Acute inhibition of AMPA receptors by perampanel reduces amyloid β‐protein levels by suppressing β‐cleavage of APP in Alzheimer's disease models

Hippocampal hyperexcitability is a promising therapeutic target to prevent Aβ deposition in AD since enhanced neuronal activity promotes presynaptic Aβ production and release. This article highlights the potential application of perampanel (PER), an AMPA receptor (AMPAR) antagonist approved for partial seizures, as a therapeutic agent for AD. Using transgenic AD mice combined with in vivo brain microdialysis and primary neurons under oligomeric Aβ‐evoked neuronal hyperexcitability, the acute effects of PER on Aβ metabolism were investigated. A single oral administration of PER rapidly decreased ISF Aβ40 and Aβ42 levels in the hippocampus of J20, APP transgenic mice, without affecting the Aβ40/Aβ42 ratio; 5 mg/kg PER resulted in declines of 20% and 31%, respectively. Moreover, PER‐treated J20 manifested a marked decrease in hippocampal APP βCTF levels with increased FL‐APP levels. Consistently, acute treatment of PER reduced sAPPβ levels, a direct byproduct of β‐cleavage of APP, released to the medium in primary neuronal cultures under oligomeric Aβ‐induced neuronal hyperexcitability. To further evaluate the effect of PER on ISF Aβ clearance, a γ‐secretase inhibitor was administered to J20 1 h after PER treatment. PER did not influence the elimination of ISF Aβ, indicating that the acute effect of PER is predominantly on Aβ production. In conclusion, acute treatment of PER reduces Aβ production by suppressing β‐cleavage of amyloid‐β precursor protein effectively, indicating a potential effect of PER against Aβ pathology in AD.


| INTRODUCTION
Alzheimer's disease (AD) is a multifactorial neurodegenerative disorder, with several mechanisms contributing to its etiology.The key initiator of AD pathogenesis is amyloid β-protein (Aβ). 1 The Aβ cascade hypothesis suggests that Aβ aggregation plays a pivotal role in the pathological cascade of AD, leading to the formation of amyloid plaques, neurofibrillary tangles, and ultimately neurodegeneration. 2 This hypothesis has guided the development of various therapeutic strategies targeting Aβ to mitigate disease pathogenesis and progression.Aβ is derived from the sequential proteolytic cleavage of amyloid-β precursor protein (APP) by βand γ-secretases. 3,4The causal mutations identified in familial AD are located in the gene encoding presenilin or APP, leading to increased Aβ production and aggregation.Additionally, a rare protective mutation against sporadic AD, an APP A673T variant, has been reported to decrease Aβ burden in the brain and preserve cognitive functions, leading further support to the Aβ cascade hypothesis. 5ecent studies shed light on hippocampal network hyperexcitability as an early event in the pathophysiology of AD.][8][9][10] Detailed studies using electroencephalography (EEG) revealed a higher prevalence of epileptiform activity in early AD patients that was profoundly associated with the onset or progression of cognitive decline than expected. 11,12As silent hippocampal seizures and epileptiform spikes were detected by using only intracranial electrodes placed adjacent to the mesial temporal lobe (called foramen ovale electrode) in two AD patients without a history of epilepsy, 13 it could be one reasonable explanation for the occult epileptiform activity that, in early AD patients, neuronal hyperexcitability can locally occur in the hippocampus where epileptiform discharges are challenging to be detected on routine scalp EEG.5][16] However, AEDs may be inappropriate for AD under certain conditions.Some AEDs and high-dose LEV were ineffective in improving cognitive function, and phenytoin and pregabalin aggravated epileptic discharges in an AD mouse model. 17,18urther, multiple AEDs reportedly can deteriorate various cognitive disorders in a type and/or dose-dependent fashion, [19][20][21][22][23][24][25][26] suggesting the importance of determining the appropriate type and dose of AED in view of therapeutic development against AD based on the "neuronal hyperexcitability hypothesis." ][32][33] While AMPARs predominantly mediate the fast excitatory transmission in the brain, direct stimulation with AMPA increases non-amyloidogenic processing of APP and inhibits Aβ production in wild-type primary neurons.Sustained stimulation with AMPA also promotes Aβ clearance in APP-transgenic mice. 34However, few reports have investigated how suppressing excessive AMPAR excitability affects Aβ metabolism.In the present study, we focused on perampanel (PER), a selective noncompetitive AMPAR antagonist, as a potential therapeutic agent against AD. 35,36xcitatory neurons upregulating AMPARs contribute substantially to epileptogenicity, and a low inhibition of AMPARs is sufficient for seizure protection in patients with epilepsy, [37][38][39][40] supporting the outstanding contribution of AMPAR in generating epileptogenicity.The previous study has shown that AMPAR modulation by PER reduces neuronal hyperactivity and synchronization while it does not affect physiological synaptic transmission. 41ER is well tolerated in elderly people aged ≥65 years, whereas low-dose PER showed high efficacy in treating elderly-onset epilepsy concomitant with AD. 42,43 Surprisingly, they also described that in 80% of 48 patients, conventional-dose PER improved cognitive function and seizure control (reported in Japanese).44 The narrow-spectrum AED targeting only excitatory neurons could be appropriate for suppressing aberrant neuronal activities without inhibiting regular neuronal transmission.However, reports on the use of PER in AD patients are limited, and a potential effect of PER on AD pathophysiology has not yet been fully elucidated at the molecular levels.In the present study, we aimed to investigate whether the acute treatment of PER could affect hippocampal Aβ metabolism in APP transgenic mice using an in vivo brain microdialysis technique.

