FLZ Alleviates the Memory Deficits in Transgenic Mouse Model of Alzheimer’s Disease via Decreasing Beta-Amyloid Production and Tau Hyperphosphorylation

Alzheimer’s disease (AD) is the most common cause of dementia worldwide and mainly characterized by the aggregated β-amyloid (Aβ) and hyperphosphorylated tau. FLZ is a novel synthetic derivative of natural squamosamide and has been proved to improve memory deficits in dementia animal models. In this study, we aimed to investigate the mechanisms of FLZ’s neuroprotective effect in APP/PS1 double transgenic mice and SH-SY5Y (APPwt/swe) cells. The results showed that treatment with FLZ significantly improved the memory deficits of APP/PS1 transgenic mice and decreased apoptosis of SH-SY5Y (APPwt/swe) cells. FLZ markedly attenuated Aβ accumulation and tau phosphorylation both in vivo and in vitro. Mechanistic study showed that FLZ interfered APP processing, i.e., FLZ decreased β-amyloid precursor protein (APP) phosphorylation, APP-carboxy-terminal fragment (APP-CTF) production and β-amyloid precursor protein cleaving enzyme 1 (BACE1) expression. These results indicated that FLZ reduced Aβ production through inhibiting amyloidogenic pathway. The mechanistic study about FLZ’s inhibitory effect on tau phosphorylation revealed t the involvement of Akt/glycogen synthase kinase 3β (GSK3β) pathway. FLZ treatment increased Akt activity and inhibited GSK3β activity both in vivo and in vitro. The inhibitory effect of FLZ on GSK3β activity and tau phosphorylation was suppressed by inhibiting Akt activity, indicating that Akt/GSK3β pathway might be the possible mechanism involved in the inhibitory effect of FLZ on tau hyperphosphorylation. These results suggested FLZ might be a potential anti-AD drug as it not only reduced Aβ production via inhibition amyloidogenic APP processing pathway, but also attenuated tau hyperphosphoylation mediated by Akt/GSK3β.


Introduction
Alzheimer's disease (AD) is the most common cause of dementia worldwide. The hallmarks of AD are mainly characterized by senile plaques (SPs) and neurofibrillary tangles (NFTs), which consist of aggregated b-amyloid protein (Ab) and hyperphosphorylated tau protein (p-tau). Ab peptides are generated by successive proteolysis of b-amyloid precursor protein (APP), a large transmembrane glycoprotein, which is initially cleaved by the bamyloid precursor protein cleaving enzyme 1 (BACE1) and subsequently by c-secretase in the transmembrane domain [1]. Aggregated Ab plays a pivotal role in the pathogenesis of AD [2]. After penetrating neuronal membrane, Ab aggregates and destroys cell membrane [3], thus inducing neuronal loss and memory deficits [4,5]. Moreover, Ab alters cellular metabolism and triggers downstream tau hyperphosphorylation/tangle formation, which plays a major role in the onset of cognitive decline and tau pathology. Aberrant hyperphosphorylated tau is reported to lose its ability to bind and stabilize microtubules, resulting in destabilization of the cytoskeleton and perturbation of axonal transport [6].
Current drugs for AD treatment such as acetylcholinesterase inhibitor and NMDA antagonist show limited benefits to most AD patients [7]. Therefore, there is an urgent need for novel therapeutic strategies that could halt the disease process and are suitable for long-term clinical management. A new viewpoint has emerged that the treatment of AD can be more effective at specific targets, such as oligomeric Ab and tau hyperphosphorylation [7]. Therefore, exploring compounds which prevent Ab accumulation or tau hyperphosphorylation has been the main focus of the drug development in recent years. We have previously reported that FLZ (Fig. 1), a novel synthetic cyclic analogue of natural squamosamide from the leaves of Anona squamosa, has neuroprotective effects on the in vivo and in vitro AD models. FLZ improved memory deficits of natural aging mice [8], demential model mice induced by Ab [25][26][27][28][29][30][31][32][33][34][35] (Fang & Liu, 2006) and D-galactose plus NaNO 2 [9]. FLZ also markedly prevented neuronal apoptosis induced by Ab 25,35 [10]. Pharmacokinetic study indicated that FLZ could cross the blood-brain barrier and reach cortex and hippocampus [11]. These previous studies suggested FLZ had potential therapeutic effect on AD-associated pathologies. However, the underling mechanism of FLZ's therapeutic effect on AD models has not yet been ascertained.
The present study was designed to investigate the beneficial effects of FLZ on learning and memory deficits of AD model mice and the possible mechanisms. Consistent with previous studies [12,13], APP/PS1 transgenic mice demonstrated memory impairment, Ab overproduction and tau hyperphosphorylation. Using APP/PS1 transgenic mice and SH-SY5Y (APP wt/swe) cells, we found that FLZ significantly improved learning and memory ability of mice, decreased Ab production and tau phosphorylation. Mechanistic study indicated that FLZ attenuated Ab production by inhibiting APP phosphorylation and BACE1 expression, inhibited tau hyperphosphorylation through regulating Akt/ glycogen synthase kinase 3b (GSK3b) pathway.

