Coffee polyphenols prevent cognitive dysfunction and suppress amyloid β plaques in APP/PS2 transgenic mouse

Epidemiological studies have found that habitual coffee consumption may reduce the risk of Alzheimer's disease. Coffee contains numerous phenolic compounds (coffee polyphenols) such as chlorogenic acids. However, evidence demonstrating the contribution of chlorogenic acids to the prevention of cognitive dysfunction induced by Alzheimer's disease is limited. The present study investigated the effect of chlorogenic acids on the prevention of cognitive dysfunction in APP/PS2 transgenic mouse model of Alzheimer's disease. Five-week-old APP/PS2 mice were administered a diet supplemented with coffee polyphenols daily for 5 months. The memory and cognitive function of mice was determined using the novel object recognition test, Morris water maze test, and the step-through passive avoidance test. Immunohistochemical analysis revealed that chronic treatment with coffee polyphenols prevented cognitive dysfunction and significantly reduced the amount of amyloid β (Aβ) plaques in the hippocampus. Furthermore, we determined that 5-caffeoylquinic acid, one of the primary coffee polyphenols, did not inhibit Aβ fibrillation; however, degraded Aβ fibrils. In conclusion, our results demonstrate that coffee polyphenols prevent cognitive deficits and reduce Aβ plaque deposition via disaggregation of Aβ in the APP/PS2 mouse.


Introduction
Alzheimer's disease (AD), a neurodegenerative disorder characterized by memory and cognitive dysfunction, is a public health problem worldwide. Its pathological features include the accumulation of amyloid ␤ (A␤) peptides (amyloid plaque) and aggregation of neurofibrillary tangles, which play causal and central roles in AD progression (Braak and Braak, 1997;Mucke and Selkoe, 2012). A␤ accumulation arises due to an imbalance between its synthesis and clearance. A␤ is generated by proteolytic processing that involves two types of proteases, namely, ␤and ␥-secretase (De Strooper et al., 2010). A transmembrane aspartic protease called ␤-site APP cleaving enzyme (BACE1) is responsible for the ␤-secretase activity. Although BACE1 cleaves amyloid-␤ precursor protein (APP) studies found that coffee consumption habits may reduce the risk of mild cognitive impairment and AD Solfrizzi et al., 2015). Coffee contains numerous phenolic compounds [coffee polyphenols (CPP)], such as chlorogenic acids (CGAs), and a single cup of coffee contains 70-350 mg of CGAs (Clifford, 1999). Nakamura et al. demonstrated that the plasma concentrations of unmetabolized CGA peaked 0.5 h after the ingestion of coffee (CGAs =300 mg) (Nakamura et al., 2006), suggesting that the unmetabolized form of CGAs might exert its aforementioned protective effect. Several studies have reported that CGAs promote neuronal differentiation (Ito et al., 2008) and protect against A␤-induced cell death by disaggregating the A␤ protein (Miyamae et al., 2012;Wei et al., 2016). In addition, CGAs possess antioxidant activity, thereby improving temporary amnesia in mice (Kwon et al., 2010). A few clinical studies have reported the effects of CGAs on cognitive function. Acute administration of coffee enriched with CGAs to healthy elderly volunteers had a positive effect on mood and moodrelated behavior. However, no improvement in cognitive function was reported (Cropley et al., 2012). Similar results were obtained by Camfield et al. (Camfield et al., 2013). However, recent studies have demonstrated that chronic (4-6-month) intake of CGAs improved cognitive function in the elderly (Kato et al., 2018;Saitou et al., 2018). These findings led us to hypothesize that CGAs may prevent memory and cognitive dysfunction induced by A␤. However, whether CGAs have beneficial effects on AD is unknown. APP/PS2 mice are double transgenic mice that overexpress mutant forms of human A␤ precursor protein (hAPP) and human presenilin-2 (hPS2) (Richards et al., 2003;Toda et al., 2011). Richards et al. reported that APP/PS2 mice exhibit AD-like impairments, such as amyloidosis, inflammation, impaired synaptic plasticity, and cognitive dysfunction (Richards et al., 2003).
In the present study, we used APP/PS2 mice as a model of AD to investigate whether the treatment with CPP may prevent/reduce progressive impairments in memory and cognitive function. These mice are a suitable model to investigate potential preventive and therapeutic strategies for AD because A␤ deposition is observed at 2-3 months of age. Thereafter, cognitive decline becomes apparent at 4-5 months of age (Toda et al., 2011). These behavioral and pathological changes in APP/PS2 mice are due to age-related cognitive dysfunction associated with amyloidosis.