| Animals
All procedures were approved by the Institutional Animal Care Committee of the Kyoto University Graduate School of Medicine in accordance with the guidelines for animal experimentation from the ethical committee of Kyoto University and with the National Research Council's Guide for the Care and Use of Laboratory Animals.We used PDGFB-APPSwInd transgenic (Tg) mice (J20) expressing a mutant form of the human APP bearing both the Swedish (K670N/M671L) and Indiana (V717F) mutations (APP Sw/Ind ) under the control of the PDGFB promoter in a C57BL/6J (B6) genetic background (Jackson Laboratory) as an AD model and the littermates as a control.All animals were derived from the same colony.Altogether, 52 J20 mice (2-4-monthold; n = 45, 9-11-month-old; n = 7) were implanted with microdialysis probes; B6 (n = 6) and J20 (n = 4) mice were used for the blood test; and J20 (n = 21) and B6 (n = 1) mice were used for the biochemical analysis of the hippocampus and synaptoneurosomes (SNSs).ICR mice were purchased from Shimizu Laboratory (Kyoto, Japan) to perform cortical primary neuronal cultures.

| Preparation of Aβ oligomers
Monomeric Aβ 42 was synthesized from the O-acyl isopeptide (Peptide Institute, Osaka, Japan).The O-acyl isopeptide was diluted with 14 mM of HCl to a concentration of 30 μM and neutralized with 30 mM of NaOH.The medium was added to achieve a final concentration of Aβ 42 of 20 μM.The adjusted monomeric Aβ 42 was incubated at 37°C for 48 h to obtain oligomeric Aβ (AβO).The AβO solution was used in experiments immediately after preparation.

| Primary cultures of mouse cortical neurons and induction of AβO-evoked neuronal hyperexcitability
Primary neuronal cultures were obtained from 15-day-old embryos of wild-type ICR mice.Cortical neurons were extracted using a Papain Dissociation System (Worthington Biochemical Corporation, NZ, USA) and plated onto tissue culture 6 well plates (Gibco, WA, USA) coated with polyethylene imine at a density of 6 × 10 6 .Cells were cultured for 28 days in vitro (DIV) in Neurobasal™ Plus Medium with B-27 Plus Supplement (Gibco, MA, USA) and a 1% penicillin-streptomycin solution mixture in a 37°C, 5% CO 2 incubator.To establish AβO-evoked neuronal hyperexcitability as in an vitro AD model, 0.5 μM AβO and 1 μM TTX were simultaneously applied to primary neuronal cultures at 28 DIV for 24 h, according to a previous report. 45Subsequently, the culture media were treated with fresh media containing 30 μM L-glutamic acid, and either 2 μM PER or the vehicle were replaced after a gentle wash with PBS.After 70 min, the conditioned media were collected, and the neuronal cells were harvested for biochemical analysis.

| In vivo Aβ microdialysis
For the continuous interstitial fluid (ISF) collection, we used an in vivo microdialysis technique, as described previously.J20 mice were anesthetized with isoflurane, and guide cannulas (BR Intracerebral Guide Cannula; BASi, Indiana, USA) were stereotaxically implanted above the left hippocampus (bregma −3.1 mm, 2.5 mm lateral to the midline, and 1.2 mm ventral to dura at a 12° angle).The cannulas were mounted with dental cement.One day after the implantation of the guide cannulas, microdialysis probes (BR-2 2 mm 38 kDa MWCO polyacrylonitrile membrane, BASi) were inserted into the hippocampus through the guide cannula while perfused with artificial cerebrospinal fluid (aCSF: 122 mM NaCl, 1.3 mM CaCl2, 1.2 mM MgSO4, 3.0 mM KCL, 4 mM KH2PO4, and 25 mM NaHCO3, pH 7.35) with 0.15% bovine serum albumin (BSA; Sigma-Aldrich) at a rate of 1.0 μL/min for 13 h before starting the sampling ISF.The samples were maintained at 4°C in a fraction collection and stored until analysis at −80°C.After sampling ISF, animals were perfused with 0.2% Evans blue in aCSF for 90 min, and the brains were rapidly detected to confirm the accuracy of probe insertion into the hippocampus.
While sampling hippocampal ISF every 90 min, J20 mice (2-4-month-old; n = 10, male 5, female 5) were orally administered with either PER (2 and 5 mg/kg) or vehicle using flexible plastic feeding tubes after the basal sampling of ISF for three-time points during 270 min.ISF was collected at 12 h after the treatment.The values of either ISF Aβ 40 or Aβ 42 were plotted as the percentage to the basal levels and Aβ 40 /Aβ 42 ratio.
2.6 | ISF Aβ elimination half-life J20 mice (2-4-month-old; n = 4) were injected with LY411575 subcutaneously at a dose of 3 mg/kg at 1 h after the oral gavage of 5 mg/kg PER, while hippocampal ISF was collected every 60 min.ISF was collected at 6 h after administering LY411575.ISF Aβ 40 levels were plotted as the percentage to those at a time point before the treatment.The elimination half-life of Aβ 40 was calculated from the slope of the ISF levels and time curve by plotting on a semilogarithmic graph. 46

