APP-BP1 inhibits Aβ42 levels by interacting with Presenilin-1

Abstract

BackgroundThe β-amyloid precursor protein (APP) is sequentially cleaved by the β- and then γ-secretase to generate the amyloid β-peptides
Aβ40 and Aβ42. Increased Aβ42/Aβ40 ratios trigger amyloid plaque formations in Alzheimer's disease (AD). APP binds to APP-BP1,
but the biological consequence is not well understood.

ResultsWe report that when the endogenous APP-BP1 was suppressed by small interfering RNAs (siRNAs), cell-associated Aβ42 was dramatically
increased in APP695 expressing primary neurons. The accumulation of Aβ42 was accompanied by significant increases in APP and APP-CTF in APP-BP1
siRNA expressing neurons. In contrast, APP-BP1 overexpression in primary neurons significantly decreased the levels of Aβ
and endogenous APP but not APLPs. We also investigated the potential mechanism of APP-BP1-mediated APP processing. APP-BP1
co-precipitated with Presenilin-1 (PS1) in native rat brain extracts, co-migrated with the γ-secretase components in brain
membrane extracts in glycerol gradient centrifugation, and colocalized in primary neurons. Further, the endogenous PS1-CTF
was significantly downregulated by APP-BP1 expression.

ConclusionOur data suggest that APP-BP1 may inhibit Aβ42 production by interacting with PS1 under physiological conditions.

Full text

Page 1

BioMed Central
Page 1 of 12
(page number not for citation purposes)
Molecular Neurodegeneration
Open Access
Research article
APP-BP1 inhibits Aβ42 levels by interacting with Presenilin-1
Yuzhi Chen*1,2, Angela M Bodles1, Donna L McPhie3, Rachael L Neve3,
Robert E Mrak4 and W Sue T Griffin1,2
Address: 1Department of Geriatrics, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA, 2Department of Neurobiology &
Developmental Sciences, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA, 3Department of Psychiatry, McLean Hospital
and Harvard Medical School, Belmont, MA 02478, USA and 4Department of Pathology, University of Arkansas for Medical Sciences, Little Rock,
AR 72205, USA
Email: Yuzhi Chen* - chenyuzhi@uams.edu; Angela M Bodles - bodlesangela@uams.edu; Donna L McPhie - mcphie@helix.mgh.harvard.edu;
Rachael L Neve - neve@helix.mgh.harvard.edu; Robert E Mrak - mrakroberte@uams.edu; W Sue T Griffin - griffinst@uams.edu
* Corresponding author
Abstract
Background: The β-amyloid precursor protein (APP) is sequentially cleaved by the β- and then
γ-secretase to generate the amyloid β-peptides Aβ40 and Aβ42. Increased Aβ42/Aβ40 ratios
trigger amyloid plaque formations in Alzheimer's disease (AD). APP binds to APP-BP1, but the
biological consequence is not well understood.
Results: We report that when the endogenous APP-BP1 was suppressed by small interfering
RNAs (siRNAs), cell-associated Aβ42 was dramatically increased in APP695 expressing primary
neurons. The accumulation of Aβ42 was accompanied by significant increases in APP and APP-CTF
in APP-BP1 siRNA expressing neurons. In contrast, APP-BP1 overexpression in primary neurons
significantly decreased the levels of Aβ and endogenous APP but not APLPs. We also investigated
the potential mechanism of APP-BP1-mediated APP processing. APP-BP1 co-precipitated with
Presenilin-1 (PS1) in native rat brain extracts, co-migrated with the γ-secretase components in
brain membrane extracts in glycerol gradient centrifugation, and colocalized in primary neurons.
Further, the endogenous PS1-CTF was significantly downregulated by APP-BP1 expression.
Conclusion: Our data suggest that APP-BP1 may inhibit Aβ42 production by interacting with PS1
under physiological conditions.
Background
The amyloid precursor protein (APP) binding protein-
1(APP-BP1) is the regulatory subunit of the activating
enzyme for the small ubiquitin-like protein Nedd8 [1,2].
APP-BP1 is homologous to the amino-terminus of E1,
and binds to Uba3 which is homologous to the carboxyl
terminus of E1 and contains the catalytic cysteine residue.
APP-BP1 forms a heterodimer with Uba3, and together
activates the small ubiquitin-like protein Nedd8 [3].
When Nedd8 is activated, it covalently modifies (ned-
dylates) Cullin family members [4]. Cullins are the subu-
nit of a class of ubiquitin E3 ligase complex. Neddylated
Cullins may become transiently stable, resulting in
enhanced ubiquitin ligase activity and increased degrada-
tion of target proteins [5].
At the cellular level, neddylation regulates cell cycle pro-
gression at two checkpoints – APP-BP1 inhibits entry into
Published: 7 February 2007
Molecular Neurodegeneration 2007, 2:3doi:10.1186/1750-1326-2-3
Received: 9 October 2006
Accepted: 7 February 2007
This article is available from: http://www.molecularneurodegeneration.com/content/2/1/3
© 2007 Chen et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0),
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Page 2