Animal and diets
APP/PS2 double transgenic mice and wild-type (WT) littermates were generated as described by Toda et al. (Toda et al., 2011). APPswe mice expressing hAPP gene mutation were purchased from Taconic Biosciences (Hudson, NY, USA). PS2M1 mice expressing hPS2 gene mutation were obtained from Oriental Yeast Co., Ltd. (Tokyo, Japan). APP/PS2 double transgenic mice were maintained by cross-breeding APPswe male mice and PS2M1 female mice using in vitro fertilization and embryo transfer techniques. Mouse genotype was confirmed by polymerase chain reaction analysis of tail-tip DNA.
Five-week-old male APP/PS2 mice and WT littermates were individually housed in a temperature-and humidity-controlled room (23 ± 3 • C, 55 ± 15% relative humidity) with an alternate 12 h light/dark cycle (lights on at 0600 h). The mice were divided into three groups (N = 12-15 mice/group). Behavioral analyses were performed on all mice (N = 12-15 mice/group), and mice were divided into two populations on the basis of their behavioral analysis scores. One population was used for brain weight measurement and RNA extraction (N = 6-8 mice/group). The other population was used for immunohistochemical analyses (N = 6-7 mice/group). The mice were provided with ad libitum access to water and either control or CPP diet. The control diet consisted of 10% (wt/wt) fat (corn oil), 20% casein, 61.5% potato starch, 4% cellulose, 3.5% vitamins, and 1% minerals. The CPP diet consisted of the control diet supplemented with 1% CPP; this amount was referred from a prior study (Murase et al., 2011). Potato starch was reduced to 59.5% to compensate for the addition of CPP. Animals were maintained on these diets for 22 weeks. Individual body weights were recorded weekly, and food intake was measured every 3-4 days. Behavioral analyses were performed after 18 weeks. After 22 weeks, the mice were anesthetized using isoflurane (Forane ® ; Abbott, Tokyo, Japan) inhalation. The mice were then transcardially perfused with 20 ml of saline, followed by 20 ml of 4% paraformaldehyde (PFA; Wako Pure Chemical, Osaka, Japan). The perfusate was maintained at 4 • C. After perfusion, the brain was dissected out, weighed, and stored in 4% PFA at 4 • C until analysis. All animal experiments were conducted in the Experimental Animal Facility of the Kao Tochigi Institute in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. The protocols were approved by the Kao Corporation Animal Care Committee (Protocol Number: F15045-0000). All surgeries were performed under isoflurane anesthesia, and all efforts were made to minimize suffering.

Behavioral analyses
Behavioral analyses were performed after 18 weeks in the following order: novel object recognition test, Morris water maze test, and step-through passive avoidance test.

Novel object recognition test
The novel object recognition test was performed in a plastic box (22 × 32 × 13 cm 3 ). In the habituation trial on the first day, the mice were allowed to explore the empty test box for 10 min. On the following day, two objects (blocks) were placed in the test box. During the training trial, each mouse was placed in the test box and allowed to explore the objects for 10 min. Mice were then returned to their home cage. After a 2 h training trial, one of the two objects (familiar) was replaced by a new one (novel), and test trial was carried out for 5 min. All sessions were video recorded. The exploration time spent by a mouse touching the object with its nose was measured. Data were presented as the exploration time spent by a mouse touching the familiar or novel object relative to the total exploration time during the test trial {[familiar or novel / (familiar + novel)] × 100}. The discrimination index was calculated as follows: {(novel -familiar) / (familiar + novel)}.

