Mitochondrial perturbation drives tau oligomers pathology in Alzheimer’s disease

Tau oligomers, prior to neurobrillary tangle formation, are toxic species responsible for tau pathology, mitochondrial and synaptic damage, and memory impairment. The underlying mechanisms of abnormal tau accumulation and strategies to eliminate them remain largely unknown. The present study addresses whether mitochondrial reactive oxygen species (ROS) are major contributing factors for tau oligomer formation and, if so, whether eliminating mitochondrial ROS reduces accumulation of tau oligomers and improves mitochondrial and cognitive function in Alzheimer’s disease (AD). First, we determined whether increased oxidative stress correlates with aggregation of tau oligomers in human AD-affected brains, Aβ/tau overexpressed mouse models, human trans-mitochondrial “cybrid” (cytoplasmic hybrid) neuronal cells containing mild cognitive impairment (MCI) and AD-derived mitochondria, and Aβ/tau expressing neuronal cells. In P301S tau and AD mice, upregulation of tau oligomers correlates with ROS accumulation. Elevated tau oligomer levels are also correlated with elevated ROS levels in the AD patient hippocampus. Importantly, human cybrid cells, whose mitochondria are derived from platelets of patients with sporadic AD or MCI, displayed aggregated tau oligomers, which also correlated with upregulated ROS levels. Application of mito-Tempo, a mitochondria-targeted antioxidant, to inhibit the generation of mitochondrial and intracellular ROS in tau and AD neurons, as well as in MCI and AD cybrids ex vivo, leads to a striking decrease in tau oligomers. Finally, in AD mice, mito-Tempo inhibited tau oligomer accumulation and improved behavioral deciency. Our work adds to the growing body of evidence that oxidative stress contributes to tau oligomer formation and that inhibition of oxidative stress ameliorates tauopathy in AD.


Introduction
Alzheimer's disease (AD) is a progressive neurodegenerative disease that is associated with abnormal upregulation of oxidative stress. A key pathological hallmark of AD is the presence of neuro brillary tangles (NFTs), which are composed of hyperphosphorylated aggregates of tau 1,2 . Elevated phosphorylation and aggregation of tau destabilizes tau-microtubule interactions, leading to microtubule instability, dysfunctional axonal transport along microtubules, and neuronal death 3,4 . Tau phosphorylation is modulated by stress conditions such as oxidative stress and alterations in glucose metabolism during hypothermia and starvation 5

. Tau oligomers, intermediate species that form prior to
NFTs, include various species of tau protein including dimeric, multimeric, granular and possibly small lamentous aggregates, have deleterious effects on synaptic function and contribute to memory de ciency [6][7][8] .
Mitochondria are a major source of reactive oxygen species (ROS). Oxidative stress is a pathological characteristic of tauopathy and evidence has shown that accumulation of ROS could directly stimulate tau hyperphosphorylation and aggregation [9][10][11] . However, the precise role of tau oligomers in the disease process is poorly understood. In the present study, we investigated accumulation of pathological tau oligomers and its relevance to AD pathogenesis including tau and amyloid pathology and dysfunctional Page 3/23 AD mitochondria to address the following key questions: does the accumulation of tau oligomers associate with mitochondrial defects and ROS production in AD and an Aβ/tau-enriched environment? If so, does blockade of mitochondrial ROS eliminate tau oligomer formation and attenuate mitochondrial dysfunction? Does suppression of mitochondrial ROS rescue Aβ-mediated cognitive dysfunction? We comprehensively analyzed levels of tau oligomers, ROS, mitochondrial function, and behavioral endpoints using AD patient hippocampus, tauopathy and human Aβ-producing AD mouse models and in vitro cultured neurons, and MCI and AD cybrids as ex vivo models for AD mitochondrial dysfunction. Our studies demonstrate that tau oligomers were signi cantly elevated in the AD patient hippocampus, P301S tauopathy and mAPP mice hippocampus and entorhinal cortex, cultured neurons, and human MCI and AD cybrids. Importantly, there is a positive correlation between tau oligomer levels and ROS levels. Scavenging mitochondrial ROS prevented accumulation of toxic tau oligomers, attenuated Aβ-and tauinduced mitochondrial perturbation, and led to improvements in learning and memory in P301S tauopathy and mAPP mice.

