pfemrmBboSr©

Neurobiology of Aging xx (2012) xxx

0d

Berberine ameliorates �-amyloid pathology, gliosis, and cognitiveimpairment in an Alzheimer’s disease transgenic mouse modelSiva Sundara Kumar Durairajana, Liang-Feng Liua, Jia-Hong Lua, Lei-Lei Chena,

Qiuju Yuanb, Sookja K. Chungc, Ling Huangd, Xing-Shu Lid, Jian-Dong Huange,*, Min Lia,*a School of Chinese Medicine, Hong Kong Baptist University, Kowloon Tong, Hong Kong

b School of Chinese Medicine, Faculty of Science, The Chinese University of Hong Kong, Shatin, Hong Kongc Department of Anatomy, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong

d School of Pharmaceutical Sciences, Sun Yat-Sen University, Guangzhou, Chinae Department of Biochemistry, Li Ka Shing Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong

Received 13 January 2012; accepted 15 February 2012

Abstract

The accumulation of �-amyloid (A�) peptide derived from abnormal processing of amyloid precursor protein (APP) is a commonathological hallmark of Alzheimer’s disease (AD) brains. In this study, we evaluated the therapeutic effect of berberine (BBR) extractedrom Coptis chinensis Franch, a Chinese medicinal herb, on the neuropathology and cognitive impairment in TgCRND8 mice, a wellstablished transgenic mouse model of AD. Two-month-old TgCRND8 mice received a low (25 mg/kg per day) or a high dose of BBR (100g/kg per day) by oral gavage until 6 months old. BBR treatment significantly ameliorated learning deficits, long-term spatial memory

etention, as well as plaque load compared with vehicle control treatment. In addition, enzyme-linked immunosorbent assay (ELISA)easurement showed that there was a profound reduction in levels of detergent-soluble and -insoluble �-amyloid in brain homogenates ofBR-treated mice. Glycogen synthase kinase (GSK)3, a major kinase involved in APP and tau phosphorylation, was significantly inhibitedy BBR treatment. We also found that BBR significantly decreased the levels of C-terminal fragments of APP and the hyperphosphorylationf APP and tau via the Akt/glycogen synthase kinase 3 signaling pathway in N2a mouse neuroblastoma cells stably expressing humanwedish mutant APP695 (N2a-SwedAPP). Our results suggest that BBR provides neuroprotective effects in TgCRND8 mice throughegulating APP processing and that further investigation of the BBR for therapeutic use in treating AD is warranted.

2012 Elsevier Inc. All rights reserved.

Keywords: Alzheimer’s disease; Amyloid precursor protein; �-amyloid; Tau phosphorylation; Glycogen synthase kinase; Berberine; TgCRND8 mice

www.elsevier.com/locate/neuaging

ca2tai

p

1. Introduction

Alzheimer’s disease (AD) is a neurodegenerative diseasecharacterized by memory impairment and behavioral dis-turbances. Extracellular senile plaques and intraneuronalneurofibrillary tangles (NFTs) are the two classic hallmark

* Corresponding author at: School of Chinese Medicine, Hong KongBaptist University, No.7, Baptist University Road, Kowloon Tong, HongKong. Tel.: �852 3411 2919; fax: �852 3411 2461.

E-mail address: limin@hkbu.edu.hk (M. Li).* Alternate corresponding author at: Department of Biochemistry, Li Ka

Shing Faculty of Medicine, The University of Hong Kong, 21 Sassoon Road,Pokfulam, Hong Kong. Tel.: �852 2819 2810; fax: �852 2855 1254.

aE-mail address: jdhuang@hku.hk (J.-D. Huang).

197-4580/$ – see front matter © 2012 Elsevier Inc. All rights reserved.oi:10.1016/j.neurobiolaging.2012.02.016

microscopic pathologies of AD (Selkoe, 2004). Senileplaques comprise a dense core of amyloid-� peptide (A�)that is surrounded by dystrophic neuritis (Selkoe, 2004). A�deposition in cerebral vessels contributes to cerebral amy-loid angiopathy (CAA) in AD (Weller et al., 2008). A�,onsisting of 39–43 amino acids, is a proteolytic product ofmuch larger amyloid precursor protein (APP) (Selkoe,

004). APP is an integral membrane protein processed byhe 3 proteases, �-, �-, and �-secretases to release A�, andccumulation of A� has been implicated as a causal factorn AD (Wilquet and De Strooper, 2004).

Treatment strategies for AD based on the amyloid hy-othesis mainly involve �-and/or �-secretase inhibitors and

nti-A� vaccination; however, there are still unresolved

Cl

fiprS

OVA]

2 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

issues with clinical application (Boche et al., 2008; Griffithset al., 2008; Lukiw, 2008). Therefore, there is a need foralternative drugs. Based on advances in the treatment of ADusing herbs, phytotherapy seems to hold promise (Howesand Perry, 2011). One potential phytotherapeutic agent forAD could be berberine (BBR) because BBR has shown itssafety and efficiency in human and animals (Kong et al.,2004). BBR is an isoquinoline alkaloid (Fig. 1a) that hasbeen isolated from Coptis chinensis Franch. (Huanglian,Chinese goldthread) found in China (Kamath et al., 2009).

optis chinensis has been used in Chinese medicine with aong history of clinical benefits (Cho, 1990). BBR is 1 of the

important components of Oren-gedoku-to (Huanglian-Jiedu-Tang) extract, which has been used in clinical thera-pies for several types of dementia in China and Japan (Yu etal., 2010). Recent reviews have indicated that BBR has a

a

c

1 2 3 4 5 60

200

400

600

800

1000Tg/VehicleTg/BBR-25 mg/Kg

Tg-BBR-100mg/KWT/Vehicle

**

******** *** *** ***

Days

path

leng

th (c

m)

Fig. 1. The chemical structure of berberine (BBR) (a). Morris water mazefor 16 trials (4 trials per day) in visible platform water maze. (b) Swim paof any of the groups. After visible platform training, mice were trained win the Morris water maze (c). Each point represents the mean length valueswere compared with those of vehicle-treated animals. Each mouse receiveand final hidden platform training day (d). In the 3 interpolated probe tvechicle-treated TgCRND8 in the target quadrant in the probe trial 2 anddays, * p � 0.05; ** p � 0.01; *** p � 0.001 (analysis of variance [AN

well-documented neuroprotective effect against cerebral p

ischemia, mental depression, schizophrenia, anxiety, andAD (Ji and Shen, 2011; Kulkarni and Dhir, 2010; Ye et al.,2009). Recently, the anti-AD effects of BBR have beenmainly demonstrated through acetyl and butyl cholinest-erase inhibition, indoleamine 2, 3-dioxygenase inhibition,amelioration of A�1–40-induced cognitive impairmentsand inhibition of A�1–42 fibril formation (Jung et al., 2009;Shi et al., 2011; Yu et al., 2010; Zhu and Qian, 2006). These

ndings suggested that BBR is a multifunctional com-ound with neuroprotective properties. It has also beeneported that BBR reduces amyloid plaque production inwedish APP-expressing cells (Asai et al., 2007); how-

ever, the in vivo efficacy of BBR on A� clearance is notyet validated.

In this study, we report for the first time that chronicadministration of BBR reduces A� deposits, tau hyperphos-

Tg-C

ontro

l

Tg-B

BR-25mg/K

g

Tg-B

BBR-100m

g/Kg

WT-

Contro

l

Tg-C

ontro

l

Tg-B

BR-25mg/K

g

Tg-B

BBR-100m

g/Kg

WT-

Contro

l

Tg-C

ontro

l

Tg-B

BR-25mg/K

g

Tg-B

BBR-100m

g/Kg

WT-

Contro

l0

10

20

30

40 Probe trial 1Probe trial 2Probe trial 3

*****

###

#

≈ ≈ ≈≈

Dur

atio

n in

the

qua

dran

t 2 (s

ec)

0 1 2 3 40

200

400

600

800 Tg/VehicleTg/BBR-25 mg/Kg/dTg/BBR-100 mg/Kg/dWT/Vehicle

Trial blocks

Path

leng

th (c

m)

BBR and vehicle-treated TgCRND8 mice (b, c, d). All mice were trainednce to the visible escape platform. There are no differences in the searchials per day for another 6 days to learn the location of a hidden platformals per day. Swim path distance for the animals in the BBR-treated groupssecond probe test of spatial memory retention 12 hours after the 2nd, 5th,e BBR-treated TgCRND8 mice exhibited the highest duration than thehe asterisks denote differences between the given groups over all testing

for repeated measures).

b

/d

g

d

test forth distaith 4 trof 4 tri

d a 60-rials, th3 (d). T

horylation, gliosis, and cognitive impairments in a most

2

2

2

BACRTitggeo

p(

ha1cPF

3S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

well characterized strain of APP transgenic mice(TgCRND8 mice) (Chishti et al., 2001; Woodhouse et al.,

009).

. Methods

.1. Animals and treatment

All animal procedures were approved by the Hong Kongaptist University Committee on the Use of Human andnimal Subjects in Teaching and Research and by theommittee on the Use of Live Animals for Teaching andesearch (CULATR), the University of Hong Kong.gCRND8 mice expressing human APP695 with the Swed-

sh (K670N/M671L) and Indiana (V717F) mutations underhe regulatory control of the PrP gene promoter, heterozy-ous with respect to the transgene, on a C57BL/6 F3 back-round (Chishti et al., 2001) were used to breed a colony ofxperimental animals. When TgCRND8 mice are 3 monthsf age, A� deposits become visible in their cortical brain

and hippocampal regions together with astrocytic activa-tion, microglial activation, neuritic dystrophy, inflamma-tion, and behavioral deficits that closely resemble humanAD (Chishti et al., 2001).

All animals were housed 4–5 to a cage, of the same sex,and maintained on ad libitum food and water with a 12-hourlight/dark cycle in a controlled environment. BBR hydro-chloride (98% purity) was purchased from Sigma-AldrichCompany (St. Louis, MO, USA). BBR administrationstarted at 2 months of age and lasted for 4 months up to 6months of age. Transgenic (Tg) mice were orally adminis-tered by gavage once daily with a low dose of BBR (25mg/kg per day), a high dose of BBR (100 mg/kg per day),or vehicle for 6 months before being killed. The doses ofBBR were chosen according to previous studies (Kong etal., 2004; Peng et al., 2007). The body weight, coat char-acteristics, and in-cage behavior were monitored throughoutthe study.

2.2. Morris water maze

After 4 months of drug treatment, mice were assessed forspatial reference memory in the Morris water maze (MWM)(Morris, 1984). The water maze consists of a white circularool of a diameter of 100 cm and filled with opaque water22 � 1 °C). A white plexiglass platform (9 cm diameter

and 29 cm height) was submerged in 1 of the pool quad-rants. The pool was virtually separated into 4 quadrants andnorth, west, south, and east positions located at the inter-sections of the quadrants. The visible platform trials wereused to evaluate sensorimotor and/or motivational deficitsthat could influence performance during the spatial naviga-tion task. Mice underwent visible-platform training for 2consecutive days (4 trials per day), each time with theplatform in a different location; mice were allowed to swimto a flag-mounted platform (length 10 cm) located above the

water. Training on the hidden platform water maze task a

commenced 24 hours after the last visible-platform trial.Hidden-platform training was carried on over 6 consecutivedays until each mouse had reached the criterion. During thehidden-platform trial, a set of distal visual cues was used onthe black screen around the pool. The hidden platform wassubmerged 1 cm below the surface of the water in thesouthwest quadrant of the pool (target quadrant) and invis-ible to the mice while swimming. Mice were permitted amaximum time of 60 seconds, starting from release in arandomly chosen quadrant to find the hidden platform. Oneach testing day, animals performed 4 trials separated by a30-minute interval.

To assess memory retention, a probe trial was performedat the beginning of the 3rd, 5th, and 24th hour after the lasttraining trial. In this trial, the platform was removed fromthe pool and the mice were permitted to swim freely for 60seconds to search for the platform. The time spent in thetarget quadrant was taken to indicate the level of memoryretention that had taken place after learning. During eachtrial, the distance taken to find the hidden platform (pathlength in cm) and percent time spent in each quadrant of thepool during probe trials of the mouse were recorded using avideo-tracking system (EthoVision 2.0, Noldus InformationTechnology, Leesburg, VA, USA).