Molecular Neurodegeneration 2007, 2:3 http://www.molecularneurodegeneration.com/content/2/1/3
Page 2 of 12
(page number not for citation purposes)
the DNA synthesis (S) phase and promotes entry into
mitosis in dividing cells [2,6], presumably by neddylation
of different Cullin proteins. Loss of APP-BP1 function in
the Chinese hamster liver cell line, ts41 cells, results in the
accumulation of cells in the S and G1 phases of the cell
cycle [6]. Expressing human APP-BP1 can rescue the cell
cycle defect in ts41 cells at the non-permissive tempera-
ture [2]. Knocking out the activation subunit of the Nedd8
activating enzyme, Uba3, leads to embryonic lethality at
the peri-implantation stage [7]. Therefore, the APP-BP1-
mediated neddylation is critical for somatic cell survival.
The function of APP-BP1 in the brain is not well under-
stood. In primary neurons, overexpression of APP-BP1
triggers DNA synthesis followed by apoptosis [2,8]. In
Alzheimer's disease (AD), the modified APP-BP1 (about
65 kDa) is elevated in the Triton-insoluble and SDS-solu-
ble fractions [8]. Neddylation appears to be activated in
AD hippocampal neurons [8]. Based on these data, we
hypothesized before that APP-BP1 was detrimental to
neuronal survival. However, APP-BP1 mRNAs are highly
expressed in brain hippocampal pyramidal and granule
cells [1]. Nedd8 and APP-BP1 proteins are also expressed
in hippocampal neurons [8]. The hippocampal brain
structure is critical for learning and memory and is
severely damaged in AD. These observations suggest that
APP-BP1 may have a regulatory function in neurons in
physiological conditions.
APP-BP1 binds to the cytoplasmic domain of APP. Data
from primary neuronal cultures and drosophila genetics
suggest that APP signaling is at least partially transmitted
through APP-BP1 [8,9]. APP undergoes cleavage within
the membrane spanning region by the γ-secretase to gen-
erate Aβ. Presenilin is an integral component of the γ-
secretase complex [10]. Presenilin-deficiency abolishes
most Aβ genesis if not all of it [11-13]. In addition to the
interactions with APP, both APP-BP1 and PS1 regulate β-
catenin levels. For example, knocking out the activation
subunit of the Nedd8 activating enzyme, Uba3 [7], or PS1
[14] in mice both result in β-catenin accumulation. Based
on these intricate connections, here we examined if APP-
BP1 played a role in APP processing. We reported that
suppression of APP-BP1 by small interfering RNAs (siR-
NAs) in primary neurons induced a dramatic increase in
Aβ42 production accompanied by increased levels of APP
and APP-CTFs. We examined the possible mechanisms of
APP-BP1-mediated Aβ42 inhibition and discovered that
APP-BP1 expression facilitated PS1-CTF degradation.
Results
APP-BP1 co-precipitated with PS1 in rat brain extracts
PS1 and APP-BP1 both mediate the degradation of β-cat-
enin and interact with the Aβ precursor protein APP.