Morris water maze test
The Morris water maze pool, which had a diameter of 148 cm, contained water (at 17-18 • C) with a platform (10 cm diameter) that was submerged 2 cm beneath the water surface. Mice were first trained (4 days, 3 sessions/day) to find the hidden platform. The platform location remained invariable during the training, however, the entry point was different for each trial. The trial ended when the mouse stayed on the platform for 30 s, or after 90 s. The platform was removed after 4 days of the training trial and the test trial was carried out for 90 s on day 5. The entry point for this trial was the opposite quadrant of the platform. The time taken for each mouse to swim to the previous quadrant of the platform was recorded. The performance was recorded and analyzed with the EthoVision XT (Version 7.0; Noldus Information Technology, Wageningen, Netherlands).

Step-through passive avoidance test
The step-through passive avoidance test chamber consisted of a light chamber (10 × 10 × 30 cm 3 ) and a dark chamber (24 × 24.5 × 30 cm 3 ), separated by a guillotine door (light and dark chamber; Nihon Bioresearch, Gifu, Japan). During the acquisition trial, each mouse was placed in the light chamber and allowed to explore freely. After 10 s, the guillotine door was opened, allowing the mouse access to the dark chamber. The door closed after the mouse entered the dark chamber, and a 0.2 mA foot shock was administered to the mouse for 3 s (Shock scrambler; UNICOM, Tokyo, Japan). The latency time for the mouse to enter the dark chamber was measured. Twenty-four hours after the acquisition trial, the mouse was placed in the light chamber for the test trial. The time taken for a mouse to enter the dark chamber was recorded.

Immunohistochemistry
To detect A␤ plaques, brain samples were fixed in 4% PFA, embedded in paraffin, and 4-m-thick sections were cut. Tissue sections were deparaffinized, treated with 90% formic acid for 5 min, and then, incubated with 0.1% hydrogen peroxide in methanol for 30 min to deactivate endogenous peroxidases. Subsequently, the tissue sections were incubated with the monoclonal anti-human A␤ antibody (#10323; Immuno-Biological Laboratories, Gunma, Japan, 1:200) at 25 • C for 1 h. The tissue sections were then incubated with HRP-conjugated streptavidin (Nichirei Biosciences, Tokyo, Japan) for 5 min, and hydrogen peroxide as substrate and diaminobenzidine (DAKO, Tokyo, Japan) was used for color development. The images were acquired with a fluorescence microscope (BZ-X710; KEYENCE, Osaka, Japan), and quantitative image analysis was determined using BZ-II application (KEYENCE). Four slices from each mouse were used to quantify the mean average value of the selected regions. The A␤ deposition in the cerebral cortex and dorsal hippocampus was analyzed as the percentage of brain regions covered by A␤ immunoreactivity.

RNA extraction and quantitative PCR (qPCR)
Total RNA was extracted using RNeasy Plus Universal Mini kit (Qiagen, Hiden, Germany). For real-time qPCR, cDNAs were synthesized with the High Capacity RNA-to-cDNA Kit (Applied Biosystems, Life Technologies, Forster City, CA, USA). qPCR assays were performed using an Applied Biosystems ViiA7 Real-time PCR system (Applied Biosystems). Commercially available polymerase chain reaction primers and FAM-labeled TaqMan probes (TaqMan Gene Expression assays; Applied Biosystems) were used for the assays. The expression of each gene was normalized to that of the gene encoding GAPDH, a housekeeping gene. The genes assessed in this study are listed in Supplemental S1 Table. 2.9. Aˇ fibrillization assays A␤ fibrillization was measured with a commercially available thioflavin T (ThT) A␤ aggregation kit (Ana Spec, San Jose, CA, USA). The A␤ fibrillization assays were performed as described in the manufacturer's procedure booklet. A␤ 1-42 peptides were dissolved in dimethyl sulfoxide (DMSO) to prepare A␤ 1-42 (2.5 mM) stock. 5-CQA (Cayman Chemical, Ann Arbor, MI, USA; 5-CQA) was dissolved in DMSO to prepare the stock of 5-CQA (1 mM), which was subsequently diluted with 50 mM Tris buffer to yield a 5-CQA (100 M) solution. A␤ 1-42 solution (Ana Spec) was diluted to a final concentration of 50 M. A␤ solution with ThT solution (200 M; Ana Spec), and 5-CQA solution (100 M) was incubated at 37 • C in a black 96-well plate. The ThT fluorescence intensity was measured every 5 min for 90 min using the EnSight plate reader (Perkin Elmer, Waltham, MA, USA) at 440 nm (ex) and 484 (em).