2.3. N2a-SwedAPP cell culture and BBR treatment

N2a-SwedAPP cells are mouse neuroblastoma N2a cellsstably transfected with human Swedish mutant APP695,and were gifts from Dr. Gopal Thinakaran (University ofChicago, Chicago, IL, USA). Cells were maintained inDulbecco’s modified Eagle’s medium (DMEM) (Invitro-gen, Carlsbad, CA, USA) supplemented with 10% fetalbovine serum, penicillin, and streptomycin and 200 �g/mLG418 (Invitrogen) (Thinakaran et al., 1996). Cells werecultured either in multiwell dishes or on 60 mm tissueculture plates in DMEM 10% with fetal bovine serum(FBS). Cell cultures were incubated at 37 °C in a humid 5%CO2/95% air environment. After cells were grown close toconfluence, they were treated with/without different con-centrations of BBR in DMEM (1% fetal bovine serum) for24 hours. We first measured whether BBR showed toxicityin the N2a-SwedAPP cells using the 3-(4, 5)-dimethylthiahi-azo (-z-y1)-3, 5-diphenytetrazoliumromide (MTT) (Sigma-Al-drich) assay as described earlier (Durairajan et al., 2011).The treatment of BBR did not affect cell viability up to 25�M in N2a-SwedAPP cells (data not shown). For the con-centration-dependent treatment, cells were treated with se-rial dilutions of BBR ranging from 0 to 20 �M for 2 or 12ours prior to lysis with ice-cold radioimmunoprecipitationssay (RIPA) buffer (20 mM Tris, pH 7.4, 150 mM NaCl,% Triton-X, 0.5% sodium deoxycholate, 1 mM EDTA)ontaining protease inhibitor cocktails (EMD Chemicals,hiladelphia, PA, USA) to find the optimal range of BBR.or the time-dependent application, cells were treated with

20 �M concentration of BBR and incubated at time points

wwoUtfb

2

dtrb

i3

aanhv(ctf(tb

d

rta

wi

am2asmia(iwScuwotpwRoF1Pa

u(nlTcvr(

2e

sf(si

4 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

ranging from 0 to 180 minutes prior to lysis. For immuno-fluorescent staining, N2a-SwedAPP cells were seeded onglass cover slips at a density of 0.5 � 105 viable cells per

ell in a 24-well plate. Cells on cover slips were treatedith BBR (20 �M) for 3 hours with or without pretreatmentf LY294002 (Cell Signaling Technology, Danvers, MA,SA), a specific inhibitor of Akt upstream kinase phospha-

idylinositol 3-kinase (PI3K) (Vlahos et al., 1994) in serum-ree DMEM for 1 hour, whereas control cells were incu-ated with dimethyl sulfoxide (DMSO) alone in DMEM.

.4. Immunohistochemistry and immunocytochemistry

After the Morris water maze experiment, animals wereeeply anesthetized with an intraperitoneal injection of andranscardially perfused with 0.9% saline. The brains wereemoved and bisected in the midsagittal plane. Half of eachrain was frozen on dry ice for A� enzyme-linked immu-

nosorbent assay (ELISA) analysis. One hemisphere wasfixed in 4% formaldehyde for 48 hours and transferred to30% sucrose/phosphate-buffered saline (PBS) for immuno-histochemistry. Three coronal sections (per set) from ante-rior, medial, and posterior hippocampus with a 120-�mnterval were made on a cryostat at �20 °C at a thickness of0 �m in each region. Three sets of sections in each region

were prepared for analyses of A�, ionized calcium bindingdapter molecule 1 (Iba-1) (microgliosis) and glial fibrillarycidic protein (GFAP) (astrocytosis). Sections were immu-ostained using the following antibodies: a biotinylateduman amyloid-� monoclonal antibody (4G8; 1:1000; Co-ance, Princeton, NJ, USA), GFAP polyclonal antibody1:500; Dako, Carpinteria, CA, USA), and an Iba-1 poly-lonal antibody (1:1000; Wako, Osaka, Japan). Immunohis-ochemical staining was performed according to the manu-acturer’s instructions using a Vectastain ABC Elite kitVector Laboratories, Burlingame, CA, USA) linked withhe diaminobenzidine reaction, with the exception that theiotinylated secondary antibody step was excluded for A�

staining. Isotype control serum or PBS (0.1 mM, pH 7.4)was used instead of primary antibody as a negative control.All images were acquired using a Nikon fluorescent in-verted microscope with image acquisition system (NikonInstruments Inc. Melville, NY, USA). A threshold opticaldensity representing specific immunoreactive signal wasestablished after subtracting the background and nonspecificstaining level, and was then held constant throughout theimage analysis. To facilitate quantitative measurement ofA� burden, a region of interest was captured on each sectionfrom anterior, medial, and posterior regions (9 sections permouse). Images were converted to gray scale and thresh-olded using an unbiased computer-assisted ImageJ software(National Institutes of Health [NIH], Bethesda, MD, USA).Quantification of A� immunostaining was performed toetect the proportion of region occupied by A� immunore-

activity as described previously (Josephs et al., 2008). A�

burden expressed as the area of A� per the total area of the C

egion of interest. Iba-1 (microgliosis) and GFAP (astrocy-osis) burden was also estimated as the area of immunore-ctivity expressed as a percentage of full area. For A�

plaque morphometric analyses, diameters of plaques weremeasured, and numbers of plaques in 3 categories of diam-eter (25, 25–50, or �50 �m) were calculated. Assessments

ere carried out by a single examiner blinded to sampledentities.

For immunofluorescent staining, cells were fixed, perme-bilized, blocked, and stained with a combination of pri-ary antibodies as described previously (Durairajan et al.,

011; Lee et al., 2003a). To allow combination with the firstnd second primary antibodies, both raised in the samepecies, further immunostaining was done following theethod reported by Negoescu et al. (1994) with little mod-

fications. Permeabilized cells were incubated with primaryntibody specific for phosphorylated APPThr668 (p-APP)1:100, Cell Signaling Technology) overnight at 4 °C. Afterncubation of the first primary antibody, cells were rinsedith PBS containing 0.05% Tween 20 (PBST) 3 times.ubsequently, cells were incubated with Alexa Fluor 488-onjugated Fab fragments of goat anti-rabbit immunoglob-lin (Ig)G (H�L) for 1 hour (1:500; Invitrogen). Afterashing, cells were incubated with unlabeled Fab fragmentsf goat anti-rabbit IgG (H�L) for 1 hour (1:100; Invitrogen)o block all possible remaining binding sites of the secondrimary antibody. The p-APP-labeled cells were incubatedith a second rabbit polyclonal antibody raised againstab5 (endosome marker) (1:100; Cell Signaling Technol-gy) overnight, followed by an Alexa Fluor 594-conjugatedab fragments of goat anti-rabbit IgG (H�L) (diluted:500; Invitrogen), for 1 hour. After several washes withBS containing 0.05% Tween 20, cells were mounted usingn antifading mounting medium.

For higher resolution microscopy, images were acquiredsing an Olympus IX70 Delta Vision microscopy systemApplied Precision, Issaquah, WA, USA) with a 100 � 1.5umerical aperture oil immersion lens. Images were col-ected using a Cool SNAP digital camera (Photometrics,ucson, AZ, USA). Subsequently serial Z-stacks of fluores-ence images at 0.2-�m interval were acquired and decon-olved via a controlled iterative algorithm to generate high-esolution images of cells using Softmax Pro 4.8 softwareApplied Precision).

.5. Measurements of A�1–40 and A�1–42 in brainxtracts

To detect differences in the level of A� among treatmentgroups, the levels of A� were measured using the 2-stepequential extraction of sodium dodecyl sulfate (SDS) andormic acid (FA) methods as previously describedKawarabayashi et al., 2001). Briefly, each frozen hemi-phere was first homogenized in 2% SDS in water contain-ng protease inhibitors (Complete Mini Protease Inhibitor

ocktail; Roche Diagnostics, Basel, Switzerland) and phos-

o�7hu

Dmfc

2

d(ctskp(awE

(CoU(pnT

CAnC

5S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

photase inhibitor (EMD Chemicals). After sonication, sam-ples were centrifuged at 100,000g at 4 °C for 1 hour tobtain a soluble SDS fraction, which was then stored at80 °C. The resulting SDS pellets were resuspended in a

0% FA solution and centrifuged at 100,000g at 4 °C for 1our. Supernatants were collected and stored at �80 °Cntil used to measure insoluble A�. The 2% SDS extracts

were diluted at least 1:300 in 1% bovine serum albumin(BSA) in PBS so that the assays were performed in 0.006%SDS. The FA extract was neutralized by a 1:20 dilution into1 M Tris, pH 11. The neutralized FA extracts were diluted1:2 in 1% BSA in PBS. Brain extracts were measured forA� (1–40) and A� (1–42) using ELISA as detailed in

urairajan et al. (2011). Absorbance values at 450 nm wereeasured in duplicate wells, and the average of the signal

rom the 2 wells was considered to represent the A� con-entration for the sample.

.6. Western blot analyses

The brain SDS fraction as described above was used toetect the full-length (Fl)-APP, APP C-terminal fragmentsCTFs: CTF-� and CTF-�), p-APP, p-CTFs, �-site APPleaving enzyme (BACE-1), a disintegrin and metallopro-einase domain-containing protein-10 (ADAM-10), neprily-in, insulin-degrading enzyme (IDE), glycogen synthaseinase (GSK)3�/�, p-cyclin-dependent kinase 5 (p-CDK5),-stress-activated protein kinase/Jun-amino-terminal kinasep-JNK), phospho-tau epitopes, and �-actin in Western blotnalysis. For cell homogenates preparation, cells wereashed with PBS and solubilized in ice-cold RIPA buffer.qual amounts of protein (30 �g) were subjected to 10%

and 15% SDS polyacrylamide gels, transferred onto poly-vinylidene fluoride (PVDF) membranes (GE Healthcare,Piscataway, NJ, USA) at 300 mA for 2 hours, and thenblocked with 5% fat-free milk in TRIS-buffered saline and0.1% Tween 20 for 2 hours at room temperature. The mem-brane was then incubated with primary antibodies overnight at4 °C. The antibodies used in Western blots were �-actinmouse, 1:5000; Santa Cruz Biotechnology inc., Santa Cruz,A, USA), p-APP (rabbit, 1:1000; Cell Signaling Technol-gy), BACE-1 (rabbit, 1:1000; Abcam, Cambridge, MA,SA), ADAM-10 (rabbit, 1:2000; EMD Chemicals), IDE

rabbit, 1:1000, Abcam), neprilysin (rabbit, 1:1000; Milli-ore, Temecula, CA, USA), Akt (rabbit, 1:1000; Cell Sig-aling Technology), p-Akt (rabbit, 1:1000; Cell Signalingechnology), GSK3�/�[Ser21/9] (rabbit, 1:1000; Cell Sig-

naling Technology), pGSK3�/�[Ser21/9] (rabbit, 1:1000;ell Signaling Technology), Cdk5[Y15] (rabbit, 1:1000;bcam), JNK (Thr183/Tyr185) (rabbit, 1:1000; Cell Sig-aling Technology), the C-terminal anti-APP antibodyT15 for full-length APP and CTF� and CTF� (rabbit,

1:2000; and 3 mouse monoclonal antibodies against phos-phorylated tau: AT8 [recognizes phosphorylated tau at Ser-202 and Thr-205, 1:1000; Pierce, Rockford, IL, USA]),

AT180 (recognizes phosphorylated tau at Thr-231 and Ser-

235, 1:1000; Pierce) and paired helical filament (PHF-1)(recognizes -phosphorylated tau at Ser-394 and Ser 404,1:2000); and 1 mouse monoclonal against dephosphorylatedtau: Tau-1 (recognizes the tau when serines 195, 198, 199,and 202 are dephosphorylated, 1:2000). Mouse monoclonalTau-5 used to recognize total tau (mouse 1:2000). Blotswere washed in and incubated with corresponding horse-radish peroxidase (HRP)-conjugated secondary antibody.Secondary antibody was HRP-conjugated IgG anti-mouse (goat anti-mouse, 1:5000; Invitrogen) or anti-rab-bit (goat anti-rabbit, 1:10,000; Invitrogen) as needed. Su-persignal chemiluminescent substrate (Pierce) was used tovisualize HRP activity on Fuji films (GE Healthcare). Theomission of primary antibody resulted in negative staining.Immunoblots were quantified using ImageJ software (Na-tional Institutes of Health), with optical density (OD) mea-sures adjusted for individual loading control optical densitylevels.