| Isolation of synaptoneurosomes (SNSs) from the hemicortex dissected after PER administration
SNSs were isolated as described previously 47,48 with minor modifications.The brains were rapidly collected from J20 mice sacrificed at 5 h after the oral administration of 5 mg/kg PER (n = 6, male 3, female 3) or vehicle (n = 6, male 3, female 3).Data obtained from a single female mouse in the vehicle group was excluded after analysis because of a genotyping error.The hemicortex was homogenized in the 1.5 mL SNS buffer (25 mM Tris-HCL, 120 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 2 mM DTT, pH 7.5) with 1% (v/v) PPI and disrupted by the Digital Homogenizer HK-1 (ASONE, Osaka, Japan) at 5000 rpm for 10 min.The sample was loaded into a 2 mL leur-lock syringe (HENKE-Sass Wolf, Tuttlingen, Germany) and filtered through two layers of 80 μm pore nylon filters held in a filter holder (Merk, Darmstadt, Hessen, Germany).The 100 μL filtrate was diluted in 100 μL DDW with 3% SDS, 4 mM DTT, and 2% (v/v) PPI.The obtained sample was the total homogenate (TH) and was stored at −80°C.The rest filtrate was again loaded into a 2 mL leur-lock syringe and filtered through a 5 μm pore Supor® Membrane (PALL, New York, USA).The filtrate was then centrifuged at 1000× g for 10 min.The pellet corresponded to the SNS fraction and was stored at −80°C after the resuspension in 50 mM Tris-HCl buffer with 1.5% (v/v) SDS, 2 mM DDT, and 1% (v/v) PPI.

| sAPPα and sAPPβ ELISA
The levels of endogenous sAPPα and sAPPβ in the conditioned media of primary neuronal cultures were quantified by Mouse sAPPα Assay Kit (IBL, Gunma, Japan) and Mouse sAPPβ-w Assay Kit (IBL, Gunma, Japan), respectively.

| BCA assay
The protein quantification of cell lysates of primary neurons and hippocampal homogenates extracted from J20 mice after the oral administration of PER (5 mg/kg) was performed by Pierce™ BCA Protein Assay kit (Thermo Fisher Scientific).The absorbance at 562 nm was measured using a Synergy™H4 Hybrid Microplate Reader.

| Cell death assay in primary neuronal cultures
To assess cytotoxicity of the cultured primary neurons, the levels of lactate dehydrogenase (LDH) released into the culture media through compromised cell membranes were measured using a WST assay kit (Dojindo Molecular technologies, Kumamoto, Japan) according to the manufacturer's instructions.
The samples were diluted with 4× LDS sample buffer (Invitrogen, MA, USA) containing 10× reducing agent (Invitrogen) and denatured at 95°C for 10 min.The samples at 20 μg protein levels were applied to 4%-10% NuPAGE Bis-Tris gel NP0335BOX (Thermo Fisher Scientific).After electrophoresis, the proteins were transferred to Novex PVDF membranes (Invitrogen).The target proteins were applied using Chemi-Lumi One Super (Nacalai tesque, Kyoto, Japan) and detected using an ImageQuant LAS500 system (GE Healthcare Bio-Sciences, Uppsala, Sweden).The band intensities were measured using ImageJ (National Institute of Health, USA) and normalized to β-actin.

| Sample preparation for liquid chromatograph-mass spectrometer (LC-MS)/MS to assess the plasma concentration of PER
Blood samples (80-100 μL/time point) were collected from the facial vein of B6 mice at 1, 3, 6, and 9 h after the oral gavage of PER at a dose of 2 (n = 4, male 2, female 2) or 5 (n = 4, male 2, female 2) mg/kg.Blood samples of J20 mice (n = 4, male 1, female 3) were collected at 1 and 5 h after the administration of 5 mg/kg PER.The obtained blood samples were immediately centrifuged at 12 000 rpm for 10 min, and the supernatants were assayed by LC-MS/MS.PER dilution in DMSO (10 mg/ mL) was dissolved in methanol to prepare the stock solutions (10 μg/mL).The standard working solution for the calibration curve was prepared by diluting the stock solutions with plasma at PER doses of 0, 0.005, 0.01, 0.05, 0.1, 0.5, 1, and 1.5 μg/mL.Twenty-five μl of the standard working solution or the plasma samples were added with 150 μL of methanol and centrifuged at 15 500× g for 10 min.The supernatants were applied to 0.22 μm filter units (#SLGV004SL; Merk).

| LC-MS/MS conditions
The samples were analyzed using a triple quadrupole LC-MS (LC-MS 8040 Shimadzu Corporation, Kyoto, Japan) system on an Acquity UPLC BEH C18 column (1.7 μm 50 × 2.1 mm; Waters, Milford, MA, USA), as previously reported with some modifications. 49The MS/MS data acquisition was performed under Multiple Reaction Monitoring in positive electrospray ionization mode.The mobile phases comprised eluents A (10 mM ammonium acetate, 0.1% (v/v) formic acid in the water) and B (methanol).The flow rate was 0.4 mL/min and the total run time was 5 min with the following gradient program: 50% eluent B (0 min), 50% (0.5 min), 65% (2 min), 95% (2.1 min), 95% (4 min), and 50% (4.1 min).The total volume was 2 μL.The ion spray voltage was set at +1000 V.The curtain gas pressures were set at 20 U and the collision gas at medium.The ion source gas pressures were set at 45 and 40 U, with temperatures at 400°C.The precursor/product ion transition for PER was m/z 350.0/219.1.