Aˇ disaggregation assays
A␤ 1-42 peptides (Peptide Institute, Osaka, Japan) were dissolved in DMSO to prepare A␤ 1-42 (5 mM) stock. Subsequently, the A␤ 1-42 stock was diluted with 50 mM Tris buffer to make A␤ 1-42 (250 M) solution. The A␤ 1-42 solution was aggregated following incubation at 37 • C for 7-10 days. The aggregated and/or oligomeric state of A␤ 1-42 solution was supplemented with either DMSO or 5-CQA solution (1, 10, 100 M, respectively) and incubated at 37 • C for Fluorescence of A␤ bound to ThT was measured on the EnSight plate reader (Perkin Elmer) at 440 nm (excitation) and 484 (emission).

Computational system
A␤-5-CQA interaction was analyzed using Schrödinger Suite 2018-3 (URL: https://www.schrodinger.com/suites/smallmolecule-drug-discovery-suite). The 3D structures of A␤ were built from the nuclear magnetic resonance (NMR) structure of A␤ (residues 17-42) (Protein Data Bank ID: 2BEG), which has 10 models with five chains each (Lührs et al., 2005). Protein Preparation Wizard was used to prepare the structures for all 10 models at pH 7.0. The N-termini were capped with the acetyl groups. The binding sites were analyzed for each model using Sitemap (URL: https://www.schrodinger.com/sitemap). The initial 3-dimensional ligand structure was constructed using Ligprep at pH 7.0 ± 2.0. A docking study was performed using Glide SP and the Sitemap results. The ligand binding pose with the best docking score of all the docking simulations was chosen and analyzed. The figure showing the structural interaction was prepared with PyMOL software (URL: https://www.pymol.org/).

Statistical analyses
Variables are expressed as the mean ± standard error of mean (SEM). Statistical analyses were conducted using one-way analysis of variance (ANOVA), followed by Bonferroni's post-hoc test or T-test (GraphPad Prism 6; GraphPad Software, La Jolla, CA, USA). Two-way repeated ANOVA, followed by Bonferroni's post-hoc test, was used to assess changes over time and between the groups (GraphPad Prism 6). Differences were considered significant when P < 0.05.

Chronic treatment with CPP does not alter body and brain weight in APP/PS2 mice
During the CPP treatment, general health status of APP/PS2 mice did not change significantly. The average body weights of the WT, APP/PS2, and APP/PS2 + CPP groups were 44.48 ± 0.46, 44.65 ± 0.71, and 42.67 ± 0.69 g (N = 12-15 mice/group), respectively. The brain wet weights in the WT, APP/PS2, and APP/PS2 + CPP groups were 0.4873 ± 0.003, 0.4989 ± 0.009, and 0.4986 ± 0.006 g (N = 6-8 mice/group), respectively. Neither body nor brain weights differed significantly among the groups.

Effect of CPP on the performance of novel object recognition test
The effect of CPP on recognition memory was investigated using the novel object recognition test. In the training trial, the exploration time dedicated to the two objects was similar among the three groups (data not shown).
The test trial was carried out after a 2 h training trial. In the WT and APP/PS2 + CPP groups, the percentage of the time spent exploring the novel object was observed to increase (P < 0.001, Bonferroni's post-hoc test) as compared to that with the familiar object. However, the time dedicated exploring familiar versus novel objects did not significantly differ in the APP/PS2 group ( Fig. 2A).
When the exploration time of the objects was analyzed as a function of the discrimination index, the APP/PS2 group showed a decrease (P < 0.001, Bonferroni's post-hoc test) compared with the WT group (Fig. 2B). By contrast, the discrimination index was increased in APP/PS2 mice treated with CPP (P < 0.001, Bonferroni's post-hoc test), indicating that the recognition memory improved in CPP-treated APP/PS2 mice (Fig. 2B). The total exploration time during the test trial did not differ among the three groups (Fig. 2C).