These connections indicated that APP-BP1 might also
interact with PS1. In these experiments, rat brain PS1 was
precipitated with the rabbit PS1 antibody which was
raised against the PS1 loop region (AB5308, Chemicon).
The immunoprecipitates were transferred to the nitrocel-
lulose membrane and blotted with the rabbit anti-APP-
BP1 antibody, BP339. The anti-PS1 antibody specifically
pulled down APP-BP1 (Fig. 1A, top). In the controls
where no antibody or an antibody against cyclin B1 was
added to the lysate, no APP-BP1 was specifically co-pre-
cipitated. Probing the same blots with anti-nicastrin
revealed that nicastrin was also in the same complex (Fig.
1A, bottom).
In reciprocal co-ip experiments, APP-BP1 was precipitated
with BP339 from membrane fractions extracted from rat
brains (Fig. 1B). PS1-full length (FL) and PS1-CTF probed
by the anti-PS1 (AB5308) co-precipitated with APP-BP1.
PS1-NTF was blotted by the PS1-NTF antibody (gift from
Huaxi Xu). Fig. 1B shows that PS1 co-precipitated with
APP-BP1. Similarly, nicastrin was present in the APP-BP1
immunoprecipitate as identified by anti-nicastrin blot.
Based on the whole cell extracts loaded onto the same
gels, we estimated that about 1% of PS1 was associated
with APP-BP1. Interestingly, also about 1% of APP-BP1
was precipitated with PS1. Together, these co-ip experi-
ments suggest that APP-BP1 is associated with the γ-secre-
tase complex, and mutagenesis may be needed to
determine the molecular interactions between APP-BP1
and the components of the γ-secretase complex.
APP-BP1 co-migrated with PS1 in glycerol gradient
APP-BP1 mainly exists as a cytoplasmic protein about 59
kDa in crude cell fractionation analyses (Chen, unpub-
lished observation). A small percentage of APP-BP1 that
migrates at about 65 kDa can be isolated by sucrose gradi-
ent centrifugation from the Triton-insoluble and SDS-sol-
uble brain protein extracts which are also enriched with
lipids [8]. APP-BP1 post-translational modifications may
allow it to interact with membrane proteins. Lipid rafts
may be the site of Aβ42 genesis [15]. Lipid rafts also con-
tain the fully assembled γ-secretase complex [16].
The PS1-containing γ-secretase complex can also be iso-
lated from membrane fractions on a glycerol gradient
[17], suggesting that the fractions from the sucrose and
glycerol gradient centrifugations have overlapping resolu-
tions. Here, we used glycerol gradient to determine if APP-
BP1 co-migrated with PS1 in brain membrane protein
extracts (Fig. 1C). Fraction 1 represented the top of the
gradient. APP-BP1 co-migrated with PS1 after glycerol gra-
dient centrifugation in fractions 2 to 4, which also con-
sisted of another γ-secretase component, nicastrin. Note
that the 65 kDa APP-BP1 specifically migrated with PS1 in
the glycerol gradient, similar to what we previously
observed in the lipid-enriched Triton-insoluble and SDS-

