Finally, confocal microscopy showed co-localization of c-secretase components and a lipid raft marker in thin sections of human brain.. We conclude that the active c-secretase complex is
Trang 1membranes in human brain
Ji-Yeun Hur, Hedvig Welander, Homira Behbahani, Mikio Aoki*, Jenny Fra˚nberg, Bengt Winblad, Susanne Frykman and Lars O Tjernberg
Karolinska Institutet (KI) Dainippon Sumitomo Pharma Alzheimer Center (KASPAC), KI-Alzheimer’s Disease Research Center, Department of Neurobiology, Care Sciences and Society, Karolinska Institutet, Novum, Huddinge, Sweden
The loss of synapses and neurons in Alzheimer’s
dis-ease (AD) is thought to be, at least partly, induced
by toxic species formed by the amyloid b-peptide
(Ab) [1] Ab is produced from the amyloid precursor
protein (APP) by sequential proteolytic cleavages
mediated by b-secretase (BACE) and c-secretase [2]
An initial cleavage by b-secretase produces soluble
APP (b-APPs) and a membrane-bound C-terminal fragment (C99) that is cleaved by c-secretase, generat-ing the APP intracellular domain (AICD) and Ab Two major forms of this amyloidogenic peptide are produced, Ab40 and Ab42, the latter being less abundant but more prone to aggregation [3–5] The polymerization of Ab into fibrils leads to formation
Keywords
Alzheimer’s disease; detergent-resistant
membranes; human brain; lipid rafts;
c-secretase
Correspondence
L O Tjernberg, Department of
Neurobiology, Care Sciences and Society,
Karolinska Institutet, Novum, KASPAC,
Plan 5, 141 57 Huddinge, Sweden
Fax: +46 8 585 83610
Tel: +46 8 585 83620
E-mail: Lars.Tjernberg@ki.se
*Present address
Genomic Science Laboratories, Functional
Genomics Group, Osaka, Japan
(Received 29 October 2007, revised 22
December 2007, accepted 8 January 2008)
doi:10.1111/j.1742-4658.2008.06278.x
Several lines of evidence suggest that polymerization of the amyloid b-pep-tide (Ab) into amyloid plaques is a pathogenic event in Alzheimer’s disease (AD) Ab is produced from the amyloid precursor protein as the result of sequential proteolytic cleavages by b-secretase and c-secretase, and it has been suggested that these enzymes could be targets for treatment of AD c-Secretase is an aspartyl protease complex, containing at least four trans-membrane proteins Studies in cell lines have shown that c-secretase is partially localized to lipid rafts, which are detergent-resistant membrane microdomains enriched in cholesterol and sphingolipids Here, we studied c-secretase in detergent-resistant membranes (DRMs) prepared from human brain DRMs prepared in the mild detergent CHAPSO and isolated
by sucrose gradient centrifugation were enriched in c-secretase components and activity The DRM fraction was subjected to size-exclusion chromato-graphy in CHAPSO, and all of the c-secretase components and a lipid raft marker were found in the void volume (> 2000 kDa) Co-immunoprecipi-tation studies further supported the notion that the c-secretase components are associated even at high concentrations of CHAPSO Preparations from rat brain gave similar results and showed a postmortem time-dependent decline in c-secretase activity, suggesting that DRMs from fresh rat brain may be useful for c-secretase activity studies Finally, confocal microscopy showed co-localization of c-secretase components and a lipid raft marker
in thin sections of human brain We conclude that the active c-secretase complex is localized to lipid rafts in human brain
Abbreviations
AD, Alzheimer’s disease; AICD, APP intracellular domain; Aph-1, anterior pharynx defective-1; APP, amyloid precursor protein; Ab, amyloid b-peptide; BACE, b-site APP cleaving enzyme; CHAPSO, 3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propanesulfonate; CTF, C-terminal fragment; CT-B, cholera toxin subunit B; DRM, detergent-resistant membranes; Endo H, endo-b-N-acetylglucosaminidase;
ER, endoplasmic reticulum; Nct, nicastrin; NTF, N-terminal fragment; Pen-2, presenilin enhancer-2; PNGase F, peptide N-glycosidase F;
PS, presenilin; SEC, size-exclusion chromatography.