| Statistical analysis
The statistical data were calculated using GraphPad Prism 8 software (Dotmatics, Boston, MA, USA).Student's t-test, two-way repeated measure analysis of variance (ANOVA) (with the Greenhouse-Geisser correction and Huynh-Feldt correction) with posthoc-Bonferroni, and analysis of covariance (ANCOVA) were used.All data are shown as means ± SEM, and the p value <.05 indicated a significant difference.

| ISF levels of monomeric Aβ 42 rather than Aβ 40 are preferentially decreased in old J20 mice with Aβ deposition
In J20 mice, Aβ oligomers in the brain significantly elevate by 2-3 months, with plaques appearing after 7 months of age. 50,51Our laboratory observations also confirm the presence of Aβ plaques in J20 mice after 7 months of age. 52hese mice exhibit synaptotoxicity and neuronal loss prior to plaque formation, with memory deficits manifesting by 4 months as assessed by the radial arm maze and Morris water maze. 51,53,54Neuronal loss in J20 mice begins around 6 weeks, particularly in the CA1 region of the hippocampus. 51][57][58] Hippocampal ISF samples before and after forming Aβ plaques were collected using in vivo microdialysis and then subjected to ELISA to determine the monomeric Aβ 40 and Aβ 42 levels, respectively.Soluble Aβ species, especially Aβ 42 , reportedly decline in ISF as a result of their sequestration into amyloid plaques. 59Consistently, the ISF Aβ 42 levels in 9-11-month-old J20 mice with Aβ deposition were significantly decreased by 51% as compared to 2-4-month-old J20 mice without Aβ deposition (p = 0.0083), whereas a difference in ISF Aβ 40 levels between the two groups was not statistically significant (Figure 1A).Indeed, the Aβ 40 /Aβ 42 ratio was 92% higher in 9-11-month-old mice than in 2-4-monthold mice (Figure 1B).In the present study, therefore, we decided to use 2-4-month-old J20 mice without Aβ deposition as an appropriate model to evaluate the effect of PER on ISF Aβ metabolism before amyloid plaque formation.

F I G U R E 2
A single oral administration of PER lowers ISF levels of Aβ 42 and Aβ 40 in a dose-dependent manner under sufficient plasm concentration.(A) 2 mg/kg or 5 mg/kg PER is orally administered to B6 mice (n = 3 per group) and 5 mg/kg PER to J20 mice (n = 4).The mean plasma concentrations of PER rapidly rise and maintain sufficient values for 9 h in B6 mice.Plasma concentrations at 1 or 5 h after the oral administration of PER (5 mg/kg) to J20 mice correspond to the plasma concentration profile obtained from the B6 mice.(B) Semi-log plot of mean plasma concentration of PER vs. time after 2 mg/kg administration to B6 mice (n = 3).The elimination of plasma PER follows first-order kinetics (y = −0.14x− 0.29, R 2 = .99).(C, D) Either 2 mg/kg PER, 5 mg/kg PER or vehicle is orally administrated to 2-3-month-old J20 mice (n = 10 per group).ISF is collected every 1.5 h.The horizontal axis indicates the time since the treatment.For example, Aβ levels of ISF collected after oral administration up to 1.5 h are plotted at "0" h.ISF Aβ levels from 4.5 to 6 h after the treatment are shown in bar charts.ISF Aβ 40 levels declined rapidly after a PER administration in a dose-dependent manner (p = .025,two-way ANOVA for repeated measures, Bonferroni post hoc test).After 4.5 to 6 h of the PER treatment, the decrease in ISF Aβ 40 levels is 15% at 2 mg/kg and 20% at 5 mg/ kg (C).ISF Aβ 42 levels declined rapidly after a PER administration in a dose-dependent manner (p = .040,two-way ANOVA for repeated measures, Bonferroni post hoc test).After 4.5 to 6 h of the PER treatment, the decrease in ISF Aβ 42 levels is 21% at the dose of 2 mg/kg and 31% at 5 mg/kg (D).(E) PER at the dose of 2 mg/kg or 5 mg/kg does not alter the ratio of ISF Aβ 40 /Aβ 42 (p = .41,two-way ANOVA for repeated measures).All data are plotted as mean ± SEM. n.s.denotes not statistically significant (p ≧ .05),*p < .05.

| The plasma concentrations of PER rapidly rise and maintain sufficient values for 9 h after a single oral administration of PER
The half-life of PER after oral administration in mice has yet to be determined.We first assessed the plasma concentration of PER in B6 littermates to determine the kinetics of oral absorption of PER in mice after a single oral administration at doses of 2 and 5 mg/kg.The mean plasma concentration-time curve of PER in B6 mice is shown in Figure 2A.The peak levels were plotted at 1 h after 2 mg/kg PER administration (0.64 μg/mL) and at 3 h after 5 mg/kg PER administration (1.0 μg/mL) in B6 mice.The plasma levels decreased gradually from the peak and were as much as 0.48 μg/mL at 9 h after 5 mg/kg treatment, compared with 0.20 μg/mL after 2 mg/kg PER administration in B6 mice.According to a study in human patients, the mean serum PER concentration in patients who achieved ≥50% reduction of seizure frequency was reported to be 0.450 μg/mL (range: 0.085-1.5 μg/mL). 60ER was reported to reduce excitatory postsynaptic field potentials with an IC50 of 0.23 μM (0.80 μg/mL) and an entire block at 3 μM (1 μg/mL) in mice, indicating that 2 or 5 mg/kg of PER should be a sufficient dose to regulate excessive neuronal activities.The elimination rate of PER in plasma followed first-order kinetics (y = −0.14× −0.29, R 2 = .99).The half-life of PER in the plasma was 4.9 h (Figure 2B).The plasma levels in J20 mice after 5 mg/kg PER treatment were approximately consistent with the expected values in the case of B6 mice (1.03 μg/ mL at 1 h and 0.675 g/mL at 5 h).We thus concluded that the plasma concentration of PER in J20 mice displays a pattern similar to that in B6 mice (data not shown in the figure).Within 30 min after administering PER, all mice exhibited ataxia-like motor impairment.This condition gradually improved as the plasma levels of PER decreased.