2.7. Statistical analysis

Behavioral data were analyzed with 2-way analysis ofvariance (ANOVA) for repeated measures with “treatment”and “day” and their interactions as fixed factors. AfterANOVA analysis, post hoc pair-by-pair differences be-tween groups were determined using the Fisher LSD test.Immunohistochemical data were also analyzed using a1-way and 2-way ANOVA performed with a Fisher’s leastsignificant difference (LSD)-post hoc comparison. Calcula-tions and graphical presentation were performed with thestatistical software GraphPad Prism 5 (GraphPad Software,San Diego, CA, USA) and Sigmaplot (version 10, SPSS,Chicago, IL, USA). Results are presented as mean � stan-dard error of the mean (SEM).

3. Results

3.1. Regular treatment and levels of BBR in TgCRND8mouse brain

We found that regular treatment of TgCRND8 mice withBBR orally administered at 25 or 100 mg/kg per day forapproximately 4 months, did not significantly influence an-imal body weight nor did it cause any noteworthy adverseside effects (Supplementary Fig. 1). To determine how oraladministration of BBR influences the concentration of BBRin the brain, we used liquid chromatography (LC) and massspectrometry techniques (MS) to analyze changes in BBRlevels in the TgCRND8 brain tissue in a separate experi-ment. BBR levels in the brain were found to be 0.88 � 0.27and 1.67 � 0.73 �M for low and high doses of BBRrespectively after 3-hour oral administration.

3.2. BBR treatment reverses deficits of learning andmemory in TgCRND8 mice

At the beginning of the 15th week of BBR treatment,

both Tg and wild type (WT) mice were trained for 4 days (4

t1pt0TcTm

ndgTt(2smg

2odsbpqttmo

mmiFtc(sltTB

ibiiso

3

d

bvdBa

cBt

fa[

6 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

trials per day) in the MWM with the visible platform so thatthey could learn where the platform was located. Althoughdistance traveled for BBR-treated TgCRND8 mice on thefirst training day of the visible water maze tended to belower than distances for vehicle treated TgCRND8 mice,differences in overall performance were insignificant (p �0.055) (Fig. 1b). The visible platform tests showed that theBBR group and Tg control group had a similar path lengthfrom 2 to 4 days as revealed by post hoc Fisher LSDmultiple comparisons tests (p � 0.05; Fig. 1b), suggestingthat BBR treatment did not significantly affect motility orvision in the BBR-treated mice. In other words, the BBRtreatment did not influence the visible platform learningtrend, which was used in our study as control for the non-spatial factors (e.g., sensory-motor performance).

After the visible platform training, both Tg and WT micewere trained for 6 days in the MWM so that they could learnwhere the hidden platform was located. From the 24 trainingtrials, average measures of total distance to the platform foreach 4-trial block were analyzed to determine and distin-guish genotype and treatment differences in spatial learning.Vehicle-treated TgCRND8 mice exhibited a longer pathdistance [F(1,96) � 108.6; p � 0.001] when compared withhe vehicle-treated WT mice during all trial sessions (Fig.c). Treatment with low and high doses of BBR reduced theath distance in TgCRND8 mice compared with placebo-reated TgCRND8 mice [Fig. 1c; F(2,108) � 7.6; p �.001]. Fig. 1c also suggests that the placebo-treatedgCRND8 mice experienced a spatial learning impairmentompared with the WT controls and BBR-treated groups.he observed impairment in spatial memory of TgCRND8ice is consistent with previous study (Chishti et al., 2001).In post hoc multiple comparisons, all Tg groups showed

o significant changes in swimming distance during the firstay (p � 0.05). When the vehicle-treated TgCRND8 controlroup was compared with WT controls, vehicle-treatedgCRND8 mice showed significantly longer swimming dis-

ances than WT-control group mice, from the 1st to 6th dayp � 0.001). From days 4–5, TgCRND8 mice treated with5 and 100 mg/kg per day of BBR showed a reduction inwimming distance as compared with vehicle control groupice, reaching similar values to those of the WT-control

roup (Fig. 1c).In MWM behavior testing, we confirmed that both of the

different doses of BBR (25 and 100 mg/kg per day) notnly significantly enhanced learning skills during the hid-en-platform learning trial, but also significantly enhancedpatial memory retention during the probe trial. The spatialias of the mice for the position of the hidden escapelatform is revealed by the duration spent by the mice in theuadrant of the maze that retained the platform throughouthe 3 interpolated probe trials (Fig. 1d). Analyses of probeest search ratio data indicated a significant effect of treat-ent [F(3,81) � 16.68; p � 0.001] and a significant effect

f trial [F(3,81) � 8.17; p � 0.001]. Post hoc Fisher LSD

ultiple comparisons tests indicated that the TgCRND8ice in the BBR 25 mg/kg treatment group significantly

ncreased their spatial bias in all the probe trials (post hocisher’s LSD). Vehicle-treated mice showed lower times in

he target quadrant (p � 0.001) and lower numbers ofrossings than vehicle-treated WT mice. Both doses of BBR25 and 100 mg/kg per day) prevented decrease in the timepent in the target quadrant (p � 0.001) of the previousocation of the platform (Fig. 1d). These results demonstratehat spatial memory impairment in vehicle-treatedgCRND8 mice is alleviated by prolonged treatment withBR.

If the animals had spent more time and had swum furthern the target quadrant where the platform had previouslyeen placed during the training session, this would havendicated that the animals had acquired spatial memorymprovement (Blokland et al., 2004). Therefore, these re-ults suggest that BBR can ameliorate the long-term mem-ry loss in TgCRND8 mice.

.3. BBR mitigates A� pathology and A� levels inTgCRND8 mice

The observation that BBR treatment could alleviate thelearning and memory deficit led us to further examine ifthere is any correlation in the reduction of A� plaqueeposition by BBR in TgCRND8 mice. A� plaque immu-

nostaining with the 4G8 antibody and thioflavine S (ThioS)staining in TgCRND8 mice disclosed marked A� depositsoth in the cerebral parenchyma and in cerebral bloodessels at 6 months of age. Examination of amyloid plaqueeposition in the brains of TgCRND8 mice revealed thatBR treatment showed significant reductions in the numbernd area occupied by A� deposits in the coronal sections of

the cortex and hippocampus (Fig. 2a). Quantification of theimmunoreactivity showed significantly lower plaque burdenin BBR-treated animals compared with controls. BBR treat-ment at concentration of 25 mg/kg per day resulted in a 61%(p � 0.001) decrease in plaque burden compared withontrol animals (Fig. 2b), whereas 100 mg/kg per day ofBR treatment reduced A� plaque burden to 43% of un-

reated control levels (p � 0.05) (Fig. 2b).To further measure which subsets of A� plaques were

reduced morphometric analyses of A� plaques were per-ormed on whole brain. The 2-way ANOVA revealed bothtreatment [F(2,68) � 17.4; p � 0.001] and a size effect

F(3,68) � 292.6; p � 0.001] in the mean number of A�plaques. Post hoc multiple analysis revealed that BBR (25mg/kg)-treated TgCRND8 mice showed a significant reduc-tion in large (�50), medium (25–50 �m), small sized (�25�m), and total amyloid plaque subsets; percentage reduc-tion in large, medium, small, and total subsets are 60% (p �0.001), 41% (p � 0.05), 35% (p � 0.055), and 37% (p �0.05), respectively (Fig. 2c). However, a 100 mg/kg per dayof BBR dose produced only a nonsignificant 21%–28%

reduction of amyloid plaque subsets (Fig. 2c).

bta(TtaistaBsC

rFdTt

c

c

* by Fis

7S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

Moreover, consistent with the indication that BBR treat-ment reduces the accumulation of total A� peptides in therain, we also found a significant reduction in ThioS-posi-ive fibrillar A� in the brains of the same TgCRND8 micet 25 mg/kg per day (40%; p � 0.05) or 100 mg/kg per day51%; p � 0.05), relative to age-matched untreated controlgCRND8 mice, as examined stereologically (Supplemen-

ary Fig. 2a and c). It has been reported that TgCRND8 micere more prone to vascular amyloid deposition in leptomen-ngeal and cortical blood vessels resembling human CAAymptoms (Hawkes and McLaurin, 2009). We also at-empted to determine whether BBR alters cerebral vascularmyloidosis in TgCRND8 mice. To determine the effect ofBR on cortical CAA severity, brain sections were

tained with ThioS to detect fibrillar A�. We found that

b

0.0

0.5

1.0

1.5

2.0

61%43%

******

VehicleBBR-25 mg/Kg/dBBR-100 mg/Kg/d

Plaq

ue lo

ad (%

)

a Vehicle BBR-25 mg/Kg/d

Fig. 2. Berberine (BBR) reduces cortico-hippocampal �-amyloid (A�) placoncentration of 25 (low dose) and 100 (high dose) mg/kg per day and sacrontrol and BBR-treated mice after immunohistochemical staining agains

captured and analyzed with ImageJ software (National Institutes of Healthalculated by the area of A� immunoreactivity expressed as a percentage

is shown for all overall half coronal hemisphere of mice (c). Values repreIn (b), *** p � 0.001; ** p � 0.01; * p � 0.05 versus control after 1-way** p � 0.001; * p � 0.05 versus control after 2-way ANOVA followed

AA-immunolabeled areas occupied 1%–1.5% in each

egion examined in vehicle-treated mice (Supplementaryig. 2b), and treatment with BBR showed a significantecrease in the intensity of staining and the number ofhioS-labeled blood vessels compared with vehicle-

reated mice (vehicle, 1.44 � 0.17% cortical area cov-ered; 25 mg/kg per day of BBR 0.70% � 0.11%; p �0.001); 100 mg/kg per day of BBR 0.89 � 0.02% (p �0.05) (Supplementary Fig. 2b and d).

The above findings of reduced 4G8-positive and ThioSA� deposits were further corroborated by A� ELISA anal-ysis in the opposite hemisphere of the brain. In particular,this assay disclosed that A� levels in both SDS-soluble andinsoluble formic acid fractions were significantly decreasedin BBR-treated TgCRND8 mice (Fig. 3a and b). Soluble A�was first extracted with 2% SDS, and the remaining A� was

c

<25 25-50 50-100 total0

20406080

100120140160

VehicleBBR-25 mg/Kg/dBBR-100 mg/Kg/d

*

35%

41%

*

***

*28%

21%60% 29%

37% 27%

Size of the plaque(µm)

BBR-100 mg/Kg/d

hology in the TgCRND8 mice. The TgCRND8 mice treated with BBR atfter 4 months of treatment. Representative figures of coronal sections fromith 4G8 antibody (a). Digital images from cortex and hippocampus wereda, MD, USA). Scale bar: 100 �m. The percentage of plaque burden wasarea. (b) Morphometric analysis (mean plaque subset number per mouse)up mean � standard error of the mean (SEM); n � 6–7 mice per group.

s of variance (ANOVA) followed by Fisher LSD post hoc analysis. In (c),her LSD post hoc analysis.