Regents and Antibodies
FLZ, a white powder with 99% purity, was synthesized by the Department of Pharmaceutical Chemistry, Institute of Materia Medica, Chinese Academy of Medical Science. It was suspended in 0.5% (w/v) sodium carboxymethyl cellulose (CMC-Na) for oral administration and dissolved in dimethylsulfoxide (DMSO) for in vitro use. Dulbecco's modified Eagle medium, Nutrient Mixture F-12 (DMEM/F-12), fetal bovine serum and G418 were purchased from Gibico (Grand Island, NY, USA). b-amyloid ELISA kit was product of Invitrogen. Phosphorylated tau at Ser396, Ser404 and Thr231, phosphorylated Akt, Akt and insulindegrading enzyme (IDE) antibodies were purchased from Signalway. Phosphorylated tau at Ser202/Thr205 (clone AT8) antibody was purchased from Pierce Biotechnology. APP, APP C-terminal fragment (APP-CTF) and Ab (clone 6E10) antibodies were from Abcam. BACE1 antibody was obtained from Millipore. Phosphorylated APP (Thr668) antibody and Ly294002 were purchased from Cell Signaling. Phosphorylated GSK3b, GSK3b antibodies were from Santa Cruz Biotechnology. Caspase-3 assay kit was purchase from Sigma.

Ethics Statement
All experimental procedures were approved by the Institutional Animal Care and Use Committee of the Peking Union Medical College and performed in accordance with the National Institutes of Health guide for the care and use of laboratory animals. All efforts were made to minimize animal suffering.

Animal and Treatment
Male B6C3 mice expressing Swedish mutant AbPP695 and the exon-9 deletion mutant PS1 (APPswe/PS1D9) were supplied by the Animal Center of Chinese Academy of Medical Sciences. Mice were maintained in a 12-hours light/dark cycle at 24uC in a relative humidity of 60% room, and received food and water ad libitum. FLZ 150 mg/kg were orally administered to APP/PS1 double-transgenic mice and the WT mice were orally given 0.5% CMC-Na at 5 months of age for continued 20 weeks.

Cell Culture and Treatment
SH-SY5Y cells stably expressed wild-type APP and Swedish mutant APP (APPwt/swe) were gifts from Prof. Bao-Lu Zhao from Institute of Biophysics, Chinese Academy of Sciences. SH-SY5Y cells transfected with empty pcDNA3.1neo vector was used as negative control. The cells were maintained in DMEM/F-12 medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin and 200 mg/ml G418 at 37uC in a humidified 5% CO 2 /95% air incubator. When the confluency reached 80%, the cells were switched to a ''stimulating medium'' containing 50% DMEM, 50% Opti-MEM, 0.5% FBS, 200 mg/ml G418 and 10 mM butyric acid sodium salt for 12 h to induce the transgene expression. FLZ (0.1, 1 and 10 mM), Ly294002 10 mM combined with FLZ 10 mM or alone were incubated with cells for 24 h, and then the cells were harvested to test the related biomarkers.