| A single oral administration of PER lowers ISF levels of Aβ 42 as well as Aβ 40 in a dose-dependent manner
To evaluate the acute effect of the noncompetitive AMPAR inhibition on Aβ metabolism, the ISF Aβ 40 and Aβ 42 levels were assessed after a single oral administration of PER at doses of 2 and 5 mg/kg or vehicle (n = 10 in each group).All animals treated by PER at a dose of either 2 or 5 mg/ kg rapidly exhibited mild motor impairment but they had gradually recovered in several hours.The motor effect seems to be necessary for the protective effects against aberrant neuronal activities because it has been reported that there was little or no separation between the PER doses that are effective in reducing seizures and doses causing motor impairment. 36The ISF samples collected every 1.5 h before and after the treatment were subject to ELISA for Aβ measurement.The Aβ levels to the basal levels, an average of three-time points before the treatment, were plotted (Figure 2C,D).
PER administration rapidly and significantly lowered both the ISF Aβ 40 and Aβ 42 levels in a dose-dependent manner, as compared with the vehicle.The ISF Aβ 40 levels were decreased by 15% at maximum after 4.5-6 h of 2 mg/ kg PER administration, while a 20% decrement was seen following 5 mg/kg PER treatment (Figure 2C).A similar trend can be detected in the ISF Aβ 42 levels; ISF Aβ 42 levels were decreased by 21% and 31% after 4.5-6 h of 2 and 5 mg/kg PER treatments, respectively (Figure 2D).Next, to determine whether PER differently affects the metabolism of Aβ 40 and Aβ 42 in the ISF, we evaluated the ratio of ISF Aβ 40 to Aβ 42 .As shown in Figure 2E, we observed no significant changes in the Aβ 40 /Aβ 42 ratio before and after PER administration, suggesting that PER can regulate the ISF Aβ levels.

| PER treatment significantly reduces the β-secretase-cleaved C-terminal fragments of APP Sw/Ind (βCTF Sw/Ind ) levels and increases the full-length APP Sw/Ind (FL-APP Sw/Ind ) levels
The ISF Aβ levels should be determined by the balance of production, release into the extracellular milieu, and clearance from the ISF.To evaluate the effect of PER on amyloidogenic APP processing, we determined the FL-APP Sw/Ind and βCTF Sw/Ind levels in the hippocampal homogenates of J20 mice after the oral administration of either 5 mg/kg PER or vehicle by Western blotting (Figure 3A).We corrected the hippocampi at 5 h after F I G U R E 3 PER treatment significantly reduces the levels of β-secretase-cleaved C-terminal fragments of APP Sw/Ind (βCTF Sw/Ind ) and increases the full-length APP Sw/Ind (FL-APP Sw/Ind ) levels in the hippocampus.Aβ levels in the hippocampus homogenates, intended to represent the hippocampal Aβ pools excluding extracellular soluble Aβ, are determined by ELISA and remain unaffected by PER administration.(A) FL-APP Sw/Ind and βCTF Sw/Ind in the hippocampus are proved with 6E10 antibody.The βCTF Sw/Ind levels decrease by 46% in the PER group (p = .0021,t-test), while the FL-APP Sw/Ind levels increase by 55% (p = .029,t-test).Note that the ratio of βCTF Sw/Ind to FL-APP Sw/Ind is significantly decreased in the PER group (p = .023,t-test).(B) Total BACE1 levels in the hippocampus are unaffected by PER treatment (p = .administration because the ISF Aβ levels were estimated to decrease at most, as shown in Figure 2C,D.Western blotting analysis revealed that the levels of βCTF Sw/Ind were significantly decreased in the PER group, while the FL-APP Sw/Ind levels were increased (Figure 3A).Note that the ratio of βCTF Sw/Ind to FL-APP Sw/Ind was significantly decreased in the PER group, indicating that acute treatment of PER clearly inhibits β-cleavage of FL-APP Sw/Ind , a preferable substrate for BACE1; meanwhile, the BACE1 levels were comparable the vehicle and PER groups (Figure 3B).As shown in Figures 2C,D and 3A, our results suggest that acute treatment of PER reduces ISF Aβ levels in vivo through downregulation of β-cleavage rather than γ-cleavage in amyloidogenic processing.