Mea

n pl

aque

num

ber

que patificed at A� w, Bethesof fullsent groanalysi

pelleted at 100,000g and extracted with 70% formic acid. In

4

(

ar

3

at5

TimA

tBbfirmp

Bcontro

8 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

the SDS-soluble fraction, the diminution of A�1–40 was4% (p � 0.011) and 32% (p � 0.051) by 25 and 100 mg/kg

per day, respectively, and there was slight but statisticallynot significant reduction of A�1–42 by BBR treatmentsFig. 3a). One-way ANOVA analyses of both A�1–40 and

A�1–42 in the formic acid fraction showed treatment ef-fects from both low and high concentrations of BBR. In theformic acid extractable fraction, the reductions of A�1–40were 49% (p � 0.01) and 54% (p � 0.01) (Fig. 3b), and thecorresponding reductions of A�1–42 were 44% (p � 0.01)nd 39% (p � 0.05) by 25 and 100 mg/kg per day of BBR,espectively.

.4. BBR ameliorates A�-associated reactive gliosis andastrocytosis in TgCRND8 mice

Next, we evaluated microgliosis and astrocytosis inTgCRND8 mice which are elevated phenotypically as aconsequence of amyloid deposition. Double immunohisto-chemical staining with Iba-1/GFAP and ThioS confirmedthat only activated microglial cells were in very close asso-ciation with compact plaques and absent in regions lackingsuch deposits (data not shown). Iba-1 is expressed in mi-croglia/macrophages, and is upregulated during activationof these cells (Sasaki et al., 2001). The degree of micro-gliosis as evaluated by Iba-1 load in the 3 brain regionsexamined was significantly amplified in vehicle-treatedTgCRND8 mice relative to WT mice (data not shown),whereas it was significantly reduced to 45% (p � 0.05) and27% (p � 0.05) in 25 and 100 mg/kg per day of BBR-treated TgCRND8 mice relative to vehicle-treatedTgCRND8 mice (Fig. 4a and c). Likewise, the magnitude ofstrocytosis as assessed by clusters of GFAP-immunoreac-ive astrocytes (GFAP burden) was significantly reduced to4% (p � 0.001) and 28% (p � 0.05) in 25 and 100 mg/kg

of BBR-treated TgCRND8 mice, respectively, relative to

a

Aβ1-40 Aβ1-420

2000

4000

6000

8000VehicleBBR-25 mg/Kg/dBBR-100 mg/Kg/d

*

44%32%

22%

SDS

solu

ble

Aβ 1

-40,

1-4

2

(pic

omol

es/g

)

Fig. 3. Berberine (BBR) treatment alters �-amyloid (A�) peptide levels inmg/kg per day) or vehicle for 4 months. SDS-soluble and formic acid-exsandwich enzyme-linked immunosorbent assay (ELISA). SDS-solubilized

oth A�40 and A�42 levels were decreased in the brains of BBR-treated(SEM). n � 6–7 mice per group. * p � 0.05, ** p � 0.01 versus vehicle

vehicle-treated TgCRND8 mice (p � 0.001) (Fig. 4b and d).

Activated microglial cells are typically associated with am-yloid plaques; thus the decrease in the Iba-1 burden corre-lates with reduced plaque burden.

3.5. BBR alters APP processing, APP, and tauphosphorylation probably through inhibition of Akt/GSK3activities in TgCRND8 mouse brain

To understand the mechanisms underlying the benefits ofBBR for cognitive function and A� neuropathology of

gCRND8 mouse, we first examined whether BBR cannfluence the processing of APP. We found that BBR treat-ent had no effect on full length APP, BACE-1, and AD-M-10 in brain homogenates (Fig. 5a and b and Supple-

mentary Fig. 3). Moreover BBR did not affect the levels ofthe A�-degrading enzymes neprilysin and IDE (Supplemen-ary Fig. 3). While BBR showed no influence on secretases,BR treatment (25 mg/kg per day) significantly reducedoth CTF-� and -� (Fig. 5a and b). Taken together, thesendings suggest that BBR’s A�-reducing effects might beelated to regulation of APP-CTF expression or intracellularaturation and distribution, rather than modulation of APP

rocessing enzymes or A� degradation.Several studies suggest that APP maturation, subcellular

distribution, and generation of A� are phosphorylation-dependent (Ando et al., 1999; da Cruz e Silva and da Cruze Silva, 2003; Iijima et al., 2000). It was also suggested thatAPP phosphorylation modulates its metabolism, resulting inincreased production of CTFs (Lee et al., 2003a). Therefore,we ascertained levels of APP phosphorylation in the brainlysates by immunoblot analysis using an antibody againstp-APP-threonine668. This study showed that levels of p-APP and p-CTF were lower in BBR-treated mice comparedwith the vehicle-treated TgCRND8 mice (Fig. 5a). Asshown in Fig. 5a, robust elevation of phosphorylated APPand CTFs were detected in vehicle-treated TgCRND8

Aβ1-40 Aβ1-400

2000

4000

6000

8000

10000VehicleBBR-25 mg/Kg/dBBR-100 mg/Kg/d

** *

** **

49% 54%

43% 39%

Form

ic a

cid

solu

ble

Aβ1-

40, 1

-42

(pic

omol

es/g

)

D8 mice. TgCRND8 mice were orally administered with BBR (25 or 100le A�40 and A�42 levels from cerebral hemispheres were measured byand A�42 levels (a). Formic acid-extractable A�40 and A�42 levels (b).

0.01) animals. Values denotes group mean � standard error of the meanl group.

b

TgCRNtractabA�40

(** p �

mice. In contrast, the APP and CTFs phosphorylation

ppdov

p

iwah

d per gro

9S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

were significantly reduced in BBR-treated TgCRND8mice (Fig. 5a). Quantitative analysis of immunoblots of

-APP and p-CTFs showed 46%–50% decrease in the-APP and p-CTFs in the low dose group, and 27%–28%ecrease in the p-APP and p-CTFs in the high dose groupf the BBR-treated TgCRND8 mice compared with theehicle (Fig. 5b).

Taking into account the inhibitory role of BBR in APPhosphorylation and A� accumulation, we sought to deter-

mine a possible role of BBR in tau hyperphosphorylation inTgCRND8 mice. It has been shown that amyloid accumu-lation in TgCRND8 mice is followed by hyperphosphory-lation of tau at different sites recognized by PHF-1, AT100,AT8, and CP13 antibodies (Bellucci et al., 2007). To eval-uate quantitatively the effects of BBR on tau hyperphos-phorylation, we assessed total tau and phosphorylationepitopes on tau, including PHF-1, AT8, and AT180, byWestern blot analysis. We observed a robust reduction inthe phosphotau recognized by PHF-1, AT8, and AT180 inthe brain homogenates of BBR-treated TgCRND8 mice(Fig. 5c). This finding is based on earlier work demonstrat-ing that these sites are phosphorylated by GSK3� (Cho andJohnson, 2003). Quantitative analysis of the Western blotbands of the phosphorylated tau showed 26%, 30%, and42% decreases at PHF-1, AT8, and AT180 phosphoe-pitopes, respectively, in the BBR-treated TgCRND8 micerelative to vehicle-treated TgCRND8 mice (Fig. 5d). Inaddition, Western blotting with a tau-1 antibody, recogniz-ing the dephosphorylated tau at Ser198/Ser199/Ser202,

b

a Vehicle BBR-25 mg/Kg/d

Fig. 4. Berberine (BBR) reduces microgliosis and astrogliosis in the TgCmolecule 1 (Iba-1) (a) and glial fibrillary acidic protein (GFAP) (b) in themg/kg per day) mice. The Iba-1 burden (%) (c) and the GFAP burden (%enotes group mean � standard error of the mean (SEM). n � 6–7 mice

showed a significant increase in tau-1 immunoreactivity in

BBR-treated TgCRND8 mice. No significant difference inthe total tau levels was observed between the groups (Fig.5c and d). Both APP and tau hyperphosphorylation weresignificantly increased in vehicle-treated TgCRND8 micecompared with the WT controls (Supplementary Fig. 4).

Cyclin-dependent kinase 5 (Cdk5), GSK3, and JNK arethe 3 key kinases involved in the phosphorylation of APP(Aplin et al., 1996; Iijima et al., 2000; Judge et al., 2011;Muresan and Muresan, 2005). Because BBR reduced theaccumulation of CTFs and inhibited the phosphorylation ofAPP and APP-CTFs, we sought to determine whether BBRplays a role in regulating Cdk5 and/or GSK3. It is knownthat GSK3 is activated through the phosphorylation atTyr216 or is inhibited when Ser9 is phosphorylated (Frameand Cohen, 2001). Cdk5 activity is enhanced when Cdk5 isphosphorylated at Y15 and associates with p35, signifi-cantly increasing its kinase activity (Patzke et al., 2003;Sato et al., 2008). Thus, we examined the levels of phos-phorylated Cdk5, GSK3�, and JNK using specific antibod-ies. As shown in Fig. 5e, a profound enhancement of inac-tive form of GSK3� phosphorylated at Ser9 was observedn the brains of the BBR-treated TgCRND8 mice comparedith vehicle-treated TgCRND8 controls. Quantitative

nalysis of the Western blot bands indicated a 50% en-ancement in the phosphorylated GSK3� and � at Ser 21

and 9 in the BBR-treated TgCRND8 mice relative to vehi-cle-treated TgCRND8 mice (Fig. 5e and f). These datasuggest that BBR treatment downregulates GSK3 activity.Immunoblot analysis with antibodies against the phosphor-

c

0.0

0.5

1.0

1.5

45%

*

27 %

Vehicle25 mg/Kg/d100 mg/kg/d

Iba-

1 bu

rden

(% )

BBR-100 mg/Kg/d

d

0

2

4

6

8

10

***

*54%

28%

VehicleBBR-25 mg/Kg/dBBR-100 mg/Kg/d

GFA

P bu

rden

(%)

mice. Immunohistochemical analysis of ionized calcium binding adaptersections of a vehicle control-treated mouse and BBR-treated (25 and 100

ere calculated by quantitative image analysis. Scale bar: 100 �m. Valuesup. * p � 0.05 and *** p � 0.001 versus vehicle control group.

RND8coronal) (d) w

ylated Cdk5 and JNK showed no significant differences in

lcs

h

ohe mean

10 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

p-Cdk5 (Y15) and p-stress-activated protein kinase/Jun-amino-terminal kinase (Thr183/Tyr185) respectively be-tween the BBR- and vehicle-treated TgCRND8 mice (Sup-plementary Fig. 3). Because the Akt/GSK3 pathway isfound to be dysregulated in AD (Beaulieu et al., 2009;Chung, 2009; Ryder et al., 2004), we determined the levelsof inactive GSK3 and active Akt in the TgCRND8 mousebrains. The levels of phosphorylated Akt (p-Akt) inTgCRND8 mice that received BBR was significantly highfrom that of the vehicle-treated TgCRND8 mice, whichshowed a significant reduction in p-Akt levels (normalizedto total Akt) (Fig. 5e and f). Increased phosphorylation ofSer9 in GSK3� indicates decreased activity of GSK3�,whereas p-Akt in Ser473 indicates increased activity of Akt.These results suggest that the level of the active GSK3� isincreased in TgCRND8 mouse brains and that BBR treat-

Vehicle BBR-25 mg/Kg/d BBR-100 mg

100-

15-

55-40-

70-

15-

10-

100-

10-

55-

40-

70-

a

55-

55-

55-c

40-

55-

55-

70-

e

Fig. 5. Alteration of amyloid precursor protein (APP)-C-terminal fragmexpression of glycogen synthase kinase (GSK)3� in TgCRND8 mice by B(Fl) APP, CTFs (CTF� and CTF�), p-APP(Thr668) and p-CTFs (a). Denomogenates of TgCRND8 mice treated with BBR (25 and 100 mg/kg per

of phosphorylated (p)-tau at PHF-1, AT8, and AT180, dephosphorylated TAT180, and Tau-1 (d). Representative immunoblots of phosphorylated GSKf pGSK3 (Ser21/9), GSK3�/�, p-Akt (Ser473), and AKT (f). *** p � 0.0

of variance (ANOVA). Values denotes group mean � standard error of t

ment increases the activity of protective Akt pathway. The

enhanced levels of pGSK3� and p-Akt were not ascribableto an augment in total GSK3�/� or total Akt, because theamount of total GSK3�/� and total Akt was unaltered byBBR treatment.