Morris Water Maze Test
Spatial learning and memory o mice was assessed in the Morris water maze (Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China) after the mice had received FLZ for 20 weeks (n = 10 in each group). The Morris water maze consisted of a circular pool (120 cm diameter and 40 cm deep) filled with nontoxic opaque water with an escape platform (10 cm diameter) hidden beneath the water (2 cm). The water temperature was maintained at 2361uC. At the beginning of each trial, the mouse was placed into the water facing the wall of the pool at one of the four quadrants. Although the starting point was randomly selected, the protocol was fixed at the beginning of each trial and was maintained throughout all four acquisition trials. Each mouse was allowed 120 s to find and mount the platform. When the mouse found the platform, it was kept on the platform for 30 s. If the mouse failed to find the platform within 120 s, it was placed on the platform where it remained for 30 s. Each mouse was given 4 trials per day, with an interval of 1 h. The time to find the platform (escape latency), the total distance traveled, and the swim speed of the animals were recorded by video tracking software. The mice were then towel dried and placed in a cage with a heating pad underneath until dry and returned to their home cage. The training session was conducted for 5 consecutive days in which the platform was never moved. The latency to escape was calculated as the average time to find the platform of the 4 trials within one day.
Memory retention was evaluated on the 6th day with probe trial in which the platform was removed. The mice were put into the pool and allowed to swim freely in the pool for 120 s. the number of crossings of the platform and the time spent in the target quadrant were recorded.

Immunohistochemical Analysis
Five mice from each group were anesthetized with pentobarbital and perfused transcardially with saline and then with cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. Brains were then removed and post-fixed for 4 h in the same fixative. The fixed brains were washed and cryoprotected in the same buffer containing 20% sucrose, and finally sectioned into 40 mm sections on a freezing microtome. Every sixth serial section was selected and processed for TH immunostainning. Coronal sections through the hippocampus were processed for Ab immunohistochemistry. Briefly, after incubation for 1 h in 10% normal swine serum with 0.25% TritonX100 in 0.02 M potassium-phosphate-buffered saline containing 1% bovine serum albumin (KPBS-BSA), sections were incubated with the primary antibody (mouse monoclonal antiserum to Ab, 1:500 in KPBS-BSA containing 2% normal swine serum and 0.25% Triton X100) and then incubated first with the corresponding biotinylated secondary antibody and subsequently for 90 min with avidin-peroxidase. Finally, the labeling was visualized with 0.04% hydrogen peroxidase and 0.05% 3,39-diaminobenzidine (DAB). The sections were observed with light microscopy (NIKON E600, Japan), and the intensity of stained area in each group was analyzed by Image-Pro plus system (Media Cybernetics, Silver Spring, MD). All evaluations were done by a researcher blind to the experimental design.

ELISA Assay
The supernatants of cells were harvested, centrifuged and 106 concentrated in presence of a protease inhibitors cocktail (2.5 mM EDTA, 10 mM leupeptin, 1 mM peptastin, 1 mM phenylmethylsulfonyl fluoride) cocktail using Centricon (Amicon) with a cutoff value of 3 kDa. For the measurement of intracellular Ab, approximately 10 7 cells were scraped in ice cold PBS. Cell pellets were solubilized in 300 ml of 70% formic acid. Formic acidsolubilized cell pellets were cleared by centrifuging at 16,000 g for 5 min at 4uC, and the supernatants were centrifuged at 21,000 g for 20 min at 4uC. The supernatants were vacuum dried and resulting pellet was resuspended in 1 ml of alkaline carbonate buffer (2% Na 2 CO 3 , 0.1N NaOH) and centrifuged at 16,000 g for 3 min at 4uC. Ab production was measured by a sensitive fluorescence based sandwich ELISA assay using a kit (Human b-Amyloid 1-40 Colorimetric Immunoassay Kit) according to the manufacturer's instructions.