Effect of CPP on the performance of Morris water maze test
The effect of CPP on spatial learning and memory was investigated using the Morris water maze test. In the training trial, the escape latency time was significantly shorter (P < 0.05, Bonferroni's post-hoc test) in the WT group than in the APP/PS2 group (Fig. 3A). The escape latency time did not differ significantly between the WT and APP/PS2 + CPP groups (Fig. 3A). The swimming speed was the same in the three groups (Fig. 3B).
Following the 4-day training trial, test trials were performed on the fifth day. The time swimming in the platform quadrant decreased (P < 0.01, Bonferroni's post-hoc test) in the APP/PS2 group compared with the WT group (Fig. 3C), whereas it improved in the APP/PS2 + CPP group compared with the APP/PS2 group (Fig. 3C). The performance of the CPP-treated group was comparable to that of the WT group.

Effect of CPP on the performance of step-through passive avoidance test
The effect of CPP on long-term memory was investigated using the step-through passive avoidance test. In the acquisition trial, the latency time was similar in all groups (Fig. 4). In the test trial, after the 24 -h acquisition trial, the latency time was found to be significantly longer (P < 0.05, Bonferroni's post-hoc test) in the WT group than in the APP/PS2 group (Fig. 4). The tendency for improvement in the latency time (P = 0.10, Bonferroni's post-hoc test) was observed in the APP/PS2 + CPP group compared to the APP/PS2 group (Fig. 4).

Effect of CPP on Aˇ deposition in the brain
The area occupied by A␤ plaques increased in the cortex and hippocampus of the APP/PS2 group compared with the WT group (Fig. 5A), whereas it decreased in those of the APP/PS2 + CPP group compared with the APP/PS2 group (Fig. 5A). The area of A␤ plaques in the hippocampus was significantly reduced (P < 0.05, T-test) in the APP/PS2 + CPP group compared with that in the APP/PS2 group (Fig. 5B).

Effect of CPP on gene expression in the mouse brain
APP/PS2 mice showed significantly higher hippocampal and cortical mRNA expression of NOX2 and p22phox than the WT mice (Table. 1). The expression of amyloid-degrading enzymes, such as CatB and neprilysin, did not differ among the CPP-fed APP/PS2 and control mice (Table. 1). The hippocampal expression of amyloid sequestration proteins, such as transthyretin, significantly increased in APP/PS2 + CPP mice relative to APP/PS2 mice (Table. 1). The expression of proinflammatory genes, such as CD68, F4/80, IL-1␤, and IL-6, significantly increased in the hippocampus and cortex of the APP/PS2 mice than in the control mice. APP/PS2 mice exhibited significantly higher gene expression of glial cell markers, such as A1, Iba1, and GFAP, than the control mice in both the cortex or hippocampus. CPP-fed APP/PS2 mice exhibited slightly reduced expression of mouse APP mRNA in the cortex, but not in the hippocampus, relative to non-treated APP/PS2 mice (Table 1).
The expression of other genes tested did not differ between CPP-fed APP/PS2 and control mice (Table 1).

Effect of 5-CQA on Aˇ formation
The fluorescence intensity of A␤ alone increased immediately and reached a plateau after 60 min of incubation. The presence of 5-CQA, the main component of CPP, did not affect the fluorescence intensity, suggesting that 5-CQA did not inhibit A␤ fibrillization (Fig. 6A).

Interaction analysis of 5-CQA for Aˇ protofilament
The eighth model of the NMR structure was used as it yielded the best 5-CQA-docking score. The docking study showed that 5-CQA interacted with the site around residues such as Phe19, Phe20, Ala21, Val36 and Gly38 (Fig. 7).