3.6. BBR alters APP processing, APP and tauphosphorylation through PI3K/Akt/GSK3 pathway in vitro

Given the inhibitory role of BBR in the phosphorylationof GSK3 in TgCRND8 mouse brain, we further evaluatedphosphorylation of both GSK3� and � at Ser21 and Ser9evel in N2a-SwedAPP cells following BBR treatment. Be-ause GSK3 activity is inhibited via phosphorylation atpecific serine residues (Ser9 for GSK3� and Ser21 for

GSK3�) by Akt, we also investigated the effect of BBR onthe protein levels of phosphorylated Akt. The amount of

0.0

0.5

1.0

1.5

2.0VehicleBBR 25 mg/Kg/dBBR 100 mg/Kg/d

Fl-APP CTFsp-APP p-CTFs

**

*** **

****

sign

al/β

-act

in ra

tio

-Fl-APP

-CTFβ-CTFα

-β-actin

-pGSK-3α(Ser21)

-pGSK-3β(Ser9)

-p-APP(T688)

-p-CTFβ-p-CTFα

-GSK-3α-GSK-3β

-p-AKT(Ser473)

-AKT

-PHF-1

-AT8

-AT180

-Tau-1

-Tau-5

b

0.0

0.5

1.0

1.5VehicleBBR 25mg/Kg/dBBR 100 mg/Kg/d

* *

* ** * *****

PHF-1 AT8 AT180 Tau-1

sign

al/T

au-5

ratio

-BACE-1

0.0

0.5

1.0

1.5

VehicleBBR 25 mg/Kg/dBBR 100 mg/Kg/d

pGSK3α(Ser21)/GSK3α

pGSK3β(Ser9)/GSK3β

pAkt/Akt

***

*

****

*

ratio

d

f

TFs), phosphorylation of APP, phosphorylated (p)-CTFs and tau, andtment. Representative immunoblot demonstrating the levels of full-lengthic analysis of the ratio of Fl-APP, CTFs, p-APP and p-CTFs in the brainvehicle (b). �-actin used as a loading control. Representative immunoblotsd total tau Tau-5 (c). Densitometric analysis of the ratio of PHF-1, AT8,1/9), GSK3�/�, p-Akt (Ser473), and AKT (e). Image analysis of the levels� 0.05 compared with vehicle-treated TgCRND8 mice by 1-way analysis(SEM). n � 6 mice per group.

/Kg/d

ents (CBR treasitometrday) orau-1 an3 (Ser201, * p

phosphorylated GSK3 increased during the first 30 minutes

occ

ptWCmrc

t

o

s inutes)

11S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

of BBR treatment, and remained higher than in the controlgroup at times exceeding 3 hours following BBR treatment(Fig. 6b and d). BBR also increased the phosphorylation

f Akt at Ser 473 and GSK3 at Ser21 and Ser9 in aoncentration-dependent manner with significant in-reases achieved at BBR concentrations of 10 and 20

�M, but it did not significantly affect cellular total Aktand GSK3� levels (Fig. 6a and c).

To study the effect of BBR on APP processing and tauhosphorylation in N2a-SwedAPP cells, we first examinedhe levels of CTFs, p-APP and PHF-1 in cell lysates by

estern blot analysis. As shown in Fig. 7a, the level ofTFs from the N2a-SwedAPP cells was reduced by treat-ent with BBR in a concentration-dependent manner,

eaching maximal reduction by 55% of basal level at aoncentration of 20 �M. At low concentration of BBR (5

�M), there was a 24% reduction in the level of CTFs but itwas not a statistically significant effect. Western blot anal-ysis for p-APP in N2a-SwedAPP cell lysates enabled us tocompare APP phosphorylation and its processing. We em-

-pGSK-3α(Ser21)

Time

-pGSK-3β-(Ser9)

-GSK3β-GSK3α

-p-AKT

-AKT

0 1.25 2. 5 5 10 20BBR (µm) :

0 1.25 2.5 5 10 200.0

0.5

1.0

1.5

Conc. of BBR (µM)

*****

*

ratio

of p

-GSK

3β/G

SK3β

0 1.25 2.5 5 10 200.0

0.5

1.0

1.5

2.0

2.5

Conc. of BBR (µM)

***

*

ratio

of p

-GSK

3α/G

SK3α

N2a-SwedAPP cells/BBR

0 1.25 2.5 5 10 200.0

0.5

1.0

1.5

Conc. of BBR (µM)

*********

ratio

of p

-AK

T/A

KT

55-

40-

55-40-

70-

70-

a

c

Fig. 6. Berberine (BBR) reduces glycogen synthase kinase (GSK)3 activaylation of Akt and inactive GSK3� and GSK3� expression in vitro. N2a-Sr with a 20-�M concentration of BBR over a range of time points (b, d).

anti-pGSK3�/� (Ser21/9) or using anti-Akt and anti-GSK3 polyclonal antibshows the ratio of pAkt to total Akt and ratio of pGSK3�/� to total GSKanalysis of variance (ANOVA) followed by post hoc comparison showedpecific concentration- or time-group and the control group (0 �M or 0 m

ployed the anti-p-APP-Thr668 antibody, which specifically

identifies only this phosphorylated site of the protein, toassess BBR’s inhibitory effect. There was an apparentdecrease in the phosphorylated level of APP by BBRtreatment in a similarly dose-dependent fashion as dem-onstrated by Western blot analysis (Fig. 7a). At 20 �M,he level of p-APP was decreased by 75% and at 10 �M

by 50% (Fig. 7a and b). It has been shown in AD modelsthat overexpression of Swedish mutant APP (SwedAPP)increases p-tau (Wang et al., 2006). We next tested theeffect of BBR on abnormal tau phosphorylation at PHF-1epitope in N2a-SwedAPP cells. Compared with the con-trol, the level of tau bound to PHF-1 was decreased to30% and 50% by treatment of 10 and 20 �M concentra-tions of BBR, respectively (Fig. 7a and b). The dataindicate that hyperphosphorylation of tau at Ser-396/404was increased in N2a-SwedAPP cells and the abnormalmodification of tau at this epitope was significantly in-hibited by BBR.

PI3K has a regulatory role in the trafficking of manyproteins including APP (PI3K inhibitors increase in-

0 30 60 90 120 150 180

0 30 60 90 120 150 1800.0

0.5

1.0

1.5

BBR treatment (min)

**

** *

ratio

of p

-GSK

3α/G

SK3α

0 30 60 90 120 150 1800.0

0.5

1.0

1.5

2.0

BBR treatment (min)

***** *** *** ***

ratio

of p

-GSK

3β/G

SK3β

N2a-SwedAPP cells/BBR (20 µM)

-p-AKT

-AKT

-pGSK-3α(Ser21)-pGSK-3β-(Ser9)

-GSK3α-GSK3β

0-

70-

5-

40-

55-40-

0 30 60 90 120 150 1800.0

0.5

1.0

1.5

2.0

**** ***

***

BBR treatments (mins)

ratio

of p

-AK

T/A

KT

R induces concentration- and time-dependent increases in the phosphor-cells were treated with different concentrations of BBR for 3 hours (a, c)

ed lysates were subjected to Western blot (WB) analysis using anti-pAkt,ich recognize total Akt and GSK3�/�, respectively. Densitometry analysiss indicated beside the figures (n � 3 for each condition) (c, d). One-wayficant difference (* p � 0.05, ** p � 0.01, and *** p � 0.001) between.

(min):

b

d

7

5

tion. BBwedAPPPreparody wh3�/� aa signi

tracellular APP-CTFs) (Petanceska and Gandy, 1999;

defP

Pwi

aa

ision).

12 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

Shineman et al., 2009). Because BBR showed a dose- andtime-dependent increase in p-Akt and inactive pGSK3��with decrease in p-APP, CTFs, and PHF-1 (Fig. 6a andb), we further determined whether pretreatment of PI3Kinhibitor (LY294002) in N2a-SwedAPP cells abolishedthe BBR-induced reduction of p-APP, CTFs, and PHF-1.To determine the CTFs, p-APP and PHF-1 reducing ef-fect of BBR, N2a-SwedAPP cells were cotreated with 20�M of BBR together with the PI3K inhibitor, LY294002,used at 20 �M. As expected, BBR-induced reduction ofp-APP, CTFs and PHF-1 were reversed by the cotreat-ment of LY294002 (Fig. 7c and d). To further analyze theBBR effect, we investigated p-APP in N2a-SwedAPP

N2a-SwedAPP cells/BBR

-PHF-1

- -actin

-Fl-APP

-p-APP

BBR (µm) : 0 5 10 20

-CTF-β-CTF-α

5baRPPAp

BBR(20µLy294002 (20

0 5 10 200

1

2

3p-APPT668CTFsPHF-1

**

***

***

*

**

Concentration of BBR (µ M)

ratio

toβ

-act

in

A

Control

LY294002+

BBR

BBR

A1 2A

B1 2B

C1 2C

100-

15-

10-

55-

100-

1

11

1

40-

a

b

e

c

Fig. 7. Treatment of N2a-SwedAPP with berberine (BBR) results in the inHF-1. Cell lysates were extracted from N2a-SwedAPP that were treated were analyzed for amyloid precursor protein (APP), p-APP, CTFs, and PH

s via PI3K/Akt activity (c). N2a-SwedAPP cells treated with PI3K inhibitowere analyzed for APP, p-APP, CTFs, and PHF-1 by Western blot. Densitos indicated in panels to the right (d). Values represent as mean � standanalysis of variance (ANOVA) revealed significant differences between B

Analysis of p-APP distribution in N2a-SwedAPP cells in the endosomal comWA, USA) (e). After treatment for 3 hours with dimethyl sulfoxide DMSpretreatment with 20 �M LY294002 (C1–C4), cells were fixed and immuusing Alexa 488 (red) and 594 (red) fluorophores respectively. Nuclei werthe Softmax 4.8 software for Delta Vision imaging system (Applied Prec

cells by immunofluorescence. We found that p-APP- t

positive vesicles could be labeled by the endosome mark-ers Rab5 (Fig. 7e [A1–A4]) as determined earlier (Lee etal., 2003a). After 3-hour exposure to 20 �M BBR, p-APPstaining was markedly reduced in the endosomal com-partments (Fig. 7e [B1–B4]). BBR-induced reduction ofp-APP was significantly inhibited by LY294002 in endo-somal compartments of N2a-SwedAPP cells (Fig. 7e[C1–C4]). This confirmed the BBR mediated p-APP re-duction seen with Western blots (Fig. 7c and d). These

ata showed that the reduction of p-APP staining insidendosomal compartments is due to BBR’s inhibitory ef-ect on p-APP. Altogether, these results indicate thatI3K/Akt/GSK3 is involved in the BBR-induced reduc-

DAPI merge

- + + - - - + +

-p-APP

- -actin

-Fl-APP

-PHF-1

-CTF-β-CTF-α

N2a-SwedAPP cells

Control

BBR

BBR+Ly2

9400

2

LY2940

02

Control

BBR

BBR+Ly2

9400

2

LY2940

02

Control

BBR

BBR+Ly2

9400

2

LY2940

020.0

0.5

1.0

1.5

2.0

2.5

p-APPCTFsPHF-1

** ##

**

#**

#

sign

al/ β

-act

in r

atio

4A3A

4B3B

4C3C

d

n of phosphorylated (p)-APP (Thr668), C-terminal fragments (CTFs) andR at different concentrations as indicated for 12 hours, and the cell lysatesWestern blot (a, b). BBR-mediated reduction of p-APP, CTFs, and PHF-14002 for 1 hour prior to the exposure of BBR (20 �M), and the cell lysatesnalysis shows the band density ratio of p-APP, CTFs, and PHF-1 to �-actinr of the mean (SEM) from 3 independent experiments (n � 3). One-wayted cells and control cultures (* p � 0.05, ** p � 0.01, *** p � 0.001).nts by Delta Vision fluorescence microscopy (Applied Precision, Issaquah,

cle control (A1–A4), 20 �M BBR (B1–B4), or 20 �M BBR after 1 houred using p-APP and Rab-5 (endosome marker) antibodies and visualizedlized using DAPI staining. Scale bar: 5 �m. Staining was analyzed using

M) : µM) :

00-

5-0-

55-

00-

40-

hibitioith BB

F-1 byr LY29metry ard erroBR-treapartme

O vehinostaine visua

ion of p-APP, CTFs, and PHF-1.

opdBpdSdwd

bm4st(simn

LBtf

b2tci

s

T

13S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

4. Discussion

Although BBR is a well known neuroprotective agent (Jiand Shen, 2011; Kulkarni and Dhir, 2010; Ye et al., 2009),the actual therapeutic role of BBR in AD pathology has notyet been evaluated. BBR has been shown to inhibit theproduction of A� in H4-SwedAPP cells (Asai et al., 2007);however, its effect on A� accumulation in vivo has not beendocumented. The data from this study represent the firstevidence that regular administration of BBR can prevent theage-related cognitive impairments and A� accumulationbserved in TgCRND8 mice with an early-onset AD-likeathology (Chishti et al., 2001). Several lines of our evi-ence suggest that BBR has beneficial effects in AD. First,BR readily passed through the blood-brain barrier, and itsresence in brain tissue of treated mice was definitivelyetermined by liquid chromatography-mass spectrometry.econd, dose-specific effects on brain A� levels, cognitiveeficits, amyloid neuropathology, and accelerated gliosisere found in the TgCRND8 mouse model of Alzheimer’sisease. Third, BBR significantly reduced the A� and CTFs,

probably by downregulating the phosphorylation of APPand of CTFs via the activation of the PI3K/Akt/GSK3pathway.