Western Blot Assay
Mouse hippocampus (four mice from each group) and cells were lysed in nondenaturing lysis buffer. The lyses were then centrifuged at 12,000 g for 15 min at 4uC, the supernatants were mixed with loading buffer and boiled for 5 min. Protein concentration was measured by Bradford protein assay. Samples containing 30 mg of protein per lane were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to PVDF membranes (For Ab assay, 16% Tris-Tricine gel was used.). The membranes were blocked in 5% skim milk-TBST (20 mM Tris-HCL, pH 7.5, 500 mM NaCl, 0.1% Tween 20) for 1 h, then respective primary antibodies were added in the same milk and incubated overnight at 4uC, and then incubated with horseradish peroxidase-conjugated secondary antibody in TBST for 2 h at room temperature. The blot was developed with LAS3000 chemiluminescence system (Fujifilm, Tokyo, Japan), and the densities of the bands were determined using Gel-Pro Analyzer 4.0 software.

Flow Cytometry Assay
The apoptosis rate of cells was assayed by flow cytometry according to the manufacture's instruction. Briefly, the cells were harvested and washed twice with PBS, followed by resuspending in 500 ml binding buffer, and then incubated with 5 ml Annexin V and 5 ml Propidium Iodide (PI) for 15 min in the dark at room temperature. Then 1610 4 cells were measured with a flow cytometer using the Cell Quest software.

Caspase-3 Activity Determination
Caspase-3 activity was determined according to the manufacturer's protocol. In brief, cell lysates were prepared by lysis buffer. Caspase-3 activity was determined by monitoring proteolysis of the colorimetric substrates. Ac-DEVD-p-nitroaniline was used as colorimetric p-nitroaniline linked substrate. The whole-cell lysate was added to a buffer containing 200 mM substrate. After 1.5 h of incubation, the cleavage of the peptide by the caspase was quantified spectrophotometrically at 405 nm in a 96 well plate. The unit of the optical density was converted to nmols of pnitroaniline using a standard curve generated with free pnitroaniline.

Statistical Analysis
Data were expressed as means 6SD. MANOVA was used for repeated measurement of escape latency data, using a general linear model (GLM) in SPSS 13.0. Other data were analyzed by one-way ANOVA followed by Dunnett's post hoc test. P,0.05 was considered to be statistically significant.

FLZ Reduced Spatial Learning and Memory Deficits of APP/PS1 Mice
FLZ treatment started when the APP/PS1 double-transgenic mice were 5 months old and exhibited memory deficits (Fig. S1 in File S1). After FLZ was administrated daily for consecutive 20 weeks, the mice were subjected for Morris water maze test. As shown in Fig. 2, APP/PS1 mice showed significantly longer escape latencies than the wide-type (WT) mice (P,0.01), confirming the impaired spatial learning in the APP/PS1 mice. FLZ treatment significantly decreased the escape latency at the 3 rd , 4 th and 5 th day of the test (P,0.05) (Fig. 2). In the probe test, the number of APP/ PS1 mice crossing the platform and the time spent in the target quadrant remarkably decreased compared with WT mice, and the travel distances were much longer than those of WT mice. FLZ treatment significantly increased the number of platform crossing (1.16-fold increase, P,0.01) and the time spent in the target quadrant (1.8-fold increase, P,0.01), and the travel distance was decreased by FLZ treatment (62.18% decrease, P,0.01). There were no differences of swimming speed among the three groups of mice. Besides, the body weights of mice in all the groups were at the same level during the test (Fig. S2 in File S1). These data indicated that FLZ had beneficial effects on the learning and memory lesion of APP/PS1 transgenic mice.

FLZ Inhibited Ab Production in APP/PS1 Mice and SH-SY5Y (APPwt/swe) Cells
The effect of systemic administration of FLZ on Ab production in APP/PS1 double-transgenic mice was tested. Consistent with previous studies [12,13], immunohistochemical assay showed that there were large areas of Ab deposits in hippocampus of APP/PS1 mice. FLZ treatment significantly reduced the areas of Ab deposits in hippocampus (62.17% decrease, P,0.01) (Fig. 4A). We also found that FLZ decreased Ab deposits in cortex of APP/PS1 mice (Fig. S3 in File S1). Similar data were obtained in Western blot assay, which showed the increased amount of Ab in hippocampus of APP/PS1 mice and the therapeutic effect of FLZ (71.62% decrease, P,0.01) (Fig. 4B).
The above in vivo data were confirmed in the following in vitro experiment, in which SH-SY5Y (APPwt/swe) cells were used. This cell line stably overexpresses the Swedish mutant form of human APP and overproduces Ab. Western blot assay and ELISA assay were applied to measure the intracellular and extracellular Ab levels, respectively. The results showed that both intracellular and extracellular Ab levels significantly increased in SH-SY5Y (APPwt/swe) cells. FLZ 10 mM significantly reduced intracellular Ab production (73.33% decrease, P,0.05), FLZ 1 mM and 10 mM markedly decreased levels of extracellular Ab (FLZ 1 mM: 44.56% decrease, P,0.05; FLZ 10 mM: 63.93% decrease, P,0.01) (Fig. 4 C, D). The above data indicated that FLZ potentially inhibited the production of Ab.

