPP1 mutations result in excess ROS generation [34] and such hyperoxidation can result in oxidative tissue damage. Moreover, excessive ROS is regarded as a causative factor in AD [35]. Hence, increased expression of TPP1 following oxygen treatment may reduce ROS levels, which would be conducive to alleviation of AD progression.
A number of AD-associated proteins were found to be downregulated in the OT group, including ENOA, TS101, PRDX6, HMGB1, LMNB1, and ALDOC. ENOA has been reported to be over-expressed in human AD brains [36]. TS101 depletion leads to an accumulation of endocytosed AβPP in early endosomes with reduced AβPP processing [37]. Meanwhile, PRDX6 over-expression has been associated with accelerated AD progression in relation to its increasing influence on amyloidogenesis through phospholipase A2 activation and increased nuclear factor erythroid 2-related factor 2 transcription [38]. Therefore, it could be that reduced PRDX6 expression in OT mice may have slowed the progression of the AD phenotype. HMGB1 augments neuroinflammation after neural injury or death [39]. Duplication of the gene that encodes LMNB1 results in a reduction in fractional anisotropy values in the corpus callosum, a characteristic that has been related to cognitive impairment severity in AD [40, 41]. Finally, ALDOC is a major target of S-nitrosylation in the hippocampus, substantia nigra, and cortex of AD brains [42]. Indeed, due to the covalent binding of NO with cysteine residues, ALDOC has been implicated in the pathology of several neurodegenerative diseases and has been suggested to play a leading role in the deterioration of neuronal function [43–45]. Although we did not quantitate the extent of S-nitrosylation of ALDOC in our samples, a significantly decreased quantity of ALDOC in the OT group may indicate that oxygen therapy has the potential to reduce the degenerative effects of protein S-nitrosylation in AD.
We have also identified differentially expressed proteins in the hippocampal mitochondria of 3xTg-AD mice and are examining oxidative-induced variation in their expression (unpublished).Interestingly, 182 of the 205 proteins we observed to be differentially expressed in the lenses of OT versus Ctrl mice in the present study were also found in hippocampal mitochondria, including several proteins that have been implicated in AD (e.g., ACOX1, ATCAY, and HS90A) and a mitochondrial respiratory chain regulator (e.g., SUV3).
Conclusion
Maintenance of a stable cellular redox state is critical for normal cell functioning. The present findings support the notion that a lack of cerebral oxygen may result in a dysregulation of oxidant levels, increasing susceptibility to AD progression. If so, oxygen supplementation therapy may prevent oxidative damage and thereby attenuate AD symptoms. The benefits of exercise on AD symptoms may be due to the increased oxygen intake associated with physical exertion.
