Revisiting the peripheral sink hypothesis: inhibiting BACE1 activity in the periphery does not alter...

Acc

epte

d A

rtic

le

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/jnc.12937

This article is protected by copyright. All rights reserved.

Received Date : 11-Jul-2014

Accepted Date : 03-Aug-2014

Article type : Original Article

Revisiting the peripheral sink hypothesis:

Inhibiting BACE1 activity in the periphery does not alter β-amyloid

levels in the CNS

Biljana Georgievska1, Susanne Gustavsson1, Johan Lundkvist1, Jan Neelissen1, Susanna

Eketjäll1, Veronica Ramberg1, Tjerk Bueters1, Karin Agerman1, Anders Juréus1, Samuel

Svensson1, Stefan Berg1, Johanna Fälting1 and Urban Lendahl2#

1Innovative Medicines AstraZeneca, CNS & Pain, SE-151 85 Södertälje, Sweden

2Department of Cell and Molecular Biology, Karolinska Institute, SE-171 77 Stockholm,

Sweden

Abbreviated title: BACE1 inhibition and Aβ load

#Corresponding author:

Urban Lendahl

Acc

epte

d A

rtic

le


Department of Cell and Molecular Biology, Karolinska Institute, SE-171 77 Stockholm,

Sweden

E-mail address: [email protected]

Phone number: +46 8 52487323

Abbreviations used:

Aβ: Amyloid beta

AD: Alzheimer’s disease

APP: Amyloid precursor protein

BACE: β-secretase

FRET: fluorescence resonance energy transfer

HCl: hydrogen chloride

LRP1: lipoprotein receptor-related protein 1

MDR1: multi-drug resistance protein 1

NaCl: sodium chloride

TRIS: tris(hydroxymethyl)aminomethane

s.c.: subcutaneous

ABSTRACT

Aggregation of amyloid beta (Aβ) peptides and the subsequent neural plaque formation is a

central aspect of Alzheimer's disease. Various strategies to reduce Aβ load in the brain are

therefore intensely pursued. It has been hypothesized that reducing Aβ peptides in the

periphery, i.e. in organs outside the brain, would be a way to diminish Aβ levels and plaque

Acc

epte

d A

rtic

le


load in the brain. In this report, we put this peripheral sink hypothesis to test by

investigating how selective inhibition of Aβ production in the periphery using a BACE1

inhibitor or reduced BACE1 gene dosage affects Aβ load in the brain. Selective inhibition of

peripheral BACE1 activity in wild type mice or mice overexpressing APP (APPswe transgenic

mice; Tg2576) reduced Aβ levels in the periphery but not in the brain, not even after chronic

treatment over several months. In contrast, a BACE1 inhibitor with improved brain

disposition reduced Aβ levels in both brain and periphery already after acute dosing. Mice

heterozygous for BACE1, displayed a 62% reduction in plasma Aβ40 whereas brain Aβ40 was

only lowered by 11%. These data suggest that reduction of Aβ in the periphery is not

sufficient to reduce brain Aβ levels and that BACE1 is not the rate-limiting enzyme for Aβ

processing in the brain. This provides evidence against the peripheral sink hypothesis and

suggests that a decrease of Aβ via BACE1 inhibition would need to be carried out in the

brain.

Keywords: neurological disorder, amyloid, Alzheimer’s disease, neural plaque, secretase,

BACE inhibitor

Introduction

Alzheimer's disease (AD) is characterized by the accumulation of neural plaques in the brain.

The amyloid hypothesis posits that aberrant aggregation of amyloid beta (Aβ) peptides is

the central cause of the disease. Cleavage of the amyloid precursor protein (APP) by the β-

and γ-secretase complexes generates several Aβ species (Okochi et al., 2013; Olsson et al.,

Acc

epte

d A

rtic

le


2013), some of which aggregate and readily form oligomers, which progress to protofibrils

and fibrils and eventually to neural plaques. Developing therapies that reduce Aβ formation

is therefore a prioritized research area (Citron, 2010). In recent years, several potent and

selective BACE1 inhibitors have been developed (Hunt and Turner, 2009; Swahn et al., 2012)

as well as γ-secretase inhibitors or modulators (Borgegård et al., 2012; Golde et al., 2013).

Efforts are also focused on reducing large toxic soluble Aß oligomers, protofibrils, using

antibody-based approaches (Lord et al., 2009).

An alternative approach to reduce Aβ load in the brain is through lowering of Aβ peptides in

peripheral organs. This strategy is referred to as the peripheral sink hypothesis. It rests on

the assumption that Aβ peptides in the brain and periphery are in equilibrium; that removal

of Aβ in the periphery, and passive diffusion down a concentration gradient, would also lead

to a reduction of monomeric Aβ in the brain (DeMattos et al., 2001; Zhang and Lee, 2011;

Sharma et al., 2012). Aβ peptides in the brain can be shuttled across the blood-brain barrier

(BBB) by the low density lipoprotein receptor-related protein (LRP1) (Zlokovic et al., 2010).

Conversely, Aβ peptide can be transported from the periphery to the brain by the receptor

of advanced glycation end products (RAGE) (Bell, 2012). The peripheral sink hypothesis is

appealing as it would only be necessary to pharmacologically intervene with Aβ production

in the periphery, and it is considerably easier to discover drugs that act on peripheral BACE1

than drugs that also need to effectively cross the BBB.

Acc

epte

d A

rtic

le


In support of the peripheral sink hypothesis, a number of studies indicate that facilitating Aβ

clearance in the periphery may reduce negative effects of Aβ in the brain. Peripheral

administration of an anti-Aβ antibody reduced brain Aβ load (DeMattos et al., 2001). The

use of an extract from the plant Withania somnifera led to increased LRP1 expression and

reversal of behavioural and pathological features in an animal model of AD (Sehgal et al.,

2012). Similarly, reduction of Aβ peptides in plasma by the Aβ binding protein gelsolin

blocked progression of cerebral amyloid angiopathy in mice (Gregory et al., 2012).

Here, we decided to more stringently test the hypothesis by investigating the effects on Aβ

peptide levels using two BACE1 inhibitors with distinct differences in disposition in

periphery and brain and by altering BACE1 gene dosage. The data show that reduction of Aβ

levels in the periphery does not lead to a corresponding reduction of Aβ levels in the brain.