Discussion
In general, the effectiveness of coffee on health is due to the abundant polyphenols in coffee beans typified by CGAs. In this study, we found that chronic consumption of CPP improved memory and cognitive function and reduced A␤ pathology in APP/PS2 mice. Moreover, we demonstrated that CPP may modulate AD phenotypes by promoting disaggregation of fibrillar A␤ species into A␤ peptide in the brain.
The CPP-treated mice showed improvements in memory and cognitive function as assessed by three behavioral analyses. The novel object recognition test is based on the spontaneous tendency of rodents to explore novel objects without reward or penalty. This test measures visual recognition memory and the spatial and temporal context of object recognition, supported by the interactions between the cortex and the hippocampus (Grayson et al., 2015). The Morris water maze test investigates spatial learning and memory. This test involves the hippocampus, cerebral cortex, and striatum (D'Hooge and De Deyn, 2001). The step-through passive avoidance test evaluates long-term memory of learned avoidance behavior by electric foot shock. These behavioral analyses may reflect not only cognitive function, but also on motivation or motor function (Webster et al., 2013). Improvements in the performance of learning and memory tasks in CPP-treated mice are unlikely to be due to the changes in activity, motivation, or motor function. The total object exploration time in the novel object recognition test, the swimming speed in the Morris water maze test, and the latency time during acquisition trial in the step-through passive avoidance test in CPP-treated mice did not differ from those in CPP-non-treated mice. These results suggest that exploratory activity and motor function, that is, the swimming ability, did not change. Therefore, the improved behavioral performance in CPPtreated mice is probably due to enhanced learning and memory. Several studies have suggested that hippocampal damage causes impaired performance in behavioral tests (D'Hooge and De Deyn, 2001;Mumby, 2001;Shamsaei et al., 2015). The CPP-treated mice showed significantly decreased A␤ plaque in the hippocampus than the CPP-non-treated mice. These findings suggest that the CPP confers protection against A␤ toxicity, particularly in the hippocampus, and the performance in the three behavioral tests. The immunohistochemical analysis conducted in the present study revealed that dietary CPP significantly decreased the amount of A␤ plaques in the hippocampus of APP/PS2 mice without changing A␤ processing-related gene expression. To study the mechanism underlying the CPP-induced reduction of A␤ deposition, we investigated the effects of CPP components on the aggregation and disaggregation of A␤-protein.
structure of the A␤ protofibril model. The catechol of 5-CQA was found to be important for this interaction. Ciaramelli et al. demonstrated that 5-CQAs are A␤ oligomer ligands and that their ability to inhibit fibrillation and neurotoxicity depends on their direct interaction with A␤ protofibrils (Ciaramelli et al., 2018.). Curcumin, a substructure similar to 5-CQA, also features the ability to degrade A␤: the catechol of curcumin interacts with the residue 12 and 17-21 of A␤ fibrils (Masuda et al., 2011). Together with these previous findings, our docking calculation analysis and ThT assay indicate that 5-CQA promotes the disaggregation of fibril A␤. Ono et al. reported that wine polyphenols (myricetin, morin, quercetin, kaempferol, (+)-catechin and (−)-epicatechin), cur-cumin, rosmarinic acid, and grape seed-derived polyphenols with phenolic groups like CGA disaggregate and destabilize the A␤ fibrils and reduce its cytotoxicity (Ono et al., 2003(Ono et al., , 2004Ono et al., 2008). Wei et al. reported that CGA has a neuroprotective effect on the cytotoxicity of A␤ (Wei et al., 2016). These studies suggest that CGA may disaggregate A␤ fibrils and could possibly reduce cytotoxicity. Conventionally, A␤ fibrils that accumulate as cerebral amyloid are thought to exert neurotoxicity. However, in recent years, oligomers, the intermediate stages of amyloid ␤ protein aggregation, have been reported to be the most toxic, and oligomer reduction was suggested as effective for treating AD (Ono et al., 2009). Additionally, it has been reported that polyphenols block A␤ oligomerization The time course of changes in the fluorescence intensity was measured. Fluorescence intensity is expressed as A␤ fibrillization. (B) The aggregation and/or oligomerization state of A␤1-42 (25 M) was incubated with either DMSO or 5-CQA (1, 10, 100 M) for 7 days before the assay. Fluorescence intensity represents A␤ disaggregation. Values are the mean ± SEM of N = 5-8. *: P < 0.05, **: P < 0.01, vs. without 5-CQA (Bonferroni's post-hoc test). and reduce the cellular toxicity and synaptic dysfunction of the A␤ oligomers Ono et al., 2012;Wang et al., 2014). We measured