FLZ Decreased the Expression of Phospho-APP and APP-CTF
Ab is a proteolytic product of APP, and phosphorylation at Thr668 of APP facilitates its processing towards Ab. So we measured the total APP and phospho-APP (Thr668) levels in the hippocampus of APP/PS1 mice and SH-SY5Y (APPwt/swe) cells. The results showed that APP and phospho-APP (Thr668) levels markedly increased in the hippocampus of APP/PS1 mice and SH-SY5Y (APPwt/swe) cells. FLZ treatment did not affect the level of total APP, but significantly reduced APP phosphorylation at Thr668 both in vivo (44.44% decrease, P,0.05) and in vitro (FLZ 1 mM: 52.94% decrease, P,0.05; FLZ 10 mM: 73.53% decrease, P,0.01) (Fig. 5), which indicated the interference of FLZ on APP processing.
The phospho-APP is cleaved by BACE1, and produces membrane-bound C-terminal fragment (CTF). The APP-CTF is then digested by c-secretase within the transmembrane region, which generates small secreted peptides, including the pathogenic Ab. So the increased level of APP-CTF will lead to increased production of Ab. To test whether FLZ inhibited APP-CTF production, APP/PS1 mouse hippocampus was subjected to Western blot analysis. As shown in Fig. 5A, a significant decrease of APP-CTF expression was observed in FLZ-treated APP/PS1 mice (61.97% decrease, P,0.01). APP-CTF expression was also reduced by FLZ treatment in SH-SY5Y (APPwt/swe) cells (FLZ 1 mM: 40.48% decrease, P,0.05; FLZ 10 mM: 69.05% decrease, P,0.01) (Fig. 5B). The above results indicated that the inhibitory effects of FLZ on Ab production may be through attenuating APP phosphorylation and APP-CTF production.

FLZ Decreased the Expression of BACE1
We further examined the effects of FLZ on the enzymes that contribute to the production of Ab. BACE1 is the key enzyme that is responsible for APP processing and Ab production. In the present study, it was observed that FLZ treatment markedly decreased the expression of BACE1 in APP/PS1 mouse hippocampus (53.33% decrease, P,0.01) and SH-SY5Y (APPwt/swe) cells (FLZ 1 mM: 49.62% decrease, P,0.05; FLZ 10 mM: 56.39% decrease, P,0.05) (Fig. 6).
The extracellular degradation of Ab is carried out primarily principally by insulin degrading enzyme (IDE). We then tested the effect of FLZ on IDE expression. The present results showed that FLZ did not affect the expression of IDE. These data suggested that FLZ decreased Ab level through inhibiting Ab production without affecting its degradation (Fig. 6).