This argues against the peripheral sink hypothesis.

Materials and methods

Compounds

The BACE1 inhibitors AZ9514 and AZ2000 were designed and synthesized at AstraZeneca

R&D Södertälje and the chemical structures are shown in Table 1.

Acc

epte

d A

rtic

le


BACE1 TR-FRET assay

Activity on human BACE1 was performed using time-resolved FRET (TR-FRET) as previously

described (Swahn et al., 2012). Briefly, recombinant human BACE1 (amino acids 1-460) was

pre-incubated with the compound in reaction buffer (sodium acetate, CHAPS, Triton X-100,

EDTA, pH 4.5) for 10 min. The substrate (europium) CEVNLDAEFK(Qsy7) was added and the

reaction was stopped 6.5 hrs later using sodium acetate, pH 9.0. The fluorescence of the

product was measured on a Victor II 1420 Multilabel Counter plate reader (Wallac) with an

excitation wavelength of 340 nm and an emission wavelength of 615 nm.

SH-SY5Y sAPPβ release assay

SH-SY5Y cells (a human neuroblastoma cell line) stably overexpressing APP695 cultured in

DMEM/F-12 with Glutamax, 10% FCS, and 1% non-essential amino acids (Invitrogen) were

used to assess cellular potency. Cells were incubated with compound for 16 hrs at 37 °C, 5%

CO2. Release of sAPPβ was detected using MSD plates (Meso Scale Discovery, Gaithersburg,

MD) according to the manufacturer's instructions, and the plates were read in a SECTOR

Imager.

Permeability assays

The Caco-2 assay was conducted as previously described (Bylund and Bueters, 2013). Briefly,

cells were grown for 14−21 days to achieve confluency and polarization. Cells were then

incubated with test compounds at a concentration of 10 μM for 90 min (pH 7.4 on both

Acc

epte

d A

rtic

le


sides). Concentrations of compound in donor and receiver samples were analysed by liquid

chromatography−tandem mass spectrometry and the apparent permeability coefficient

(Papp) was calculated as previously described (Swahn et al., 2012). The efflux ratio is the ratio

Papp(B−A)/Papp(A−B).

The MDCK-MDR1 cell assay was conducted as previously described (Gravenfors et al., 2012).

MDCK cells expressing the multi-drug resistance protein 1 (MDR1) were grown for 4 days

and the test compounds were investigated at a concentration of 1 μM for 120 min (pH 7.4

on both sides). Concentrations of compound in donor and receiver samples were analysed

by liquid chromatography−tandem mass spectrometry.(Gravenfors et al., 2012).

Animals and animal handling

All in vivo experiments were performed in accordance with guidelines and regulations

provided by the Swedish Board of Agriculture and were approved by the local animal ethics

committee (Stockholm North Animal Research Ethical Board). Female C57BL/6 mice (11-16

weeks old) were purchased from Harlan Laboratories (Boxmeer, Netherlands) and Tg2576

transgenic mice overexpressing a human APP cDNA transgene with the K670M/N671L

double mutation (APPSWE) under the control of the hamster prion promoter (17-19 weeks

old) were purchased from Taconic (Hudson, NY). Breeding of the BACE1 knock-out mice

(Roberds et al., 2001) was performed at the Karolinska Institute (Stockholm, Sweden) as

previously described (Jin et al., 2010) to generate homozygous (BACE1-/-), heterozygous

Acc

epte

d A

rtic

le


(BACE1+/-) and wild-type (BACE1+/+) littermates. The mice were kept in conventional housing

and fed standard rodent chew and tap water ad libitum.

In vivo pharmacology

AZ9514 was administered to female C57BL/6 mice in four separate studies. For

acute effects of AZ9514, mice received a single subcutaneous (s.c.) injection of

AZ9514 at 500 µmol/kg (n=20) or vehicle (0.1 M gluconic acid, 0.8% glycerol, pH

3.15; n=20). At 1.5 and 3 hrs after administration, blood samples were collected and

brains dissected. In the dose-response study, mice received a single s.c. injection of

AZ9514 at 3 (n=22), 10 (n=22), 30 (n=7), 100 (n=10) or 300 (n=8) µmol/kg or vehicle

(40% hydroxypropyl-β-cyclodextrin in 0.3 M gluconic acid, pH 3; n=31). In the time-

response study, AZ9514 at 100 µmol/kg (n=42) or vehicle (40% hydroxypropyl-β-

cyclodextrin in 0.3 M gluconic acid, pH 3; n=44) was administered as a single s.c.

injection. Blood samples were collected at 3 hrs (dose-response) or 0.5, 1.5, 4, 6, 8,

11, 16, and 24 hrs (time-response) after administration. For subchronic effects of

AZ9514, mice received vehicle (2.5% Glycerol in dH2O; n=20) or AZ9514 at 3

(n=20), 10 (n=20) or 30 (n=10) µmol/kg twice daily via s.c. injections for 7

consecutive days. At 3 hrs after last administration, blood samples were collected

and brains dissected. For chronic effects of AZ9514, Tg2576 mice received vehicle

(dH2O; n=32) or AZ9514 at 150 µmol/kg (n=32) twice daily via oral gavage for 1, 2 or

3 months. At 3 hours after last administration, blood samples were collected and

brains dissected.

Acc

epte

d A

rtic

le


AZ2000 was administered to C57BL/6 mice in acute dose-and time-response

studies. In the dose-response study, AZ2000 at doses 30, 100 or 300 µmol/kg or

vehicle (0.3 M gluconic acid) was administered as a single dose via oral gavage

(n=14/group). In the time-response study, animals received AZ2000 at 300 µmol/kg

or vehicle (0.3 M gluconic acid) as a single dose via oral gavage (n=7/group). Blood

samples were collected and brains dissected at 1.5 hrs (dose-response) or 0.5, 1.5,

3, 6 and 8 hrs (time-response) after dose. For the treatment study 14-17 weeks old

mice, received vehicle (0.3 M gluconic acid; n=18) or AZ2000 at 30 or 100 µmol/kg

(n=9-10/dose and genotype) as a single dose via oral gavage. Blood samples were

collected and brains dissected at 1.5 hrs after dose.