Results
Accumulation of tau oligomers is associated with ROS in the AD brain Given the accumulation of amyloid beta (Aβ) and abnormal tau in AD-affected brain regions including the hippocampus, we rst tau oligomers (oTau) in the AD-affected hippocampus. Immunostaining with speci c tau oligomeric complex I (TOC1) antibody showed that intensities of oTau-positive signals were signi cantly elevated by 3-4 folds in the AD hippocampus ( Fig. 1A-B). Consistent with the immunostaining results, immunodot blotting demonstrated that oTau levels were greatly elevated in the AD hippocampus but not in the cerebellum as compared with non-AD brains ( Fig. 1C-F). To determine whether elevation of oTau was correlated to oxidative stress, we quantitatively measured the intracellular reactive oxygen species (ROS) levels in the hippocampus by highly speci c electron paramagnetic resonance (EPR) spectroscopy. Intracellular ROS levels as indicated by EPR peaks were signi cantly elevated in the AD hippocampus ( Fig. 1G-H). Furthermore, oTau levels were positively correlated with ROS (Figs. 1I), suggesting a possible link between oTau accumulation and ROS production/accumulation relevant to AD pathology.
Accumulation of tau oligomers associates with ROS in P301S tauopathy mice Next, we determined tau oligomers, ROS levels, and their association with tau pathology in tauopathy mice. Immunostaining revealed increased oTau in the hippocampus and cortex of P301S mice and their presence in cortical neurons ( Fig. 2A-C). Compared to age-matched nonTg mice, P301S mice displayed robustly elevated oTau in the hippocampus and entorhinal cortex at 9-months-old, as shown by immunodot blotting (Fig. 2D-E), but not at 1-month-old (Supplementary Fig. 1A and D). Similarly, phosphorylation of tau (Ser202 and Thr205) was signi cantly elevated in the hippocampus and entorhinal cortex of 9-month old P301S mice (Figs. 2F-G), but not in 1-month-old P301S mice (Supplementary Fig. 1C and F). The intracellular ROS levels indicated by EPR peaks were signi cantly elevated in the 9-month-old P301S hippocampus and entorhinal cortex compared to nonTg mice (Figs. 2H-K), but not in 1-month-old P301S mice (Supplementary Fig. 1B and E). Levels of oTau were positively correlated with ROS levels (Figs. 2L-M). These data suggest that tau oligomers are associated with aging and oxidative stress relevant to tau pathology.
Tau oligomer accumulation correlates with elevated oxidative stress in an Aβ-producing AD mouse model Given the potential relationship between abnormal tau accumulation and excessive Aβ and oxidative stress [11][12][13] , we assessed the effect of Aβ on tau oligomer levels in Aβ-producing transgenic mice. In mAPP mice at 12 months of age, a timepoint with tremendous accumulation of cerebral Aβ, we observed signi cant accumulation of oTau staining in the hippocampus ( Fig. 3A and B) and entorhinal cortex ( Fig. 3A and C). Similarly, immunodot blotting showed, respectively, a ~ 4.5 and ~ 5.5-fold increase of oTau in the hippocampus (Fig. 3D) and entorhinal cortex (Fig. 3E). Furthermore, we studied the progression of oTau deposition by assessing expression of oTau in different ages of mAPP mice. As shown in Fig. 4, oTau were signi cantly elevated in the hippocampus (Fig. 4A) and the entorhinal cortex ( Fig. 4B) of mAPP mice compared to nonTg controls in an age-dependent manner starting at 6-9 months of age, with a greater degree of accumulation at 12-18 months of age, a time of tremendous accumulation of cerebral Aβ. The ROS levels in the hippocampus of mAPP were signi cantly elevated starting at 6 months of age, prior to the start of tau oligomers accumulation ( Fig. 4C-D), and importantly, tau oligomers content was signi cantly correlated to ROS levels ( Fig. 4E-F). These studies indicate the impact of Aβ on tau oligomer accumulation in an Aβ-rich environment.