A␤ oligomer levels in the mice hippocampus using an immuno-assay and found them to have significantly increased in the APP/PS2 group relative to the WT group; however, they did not differ between the APP/PS2 + CPP and APP/PS2 groups (data not shown). Our findings indicate that CGAs did not affect A␤ oligomerization, despite interacting with and degrading A␤ protofibrils. Further, CGAs decreased A␤ plaque deposition in the hippocampus without an observable change in the amount of A␤ oligomers. These results inform our speculation that CGAs may not only degrade A␤ protofibrils but may also exert a clearance effect: Our preliminary data evinced significant A␤-clearance mediated by an increase of LRP-1 mRNA expression in the hippocampi of CPP-fed mice. Further research on the influence of the degradation of oligomers and the clearance of amyloid ␤ protein in APP/PS2 mice is therefore warranted. As aforementioned, other polyphenols, such as curcumin and grape seed-derived polyphenols, reportedly reduce A␤ toxicity and improve cognitive dysfunction. Additionally, cognitive dysfunction in AD involves not only A␤ but also tau; therefore, Fig. 7. The docking simulation of the predicted model of 5-CQA binding with the A␤ protofilament. The A␤ NMR structure (Protein Data Bank ID: 2BEG) permitted the docking of 5-CQA. 5-CQA is displayed as sticks, and the A␤ residues within 4Å from 5-CQA are depicted as lines. Carbon atoms from the A␤ structure are shown in gray, and the docking pose that achieved the best docking score is shown in green. Red, blue, and white represent oxygen, nitrogen, and hydrogen atoms, respectively. Yellow dashed lines represent hydrogen bonds between A␤ and 5-CQA (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).
it is necessary to consider the effects of polyphenols on tau. Ho et al. reported that grape seed-derived polyphenols disaggregates tau peptide and dissociate tau peptide aggregates (Ho et al., 2009). It is important to consider the pharmacokinetics of polyphenols, including its blood-brain barrier (BBB) permeability for polyphenols to be effective in the brain. For example, the polyphenol proanthocyanidin constitutes monomeric, oligomeric, and polymeric forms, and the monomeric metabolites have been reported to selectively reach and accumulate in the brain and exert their effects (Wang et al., 2012). However, the present study was unable to reveal the differential effects of CPP and other polyphenols on A␤ or tau, and their pharmacokinetics including BBB permeability. Thus, the mechanism underlying CPP action should also be explored by future research.
Transthyretin has been shown to interact with A␤ peptides and suppress A␤-oligomer toxicity (Buxbaum et al., 2008.). Further, serum concentrations of transthyretin reportedly decrease as AD progresses (Uchida et al., 2015.). Hop-derived components that prevent cognitive impairment of AD-model mice increase the expression of transthyretin in the hippocampus (Fukuda et al., 2018.). CPP may have decreased the toxicity of A␤ by increasing the expression of transthyretin, possibly helping to prevent cognitive impairment thereby.
A␤ deposition is observed in APP/PS2 mice from 2 to 3 months of age and it increases with time; cognitive function decreases after 4-5 months of age (Richards et al., 2003;Toda et al., 2011). Fontana et al. reported that synaptic excitability of the hippocampus changes from 3 months of age in APP/PS2 mice (Fontana et al., 2017). In addition, the increase in A␤ deposition in humans begins 15-20 years before AD onset (Bateman et al., 2012). Therefore, it is important to protect the brain from disorders caused by A␤ and prevent the onset of AD through early intervention. The clearance of A␤ in the brain was insufficient to improve cognitive dysfunction in AD (Salloway et al., 2014). In this study, CGA was given to 5-weekold mice before the changes in A␤ deposition, cognitive function, and synaptic function occurred, and therefore, the preventive effect of CGA on AD onset was clearly demonstrated. Further studies are needed to precisely analyze the mechanism underlying the effect of CGA on brain functions and its effectiveness on existing A␤ deposits and cognitive decline. CGA not only reduces A␤ deposition, but also alleviates the inhibitory actions on antioxidant and AChE activity (Kwon et al., 2010). Thus, these findings suggest that CGA can improve cognitive dysfunction through several pathways and may exert a therapeutic effect in AD.

Conclusion
We demonstrated that CPP treatment significantly attenuated impairments in the decline of recognition and spatial memory in a mouse model of AD. The reduction of A␤ plaque deposition in the hippocampus may account for these findings. Consequently, CPP provided a protective effect against AD progression through A␤ in mice. These findings strongly suggest CPP as an effective therapeutic and prophylactic agent for cognitive deficits in AD.