Blood sampling and brain dissection

Blood was sampled from anaesthetized mice by heart puncture and plasma was prepared by

centrifugation for 10 min at 3000g at +4°C within 20 min from sampling. Plasma was used

for analysis of compound exposure (bioanalysis) and Aβ levels. After blood sampling, the

mice were sacrificed by decapitation and the brain was dissected. Cerebellum and olfactory

bulbs were removed and the forebrain was divided into left (for Aβ analysis) and right (for

bioanalysis) hemispheres. The brain hemispheres were snap-frozen on dry ice and stored at

-70°C until analysis.

Extraction and analysis of mouse and human Aβ

The left brain hemispheres were homogenized in 0.2% diethylamine with 50 mM NaCl,

followed by ultracentrifugation. Recovered supernatants were neutralized to pH 8.0 with 2

Acc

epte

d A

rtic

le


M Tris-HCl. Mouse Aβ40 in plasma and brain extracts was analysed using a commercial Aβ1-

40 ELISA kit (KMB3481; Invitrogen, Camarillo, CA). For analysis of human Aβ40 and Aβ42 in

plasma and brain extracts from Tg2576 mice, commercial ELISA kits for measuring Aβ1-40

(#KHB3482, Invitrogen) and Aβ1-42 (RUO80177; Innogenetics, Gent, Belgium) were used.

Bioanalysis

Bioanalysis of plasma and brain homogenate samples was performed as previously

described (Borgegård et al., 2012). Briefly, the right brain hemisphere was homogenised in 2

volumes (w/v) of Ringer solution using a multi-element sonication probe. Plasma (25 μl)

and brain homogenate (50 μl) samples were precipitated with 150 μl acetonitrile containing

internal standard. Samples were mixed, centrifuged (+4°C, 4000 rpm, 20 min) and diluted

with mobile phase. Drug concentration was determined by LC-MS/MS. Correction for blood

content in brain was made by subtracting 1.3% of the plasma concentration from the total

brain concentration.

Statistical analysis

Data were analysed using Prism 5.0 software (GraphPad, San Diego, CA). Statistical analysis

was performed using equal variance two-tailed t-test or one-way ANOVA, followed by

Dunnett’s Multiple Comparison Test (dose-response studies) or Bonferroni’s (time-response

studies) Multiple Comparison Test (selected pairs), where P<0.05 was considered significant.

For the pharmacological study in BACE1 knock-out mice, pair-wise comparisons as t-tests

Acc

epte

d A

rtic

le


within a one-way ANOVA model on log transformed data were performed, with the level of

significance set at P<0.05.

Results

AZ9514 and AZ2000 – two potent BACE1 inhibitors

To assess the peripheral sink hypothesis, BACE1 inhibitors with different permeability

properties and distribution between brain and periphery were designed in order to allow

inhibition of BACE1 activity in the entire body, including the brain versus exclusively in the

periphery. The BACE1 inhibitors, AZ9514 and AZ2000, were first evaluated in vitro (Table 1).

AZ9514 is a potent inhibitor of recombinant human BACE1 (IC50=33 nmol/L) and efficiently

decreased sAPPβ formation in cells overexpressing APP695 (IC50=97 nmol/L). The in vitro

potency of AZ2000, which was derived from a different chemical class of inhibitors (Swahn

et al., 2012), is comparable to AZ9514, with an IC50 of 124 nmol/L on human BACE1 and a

high potency in the cellular assay (IC50=8 nmol/L) (Table 1). The in vitro data demonstrate

that AZ9514 and AZ2000 are both potent inhibitors of BACE1.

AZ9514 and AZ2000 display distinct brain-periphery distributions

We next analysed the in vitro permeability properties of the BACE1 inhibitors using Caco-2

cells to measure the permeability (Papp) from apical (A) to basolateral (B) side. AZ2000

displayed a 5-fold higher permeability compared to AZ9514 (Table 1). The efflux ratio was

also estimated in the Caco-2 cells (Papp B-A/Papp A-B) and AZ2000 had a 5-fold lower efflux

ratio compared to AZ9514. We then assessed whether AZ9514 and AZ2000 were substrates

Acc

epte

d A

rtic

le


for the BBB efflux transporter protein P-gp (also known as MDR1) (Padowski and Pollack,

2010). Efflux ratios obtained from MDCK cells expressing human MDR1 showed that both

compounds were P-gp substrates, although AZ9514 displayed a 7-fold higher efflux ratio

(Table 1), suggesting that this compound has stronger affinity for P-gp. Together, the in vitro

data suggest that the penetration across the BBB would be lower for AZ9514 as compared

to AZ2000.

To examine whether the different in vitro profiles translated to the in vivo situation, we

analysed the distribution of AZ9514 and AZ2000 in the brain in C57BL/6 mice. The

concentrations measured in plasma and brain 1.5 h after single administration of AZ2000

(30-300 μmol/kg) are reported in Table 2. The calculated brain/plasma ratios for AZ2000

ranged between 0.09-0.26 at tmax (time to reach maximal concentration in plasma). In

contrast, single administration of AZ9514 at a high dose of 500 μmol/kg resulted in

brain/plasma ratios of only 0.02-0.03 at 1.5-3 hours after dose (Table 2). In sum, these data

demonstrate that AZ9514 and AZ2000 display distinct characteristics with regard to

distribution into the brain, and that AZ9514 is largely confined to the periphery.

AZ9514 affects plasma but not brain Aβ levels in a dose- and time-dependent manner

The potency of AZ9514 to inhibit BACE1 in vivo was evaluated in C57BL/6 mice. Following

administration of a single dose of AZ9514 (500 μmol/kg, s.c.) a significant reduction in Aβ40

levels in plasma, but not in the brain, was observed at 3 hrs after administration (74%

reduction vs. vehicle, P<0.0001, Fig. 1A; P>0.05, Fig. 1B).