Both low and high doses of BBR significantly reducedthe cognitive impairment characteristic of AD, both in theconventional reference memory MWM task and memoryretention task (probe trial) of the mice (Fig. 1c and d). Theseresults are consistent with the amelioration of hippocampal-dependent memory function (McDonald and White, 1994).The cognitive benefits of long-term BBR administration didnot involve significant side effects on sensorimotor func-tion, as evidenced by visible platform training (Fig. 1b).

Notably, BBR-induced decreases in memory deficits areaccompanied not only by a significant reduction in theamyloid burden but also by an evident plaque fragmentationin the brains of transgenic mice. In particular, A� plaques ofall 3 size subsets (�25, 25–50, and �50 �m) were signif-icantly decreased in the brains of BBR-treated TgCRND8mice (Fig. 2c). The magnitude of this decrease in the num-er of plaques is similar to the decrease found in brains ofice treated with arundic acid (Mori et al., 2006). The

3%–63% reduction in amyloid burden observed in thistudy is also comparable with that found in studies of micereated with curcumin or epigallocatechin 3-gallate (ECGC)Lim et al., 2001; Rezai-Zadeh et al., 2005). The presenttudy also includes the important finding that BBR admin-stration can reduce vascular amyloids as well as parenchy-al amyloids (Supplementary Fig. 2a and b). Apart from

euronal A� deposition, previous studies have shown thatthe pattern of vascular amyloid deposition in leptomenin-geal and small cortical blood vessels observed in TgCRND8mice mirrors that typically found in human age-relatedCAA (Chishti et al., 2001; Hawkes and McLaurin, 2009).

BBR treatment had significant effects in reducing ThioS-

positive vascular amyloids (Supplementary Fig, 2b and d),and might reduce CAA through reduction of A� production.Effects reported in the present study are similar to thereported effects of nicotine administration where percentageof vessels with A� was significantly reduced (Hellström-

indahl et al., 2004). Because regular administration ofBR reduces ThioS-positive plaques and insoluble pep-

ides, its effect may be on amyloid fibril formation. Weound there was a mild inhibition of A�1–42 fibril forma-

tion with an IC50 (the half maximal inhibitory concentra-tion) value of 23.5 �M in vitro (data not shown). It has alsoeen recently demonstrated that BBR at a concentration of0 �M showed 36.3% inhibition of A�1–42 fibril forma-ion in vitro (Shi et al., 2011). However, the effectiveoncentration of BBR used in the in vitro studies, approx-mately 20 �M, were 23 times less than those found in the

brain. Therefore, the exact molecular mechanism of BBR-mediated reduction of A� accumulation needs further study.

The strong association of reactive astrocytes and acti-vated microglia with amyloid deposition contributes to theprogressive course of AD because of amplification of alarge range of proinflammatory molecules that mediate, inpart, the neuronal loss detected in AD (Akiyama et al.,2000). We found that treating TgCRND8 mice with BBRresulted in a 45% reduction in microgliosis and a 54%decrease in astrocytosis (Fig. 4a and b).

A similar marked decrease in activated microglia (57%)was also observed in ibuprofen-treated animals (Yan et al.,2003). Our results showing that BBR significantly reducedamyloid deposits along with amyloid plaque-associated re-active microgliosis and astrocytosis are in full agreementwith the above view. However, this does not essentiallyentail a unidirectional relationship between the 2 events.The observed reduction in gliosis by BBR is more likely tobe linked to its inhibition of A�. However, BBR appears touppress neuroinflammatory responses independent of A�

(Jeong et al., 2009; Lu et al., 2010). Therefore, the putativeaction of BBR needs to be determined in relation to themechanisms underlying both gliosis and amyloidosis.Nonetheless, because neuroinflammation is a risk factor forneurodegenerative disease, the anti-inflammatory effect ofBBR in the TgCRND8 mice supports the evidence for itstherapeutic potential for AD.

It has been demonstrated that successive extraction ofbrain homogenates using SDS and FA solubilizes the ma-jority of the total amyloid species from brains of Tg2576mice (Kawarabayashi et al., 2001). We have noted both lowand high doses of BBR significantly reduced both solubleand insoluble A� peptides (FA fraction) in the brain of

gCRND8 mice. This reduction in insoluble A� peptides iscomparable with earlier observed reduction after chronicnicotine administration in Swedish APP mice (Nordberg etal., 2002). The earlier observation that the nicotine treat-ment reduced the insoluble but not soluble A� peptides in

the brain of Tg2576 is in accordance with this result. How-

(p

m

m

At

Cn

AB

epipakc2blclmtp

usAA

14 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

ever, a low dose of BBR reduces SDS-soluble A�1–4044% reduction); therefore, BBR may also have an effect onotentially neurotoxic soluble A� oligomers. Overall, the

low dose BBR-induced reduction of total extracted A�40(44%–49%) and A�42 (22%–43%) in the brain hemi-spheres corresponds well with the decrease in plaque num-ber (37%) and plaque load (61%) in other brain hemispheresevaluated by immunostaining, indicating that these bio-chemical measures are an exact reflection of overall amy-loid load (Figs. 2 and 4). BBR treatment prevents spatial

emory reference deficits and A� pathology in vivo atdoses that are equivalent to or lower than the doses pre-scribed for humans for hypercholestermia (Kong et al.,2004). In this study we found that the low dosage of BBRshowed more significant effect than the high dosage inreducing A� pathology and gliosis (Figs. 2 and 4). A similarcase has been published, reporting the A�-reducing effect of

emantine in Tg2567 AD mice (Dong et al., 2008). Thus,it is possible that lower BBR doses may obstruct justenough neuronal activity to reduce amyloid generation,while higher dose obstructs so much neuronal activity thatA� clearance mechanisms are also reduced.

A possible mechanism of the reducing effect of BBR on� deposition is modulation of APP processing, because

he levels of APP-CTFs, the direct precursor of A� (DeStrooper and Annaert, 2000), were decreased by BBR treat-ment (Fig. 5a). APP processing can be modulated either byan altered APP expression or by the function of �-secretase(BACE-1). However, we did not notice a significant differ-ence in the protein levels of Fl-APP or BACE-1 between thegroups (Fig. 5a). It has been shown that APP phosphoryla-tion at position Thr668 facilitates the accumulation of CTFsand increases A� generation (Lee et al., 2003a). Severalstudies have revealed that APP maturation and targeting forproteolysis requires APP Thr668 phosphorylation (da Cruze Silva and da Cruz e Silva, 2003; Lee et al., 2003a).Phosphorylated APP undergoes fast anterograde axonaltransport to the nerve terminals, where �- and �-secretasemediated-cleavage occurs, resulting in A� release (Kamal etal., 2000; Lazarov et al., 2005; Lee et al., 2005). Thisindicates that BBR-mediated reduction of APP phosphory-lation might subsequently result in a reduction of maturep-APP in the trans-Golgi compartment and a lesser amountof APP directed to the distal axon, which would prevent A�production at synaptic terminals. These findings suggestthat BBR modulates APP processing through a mechanismbeyond regulating the expression of APP and BACE-1. Thisis consistent with a previous study showing that retionicacid reduced the A� pathology mainly via reducing CTFs,p-APP without influencing the levels of Fl-APP andBACE-1 (Ding et al., 2008). Similarly, indirubin-3-mon-oxime, a GSK3� inhibitor isolated from the traditional

hinese medicinal herb Indigo naturalis (Qing Dai in Chi-

ese) reduced A� accumulation and tau phosphorylation in

PP/PS-1 mice without affecting the levels of Fl-APP andACE-1 (Ding et al., 2010).

Based on the result that BBR treatment decreased thelevation of APP phosphorylation in TgCRND8 mice, weropose that BBR may prevent APP processing by inhibit-ng its phosphorylation. Among the several protein kinaseshosphorylating APP at Thr668 (Aplin et al., 1996; Iijima etl., 2000; Standen et al., 2001), GSK is considered to be aey kinase responsible for APP phosphorylation in neuronalells (Aplin et al., 1996; Chang et al., 2006; Judge et al.,011; Ploia et al., 2010). Suppressing GSK3 activity haseen demonstrated to decrease the generation and accumu-ation of A� in APP mice (Rockenstein et al., 2007). Thisoincides with our result showing a concomitant downregu-ation of GSK activity by BBR treatment in TgCRND8ice. In TgCRND8 mice and N2a-SwedAPP cells, BBR

reatment led to a trend of increased p-Akt and inactiveGSK3�/� protein levels (Figs. 5e and f, and 6). GSK3

activity is inhibited via phosphorylation at specific serineresidues (Ser 9 and Ser21) by Akt (Frame and Cohen,2001). It is also possible that the improved spatial learningcan be ascribed to the BBR-induced activation of the Aktpathway because activation of the Akt pathway is essentialfor the expression of long-term potentiation (Karpova et al.,2006; Sanna et al., 2002). There are several studies indicat-ing that the PI3K/Akt/GSK3� signaling pathway contrib-tes to a number of aspects of AD pathology. In vitrotudies suggest that PI3K and Akt protect neurons against� toxicity (Martín et al., 2001). GSK3� also modulatesPP processing thereby influencing the production of A�s

(Phiel et al., 2003). Liu et al. (2003) showed that inhibitionof PI3K contributes to tau phosphorylation and impairmentof spatial memory. GSK3� is strongly involved in tauphosphorylation at various sites including PHF-1, Tau-1,AT8, and AT270 epitopes (Lee et al., 2003b). It has beenshown that GSK3� activity increases in cell expressingSwedish APP mutation and in AD presenilin-1 and presni-lin-2 mutation lymphoblast cells via a downregulation ofboth Ser473 Akt and inactive Ser9 phosphorylated GSK3�(Ryder et al., 2004). Although underlying mechanisms ofPI3K/Akt/GSK3� signaling in AD are still unexplainedthey are presently being intensively investigated. We stud-ied the involvement of PI3K/Akt within the complex intra-cellular signaling pathway involved in the inhibition of APPprocessing. Pharmacological inhibition of the kinase activ-ity of PI3K not only reversed the BBR-induced reduction ofp-APP and PHF-1, but it also blocked the BBR-mediatedreduction of CTFs (Fig. 7b). We have been able to detectreduced p-APP staining in the endosomal compartment ofN2a-SwedAPP cells in immunohistochemistry (Fig. 7c),suggesting that the phosphorylation state of APP is animportant factor due to its association with BACE-1, whichmediates the amyloidogenic processing of APP to increaseA� through increased generation of the CTFs (Lee et al.,

2003a). Interference in PI3K/Akt/GSK3 pathway thus rep-

b

tp

D

BACR

A

Cc(pLpp1R(fKKcJpl

A

f2

R

15S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

resents a key mechanism through which BBR may decreaseAD pathology and improve cognitive function.