Accumulation of tau oligomers is associated with ROS in human MCI and AD cybrids
Mitochondrial dysfunction is one of the early pathological features of AD. Dysfunctional mitochondria produce excessive ROS. Human trans-mitochondrial "cybrid" (cytoplasmic hybrid) neuronal cells whose mitochondria are derived from platelets of patients with sporadic AD or mild cognitive impairment (MCI) exhibit signi cant changes in mitochondrial structure and function and increases in ROS generation/accumulation [14][15][16][17] . We therefore utilized MCI and AD cybrids as an ex vivo model to determine the potential impact of MCI-and AD-derived mitochondrial defects on pathological tau oligomers. Levels of tau oligomers were signi cantly increased in differentiated MCI and AD cybrids compared to non-AD controls. AD cybrid cells exhibited higher levels of tau oligomers than MCI cybrids (Fig. 5A), suggesting that accumulation of tau oligomers is associated with progression of mitochondrial perturbation in AD. Similar results were obtained from MCI and AD platelets (

Scavenging mitochondrial ROS eliminates tau oligomers in Aβ and tau neurons in vitro
To further evaluate the contribution of mitochondrial ROS to tau aggregation, Aβ-and tau-producing neurons cultured from mAPP mice and P301S tau mice, respectively, were treated with mito-TEMPO, a scavenger for mitochondria-derived ROS. Compared to nonTg neurons, mAPP and tau neurons display signi cantly elevated levels of tau oligomers, with TOC1 staining and expression signi cantly distributed along the cellular bodies and processes; mito-TEMPO treatment strikingly inhibited these TOC1 positive staining signals (Figs. 6A-B). Treatment with mito-TEMPO almost completely reduced tau oligomer levels to those of the vehicle controls as demonstrated by TOC1 immunodot blotting (Fig. 6C). Functionally, mito-TEMPO treatment not only alleviated mitochondrial defects, as shown by increased activity of key mitochondrial respiratory enzymes (complex IV in Fig. 6D and complex I in Fig. 6E) and ATP levels ( Fig. 6F), but also suppressed ROS levels ( Fig. 6G-H). Similarly, mito-TEMPO abolishment of Aβ-and taumediated oTau formation was positively correlated with reduction in ROS levels in Aβ-and Tau-producing neurons (Fig. 6I). These data indicate that scavenging mitochondrial ROS blocks Aβ-, tau-and ADmediated pathological tau oligomer accumulation and restores mitochondrial function.

Scavenging mitochondrial ROS reduces tau oligomers and improves learning and memory in AD mice
In view of the association between elevation of tau oligomers and oxidative stress in Aβ-producing AD mice (Fig. 4), we assessed the effects of Aβ-mediated mitochondrial defects and excessive ROS on tau oligomer accumulation and cognitive function (Fig. 7). mAPP mice overexpressing Aβ were treated with mito-TEMPO starting at 6-7 months of age for three months and then analyzed for tau oligomers