Acc

epte

d A

rtic

le


The inhibition of peripheral Aβ production was further evaluated in dose- and time-

response studies in C57BL/6 mice. Dose-dependent reductions of Aβ40 in plasma were

observed in mice treated with AZ9514 at 3-300 μmol/kg as a single s.c. injection (~20-80% of

vehicle), with statistically significant reductions at 30 μmol/kg (38% vs. vehicle; P<0.01), 100

μmol/kg (70% vs. vehicle; P<0.01) and 300 μmol/kg (>79% vs. vehicle; P<0.01) at 3 hrs after

dose (Fig. 1C). The mean concentrations of AZ9514 measured in plasma 3 hrs after dose

were 0.2, 1.3, 6.6, 28.1 and 96.9 µM at 3, 10, 30, 100, and 300 μmol/kg, respectively. In the

time-response study, the level of Aβ40 in plasma was significantly reduced by approximately

40-50% at 0.5-8 hrs after a single s.c. injection of 100 μmol/kg AZ9514 (Fig. 1D). The levels of

Aβ40 returned to baseline at the later time-points, 11-24 hrs after dose (Fig. 1D). The

maximal exposure of AZ9514 in plasma was reached at 1.5 hrs with a mean concentration of

50.6 µM and returned to baseline by 24 hrs (Fig. 1D).

Chronic treatment with AZ9514 does not lower brain Aβ in wild type or Tg2576 mice

To assess whether repeated dosing of C57BL/6 mice with AZ9514 affected the Aβ

production in the brain, the inhibitor was given s.c. twice daily during 7 days at doses 3, 10

and 30 μmol/kg. The dosing interval was chosen to minimize drug holiday based on the

pharmacokinetics of AZ9514. Significant reduction of Aβ40 in plasma was observed at 10

(12% vs. vehicle, P<0.05) and 30 (26% vs. vehicle, P<0.001) μmol/kg (Fig. 2A), while no effect

on brain Aβ40 was observed at any dose, despite repeated dosing (Fig. 2B). The

concentrations of AZ9514 in plasma 3 hrs after the last administration were 0.2, 1.2 and 4.5

µM at 3, 10 and 30 μmol/kg, respectively. At the same time-point the brain exposure was

Acc

epte

d A

rtic

le


below limit of quantification for the lowest dose, whereas the intermediate and high doses

resulted in an exposure of 0.07 and 0.2 µM, respectively.

We next explored the efficacy of chronic treatment of AZ9514 in Tg2576 transgenic mice,

which overexpress APPswe and exhibit readily detectable human Aβ40 and Aβ42 levels

(Hsiao et al., 1996). Tg2576 mice were given vehicle or AZ9514 (150 μmol/kg) twice daily via

oral gavage during 1, 2 or 3 months. Similar to the effects observed in C57BL/6 mice, the

levels of Aβ40 in plasma were significantly reduced with 47-59% in mice treated with

AZ9514 during 1 (P<0.01 vs. vehicle), 2 (P<0.001 vs. vehicle) or 3 months (P<0.001 vs.

vehicle) (Fig. 3A-C). Similar reductions of plasma Aβ42 levels (41-65%) were observed after 1

(P<0.001 vs. vehicle), 2 (P<0.05 vs. vehicle) or 3 months (P<0.001 vs. vehicle) treatment (Fig.

3D-F). Despite a consistent decrease of Aβ in plasma, AZ9514 failed to reduce the level of

Aβ40 and Aβ42 in the brain (Fig. 3G-L). The concentrations of AZ9514 were determined in

plasma and brain 3 hrs after the final dose following repeated administration during 1, 2 or

3 months. Mean plasma concentrations of AZ9514 were 39, 31 or 21 μM after 1, 2 or 3

months of treatment, while the brain concentrations were 0.9, 0.7 or 0.5 μM, respectively.

The corresponding brain/plasma ratios were 0.02-0.03 after 1-3 months of treatment,

demonstrating that there is no accumulation of the compound in the brain after chronic

treatment.

Acc

epte

d A

rtic

le


AZ2000 affects Aβ levels in the brain and periphery

The potential of AZ2000 to reduce brain Aβ40 levels in vivo was evaluated in dose- and

time-response studies in C57BL/6 mice following acute administration via oral gavage at 30-

300 μmol/kg. Dose-dependent reductions in Aβ40 levels were observed in both plasma (Fig.

4A) and brain (Fig. 4B) of mice treated with AZ2000. A significant reduction in plasma Aβ40

was seen at all doses tested (41-70%, P<0.001 vs. vehicle; Fig. 4A) and a reduction in brain

Aβ40 was observed at 100 and 300 μmol/kg (35-61%, P<0.001 vs. vehicle; Fig. 4B) at 1.5 hrs

after dose. The concentrations measured in plasma and brain at 1.5 hrs following

administration of AZ2000 are reported in Table 2. In the time-response study, the level of

Aβ40 in plasma was significantly reduced at 1.5 hrs (67%, P<0.001), 6 hrs (58%, P<0.01) and

8 hrs (58%, P<0.05) after single oral administration of 300 μmol/kg AZ2000 (Fig. 4C). AZ2000

also significantly reduced brain Aβ40 at 0.5 hrs (20%, P<0.001), 1.5 hrs (55%, P<0.001) and 3

hrs (47%, P<0.001) after dose (Fig. 4D). The levels of Aβ40 in the brain returned to baseline

at 6-8 hrs after dose. The maximal exposure of AZ2000 was reached at 1.5 hrs with a mean

concentration of 51.2 µM in plasma (Fig. 4C) and 5.9 µM in brain (Fig. 4D). Taken together,

these data show that AZ2000 affects Aβ levels both in the brain and in the periphery.

Aβ levels are reduced in the periphery but not in the brain in BACE1 heterozygous mice

We next addressed how a reduced gene dosage of BACE1 would affect Aβ levels in the brain

and periphery, and what would be the response to BACE1 inhibition in the periphery and

the brain. To this end, Aβ40 levels in brain and plasma from BACE1 heterozygous (BACE1+/-),

homozygous (BACE1-/-) and wild-type littermates (BACE1+/+) were analysed. The Aβ40 levels

in plasma were reduced in BACE1+/- mice by 62% compared to wild-type littermates

Acc

epte

d A

rtic

le


(P<0.001 vs. BACE1+/+; Fig. 5A). In contrast, there was no significant reduction of brain Aβ40

levels in BACE1+/- mice (Fig. 5B; P>0.05). The Aβ40 levels in plasma and brain from BACE1-/-

mice were below the lower limit of quantification (Fig. 5A-B), indicating that the

contribution of other enzymes such as BACE2 for Aβ production in the brain is very limited.