| Acute administration of PER did not influence Aβ pools in the hippocampus, except for extracellular soluble Aβ
In the process of preparing the hippocampal homogenates, as detailed in the Materials and Methods section, hippocampi were sonicated in surfactant-free buffer.The pellets from the ultracentrifugation were then dissolved in RIPA buffer.As such, Aβ levels in the hippocampal homogenates should represent the entirety of hippocampal Aβ pools, excluding extracellular soluble Aβ.These homogenates were treated with 0.5 M guanidine hydrochloride (Nacalai tesque, Kyoto, Japan), to facilitate the solubilization and monomerization of Aβ species.Aβ 40 and Aβ 42 levels were subsequently measured by ELISA.Notably, neither the levels of Aβ 40 and Aβ 42 nor their ratio in the hippocampal homogenates were affected by acute PER treatment (Figure 3C).This result indicates that the decline of ISF Aβ by PER is not due to their sequestration into other Aβ pools.

| PER administration does not affect the half-life of ISF Aβ levels
Aβ is derived from the sequential proteolytic cleavage of APP by βand γ-secretases.To evaluate the alternation of Aβ clearance from ISF by PER treatment, the ISF samples were collected hourly using the in vivo microdialysis technique from the time of the subcutaneous injection of 3 mg/kg LY411575, a potent γ-secretase inhibitor (GSI), following oral administration of either 5 mg/kg PER or vehicle an hour earlier (Figure 4A).LY411575 can almost completely inhibit Aβ production, and we can only evaluate the Aβ clearance and estimate the ISF Aβ half-life periods by rapidly inhibiting Aβ production.The ISF Aβ 40 levels were measured as a representative of Aβ species because the ISF Aβ 40 levels after PER treatment showed similar kinetics with those of ISF Aβ 42 .The ISF Aβ 40 levels after the LY411575 injection are plotted in Figure 4B, and semi-log plot slopes are presented in Figure 4C.The elimination of ISF Aβ 40 followed first-order kinetics (vehicle: y = −0.111x+ 2.40, R 2 = 0.98, PER: y = −0.130x+ 2.10, R 2 = .91),and there was no significant change in the slopes of ISF between PER treatment and vehicle.The ISF Aβ 40 elimination half-life calculated from the semi-log plot slope under PER treatment was approximately 2.3 h, which was consistent with that under vehicle treatment (2.7 h), indicating that the rapid decrement in ISF Aβ levels after the PER treatment was not due to enhanced clearance.

| Acute treatment of PER does not affect the levels of GluA1 and GluA2
AMPARs are highly mobile and are transported to postsynapses in a structure-or neuronal activity-dependent manner, followed by recycling and degradation. 61MPARs are heterotetramers composed of combinations of GluA1-GluA4 subunits, and GluA2, the determinant of Ca 2+ permeation, is preferentially expressed in the adult brain. 62,63GluA1 is inserted at extrasynaptic sites and is potentiated at the synaptic surface in response to neuronal activity, whereas GluA2 structurally targets the synapses. 64,65Through the dynamics in the properties and abundance of synaptic AMPARs, functional changes, such as LTP and long-term depression (LTD), occur. 66,67To evaluate the effects of acute administration of PER on activity-or structure-dependent synaptic transport of AMPARs, we determined the GluA1 and GluA2 levels in SNS fractions by Western blotting.To first confirm the enrichment of synaptic proteins in the SNS fractions, the synaptotagmin 1 (Syt1) and postsynaptic density protein 95 (PSD95) levels in the THs and SNS fractions extracted from the cerebral cortex of J20 mice sacrificed at 5 h after vehicle administration were determined by Western blotting.The Syt1 and PSD95 levels were approximately 2.4-fold higher in the SNS fractions than in the THs, indicating sufficient enrichment of synaptic components in the SNS fractions (Figure 5A).Next, we determined the GluA1 and GluA2 levels in the SNS fractions and TH obtained from either vehicle or PER treatment mice by Western blotting.As shown in Figure 5B, the levels of GluA1 and GluA2 were not affected by PER administration, suggesting that PER did not downregulate the postsynaptic abundance of AMPARs, at least under the acute experimental condition.Since there is considerable variation in the levels of GluA1 and GluA2 in SNS and TH, the significance of the difference between PER and control cannot be fully dismissed.

| PER treatment significantly reduces the levels of sAPPβ released into the conditioned media of primary neuronal cultures under AβO-evoked neuronal
To corroborate the above in vivo results showing that acute treatment of PER can suppress β-cleavage of APP under high AβO concentration, we applied primary neurons preincubated with AβO and TTX 24 h prior to co-treatment of glutamate with PER or vehicle to evoke neuronal hyperexcitability as a cellular in vitro AD model.The LDH levels released into the conditioned media did not display any significant differences between the PER and vehicle groups, indicating that PER treatment did not induce significant cytotoxicity under the experimental condition (Figure 6A).We detected a significant reduction in the sAPPβ levels released to the media in the PER group compared to vehicle group (p = 0.037), while the sAPPα levels remained unaltered between both (Figure 6B).To determine the levels of endogenous full-length APP (FL-APP) and APP-CTF, the total lysates were subjected to Western blotting with the APP C-terminal antibody.Consistent with the above in vivo results, the FL-APP levels were significantly higher in the PER group (p = 0.0016), whereas the αCTF levels were unchanged.Note that the blots of endogenous APP-βCTF could not be detected due to low expression levels.These findings suggest that PER treatment may suppress β-cleavage of APP in primary neurons exhibiting AβO-evoked neuronal hyperexcitability.