In conclusion, our data demonstrate that BBR is able toreduce cerebral A� levels, glial activation, and cognitiveimpairment in the TgCRND8 mouse model. In addition, wefound that BBR suppresses both CTFs and p-APPs levelsvia activating the PI3K/Akt/GSK3 signaling pathway,thereby precluding A� generation. Based on the safety andrain bioavailability of BBR, and its ability suppress to A�

levels, it appears to be a promising drug for the preven-tion and/or treatment of Alzheimer’s disease. The out-comes from our study will warrant further investigationof BBR-like alkaloids as candidates for A� and tau-basedherapeutics to modify or delay the onset of A� and tauathology in AD.

isclosure statement

The authors disclose no conflicts of interest.All animal procedures were approved by the Hong Kong

aptist University Committee on the Use of Human andnimal Subjects in Teaching and Research and by theommittee on the Use of Live Animals for Teaching andesearch (CULATR), the University of Hong Kong.

cknowledgements

The authors thank Prof. Edward Koo (UCSD, San Diego,A, USA) for providing the C-terminal anti-amyloid pre-ursor protein (APP) antibody CT15, Prof. Peter DaviesAlbert Einstein College of Medicine, Bronx, NY, USA) forroviding PHF-1 antibody against phosphorylated tau, and. Binder, (Northwestern University, Chicago, IL, USA) forroviding Tau-1 and Tau-5 antibodies. The study was sup-orted by research grants FRG-I/10-11/028 and FRG-II/10-1/035 from Hong Kong Baptist University to M. Li, and byGC GRF grant to J.D. Huang and Prof. Sookja K. Chung

GRF 732703). This work was partly supported by grantsrom an anonymous donor and the Lions Club of Southowloon, Hong Kong. The authors especially thank Prof.ei Maruyama and Masashi Asai (Department of Pharma-

ology, Faculty of Medicine, Saitama Medical University,apan) for their helpful comments and suggestions on thisroject. The authors also thank Martha Dahlen for her Eng-ish editing of this manuscript.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at doi:10.1016/j.neurobiolaging.012.02.016.

eferences

Akiyama, H., Barger, S., Barnum, S., Bradt, B., Bauer, J., Cole, G.M.,

Cooper, N.R., Eikelenboom, P., Emmerling, M., Fiebich, B.L., Finch,

C.E., Frautschy, S., Griffin, W.S., Hampel, H., Hull, M., Landreth, G.,Lue, L., Mrak, R., Mackenzie, I.R., McGeer, P.L., O’Banion, M.K.,Pachter, J., Pasinetti, G., Plata-Salaman, C., Rogers, J., Rydel, R., Shen,Y., Streit, W., Strohmeyer, R., Tooyoma, I., Van Muiswinkel, F.L.,Veerhuis, R., Walker, D., Webster, S., Wegrzyniak, B., Wenk, G.,Wyss-Coray, T., 2000. Inflammation and Alzheimer’s disease.Neurobiol. Aging 21, 383–421.

Ando, K., Oishi, M., Takeda, S., Iijima, K., Isohara, T., Nairn, A.C.,Kirino, Y., Greengard, P., Suzuki, T., 1999. Role of phosphorylation ofAlzheimer’s amyloid precursor protein during neuronal differentiation.J. Neurosci. 19, 4421–4427.

Aplin, A.E., Gibb, G.M., Jacobsen, J.S., Gallo, J.M., Anderton, B.H., 1996.In vitro phosphorylation of the cytoplasmic domain of the amyloidprecursor protein by glycogen synthase kinase-3beta. J. Neurochem.67, 699–707.

Asai, M., Iwata, N., Yoshikawa, A., Aizaki, Y., Ishiura, S., Saido, T.C.,Maruyama, K., 2007. Berberine alters the processing of Alzheimer’samyloid precursor protein to decrease Abeta secretion. Biochem. Bio-phys. Res. Commun. 352, 498–502.

Beaulieu, J.M., Gainetdinov, R.R., Caron, M.G., 2009. Akt/GSK3 signal-ing in the action of psychotropic drugs. Annu. Rev. Pharmacol. Toxi-col. 49, 327–347.

Bellucci, A., Rosi, M.C., Grossi, C., Fiorentini, A., Luccarini, I., Casa-menti, F., 2007. Abnormal processing of tau in the brain of agedTgCRND8 mice. Neurobiol. Dis. 27, 328–338.

Blokland, A., Geraerts, E., Been, M., 2004. A detailed analysis of rats’spatial memory in a probe trial of a Morris task. Behav. Brain Res. 154,71–75.

Boche, D., Zotova, E., Weller, R.O., Love, S., Neal, J.W., Pickering, R.M.,Wilkinson, D., Holmes, C., Nicoll, J.A., 2008. Consequence of A�

immunization on the vasculature of human Alzheimer’s disease brain.Brain 131, 3299–3310.

Chang, K.A., Kim, H.S., Ha, T.Y., Ha, J.W., Shin, K.Y., Jeong, Y.H., Lee,J.P., Park, C.H., Kim, S., Baik, T.K., Suh, Y.H., 2006. Phosphorylationof amyloid precursor protein (APP) at Thr668 regulates the nucleartranslocation of the APP intracellular domain and induces neurodegen-eration. Mol. Cell. Biol. 26, 4327–4338.

Chishti, M.A., Yang, D.S., Janus, C., Phinney, A.L., Horne, P., Pearson, J.,Strome, R., Zuker, N., Loukides, J., French, J., Turner, S., Lozza, G.,Grilli, M., Kunicki, S., Morissette, C., Paquette, J., Gervais, F.,Bergeron, C., Fraser, P.E., Carlson, G.A., George-Hyslop, P.S., West-away, D., 2001. Early-onset amyloid deposition and cognitive deficitsin transgenic mice expressing a double mutant form of amyloid pre-cursor protein 695. J. Biol. Chem. 276, 21562–21570.

Cho, E., 1990. Berberine hydrochloride, in: Pharmacopeia of People Re-public of China. Pharmacopoeia Commission of the Ministry of Healthof the People’s Republic of China, China, pp. 437–439.

Cho, J.H., Johnson, G.V., 2003. Glycogen synthase kinase 3beta phospho-rylates tau at both primed and unprimed sites. Differential impact onmicrotubule binding. J. Biol. Chem. 278, 187–193.

Chung, S.H., 2009. Abberant phosphorylation in the pathogenesis of Alz-heimer’s disease. BMB Report 42, 467–474.

da Cruz e Silva, E.F., da Cruz e Silva, O.A., 2003. Protein phosphorylationand APP metabolism. Neurochem. Res. 28, 1553–1561.

De Strooper, B., Annaert, W., 2000. Proteolytic processing and cell bio-logical functions of the amyloid precursor protein. J. Cell Sci. 113,1857–1870.

Ding, Y., Qiao, A., Fan, G.H., 2010. Indirubin-3’-monoxime rescues spa-tial memory deficits and attenuates �-amyloid-associated neuropathol-ogy in a mouse model of Alzheimer’s disease. Neurobiol. Dis. 39,156–168.

Ding, Y., Qiao, A., Wang, Z., Goodwin, J.S., Lee, E.S., Block, M.L.,Allsbrook, M., McDonald, M.P., Fan, G.H., 2008. Retinoic acid atten-uates beta-amyloid deposition and rescues memory deficits in an Alz-heimer’s disease transgenic mouse model. J. Neurosci. 28, 11622–

11634.

F

G

H

H

H

I

J

J

J

J

J

K

K

K

K

K

L

L

L

L

L

L

L

L

M

M

M

M

M

N

N

P

P

P

P

16 S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

Dong, H., Yuede, C.M., Coughlan, C., Lewis, B., Csernansky, J.G., 2008.Effects of memantine on neuronal structure and conditioned fear in theTg2576 mouse model of Alzheimer’s disease. Neuropsychopharmacology33, 3226–3236.

Durairajan, S.S., Liu, L.F., Lu, J.H., Koo, I., Maruyama, K., Chung, S.K.,Huang, J.D., Li, M., 2011. Stimulation of non-amyloidogenic process-ing of amyloid-� protein precursor by cryptotanshinone involves acti-vation and translocation of ADAM10 and PKC-�. J. Alzheimers Dis.25, 245–262.

rame, S., Cohen, P., 2001. GSK3 takes centre stage more than 20 yearsafter its discovery. Biochem. J. 35, 1–16.

riffiths, H.H., Morten, I.J., Hooper, N.M., 2008. Emerging and potentialtherapies for Alzheimer’s disease. Expert Opin. Ther. Targets. 12,693–704.

awkes, C.A., McLaurin, J., 2009. Selective targeting of perivascularmacrophages for clearance of beta-amyloid in cerebral amyloid angi-opathy. Proc. Natl. Acad. Sci. U. S. A. 106, 1261–1266.

ellström-Lindahl, E., Court, J., Keverne, J., Svedberg, M., Lee, M.,Marutle, A., Thomas, A., Perry, E., Bednar, I., Nordberg, A., 2004.Nicotine reduces A beta in the brain and cerebral vessels of APPswmice. Eur. J. Neurosci. 19, 2703–2710.

owes, M.J., Perry, E., 2011. The role of phytochemicals in the treatmentand prevention of dementia. Drugs Aging. 28, 439–468.

ijima, K., Ando, K., Takeda, S., Satoh, Y., Seki, T., Itohara, S., Greengard,P., Kirino, Y., Nairn, A.C., Suzuki, T., 2000. Neuron-specific phos-phorylation of Alzheimer’s beta-amyloid precursor protein by cyclin-dependent kinase 5. J. Neurochem. 75, 1085–1091.

eong, H.W., Hsu, K.C., Lee, J.W., Ham, M., Huh, J.Y., Shin, H.J., Kim,W.S., Kim, J.B., 2009. Berberine suppresses proinflammatory re-sponses through AMPK activation in macrophages. Am. J. Physiol.Endocrinol. Metab. 296, 955–964.

i, H.F., Shen, L., 2011. Berberine: a potential multipotent natural productto combat Alzheimer’s disease. Molecules 16, 6732–6740.

osephs, K.A., Whitwell, J.L., Ahmed, Z., Shiung, M.M., Weigand, S.D.,Knopman, D.S., Boeve, B.F., Parisi, J.E., Petersen, R.C., Dickson,D.W., Jack, C.R., Jr., 2008. Beta-amyloid burden is not associated withrates of brain atrophy. Ann. Neurol. 63, 204–212.

udge, M., Hornbeck, L., Potter, H., Padmanabhan, J., 2011. Mitosis-specific phosphorylation of amyloid precursor protein at Threonine 668leads to its altered processing and association with centrosomes. Mol.Neurodegener. 6, 80.

ung, H.A., Min, B.S., Yokozawa, T., Lee, J.H., Kim, Y.S., Choi, J.S.,2009. Anti-Alzheimer and antioxidant activities of Coptidis Rhizomaalkaloids. Biol. Pharm. Bull 32, 1433–1438.

amath, S., Skeels, M., Pai, A., 2009. Significant differences in alkaloidcontent of Coptis chinensis (Huanglian), from its related Americanspecies. Chin. Med. 4, 17.

arpova, A., Sanna, P.P., Behnisch, T., 2006. Involvement of multiplephosphatidylinositol 3-kinase-dependent pathways in the persistence oflate-phase long term potentiation expression. Neuroscience 137, 833–841.

awarabayashi, T., Younkin, L.H., Saido, T.C., Shoji, M., Ashe, K.H.,Younkin, S.G., 2001. Age-dependent changes in brain, CSF, andplasma amyloid (beta) protein in the Tg2576 transgenic mouse modelof Alzheimer’s disease. J. Neurosci. 21, 372–381.