Page 3

Molecular Neurodegeneration 2007, 2:3http://www.molecularneurodegeneration.com/content/2/1/3
Page 3 of 12
(page number not for citation purposes)
soluble brain extracts after sucrose gradient centrifugation
[8]. The fainter band right below the major band of APP-
BP1 in fractions 2 and 3 was the non-modified 59 kDa
APP-BP1. This 59 kDa APP-BP1 was probably reduced
from the 65 kDa form due to frequent handling judging
from our experience of working with these samples. The
distribution of β-catenin was more spread out in the glyc-
erol gradient with fractions 4 to 6 containing the highest
amount (Fig. 1C). The distribution of PS1 was similar to
those reported in glycerol gradients under similar condi-
tions [18]. A PS1-CTF-specific antibody, PS1-C [19], also
recognized the same band about 22 kDa in Fig. 1C (Fig.
1D). Together, our data suggested that the modified APP-
BP1 present in a lipid-enriched membrane fraction might
modulate PS1 γ-secretase activity. We are currently analyz-
ing the type of modification that APP-BP1 may undergo
for membrane targeting.
APP-BP1 and PS1 colocalized with each other
The PS1 and APP-BP1 antibodies used in the above exper-
iments are well documented. To examine the interaction
of APP-BP1 and PS1 even further, primary neurons were
labelled with mouse anti-myc antibody for myc-APP-BP1
and rabbit anti-PS1-C. As shown in Figure 2, APP-BP1 and
PS1 were colocalized as indicated by the orange color
when the two colors were merged. Similar results were
also obtained using the mouse anti-APP-BP1 and the rab-
bit anti-PS1 (AB5308) antibodies for the endogenous
APP-BP1 and PS1 (not shown). APP-BP1 appeared to co-
localize with PS1 outside the nucleus. Future analyses
may determine the precise location of these interactions.
Suppression of APP-BP1 by siRNAs lead to the
accumulation of APP and APP-CTFs, and a dramatic
increase in the levels of cell-associated Aβ42
We showed that APP-BP1 interacted with PS1 (Fig. 1 &2),
indicating that APP-BP1 might functionally regulate γ-
secretase activity. To determine if APP-BP1 affected the γ-
secretase cleavage of APP, we examined the effect of APP-
BP1 siRNAs in APP processing by western blots and Aβ
ELISA in primary neurons. Neurons were infected with
human APP695 virus together with the APP-BP1 siRNA
virus. As shown in Figure 3A, the endogenous APP-BP1 in
primary neurons was suppressed by APP-BP1 siRNAs but
not by the missense siRNAs. γ-Tubulin levels did not
change under these experimental infection conditions.
In another set of experiments, the human APP695 was
expressed in primary neurons with or without a siRNA
virus (Fig. 3B). Immunoblots with the anti-APP antibody,
369, showed that APP siRNAs suppressed APP protein
expressions. In contrast, in the presence of APP-BP1 siR-
NAs, APP was increased, opposite to the effect of APP siR-
NAs. The control missense siRNAs did not have major
effects on APP metabolism. Quantitative western blot
APP-BP1, PS1 and nicastrin co-immunoprecipitated and co- migrated in brain protein extracts
Figure 1
APP-BP1, PS1 and nicastrin co-immunoprecipitated
and co-migrated in brain protein extracts. A. APP-BP1
co-immunoprecipitated with PS1 in brain lysates. Adult rat
brain lysates were immunoprecipitated with the rabbit anti-
PS1 antibody (middle lane), or a non-related antibody (rabbit
anti-cyclin B1). The blot was probed with the anti-APP-BP1
antibody, BP339 or with rabbit anti-nicastrin. Anti-PS1 specif-
ically precipitated APP-BP1 and nicastrin which were absent
in control lanes. This experiment has been repeated three
times. B. PS1 co-precipitated with APP-BP1 in membrane
fractions. Proteins extracted from membrane fractions were
precipitated with the APP-BP1 antibody, BP339 or with the
BP339 preimmune serum (pre) or beads alone. The blot was
probed with the PS1 antibody (AB5308) for PS1-FL and PS1-
CTF, or with antibodies against PS1-NTF or nicastrin. C.
APP-BP1 co-migrated with PS1 in glycerol gradient of mem-
brane protein extracts isolated from adult rat brains. Brain
membrane proteins were subjected to glycerol gradient cen-
trifugation. Proteins in equal volumes from each fraction
were resolved by SDS-PAGE gels. The blots were incubated
with specific antibodies followed by ChemiGlow detections.
Fractions 2 to 4 contain the majority of each protein ana-
lyzed. APP-BP1 was found to co-migrate with PS1 (FL, NTF,
and CTF), and nicastrin. The location of β-catenin was also
shown in the glycerol gradient. BioRad Precision Plus protein
standards are shown on the right. PS1-CTF identified by
AB5308 antibody was verified by a PS1-C-specific antibody
(D).