FLZ Attenuated Tau Phosphorylation Through Akt/GSK3b Pathway
GSK3b has been reported to phosphorylate tau protein at several sites in vitro and in vivo [14]. Akt is one of the critical regulators of GSK3b, and the phosphorylation by Akt on the Ser9 residue of GSK3b leads to decreased enzymatic activity of GSK3b [15], then eventually affects tau phosphorylation. To study the mechanism underlies FLZ's inhibitory effect on tau phosphorylation, the effect of FLZ on Akt/GSK3b signal pathway was then investigated. The results showed that FLZ treatment significantly increased the phosphorylation of Akt at Ser473, which is an indication of its activation (mouse: P,0.01; cell: FLZ 10 mM: P,0.01). The Akt activation induced by FLZ treatment was accompanied by the de-activation of GSK3b, revealed by the increased phosphorylation of GSK3b at Ser9 in APP/PS1 mice (P,0.01) and SH-SY5Y (APPswe/wt) cells (FLZ 1 mM: P,0.05; FLZ 10 mM: P,0.01) (Fig. 8 A, B). We also found that using Ly294002, an inhibitor of phosphoinositide 3-kinases(PI3K)/Akt pathway, the phosphorylation of GSK3b at Ser9 was markedly reduced (P,0.01), so was the down-regulative effect of FLZ on tau phosphorylation (Ser396: P,0.05; Ser/Thr202/205: P,0.01; Thr231: P,0.05, Ser404: P,0.05) (Fig. 8 C, D). These results suggested that FLZ might attenuate tau phosphorylation through regulating Akt/GSK3b signal pathway. Besides AKT, there are many other enzymes reported to regulate GSK3b. The study of FLZ's effect on other enzymes involved is still undergoing.