Next, the inhibition of Aβ production in plasma and brain was evaluated after

administration of AZ2000 at 30 or 100 μmol/kg via oral gavage to BACE1+/- mice. Similar

relative reductions in plasma Aβ40 levels were observed in BACE1+/- (82-86%, P<0.001 vs.

vehicle) and BACE1+/+ (81-88%, P<0.001 vs. vehicle) mice at 1.5 hrs after acute treatment

with AZ2000 (Fig. 6A). A significant reduction in brain Aβ40 levels was only observed at the

higher dose (100 μmol/kg) in both BACE1+/- (32% P<0.001 vs. vehicle) and wild-type (14%,

P<0.001 vs. vehicle) mice (Fig. 6B). No significant effect was seen at the lower dose in either

genotype. Although the effect on brain Aβ40 was larger in BACE1+/- mice (32% vs. 14% in

wild-type mice), this difference was not statistically significant. The mean concentrations of

AZ2000 were 0.37-0.38 μM at 30 μmol/kg and 8.8-10.3 μM at 100 μmol/kg in plasma, and

0.06 μM at 30 μmol/kg and 0.50-0.57 μM at 100 μmol/kg in the brain, respectively, and

similar in both genotypes. Together, these data show that the lowering of Aβ levels

observed in the periphery of BACE1 heterozygous mice did not lead to a corresponding

reduction of brain Aβ levels, which suggests that BACE1 is not the rate-limiting enzyme in

the brain.

Acc

epte

d A

rtic

le


Discussion

The peripheral sink hypothesis is an attractive strategy from a drug development

perspective, since it would not require the therapeutic molecule to pass the BBB in order to

be effective. Recent studies with different Aβ binding molecules, trapping free Aβ in the

plasma, support this concept (Gregory et al., 2012; Sehgal et al., 2012). However, it remains

to be established whether this reflects a peripheral sink-based mechanism and thus would

work for any approach that selectively lowers peripheral Aβ levels. A more direct and

conclusive way to test the peripheral sink hypothesis would be to explore the effect of

differentially localized synthesis and inhibition of Aβ generation on the distribution of Aβ

load in the brain and periphery.

In this report, we assessed the peripheral sink hypothesis by studying the impact of reduced

BACE1 gene dosage on brain and peripheral Aβ levels, and by exploring the effect of two

BACE1 inhibitors that differ with regard to penetration into the brain. A sensitive ELISA for

mouse Aβ40 allowed us to monitor endogenous plasma and brain Aβ40 levels in wild-type

and BACE1+/- mice. The basal brain Aβ40 levels were only decreased by 11% in BACE1+/- as

compared to wild-type mice, which is in agreement with data from two other strains of

BACE1-targeted mice (Nishitomi et al., 2006; McConlogue et al., 2007). In contrast to brain,

we observed that basal plasma Aβ40 levels were affected to a much higher extent in the

BACE1+/- mice and were decreased by approximately 62% as compared to wild-type mice.

The results suggest that BACE1 is rate limiting for Aβ production in the plasma but not in the

brain. This is important to consider for the development of BACE1 inhibitors, since it

Acc

epte

d A

rtic

le


suggests that very potent BACE1 inhibitors need to be developed in order to achieve

efficacy on brain Aβ levels.

The differential impact of reduced BACE1 gene dosage on basal Aβ40 levels in the brain and

the periphery argues against the notion that steady state brain and peripheral Aβ40 levels

are set by a direct chemical equilibrium between the two compartments. Consistent with

these findings, neither single nor repeated dosing with the BACE1 inhibitor AZ9514, which

was largely confined to the periphery, had any impact on the levels of Aβ in the brain of

wild-type mice, but did reduce plasma Aβ levels with up to 80% at the highest dose.

Interestingly, a similar picture emerged also in mice with elevated Aβ levels: even after

repeated dosing over 3 months AZ9514 reduced Aβ only in the plasma and not in the brain

in Tg2576 transgenic mice. This was in contrast to the AZ2000 inhibitor, which

demonstrated a higher degree of BBB penetration, and efficiently reduced Aβ levels in both

compartments. This is in line with other BACE1 inhibitors that distribute into the brain and

efficiently inhibit Aβ production in the brain (May et al., 2011; Jeppsson et al., 2012; Eketjäll

et al., 2013). In the pharmacological studies with AZ2000, a brain exposure equivalent to

0.71 µM was sufficient to achieve a significant reduction of brain Aβ40 levels following

administration to C57BL/6 mice. This would suggest that the concentration of AZ9514

measured in brain homogenate (1.4 µM) after administration of a high dose (500 μmol/kg)

did not reach its target in the brain parenchyma and most likely the compound was present

in the vasculature of the brain.

Acc

epte

d A

rtic

le


The peripheral sink hypothesis stems from experiments analysing the role of RAGE in the

trafficking of Aβ. Originally, a soluble engineered form of RAGE (sRAGE) was shown to

effectively trap plasma Aβ and to combat Aβ amyloidosis in APP transgenic mice (Deane et

al., 2003). Other experimental approaches, based on the peripheral administration of anti-

Aβ antibodies, gelsolin or soluble forms of Nogo and LRP, have in a similar manner resulted

in reduced free plasma Aβ levels and a subsequent lowering of brain Aβ in APP

overexpressing mice (DeMattos et al., 2001; Jaeger et al., 2009; McDonald et al., 2011;

Gregory et al., 2012). Common to all these studies is that free plasma Aβ levels are lowered

whereas the total plasma Aβ levels are increased, as a result of the trapping of free Aβ to

the Aβ binding molecule administered. It may therefore be possible to lower brain Aβ levels

via targeting plasma Aβ, but that would require trapping and binding of Aβ in the periphery

resulting in a CNS Aβ clearance process. Such a mode of action is likely to be more complex

and distinct from a basal chemical equilibrium linked to free diffusion of Aβ between the

brain and plasma compartments.