Page 4

Molecular Neurodegeneration 2007, 2:3http://www.molecularneurodegeneration.com/content/2/1/3
Page 4 of 12
(page number not for citation purposes)
analyses revealed a significant increase in APP levels in
APP-BP1 siRNA-expressing neurons (Fig. 3D, n = 3, p <
0.003, two-tail t-Test).
In neurons infected with APP695 with or without the
APP-BP1 siRNA virus, APP-CTF levels were analyzed in
16% Tris-Tricine gels using the antibody 369. A represent-
ative blot is shown in Fig. 3C. Based on a two-tail t-Test,
APP-CTFs significantly accumulated in the APP-BP1
siRNA-expressing cells (n = 4, p < 0.001) (Fig. 3D). The
accumulation of APP-CTFs in these experimental condi-
tions was similar to the effect of the γ-secretase inhibitor
L685459 (Sigma) (Fig. 3C). Increases in APP-CTFs have
been shown in familial APP mutants [20] and in PS1
experimental deletion mutants [21]. Accumulation of
APP and APP-CTFs in our experiments indicated that APP-
BP1 might modulate APP cleavage by the γ-secretase.
To determine if APP-BP1 affected the γ-secretase cleavage
of APP, neurons were infected with APP695 with or with-
out a siRNA virus. The levels of cell-associated and
secreted Aβ40 and Aβ42 was analyzed by ELISA. Aβ levels
were normalized to the sample that expressed only APP695
(100%) (Fig. 4A&B). In the presence of APP-BP1 siRNAs,
the cell-associated Aβ42 was dramatically increased, but
Aβ40 was elevated to a much lesser extent (Fig. 4A&B, left,
respectively). Increases in cell-associated Aβ42 were
accompanied by higher Aβ42 secretions in the medium
(Fig. 4A, right). Aβ40 secretion was largely unaffected (Fig.
4B, right). These results obtained with the ELISA kits from
Biosource International were confirmed in independent
experiments using the Aβ ELSIA kits from the Genetics
Company. These data suggested that APP-BP1 might spe-
cifically inhibit APP cleavage into Aβ42.
APP-BP1 expression resulted in a reduction of rat
endogenous APP but did not affect APLPs in primary
neurons
We showed that suppression of the rat endogenous APP-
BP1 caused a dramatic increase in cell-associated Aβ42
(Fig. 4A, left) in addition to the accumulation of APP and
CTFs (Fig. 3B–D). To further validate these APP-BP1
shRNA experiments, we inspected if overexpression of
APP-BP1 in primary neurons would inhibit APP process-
ing opposite to the effect of APP-BP1 suppression by siR-
NAs. In Fig. 5, a myc-tagged human APP-BP1 was
expressed in primary neurons and the rat endogenous
APP levels were analyzed. A representative western blot
was shown in Fig. 5B. Quantitative western blot analyses
showed that APP was significantly reduced in mycAPP-
BP1 expressing neurons (Fig. 5C). The APP-like proteins
can also undergo γ-secretase cleavage [22]. However, there
were no significant changes in the levels of the APP-like
proteins, APLP1 or APLP2, due to mycAPP-BP1 expression
in these neurons. The decreases in the rat endogenous
APP-CTFs were not analyzed because they were difficult to
detect, unlike in experimental conditions where the
human APP was expressed in primary neurons (Fig. 3C).
The decrease in APP and not in APLPs indicated that APP-
BP1 specifically affected APP cleavage although APLPs,
especially APLP2, are highly homologous to APP. During
our analyses, we found that APP-BP1 also interacted with
APP upstream of C31 near the inner plasma membrane
insertion site (see Additional file 1, Fig. 1) [1,8]. The long
APP-BP1 and PS1 molecules were colocalized in primary neurons
Figure 2
APP-BP1 and PS1 molecules were colocalized in primary neurons. Primary neurons were fixed, permeablized,
blocked and stained with 9E10 for mycAPP-BP1 and the rabbit anti-PS1 (AB5308). Primary antibodies were coupled to Alexa
Fluor 594 and Alexa Fluor 488 goat secondary antibodies. The images presented were obtained under a 40× objective with a
Nikon confocal microscope using the EZ-C12.20 scanning program. Independent experiments using DAPI staining along with
the above two antibodies did not show nuclear co-localization of APP-BP1 and PS1 (not shown).