ong, W., Wei, J., Abidi, P., Lin, M., Inaba, S., Li, C., Wang, Y., Wang,Z., Si, S., Pan, H., Wang, S., Wu, J., Wang, Y., Li, Z., Liu, J., Jiang,J.D., 2004. Berberine is a novel cholesterol-lowering drug workingthrough a unique mechanism distinct from statins. Nat. Med. 10,1344–1351.

ulkarni, S.K., Dhir, A., 2010. Berberine: a plant alkaloid with therapeuticpotential for central nervous system disorders. Phytother. Res. 24,317–324.

azarov, O., Morfini, G.A., Lee, E.B., Farah, M.H., Szodorai, A., DeBoer,S.R., Koliatsos, V.E., Kins, S., Lee, V.M., Wong, P.C., Price, D.L.,

Brady, S.T., Sisodia, S.S., 2005. Axonal transport, amyloid precursor

protein, kinesin-1, and the processing apparatus: revisited. J. Neurosci.25, 2386–2395.

ee, C.W., Lau, K.F., Miller, C.C., Shaw, P.C., 2003b. Glycogen synthasekinase-3 beta-mediated tau phosphorylation in cultured cell lines. Neu-roreport 14, 257–260.

ee, E.B., Zhang, B., Liu, K., Greenbaum, E.A., Doms, R.W., Tro-janowski, J.Q., Lee, V.M., 2005. BACE overexpression alters thesubcellular processing of APP and inhibits Abeta deposition in vivo.J. Cell Biol. 168, 291–302.

ee, M.S., Kao, S.C., Lemere, C.A., Xia, W., Tseng, H.C., Zhou, Y., Neve,R., Ahlijanian, M.K., Tsai, L.H., 2003a. APP processing is regulated bycytoplasmic phosphorylation. J. Cell Biol. 163, 83–95.

im, G.P., Chu, T., Yang, F., Beech, W., Frautschy, S.A., Cole, G.M.,2001. The curry spice curcumin reduces oxidative damage and amyloidpathology in an Alzheimer transgenic mouse. J. Neurosci. 21, 8370–8377.

iu, S.J., Zhang, A.H., Li, H.L., Wang, Q., Deng, H.M., Netzer, W.J., Xu,H., Wang, J.Z., 2003. Overactivation of glycogen synthase kinase-3 byinhibition of phosphoinositol-3 kinase and protein kinase C leads tohyperphosphorylation of tau and impairment of spatial memory.J. Neurochem. 87, 1333–1344.

u, D.Y., Tang, C.H., Chen, Y.H., Wei, I.H., 2010. Berberine suppressesneuroinflammatory responses through AMP-activated protein kinaseactivation in BV-2 microglia. J. Cell. 110, 697–705.

ukiw, W.J., 2008. Emerging amyloid beta (Ab) peptide modulators for thetreatment of Alzheimer’s disease (AD). Expert Opin. Emerg. Drugs 13,255–271.

artín, D., Salinas, M., López-Valdaliso, R., Serrano, E., Recuero, M.,Cuadrado, A., 2001. Effect of the Alzheimer amyloid fragmentAbeta(25-35) on Akt/PKB kinase and survival of PC12 cells. J. Neu-rochem. 78, 1000–1008.

cDonald, R.J., White, N.M., 1994. Parallel information processing in thewater maze: evidence for independent memory systems involving dor-sal striatum and hippocampus. Behav. Neural Biol. 61, 260–270.

ori, T., Town, T., Tan, J., Yada, N., Horikoshi, Y., Yamamoto, J.,Shimoda, T., Kamanaka, Y., Tateishi, N., Asano, T., 2006. ArundicAcid ameliorates cerebral amyloidosis and gliosis in Alzheimer trans-genic mice. J. Pharmacol. Exp. Ther. 318, 571–578.

orris, R., 1984. Developments of a water-maze procedure for studyingspatial learning in the rat. J. Neurosci. Methods 11, 47–60.

uresan, Z., Muresan, V., 2005. c-Jun NH2-terminal kinase-interactingprotein-3 facilitates phosphorylation and controls localization of amy-loid-beta precursor protein. J. Neurosci. 25, 3741–3751.

egoescu, A., Labat-Moleur, F., Lorimier, P., Lamarcq, L., Guillermet, C.,Chambaz, E., Brambilla, E., 1994. F(ab) secondary antibodies: a gen-eral method for double immunolabeling with primary antisera from thesame species. Efficiency control by chemiluminescence. J. Histochem.Cytochem. 42, 433–437.

ordberg, A., Hellström-Lindahl, E., Lee, M., Johnson, M., Mousavi, M.,Hall, R., Perry, E., Bednar, I., Court, J., 2002. Chronic nicotine treat-ment reduces beta-amyloidosis in the brain of a mouse model ofAlzheimer’s disease (APPsw). J. Neurochem. 81, 655–658.

atzke, H., Maddineni, U., Ayala, R., Morabito, M., Volker, J., Dikkes, P.,Ahlijanian, M.K., Tsai, L.H., 2003. Partial rescue of the p35-/- brainphenotype by low expression of a neuronal-specific enolase p25 trans-gene. J. Neurosci. 23, 2769–2778.

eng, W.H., Lo, K.L., Lee, Y.H., Hung, T.H., Lin, Y.C., 2007. Berberineproduces antidepressant-like effects in the forced swim test and in thetail suspension test in mice. Life Sci. 81, 933–938.

etanceska, S.S., Gandy, S., 1999. The phosphatidylinositol 3-kinase in-hibitor wortmannin alters the metabolism of the Alzheimer’s amyloidprecursor protein. J. Neurochem. 73, 2316–2320.

hiel, C.J., Wilson, C.A., Lee, V.M., Klein, P.S., 2003. GSK-3alpharegulates production of Alzheimer’s disease amyloid-beta peptides.

Nature 423, 435–439.

R

R

R

S

S

S

S

S

S

S

T

V

W

W

W

W

Y

Y

Y

Z

17S.S.K Durairajan et al. / Neurobiology of Aging xx (2012) xxx

Ploia, C., Sclip, A., Colombo, A., Repici, M., Gardoni, F., Di Luca, M.,Forloni, G., Antoniou, X., Borsello, T., 2010. Role of glycogen Syn-thase Kinase-3� in APP hyperphosphorylation induced by NMDAstimulation in cortical neurons. Pharmaceuticals 3, 42–58.

ezai-Zadeh, K., Shytle, D., Sun, N., Mori, T., Hou, H., Jeanniton, D.,Ehrhart, J., Townsend, K., Zeng, J., Morgan, D., Hardy, J., Town, T.,Tan, J., 2005. Green tea epigallocatechin-3-gallate (EGCG) modulatesamyloid precursor protein cleavage and reduces cerebral amyloidosis inAlzheimer transgenic mice. J. Neurosci. 25, 8807–8814.

ockenstein, E., Torrance, M., Adame, A., Mante, M., Bar-on, P., Rose,J.B., Crews, L., Masliah, E., 2007. Neuroprotective effects of regula-tors of the glycogen synthase kinase-3beta signaling pathway in atransgenic model of Alzheimer’s disease are associated with reducedamyloid precursor protein phosphorylation. J. Neurosci. 27, 1981–1991.

yder, J., Su, Y., Ni, B., 2004. Akt/GSK3beta serine/threonine kinases:evidence for a signaling pathway mediated by familial Alzheimer’sdisease mutations. Cell. Signal. 6, 187–200.

anna, P.P., Cammalleri, M., Berton, F., Simpson, C., Lutjens, R., Bloom,F.E., Francesconi, W., 2002. Phosphatidylinositol 3-kinase is requiredfor the expression but not for the induction or the maintenance oflong-term potentiation in the hippocampal CA1 region. J. Neurosci. 22,3359–3365.

asaki, Y., Ohsawa, K., Kanazawa, H., Kohsaka, S., Imai, Y., 2001. Iba1is an actin-cross-linking protein in macrophages/microglia. Biochem.Biophys. Res. Commun. 286, 292–297.

ato, S., Xu, J., Okuyama, S., Martinez, L.B., Walsh, S.M., Jacobsen,M.T., Swan, R.J., Schlautman, J.D., Ciborowski, P., Ikezu, T., 2008.Spatial learning impairment, enhanced CDK5/p35 activity, and down-regulation of NMDA receptor expression in transgenic mice expressingtau-tubulin kinase 1. J. Neurosci. 28, 14511–14521.

elkoe, D.J., 2004. Cell biology of protein misfolding: the examples ofAlzheimer’s and Parkinson’s diseases. Nat. Cell Biol. 6, 1054–1061.

hi, A., Huang, L., Lu, C., He, F., Li, X., 2011. Synthesis, biologicalevaluation and molecular modeling of novel triazole-containing ber-berine derivatives as acetylcholinesterase and �-amyloid aggregationinhibitors. Bioorg. Med. Chem. 19, 2298–2305.

hineman, D.W., Dain, A.S., Kim, M.L., Lee, V.M., 2009. Constitutivelyactive Akt inhibits trafficking of amyloid precursor protein and amyloidprecursor protein metabolites through feedback inhibition of phospho-

inositide 3-kinase. Biochemistry 48, 3787–3794.

tanden, C.L., Brownlees, J., Grierson, A.J., Kesavapany, S., Lau, K.F.,McLoughlin, D.M., Miller, C.C., 2001. Phosphorylation of thr(668) inthe cytoplasmic domain of the Alzheimer’s disease amyloid precursorprotein by stress-activated protein kinase 1b (Jun N-terminal kinase-3).J. Neurochem. 76, 316–320.

hinakaran, G., Teplow, D.B., Siman, R., Greenberg, B., Sisodia, S.S.,1996. Metabolism of the “Swedish” amyloid precursor protein variantin neuro2a (N2a) cells. Evidence that cleavage at the “beta-secretase”site occurs in the golgi apparatus. J. Biol. Chem. 271, 9390–9397.

lahos, C.J., Matter, W.F., Hui, K.Y., Brown, R.F., 1994. A specificinhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002). J. Biol. Chem. 269, 5241–5248.

ang, Y.P., Wang, X.C., Tian, Q., Yang, Y., Zhang, Q., Zhang, J.Y.,Zhang, Y.C., Wang, Z.F., Wang, Q., Li, H., Wang, J.Z., 2006. Endog-enous overproduction of beta-amyloid induces tau hyperphosphoryla-tion and decreases the solubility of tau in N2a cells. J. Neural Transm.113, 1723–1732.

eller, R.O., Subash, M., Preston, S.D., Mazanti, I., Carare, R.O., 2008.Perivascular drainage of amyloid-beta peptides from the brain and itsfailure in cerebral amyloid angiopathy and Alzheimer’s disease. BrainPathol. 18, 253–266.

ilquet, V., De Strooper, B., 2004. Amyloid-beta precursor proteinprocessing in neurodegeneration. Curr. Opin. Neurobiol. 14, 582–588.

oodhouse, A., Vickers, J.C., Adlard, P.A., Dickson, T.C., 2009. Dystro-phic neurites in TgCRND8 and Tg2576 mice mimic human patholog-ical brain aging. Neurobiol. Aging 30, 864–874.

an, Q., Zhang, J., Liu, H., Babu-Khan, S., Vassar, R., Biere, A.L., Citron,M., Landreth, G., 2003. Anti-inflammatory drug therapy alters beta-amyloid processing and deposition in an animal model of Alzheimer’sdisease. J. Neurosci. 23, 7504–7509.

e, M., Fu, S., Pi, R., He, F., 2009. Neuropharmacological and pharma-cokinetic properties of berberine: a review of recent research. J. Pharm.Pharmacol. 61, 831–837.

u, C.J., Zheng, M.F., Kuang, C.X., Huang, W.D., Yang, Q., 2010.Oren-gedoku-to and its constituents with therapeutic potential in Alz-heimer’s disease inhibit indoleamine 2, 3-dioxygenase activity in vitro.J. Alzheimers Dis. 22, 257–266.

hu, F., Qian, C., 2006. Berberine chloride can ameliorate the spatialmemory impairment and increase the expression of interleukin-1betaand inducible nitric oxide synthase in the rat model of Alzheimer’s

disease. BMC Neurosci. 7, 78.