Our findings argue against the peripheral sink hypothesis, and are also corroborated by a

recent study showing that repeated intravenous administration of the Aβ degrading enzyme

neprilysin for up to 4 months in Tg2576 transgenic mice or for 1 month in monkeys

effectively degraded Aβ in the periphery but did not alter brain or cerebrospinal fluid Aβ

levels (Henderson et al., 2013). Additionally, early studies with the anti-Aβ binding

monoclonal antibody 266 reported reduced Aβ deposition in the brains of APP transgenic

mice due to binding of soluble plasma Aβ, thereby accelerating the Aβ efflux from the brain

(DeMattos et al., 2001; Dodart et al., 2002). More recent studies have demonstrated that

Acc

epte

d A

rtic

le


the 266 antibody acts within the brain parenchyma by binding to and stabilizing the soluble,

monomeric form of Aβ (Yamada et al., 2009). This has led to the proposal of a novel

mechanism whereby anti-Aβ antibodies act by sequestering soluble Aβ within the CNS, not

in the peripheral blood stream.

In conclusion, our data provide evidence against the peripheral sink hypothesis and suggest

that Aβ production needs to be controlled in the brain rather than in the periphery. This

information, combined with the finding that BACE1 is not the rate-limiting enzyme in the

brain, is useful as a guide for developing future therapies for AD.

Acknowledgements and conflict-of-interest disclosure:

The financial support from the Swedish Research Council (project grant, Linnaeus Center

and SFO support), Hjärnfonden, the Swedish Cancer Society, Karolinska Institutet and Knut

and Alice Wallenbergs Stiftelse is gratefully acknowledged (UL). We also thank Daniel

Bergström, Kristina Eliason, Ann Staflund and Anette Stålebring-Löwstedt for in vivo

support, Carina Stephan for formulation support, Paulina Appelkvist, Anna Bogstedt, Gunilla

Ericsson and Anja Finn for biomarker analysis, Anders Christensen, Eivor Eklund and Stefan

Martinsson for bioanalysis. All authors except UL were at the time of the work employees of

AstraZeneca.

Acc

epte

d A

rtic

le


References

Bell R.D. (2012) The imbalance of vascular molecules in Alzheimer’s disease. J Alzheimers Dis 32:699–709

Borgegård T. et al. (2012) Alzheimer’s disease: presenilin 2-sparing γ-secretase inhibition is a tolerable Aβ peptide-lowering strategy. J Neurosci 32:17297–17305

Bylund J. and Bueters T. (2013) Presystemic Metabolism of AZ ’ 0908 , A Novel mPGES-1 Inhibitor : An In Vitro and In Vivo Cross-Species Comparison. J Pharm Sci 102:1106–1115.

Citron M. (2010) Alzheimer’s disease: strategies for disease modification. Nat Rev Drug Discov 9:387–398

Deane R. et al. (2003) RAGE mediates amyloid-beta peptide transport across the blood-brain barrier and accumulation in brain. Nat Med 9:907–913

DeMattos R.B., Bales K.R., Cummins D.J., Dodart J.C., Paul S.M. and Holtzman D.M. (2001) Peripheral anti-A beta antibody alters CNS and plasma A beta clearance and decreases brain A beta burden in a mouse model of Alzheimer’s disease. Proc Natl Acad Sci U S A 98:8850–8855

Dodart J.-C., Bales K.R., Gannon K.S., Greene S.J., DeMattos R.B., Mathis C., DeLong C.A., Wu S., Wu X., Holtzman D.M. and Paul S.M. (2002) Immunization reverses memory deficits without reducing brain Abeta burden in Alzheimer’s disease model. Nat Neurosci 5:452–457

Eketjäll S., Janson J., Jeppsson F., Svanhagen A., Kolmodin K., Gustavsson S., Radesäter A.-C., Eliason K., Briem S., Appelkvist P., Niva C., Berg A.-L., Karlström S., Swahn B.-M. and Fälting J. (2013) AZ-4217: a high potency BACE inhibitor displaying acute central efficacy in different in vivo models and reduced amyloid deposition in Tg2576 mice. J Neurosci 33:10075–10084

Golde T.E., Koo E.H., Felsenstein K.M., Osborne B.A. and Miele L. (2013) γ-Secretase inhibitors and modulators. Biochim Biophys Acta 1828:2898–2907

Gravenfors Y. et al. (2012) New aminoimidazoles as β-secretase (BACE-1) inhibitors showing amyloid-β (Aβ) lowering in brain. J Med Chem 55:9297–9311

Gregory J.L., Prada C.M., Fine S.J., Garcia-Alloza M., Betensky R. A., Arbel-Ornath M., Greenberg S.M., Bacskai B.J. and Frosch M.P. (2012) Reducing available

Acc

epte

d A

rtic

le


soluble β-amyloid prevents progression of cerebral amyloid angiopathy in transgenic mice. J Neuropathol Exp Neurol 71:1009–1017

Henderson S.J., Andersson C., Narwal R., Janson J., Goldschmidt T.J., Appelkvist P., Bogstedt A., Steffen A.-C., Haupts U., Tebbe J., Freskgård P.O., Jermutus L., Burrell M., Fowler S.B. and Webster C.I. (2013) Sustained peripheral depletion of amyloid-β with a novel form of neprilysin does not affect central levels of amyloid-β. Brain 137: 553-564.

Hsiao K., Chapman P., Nilsen S., Eckman C., Harigaya Y., Younkin S., Yang F. and Cole G. (1996) Amyloid Plaques in Transgenic Mice. Science 274:7–10.

Hunt C.E. and Turner A.J. (2009) Cell biology, regulation and inhibition of beta-secretase (BACE-1). FEBS J 276:1845–1859

Jaeger L.B., Dohgu S., Hwang M.C., Farr S.A., Murphy M.P., Fleegal-DeMotta M.A., Lynch J.L., Robinson S.M., Niehoff M.L., Johnson S.N., Kumar V.B. and Banks W.A. (2009) Testing the neurovascular hypothesis of Alzheimer’s disease: LRP-1 antisense reduces blood-brain barrier clearance, increases brain levels of amyloid-beta protein, and impairs cognition. J Alzheimers Dis 17:553–570

Jeppsson F., Eketjäll S., Janson J., Karlström S., Gustavsson S., Olsson L.-L., Radesäter A.-C., Ploeger B., Cebers G., Kolmodin K., Swahn B.-M., von Berg S., Bueters T. and Fälting J. (2012) Discovery of AZD3839, a potent and selective BACE1 inhibitor clinical candidate for the treatment of Alzheimer disease. J Biol Chem 287:41245–41257