Discussion
Tauopathies are a class of neurodegenerative disorders characterized by hyperphosphorylation and aggregation of the microtubule-associated protein tau (MAPT) into paired helical laments (PHFs) or straight laments (SFs), leading to the formation of neuro brillary tangles (NFTs). Tau monomers bind to each other to form oligomeric tau when hyperphosphorylated tau dislodges from microtubules. Tau oligomers potentiate neuronal damage, leading to traumatic brain injury and neurodegeneration [19][20][21] .
Recent studies have demonstrated that tau oligomers are responsible for progression of tau pathology 6 . Up-regulated granular tau oligomer levels occur prior to NFT formation and clinical symptoms of AD.
Reduction of tau by doxycycline treatment improved memory impairment in P301L tau mice without affecting NFT formation, suggesting an early role for tau oligomers in AD pathogenesis relevant to cognitive decline 22,23 . However, the causes and co-stimulating factors that enhance pathological tau formation and accumulation relevant to the pathogenesis of AD remain largely unknown.
In the present study, we analyzed levels of tau oligomers in the human AD hippocampus, AD-related Aβ and tauopathy mouse models, and MCI-and AD-derived mitochondria. Levels of tau oligomers were signi cantly elevated in AD brains, Aβ and tau overexpressed mice, and in vitro in Aβ and tau neurons. Interestingly, tau oligomers accumulated in an age-dependent manner in mAPP mice, with signi cant elevation starting at 6-7 months of age, a time point corresponding to the appearance of behavioral abnormalities in mAPP mice. Our ndings are consistent with a previous study, which revealed that brief exposure to extracellular recombinant human tau oligomers, rather than monomers or Aβ, results in impairment in long-term potentiation (LTP) and learning and memory 6 .
Overproduction of ROS and oxidative stress-mediated cellular perturbation are known to be key players in AD pathogenesis, including tauopathy 9,24 . However, it is the causes and consequences of pathological tau metabolism such as toxic tau oligomers remain unclear. We observed that accumulation of tau oligomers is signi cantly elevated and positively correlated to ROS levels in human AD brains, MCI-and AD-derived mitochondria, and amyloid and tauopathy mice. Importantly, scavenging mitochondrial ROS by treatment with mito-TEMPO, a mitochondria-targeted antioxidant, strikingly reduced tau oligomers accumulation in cultured tau neurons. Intriguingly, mito-TEMPO also suppressed Aβ-and AD mitochondria-induced tau oligomers accumulation and rescued mitochondrial respiratory function. In Aβproducing AD mice, treatment with mito-TEMPO signi cantly diminished tau oligomers and improved cognition. These results indicate the contribution of mitochondrial ROS to tau oligomer formation and accumulation. Given that Aβ oligomers can seed and promote tau oligomerization and uptake of tau brils [25][26][27] , we propose that Aβ-mediated sustained mitochondrial stress and excessive ROS production could be a potential mechanism underlying Aβ-mediated/enhanced tau oligomers accumulation. Our studies suggest the role of the Aβ/ROS/mitochondria axis in aberrant tau oligomer accumulation, which links to cognitive dysfunction. The detailed mechanisms require further investigation in the near future. It is noted that tau oligomers were not signi cantly increased in AD-spared regions such as the cerebellum when compared to non-AD controls, nor were there signi cant changes in ROS levels in these spared regions (data not shown). Thus, accumulation of tau oligomers is associated with AD-pathology.
Taken together, using multiple AD models, we have provided substantial evidence of the connection between mitochondrial ROS and the accumulation of tau oligomers. Blocking mitochondrial oxidative stress eliminates accumulation of tau oligomers and improves mitochondrial function relevant to amyloid and tau pathology. Thus, our ndings support the possibility that inhibiting mitochondrial oxidative stress and dysfunctions could be a promising therapeutic target to prevent and treat tauopathies.

Materials And Method
Animal studies. Animal studies were carried out with the approval of the Institutional Animal Care and Use Committee of the University of Kansas Lawrence and Columbia University in New York in accordance with the National Institutes of Health guidelines for animal care.
Human subjects. We obtained hippocampal and cerebellar tissues from individuals with Alzheimer's disease and age-matched, non-Alzheimer's disease controls from the New York Brain Bank at Columbia University. Detailed information for each of the cases studied is shown in Supplementary Table 1 online. Informed consent was obtained from all subjects.

Primary neuronal culture
Hippocampal neurons from day 1 nonTg, mAPP, or P301S tau mice were prepared as described previously 28 . Neurons were cultures in neurobasal medium supplemented with 1 × B27, 600 μM L-Glutamine and penicillin-streptomycin. At day 14 in vitro (DIV), neurons from the indicated mice were used for experiments.

Evaluation of intracellular reactive oxygen species (ROS)
Evaluation of intracellular ROS levels was conducted by electron paramagnetic resonance (EPR) spectroscopy. Brain tissues or cultured neurons was incubated with CMH (cyclic hydroxylamine 1hydroxy-3-methoxycarbonyl-2, 2, 5, 5-tetramethyl-pyrrolidine, 100 μM) for 30 minutes and then washed three times with cold PBS. Subsequently, brain tissues and neurons were collected and homogenized with 100 μl of PBS for EPR measurement. The EPR spectra were recorded, stored, and analyzed with a Bruker EleXsys 540 X-band EPR spectrometer (Billerica, MA) using Bruker Xepr software Xepr (Billerica, MA) 29 .