| DISCUSSION
Recent clinical research has demonstrated that comorbid epilepsy AD, including subclinical seizures, can occur earlier than expected, even before the onset of cognitive impairment.Consistently, it has been previously reported that subclinical ictal activity and epileptic discharges are detected in EEG recordings before amyloid deposition in a mouse model of AD. 55 Interestingly, using two-photon in vivo calcium imaging, Busche et al. revealed that "hyperactive" neurons indicated by increased intracellular calcium concentrations clustered near amyloid plaques and were very frequently detected even before Aβ deposition in the brains of APP Tg mice. 58,68Further, in a similar approach, local administration of a synthetic Aβ dimer supposed to be neurotoxic to the CA1 region of the F I G U R E 5 Acute treatment of PER does not affect the levels of GluA1 and GluA2.(A) Total homogenate (TH) and synaptoneurosomes (SNS) of J20 hemicortex in the PER or vehicle-treated groups (n = 6 and 5, respectively) were electrophoresed in a single gel, and the lanes of synaptotagmin 1 (Syt1) and postsynaptic density protein 95 (PSD95) in the vehicle group were extracted.Syt1 levels were 2.4-fold higher in the SNS fraction than in total homogenates (TH) (p < .0017,t-test).Similarly, PSD95 levels were also 2.4-fold higher in the SNS fraction than in total homogenates (p = .0006,t-test).(B) No significant changes in the levels of GluA1 and GluA2 in SNS fractions and TH were observed between the vehicle and PER treatment (p = .61and .70,respectively, t-test).All data are plotted as mean ± SEM. n.s.denotes not statistically significant (p ≧ .05),**p < .005.
F I G U R E 6 PER treatment significantly reduces the soluble APPβ (sAPPβ) levels in the medium and increases the full-length APP (FL-APP) levels in the lysates of primary neurons pretreated by TTX and AβO.(A) LDH levels in the medium were not affected by PER (n = 6, p = .67,t-test).(B) sAPPβ levels were decreased by 49% in the PER group (n = 6, p = .037,t-test), while sAPPα levels tended to be increased but not significantly (p = .068,t-test).(C) FL-APP levels in the lysates were increased in the PER group by 79% (n = 6, p = .0016,t-test).αCTF levels did not significantly change (p = .18,t-test).All data are plotted as mean ± SEM. n.s.denotes not statistically significant (p ≧ .05),*p < .05,**p < .005.
hippocampus in wild-type mice also induced hyperactive neurons. 69these findings suggest that Aβ pathology may directly induce neuronal hyperexcitability prior to Aβ deposition in the pathophysiology of AD.Based on the "neuronal hyperexcitability hypothesis" in AD, various antiepileptic drugs have been tested in AD model mice and AD patients.Sanchez et al. administered seven antiepileptic drugs (levetiracetam, ethosuximide, gabapentin, phenytoin, pregabalin, valproic acid, and vigabatrin) to APP Tg J20 mice before the appearance of Aβ deposition and found that only levetiracetam (LEV), which targets synaptic vesicle protein 2A (SV2A), showed a significant reduction in epileptic discharges, leading to improved synaptic transmission and cognitive function, whereas phenytoin and pregabalin significantly increased epileptic discharges. 181][72][73][74] Intriguingly, Yamamoto et al. reported that using an optogenetic approach, a longterm synaptic activation into the hippocampal perforant pathway originating from the lateral entorhinal cortex clearly promoted Aβ deposition in the projection area (the dentate gyrus) in APP Tg mice, suggesting that aberrant chronic synaptic activation accelerated Aβ burden.Further, in human studies, it has been shown that Aβ deposition starts in the brain regions in the default-mode network, which is considered to be highly active during rest, 75 and that Aβ deposition is more prominent in surgically resected brain sections from intractable epilepsy patients than age-matched controls. 76,77Based on a bidirectional association between Aβ pathology and epilepsy, we focused on a potential impact of AMPAR antagonist on Aβ pathology for the following reasons; (1) the expression levels of AMPAR 78,79 and ISF glutamate levels 80 are concentrated at the epileptic foci even in interictal phases, (2) the levels of the GluA1 subunit to induce the Ca 2+permeable AMPA receptor is significantly increased in postsynaptic density-rich fractions from hippocampi of AD patients compared to controls 81 injection of synthetic Aβ 42 oligomers into rat hippocampal slice cultured neurons rapidly induced excitatory postsynaptic currents and increased the expression levels of synaptic Ca 2+ -permeable AMPA receptors. 31n the present study, we aimed to evaluate the therapeutic potential of PER targeting AMAPAR toward Aβ pathology in AD.To this end, we investigated how PER affects Aβ pathology in J20 mice using in vivo microdialysis techniques.We found that a single oral administration of PER lowered ISF Aβ 40 and Aβ 42 levels (up to 21% and 31%, respectively) in the hippocampi of young J20 mice without Aβ deposition.Acute administration of PER did not induce any alterations in Aβ pools within the hippocampus, except for extracellular soluble Aβ, suggesting that the reduction of ISF Aβ by a single administration of PER is not due to their sequestration into other Aβ pools.This result is in line with a previous study that highlighted how the ISF pools of Aβ collected using in vivo microdialysis reflect the regulation of Aβ metabolism. 46In young APP mice without plaque formation, ISF Aβ pools should remain independent of other compartments, while in older mice with Aβ aggregation (plaques), insoluble Aβ could potentially contribute to changes in ISF Aβ levels, leading to an equilibrium between Aβ in ISF and deposited insoluble Aβ.
ISF soluble Aβ levels should be determined by balancing production and clearance from the ISF before Aβ aggregation starts.To evaluate the inhibitory effects of acute PER treatment on in vivo Aβ production, we assessed the APP substrates for β-secretase and γ-secretase to generate Aβ, FL-APP, and βCTF, respectively.We showed that 5 mg/kg PER treatment significantly reduced the levels of βCTF in hippocampal homogenates as well as the hippocampal ISF Aβ levels, indicating inhibited β-cleavage of APP.Indeed, we detected a significant reduction of sAPPβ levels released from primary neurons under AβO-evoked neuronal hyperexcitability by acute treatment of PER, supporting the above in vivo findings.Regarding the Aβ clearance from ISF, the calculation of the half-life of ISF Aβ clarified that a single oral administration of PER did not alter the Aβ elimination from the hippocampal ISF.Another study that contradicts our results reported the positive effect of AMPAR activation on AD pathology: artificial activation of AMPARs can promote Aβ clearance. 34We speculate that there may be an appropriate therapeutic window in which aberrant neural excitability is modified with little or no suppression of chronic AMPAR activation.
Since it has been shown that the appearance of regional Aβ deposition was preceded by an increase in soluble ISF Aβ levels in the corresponding regions of young APP Tg mice, 46,82 a long-term modest reduction of soluble ISF Aβ could reduce future Aβ deposition under the pathogenic condition of AD.
We demonstrated for the first time that antagonism of AMPARs may reduce Aβ pathology by suppressing β-cleavage of APP in both in vivo and in vitro AD models.AMPARs are predominantly located at postsynapses, in contrast with the β-cleavage of APP, which mostly observed at presynapses. 83Thus, PER is unlikely to directly affect BACE1; however, it might indirectly affect APP processing through neural activity modulation.
PER is already used clinically as an AED and an option in epilepsy complicated by AD.Our study highlights PER as a therapeutic agent for AD, contrary to previous findings that activation reduces Aβ pathology. 34,84This study has potential limitations in that we only examined the acute effects of a single oral administration.
In this study, all mice displayed ataxia-like motor impairment as the plasma concentration of PER reached sufficiently high levels.PER has been documented to induce dose-dependent motor incoordination in rodents, with a small therapeutic window. 35However, PER rarely induces severe motor symptoms in human patients, potentially due to its longer half-life and the divergence in AMPAR distribution compared with rodents. 356][87] Conversely, it has been indicated that there were no differences in the occurrence of behavioral and psychiatric adverse events according to the history of cognitive decline. 43Furthermore, PER is well tolerated in patients aged ≥65 years, and low-dose PER has shown high efficacy in treating elderly-onset epilepsy concomitant with AD.Even conventional doses of PER have been observed to improve cognitive function, based on a limited study in Japan. 44In a case report, PER improved both seizures and psychiatric symptoms in a patient with severe dementia who developed myoclonus epilepsy due to AD. 89 PER has also been reported to exert anxiolytic effects in mice via AMPAR antagonism. 90Overall, the use of PER does not appear to increase the risk of psychiatric side effects in dementia patients, though the therapeutic window should be carefully determined considering background factors and pathological conditions.
In this study, we have highlighted the acute effect of PER on Aβ production, but neuronal hyperexcitability likely has multifactorial effects on AD pathophysiology.Aberrant neuronal excitability, such as epilepsy, itself affects cognitive function. 913][94] Other studies have shown that PER might also reduce inflammatory activation, tau-related excitotoxic synaptic signaling, and memory deficits in models with Aβ pathology, suggesting a broader impact of PER on AD pathophysiology. 41,95Further research with long-term administration and dose titration is needed to elucidate the multifactorial effects of PER in AD.