Jin S., Agerman K., Kolmodin K., Gustafsson E., Dahlqvist C., Jureus A., Liu G., Fälting J., Berg S., Lundkvist J. and Lendahl U. (2010) Evidence for dimeric BACE-mediated APP processing. Biochem Biophys Res Commun 393:21–27

Lord A., Gumucio A., Englund H., Sehlin D., Sundquist V.S., Söderberg L., Möller C., Gellerfors P., Lannfelt L., Pettersson F.E. and Nilsson L.N.G. (2009) An amyloid-beta protofibril-selective antibody prevents amyloid formation in a mouse model of Alzheimer’s disease. Neurobiol Dis 36:425–434

May P.C. et al. (2011) Robust central reduction of amyloid-β in humans with an orally available, non-peptidic β-secretase inhibitor. J Neurosci 31:16507–16516

McConlogue L., Buttini M., Anderson J.P., Brigham E.F., Chen K.S., Freedman S.B., Games D., Johnson-Wood K., Lee M., Zeller M., Liu W., Motter R. and Sinha S. (2007) Partial reduction of BACE1 has dramatic effects on Alzheimer plaque and synaptic pathology in APP Transgenic Mice. J Biol Chem 282:26326–26334

Acc

epte

d A

rtic

le


McDonald C.L., Bandtlow C. and Reindl M. (2011) Targeting the Nogo receptor complex in diseases of the central nervous system. Curr Med Chem 18:234–244

Nishitomi K. et al. (2006) BACE1 inhibition reduces endogenous Abeta and alters APP processing in wild-type mice. J Neurochem 99:1555–1563

Okochi M., Tagami S., Yanagida K., Takami M., Kodama T.S., Mori K., Nakayama T., Ihara Y. and Takeda M. (2013) γ-secretase modulators and presenilin 1 mutants act differently on presenilin/γ-secretase function to cleave Aβ42 and Aβ43. Cell Rep 3:42–51

Olsson F., Schmidt S., Althoff V., Munter L., Jin S., Rosqvist S., Lendahl U., Multhaup G. and Lundkvist J. (2013) Characterization of intermediate steps in Abeta production under near-native conditions. J Biol Chem 289: 1540-1550.

Padowski J.M. and Pollack G.M. (2010) Multi-Drug Resistance in Cancer Zhou J, ed. Methods Mol Biol 596:359–384

Roberds S.L. et al. (2001) BACE knockout mice are healthy despite lacking the primary β -secretase activity in brain : implications for Alzheimer ’ s disease therapeutics. Hum Mol Genet 10:1317–1324.

Sehgal N., Gupta A., Valli R.K., Joshi S.D., Mills J.T., Hamel E., Khanna P., Jain S.C., Thakur S.S. and Ravindranath V. (2012) Withania somnifera reverses Alzheimer’s disease pathology by enhancing low-density lipoprotein receptor-related protein in liver. Proc Natl Acad Sci U S A 109:3510–3515

Sharma H.S., Castellani R.J., Smith M.A. and Sharma A. (2012) The blood-brain barrier in Alzheimer’s disease: novel therapeutic targets and nanodrug delivery., 1st ed. Elsevier Inc.

Swahn B.-M. et al. (2012) Design and synthesis of β-site amyloid precursor protein cleaving enzyme (BACE1) inhibitors with in vivo brain reduction of β-amyloid peptides. J Med Chem 55:9346–9361

Yamada K., Yabuki C., Seubert P., Schenk D., Hori Y., Ohtsuki S., Terasaki T., Hashimoto T. and Iwatsubo T. (2009) Abeta immunotherapy: intracerebral sequestration of Abeta by an anti-Abeta monoclonal antibody 266 with high affinity to soluble Abeta. J Neurosci 29:11393–11398

Zhang Y. and Lee D.H.S. (2011) Sink hypothesis and therapeutic strategies for attenuating Abeta levels. Neuroscientist 17:163–173

Zlokovic B.V., Deane R., Sagare A.P., Bell R.D. and Winkler E.A. (2010) Low-density lipoprotein receptor-related protein-1: a serial clearance homeostatic

Acc

epte

d A

rtic

le


mechanism controlling Alzheimer’s amyloid β-peptide elimination from the brain. J Neurochem 115:1077–1089

FIGURE LEGENDS

Fig. 1 Acute treatment of C57BL/6 mice with BACE1 inhibitor AZ9514. Significant reductions

in plasma Aβ40 levels were seen 3 hrs after single subcutaneous (s.c.) administration of

AZ9514 at 500 µmol/kg (A), with no significant effect on brain Aβ40 levels (B). A dose-

dependent inhibition of Aβ40 production in plasma was seen at 3 hrs following AZ9514

administration (s.c), with significant effects at doses 30-300 µmol/kg (C). Time-dependent

reductions of plasma Aβ40 levels were studied following a single dose of AZ9514 at 100

µmol/kg and significant reductions were observed at 0.5-8 hrs after dose (D). Corresponding

plasma concentrations of AZ9514 are plotted in D. Data are presented as mean values ±

SEM (**P<0.01; ***P<0.001, compared to vehicle).

Fig. 2 Subchronic treatment of C57BL/6 mice with BACE1 inhibitor AZ9514. AZ9514 was

administered subcutaneously twice daily for 7 days and significant reductions in plasma

Aβ40 levels were seen at doses 10-30 µmol/kg at 3 hrs following the last administration (A),

while no significant effect on brain Aβ40 was observed (B). Data are presented as mean

values ± SEM (*P<0.05; ***P<0.001, compared to vehicle).

Fig. 3 Chronic treatment of Tg2576 mice with BACE1 inhibitor AZ9514. AZ9514 (150

µmol/kg) was administered via oral gavage twice daily for 1, 2 or 3 months. Significant

Acc

epte

d A

rtic

le


inhibition of Aβ40 (A, B, C) and Aβ42 (D, E, F) production in plasma was observed after 1, 2

and 3 months treatment. No significant effects on brain Aβ40 (G, H, I) or Aβ42 (J, K, L) levels

were observed after chronic treatment with AZ9514. Data are presented as mean values ±

SEM (*P<0.05; **P<0.01; ***P<0.001, compared to vehicle).