Measurement of respiratory chain complex activities and ATP levels
Mitochondrial respiratory complex I activities were measured in neuronal homogenates as described previously 28,30 . NADH: ubiquinone oxidoreductase (COX I) enzyme activity was determined in 25 mM potassium buffer containing KCl, Tris-HCl and EDTA (pH 7.4). Homogenized samples (50 μg protein) were incubated with with 2 μg/ml antimycin, 5mM MgCl 2 , 2mM KCN and 65 μM co-enzyme Q1 were and the oxidation of NADH was recorded for 3 min. Subsequently, 2 μg /ml rotenone was added and the absorbance was measured for another 3 min. The change in absorbance was monitored at 340 nm using an Amersham Biosciences Ultrospect 3100 Pro spectrophotometer.
Cytochrome c oxidase (CcO, complex IV) activity was spectrophotometrically determined using the Cytochrome c Oxidase Assay Kit (Sigma) as described in our previous study 28,30 . In brief, neurons were collected using lysis buffer, incubated on ice for 15 minutes, and centrifuged at 12,000g for 10 minutes. Suitable volumes of lysates and enzyme solutions were added into 475 μl of assay buffer. The reaction was triggered by the addition of 25 μl of ferrocytochrome c substrate solution (0.22 mM). Changes in absorbance of cytochrome c at 550 nm were immediately recorded using a kinetic program with 5 second delay, 10 second intervals, for a total of 6 readings on an Ultrospect 3100 Pro spectrophotometer. ATP levels were determined using an ATP Bioluminescence Assay Kit (Roche) following the manufacturer's instruction. Brie y, neurons were collected in the provided lysis buffer, incubated on ice for 30 minutes, and centrifuged at 12,000g for 10 minutes. ATP levels were then measured in the subsequent supernatants using a luminescence plate reader (Molecular Devices). A 1.6 second delay after substrate injection and 10 seconds integration time were used.

Immunoblotting analysis
Brain tissue lysates for immunoblotting were prepared following the method described in our previous study 28 . Protein lysates were subjected to 10% Bis-Tris gel electrophoresis (Invitrogen, Grand Island, NY, USA), transferred to nitrocellulose membrane, incubated with 5% non-fat dry milk in TBST buffer (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20) for 1 hour at room temperature, followed by incubation with primary antibodies under gentle agitation overnight at 4ºC. The following primary antibodies were used: Toc1 (oligomeric tau, mouse IgM, provided by Dr. Kanaan (Nicholas M. Kanaan, Department of Translational Neuroscience, Michigan State University), AT8: anti-phospho-Tau pSer202/Thr205 (MN1020, Thermo Fisher Scientifc), and β-actin (A5441; Sigma-Aldrich). ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used for quanti cation of the intensity of the immunoreactive bands in the developed blots.
Immunodot blotting Brain or cell lysates were prepared and analyzed as described for immunoblotting with the following modi cations. Samples were spotted onto the nitrocellulose membrane using a Whatman Minifold I immunodot blotting apparatus. The membranes were blocked, and probed with TOC1 and β-actin (A5441; Sigma-Aldrich). Signal intensity measurements for each dot were expressed as the ratio of oligomeric tau (TOC1 signal) to β-actin. ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used for quanti cation of the intensity of the developed blots.
Immunohistochemical staining Brain slices from the indicated Tg mice were subjected to double immuno-staining with anti-Toc1 (oligomeric tau, mouse IgM) and mouse anti-MAP2 (1:5000, sc-33796, Santa Cruz Biotechnology) at 4°C overnight followed by Alexa Fluor 488 and 594 goat anti-mouse, respectively. Images were acquired on a Leica SP5 confocal microscope and analyzed using Leica LAS AF software (Leica Wetzlar) and MetaMorph (Molecular Devices) Program.

Behavioral Test
The Morris Water Maze (MWM) test was performed according to the method described in previous publications 28,31 . During the spatial acquisition session, mice were trained for 6 consecutive days with 4 trials per mouse per day. Maximum time for each trial was capped at 60 s. Escape latency was analyzed by the HVS Image software (2015). On day 7, a probe trial was performed to assess the spatial memory of the mice. The platform was removed from the pool and the mice were allowed to swim freely for 60 s. Traces of the mice swim paths were recorded and analyzed by HVS Image. Investigators were blind to mouse genotypes and treatment groups.
Statistical Analysis. All data were expressed as the mean ± SEM. Student t-tests were performed for analysis and comparisons between two groups. Data were analyzed by one-way ANOVA for repeated measure analysis using commercially available software (       Complex IV (C) activity and ATP levels (D) were determined in the entorhinal cortex from the indicated