Discussion
Our present finding indicated that FLZ treatment significantly improved memory deficits of APP/PS1 transgenic mice and reduced apoptosis of SH-SY5Y (APPwt/swe) cells. The beneficial effects of FLZ may attribute to its inhibition of Ab production mediated by decreasing APP processing, as well as attenuation of tau hyperphosphorylation through regulating Akt/GSK3b signal pathway. To find the time when transgenic mice start to show behavioral damage, we detected the memory deficits with Morris water maze test at the beginning (0 week), 5 weeks, 10 weeks, 15 weeks and 20 weeks during FLZ treatment. The results showed that FLZ started to show beneficial effect on memory deficits of APP/PS1 mice from 15th week (data not shown). After several times of test, we assume the mice have built memory of the water maze, and that's the possible reason that the escape latency of WT group mice was so obviously different from that of transgenic mice at the first day of water maze training and at 20th week of FLZ treatment, which were shown in Figure 2.
The ''amyloid cascade hypothesis'' suggests that increased production of amyloidogenic Ab peptide is a major cause of neuronal and synaptic loss in AD [16]. Drugs targeted at Ab production have been proved to show beneficial effects on neural system. Although active Ab immunotherapy was stopped because some patients developed brain inflammation, recent alternative approaches are investigated as passive immunotherapy, basing on shorter Ab immunogens that target the N-terminus without affecting the mid-region and C-terminus of Ab [17]. The passive immunotherapy and advanced Ab targeted antibody showed potential therapeutic value for AD patients in clinical trials [18]. Besides the immunotherapy, some chemical synthetic compounds were also reported to decrease Ab deposition and modify the AD progression [19]. FLZ is a new drug candidate for AD treatment, developed by our institute. Previous studies have confirmed that FLZ ameliorates learning and memory ability of AD model mice [8,9,20]. The application for using FLZ in clinical trials is being evaluated by China FDA right now. To better understand the mechanism of FLZ's neuroprotective effect, we applied FLZ on APP/PS1 transgenic mice and APP transgenic cells in the present study. The administration of FLZ lasted for 20 weeks and did not cause any changes of the mouse body weight and behavior. In our pre-clinical toxicity analysis, consecutive administration of FLZ to rats for 30 days did not show any toxicity. Our data demonstrated that FLZ attenuated Ab production both in vitro and in vivo. APP is processed by at least two distinct proteolytic pathways, the amyloidogenic and non-amyloidogenic APP processing pathway. The amyloidogenic pathway involves cleavage by BACE1, a membrane-bound aspartyl protease that generates the N terminus of the Ab peptide. It has been shown that APP phosphorylation at Thr668 facilitates BACE1 cleavage and increases Ab production [21]. Thus, BACE1 represents a key target protein in the development of new potential drugs for the treatment of AD. CTS-21166 is the first BACE1 inhibitor tested in clinical trials, which shows dose-related reduction in Ab 40 [22]. In the present study, it was showed that FLZ markedly reduced APP phosphorylation at Thr668 and BACE1 expression in vitro and in vivo, and this may be one of the mechanisms that lead to the attenuation of Ab production by FLZ. The non-amyloidogenic pathway involves a-secretase, which cleavages APP to generate a soluble fragment of APP (sAPP) that prevents Ab production and accumulation. Previous data from our laboratory indicated that FLZ reduced Ab production by stimulating nonamyloidogenic sAPP production, which was mediated by increasing a-secretase ability [23]. Combining these results, we postulated that FLZ reduced Ab production through regulating APP processing, i.e., FLZ promoted non-amyloidogenic pathway by increasing a-secretase activity and inhibited amyloidogenic pathway by decreasing BACE1 activity. Another approach to reduce Ab deposition is to elevate its degradation. FLZ did not affect Ab degradation, as we found FLZ has no effect on IDE, a major endopeptidase involves in Ab enzymatic degradation. The above results indicated that regulating Ab production might be one of the mechanisms of FLZ to treat AD.
Tau oligomer is used as a potential target for immunotherapy of AD and other tauopathies as well [24]. Compounds that inhibit tau hyperphosphorylation showed positive effects on cognitive function in clinical trials, such as AL108 [25]. Studies have shown that amyloid deposits and neurofibrillary tangles development may be related [26]. The Ab-tau relationship is supported by the facts that there is increased tau phosphorylation in mutant APP transgenic mice, besides overexpression of Ab peptides [27]. Furthermore, intracerebral injection of Ab to mutant tau transgenic mice was reported to cause elevated tau phosphorylation [28]. Consistently, inhibition of Ab accumulation markedly reduces the onset of tau pathology [29]. In the present study, we found that tau phosphorylation markedly decreased after FLZ treatment, which might at least partially be attributed to FLZ's inhibitory effect on Ab. Neurofibrillary tangle formation is regulated by several protein kinases, which cause phosphorylation at over 79 serine/threonine residues on tau [30]. In this study, the data demonstrated that FLZ attenuated tau hyperphosphorylation at several sites. We further investigated the enzymes that induced tau phosphorylation, and focused on GSK3b, which is an important kinase responsible for APP phosphorylation in neuronal cells [14,31]. Many upstream kinases, such as protein kinase A (PKA), Akt, protein kinase C (PKC) and p70 ribosomal S6 kinase (p70S6K) are known to phosphorylate Ser9 of GSK3b. Among these kinases, the activity of GSK3b is negatively regulated by Akt through phosphorylation at Ser9 epitope in neurons [15]. Our present results suggested that FLZ attenuated tau hyperphosphorylation primarily through inhibition of Akt/GSK3b pathway. There are several evidences that support this hypothesis. First, FLZ treatment increased Akt activity and inhibited GSK3b activity both in vivo and in vitro. In addition, the inhibitory effect of FLZ on GSK3b was suppressed by inhibiting Akt activity. Furthermore, inhibiting the activity of Akt suppressed FLZinduced decrease of tau phosphorylation. These data suggested that besides the indirect effect mediated by inhibiting Ab accumulation, Akt/GSK3b pathway might be the other possible mechanism involved in the inhibitory effect of FLZ on tau hyperphosphorylation. However, how the Akt/GSK3b pathway is regulated by FLZ still needs further investigation.
The focus of the present work is the mechanism of the neuroprotective effects of FLZ on AD pathology. Our study provides the first evidence that FLZ treatment decreases APP processing by BACE1 and inhibits tau hyperphosphorylation mediated by the Akt/GSK3b pathway, highlighting the potential of FLZ as a therapy agent against AD.

Supporting Information
File S1 File includes Fig. S1-S3. Figure S1. Morris water maze test of learning and memory deficits of APP/PS1 mice. (A) The latencies of mice to find the destination. (B) The number of platform crossing of mice. Results were expressed as mean 6 SD.**P,0.01 vs, WT mice, n = 10 in WT group, n = 20 in APP/ PS1 group. Figure S2. Body weight of APP/PS1 mice during Water maze test. Figure S3. Ab production in cortex of APP/PS1 mice. Immunohistochemistry of Ab deposits in cortex. Representative sections of cortex from 5 mice were shown. Results were expressed as mean 6 SD. **P,0.01 vs. WT mice; ## P,0.01 vs. APP/PS1 mice. (DOC)