Fig. 4 Acute treatment of C57BL/6 mice with BACE1 inhibitor AZ2000. A dose-dependent

inhibition of Aβ40 production in plasma was seen at 1.5 hrs following oral administration of

AZ2000 (A). Significant reductions in brain Aβ40 levels were also observed at doses 100 and

300 µmol/kg (B). Time-dependent reductions of plasma Aβ40 levels were studied following

a single dose of AZ2000 at 300 µmol/kg and significant reductions were observed at 1,5, 6

and 8 hrs after dose (C). Significant reductions in brain Aβ40 levels were seen at 0.5-3 hrs

after dose (D). Corresponding plasma and brain concentrations of AZ2000 are plotted in C

and D. Data are presented as mean values ± SEM (*P<0.05; **P<0.01; ***P<0.001,

compared to vehicle).

Fig. 5 Basal Aβ levels in BACE1 knock-out mice. The levels of Aβ40 in plasma were

significantly reduced in BACE1 heterozygous (+/-) mice (A), while no significant reduction in

brain Aβ40 levels was observed (B). The Aβ40 levels in plasma and brain in homozygous

mice (-/-) were below the lower limit of quantification (A, B). Data are presented as mean

values ± SEM with n=5 mice per group (***P<0.001, compared to wild-type; +/+).

Acc

epte

d A

rtic

le


Fig. 6 Acute treatment of BACE1 knock-out mice with BACE1 inhibitor AZ2000. Single oral

administration of AZ2000 (30 or 100 µmol/kg) resulted in significant reductions in plasma

Aβ40 levels in BACE1 heterozygous (+/-) and wild-type (+/+) mice (A). Significant reductions

in brain Aβ40 levels were only observed in heterozygous and wild-type mice treated with

100 µmol/kg AZ2000 (B). Data are presented as mean values ± SEM (*P<0.05; ***P<0.001,

compared to vehicle).

TABLES

Table 1

In vitro potency and permeability

Compound AZ9514 AZ2000

Structure

BACE1 TR-FRET (IC50) 33 nmol/L 124 nmol/L

SH-SY5Y, sAPPβ release (IC50) 97 nmol/L 8 nmol/L

Caco-2 Papp (10-6 cm/s)a 2.9 15.7

Caco-2 Efflux ratiob 6.3 1.3

MDCK-MDR1 Efflux ratioc 44.6 6.3

a Papp is the measured permeability (apical (A) to basolateral (B)) through Caco-2 cells.

b The efflux ratio is Papp(B−A)/Papp(A−B) in Caco-2 cells.

c The efflux ratio is Papp(B−A)/Papp(A−B) in MDCK cells expressing human MDR1.

N

N

O

N

N

OS

O

O

F

F

a

a = unknown absolute

ISOMER 1

N

N

N

N N

F

ISOMER 1

a

a = unknown absolute

Acc

epte

d A

rtic

le


Table 2

Compound concentrations in plasma and brain

Compound Dose

(µmol/kg)

Time-point

(hours after dose)

Cp (µM) Cbr (µM) Cbr/Cp ratio

AZ9514 500 1.5 44.1±8.9 0.9±0.2 0.02±0.01

AZ9514 500 3 50.8±12.9 1.4±0.3 0.03±0.01

AZ2000 300 1.5 45.7±8.6 5.0±1.1 0.26±0.09

AZ2000 100 1.5 8.8±3.4 0.71±0.20 0.09±0.02

AZ2000 30 1.5 0.39±0.44 0.08±0.05 0.11±0.02

All values are reported as mean ± SD

Cp: measured concentration in plasma

Cbr: measured concentration in brain

Cbr/Cp ratio is calculated using individual values

Acc

epte

d A

rtic

le


Figure 1

Vehicle AZ95140

20

40

60

80

***

Aβ4

0 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

1000

2000

3000

4000

Aβ4

0 in

bra

in (

pg

/g)

A B

0 3 10 30 100 3000

20

40

60

80

100

120

**

****

Dose (μmol/kg)

Pla

sma

Aβ 4

0 (%

of

veh

icle

)

C D

Acc

epte

d A

rtic

le


Figure 2

A B

0 3 10 300

20

40

60

80

100

120

****

Dose (μmol/kg)

Aβ

40 in

pla

sma

(pg

/mL)

0 3 10 300

500

1000

1500

2000

Dose (μmol/kg)

Aβ

40 in

bra

in (

pg

/g)

Acc

epte

d A

rtic

le


Figure 3

Bra

in A

β42

Bra

in A

β40

Pla

sma

Aβ4

2P

lasm

a A

β40

Vehicle AZ95140

2000

4000

6000

80002 months

***

Aβ4

0 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

2000

4000

6000

80003 months

***

Aβ4

0 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

2000

4000

6000

80001 month

**

Aβ4

0 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

500

1000

1500

20001 month

***

Aβ4

2 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

500

1000

1500

20003 months

***

Aβ4

2 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

500

1000

1500

20002 months

*

Aβ4

2 in

pla

sma

(pg

/mL

)

Vehicle AZ95140

10

20

30

40

501 month

Vehicle AZ95140

10

20

30

40

503 months

Aβ4

0 in

bra

in (p

g/m

g)

Vehicle AZ95140

10

20

30

40

502 months

Aβ 4

0 in

bra

in (p

g/m

g)

Vehicle AZ95140

10

20

30

40

501 month

Aβ4

2 in

bra

in (p

g/m

g)

Vehicle AZ95140

10

20

30

40

503 months

Aβ4

2 in

bra

in (p

g/m

g)

Vehicle AZ95140

10

20

30

40

502 months

Aβ 4

2 in

bra

in (p

g/m

g)

A B C

D E F

G H I

J K L

Acc

epte

d A

rtic

le


Figure 4

A B

C D

Acc

epte

d A

rtic

le


Figure 5

Figure 6

A B

(+/+) (+/+) (+/-) (+/-)0

20

40

60

80

100

120 30 μmol/kg100 μmol/kg

genotype

Pla

sma

Aβ

40(%

of

veh

icle

)

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

(+/+) (+/+) (+/-) (+/-)0

20

40

60

80

100

120

genotype

Bra

in A

β40

(% o

f ve

hic

le) *

***

A B

Revisiting the peripheral sink hypothesis: inhibiting BACE1 activity in the periphery does not alter...

Documents

Transcript of Revisiting the peripheral sink hypothesis: inhibiting BACE1 activity in the periphery does not alter...