Page 5

Molecular Neurodegeneration 2007, 2:3 http://www.molecularneurodegeneration.com/content/2/1/3
Page 5 of 12
(page number not for citation purposes)
interacting site in APP cytoplasmic tail by APP-BP1 might
be necessary for the specificity of APP-BP1.
APP-BP1 expression inhibited cell-associated Aβ levels
We showed that APP-BP1 expression specifically down-
regulated rat endogenous APP (Fig. 5), but it is difficult to
detect the rat endogenous Aβ levels. Here we examined if
expressing human APP-BP1 can reduce the levels of cell-
associated Aβ in primary neurons that co-expressed
human APP695 Equal amount of cell extracts was precipi-
tated by the antibody 6E10 and the precipitates were ana-
lyzed by western blotting using the antibody 4G8. The
locations of Aβ peptides were determined by the standard
on the same blot. Semi-quantitative analyses of Aβ levels
from four co-ip experiments showed that APP-BP1 expres-
sion significantly reduced Aβ levels compared to the con-
trol (vector virus p1005+) (Fig. 5D. p < 0.04, one tail t-
Test). A representative blot is shown as an insert (Fig. 5D).
Downregulation of APP-BP1 by siRNAs in primary neurons affected APP processing
Figure 3
Downregulation of APP-BP1 by siRNAs in primary
neurons affected APP processing. A. APP-BP1 was sup-
pressed by APP-BP1 siRNAs but not by missense siRNAs.
Primary neurons were infected with 0.5 IU of siRNA virus
per cell. Equal amount of proteins extracted by RIPA buffer
was analyzed by western blotting for APP-BP1 expression.
The same blot was reprobed with a mouse anti-γ-tubulin
antibody. No obvious changes in γ-tubulin levels were
detected. BP1siRNA, APP-BP1siRNA; mis-siRNA, misssense
siRNA. B. Suppression of APP-BP1 by siRNAs caused
increases in exogenous human APP. Primary neurons were
infected with or without APP and siRNA constructs. APP
was probed with the antibody 369. The same membrane was
reprobed for γ-tubulin with a mouse anti-γ-tubulin for load-
ing controls. In each sample, 10 μg of proteins was analyzed.
The data is representative of three-independent experi-
ments. C. APP-BP1 siRNA inhibited APP-CTF processing in
primary neurons expressing human APP. APP-CTFs were
resolved on 16% Tris-Tricine gels. The blot was probed with
369. A representative of 4 independent experiments was
shown. The intervening lanes between mis-siRNA and
L685459 were cut off. D. APP-BP1 siRNA expression signifi-
cantly blocked APP processing. Quantitative western blot
analyses of 3 (APP) or 4 (CTF) independent experiments
were carried out using two-tail t-Test (APP: BP1siRNA vs
mis-siRNA, p < 0.003; CTF: BP1siRNA vs mis-siRNA, p <
0.001).
Suppression of APP-BP1 protein expression by APP-BP1 siR-NAs in primary neurons resulted in increases of intracellular Aβ42
Figure 4
Suppression of APP-BP1 protein expression by APP-
BP1 siRNAs in primary neurons resulted in increases
of intracellular Aβ42. Intracellular (from 50 μg of protein
from lysates) and secreted (from 1/15 volume of conditioned
medium) Aβ42 in primary neurons were determined by
Aβ42 ELISA (A). Intracellular (from 50 μg of protein) and
secreted (from 1/30 volume of medium of conditioned
medium) Aβ40 in primary neurons was determined by Aβ40
ELISA (B). The amount of Aβ in samples that expressed APP
without any siRNA interference was used as 100% of Aβ
production. All the rest of the samples were normalized to
this sample, and presented as percentage in respective exper-
imental conditions. Data is representative of three independ-
ent experiments. BP1siRNA, APP-BP1 siRNA; NegsiRNA,
Negative siRNA (Stratagene).