| CONCLUSIONS
A single oral administration of PER rapidly reduced ISF monomeric Aβ levels in the hippocampus of young J20 mice without altering Aβ clearance from ISF.Further, we found that acute treatment of PER reduced the βCTF levels in the hippocampi of APP Tg mice and the sAPPβ levels secreted from the primary cortical neurons under Aβ-evoked neuronal hyperexcitability.Taken together, we concluded that the acute treatment of PER reduces Aβ production through suppressing β-cleavage of APP without altering synaptic AMPAR levels in both in vivo and in vitro AD models.This is the first report to reveal the inhibitory effect of PER on the amyloidogenic processing of APP.

F I G U R E 4
PER administration does not affect the half-life of ISF Aβ levels.(A) An experimental scheme to determine the elimination half-life of Aβ from ISF is shown.ISFs were collected every hour in J20 mice (n = 4) before and after the oral administration of PER, followed by a subcutaneous injection of LY411575, a γ-secretase inhibitor (GSI), after 1 h.(B) ISF Aβ 40 levels after GSI injection were plotted.(C) The half-life of Aβ 40 was calculated from the slope of the ISF levels and time curve by plotting it on a semi-logarithmic graph.The elimination of ISF Aβ 40 followed first-order kinetics (vehicle: y = −0.111x+ 2.40, R 2 = .98,PER: y = −0.130x+ 2.10, R 2 = .91),and there was no significant change in the slopes of ISF between the PER treatment and vehicle (p = .46,ANCOVA).The half-life was 2.3 h in PER treatment, almost the same as 2.7 h in the vehicle.All data are plotted as mean ± SEM. | 13 of 18 UEDA et al.