Trang 21 : 1 : 1 : 1 (PS : nicastrin : Aph-1 : Pen-2) c-Secretase
is believed to be an aspartyl protease, as aspartate
resi-dues at positions 257 and 385 within transmembrane
domains 6 and 7 of PS seem to constitute the active site
of the protease [8] Assembly of the complex is initiated
in the ER, where Aph-1 and nicastrin interact, followed
by binding of PS Thereafter, Pen-2 binds to the
com-plex and facilitates endoproteolysis of PS into N- and
C-terminal fragments (PS-NTF and PS-CTF
respec-tively), resulting in an active c-secretase complex [9]
c-Secretase activity can be reconstituted in
Saccaro-myces cerevisiae, which lacks endogenous c-secretase
activity, by co-expressing PS, nicastrin, Aph-1 and
Pen-2 [10] Thus, these four proteins appear to be sufficient
for c-secretase activity, but it is possible that other
pro-teins could play a regulatory role For instance, recent
studies have shown that TMP21, a protein involved in
protein transport and quality control in the ER and
Golgi, as well as the transmembrane glycoprotein
CD147, interact with c-secretase and decrease Ab
production [11,12] Importantly, c-secretase has several
other substrates in addition to APP, all of which
are type 1 transmembrane proteins The multitude of
c-secretase substrates [13] has made development of
clinically useful inhibitors for the treatment of AD
diffi-cult For instance, gastrointestinal side-effects related
to decreased Notch signaling have been reported [14]
Therefore, it is necessary to obtain detailed knowledge
on how c-secretase activity is regulated and how the
complex selects its substrate in order to design drugs
that selectively modify the cleavage of APP
Not only protein–protein interactions but also the
lipid membrane environment can affect the activity of
proteins High cholesterol levels increase Ab
produc-tion, and high cholesterol levels in mid-life are
corre-lated with the incidence of AD at older ages [15]
Apolipoprotein E (ApoE) is involved in cholesterol
transport, and the ApoE4 isoform is a risk factor for
AD [16] Thus, cholesterol seems to have an important
role in APP processing and AD pathogenesis
Choles-terol and sphingolipids are the major lipid constituents
of ordered microdomains in cell membranes These
microdomains are called lipid rafts and are considered
the composition of lipid rafts However, different detergents give different results [20], and DRM prepa-rations do not capture the dynamics of lipid rafts Thus, the occurrence of a protein in DRMs indicates that it could be localized to lipid rafts, but further studies in intact cells or tissue sections are needed to confirm such localization
Certain proteins are concentrated to lipid rafts, and several studies have suggested that the trafficking and processing of APP partly depends on lipid rafts [21– 25] APP, BACE and c-secretase have been shown to localize to lipid rafts, but the degree of localization differs between studies [21–27] Possible explanations for the different results include choice of cell lines, whether the cells overexpress the proteins of interest, and the various detergents used for preparation of DRMs As the majority of studies on c-secretase have been performed using cell lines (in many cases trans-fected cell lines), further studies in brain material are warranted
Here, we show that c-secretase components, as well
as c-secretase activity, are highly enriched in DRMs prepared from human brain The size of the DRMs containing c-secretase was estimated by size-exclusion chromatography (SEC) to be > 2000 kDa, indicating the presence of other proteins and lipids Preparations
of DRMs from rat brain showed a similar distribution
of the c-secretase components and a postmortem time-dependent decline in c-secretase activity Finally we used confocal microscopy and verified the co-localiza-tion of c-secretase components and a lipid raft marker
in thin sections of human brain In summary, our data indicates that the active c-secretase complex is local-ized to lipid rafts in human brain
Results
The c-secretase complex is present in DRMs Previous studies have suggested that BACE1, c-secre-tase and APP are located in lipid rafts in cultured cells and mouse brain [21–24] However, the association
of c-secretase with DRMs in human brain has not
Trang 3previously been reported, and there are no studies on
the activity of c-secretase in DRMs from mammalian
brain Here, we studied the co-localization of active
c-secretase with DRMs in human brain We also included
preparations from rat brain in our study, because we
wished to determine whether there are any significant
differences between the two species regarding
c-secre-tase activity and distribution in DRMs To investigate
association of the c-secretase complex with lipid rafts in
brain, we used a procedure based on centrifugation in a
stepped sucrose gradient in which the DRMs float
to the interface between 5% and 35% sucrose In the
initial experiment, we used freshly prepared membranes
(P3, 100 000 g pellet) from rat brain as well as from
SH-SY5Y neuroblastoma cells We chose
3-[(3-chol-
amidopropyl)dimethylammonio]-2-hydroxy-1-propane-sulfonate (CHAPSO) to dissolve the membranes as it is
the detergent that best preserves c-secretase activity
[28–31] A concentration of around 0.4% CHAPSO
gives the highest activity [31], but separation between
DRMs and soluble components using 0.25–1.0%
CHAPSO was poor (data not shown) The separation
was improved when DRMs were prepared from
mem-branes solubilized in 2.0% CHAPSO Western blot
analysis showed that PS1-NTF and caveolin-1 (a lipid
raft marker) were localized to a large extent to the
inter-face between 5% and 35% sucrose (fraction 2), while
calnexin (a non-raft marker) was found in the 45%
sucrose fraction (fraction 5) (Fig 1A) Another lipid
raft marker, flotillin-1, showed poor separation in rat
brain (Fig 1A) In contrast, CHAPSO DRMs prepared
from SH-SY5Y cells showed a distinct localization of
flotillin-1 to the 5–35% interface (fraction 3, Fig 1B)
The same pattern was observed for another lipid raft
marker, GM1, which is labeled by the cholera toxin
subunit B (Fig 1B) The pronounced separation of
lipid raft markers from a non-raft marker in SH-SY5Y
cells indicates that it is easier to prepare DRMs from
SH-SY5Y cells than from brain tissue The c-secretase
components PS1-NTF and Pen-2 were also found in the
5–35% interface in SH-SY5Y cells For comparison, we
also prepared DRMs in 1% Triton X-100, a detergent
that is frequently used for isolating DRMs, but no
PS1-NTF was found in the DRM fraction (Fig 1C,D)
Thus, 2.0% CHAPSO is suitable for separation of
DRMs containing c-secretase components from
solu-ble material The 5–35% interface, which contains
DRMs and c-secretase, will be referred to as the DRM
fraction
Using the protocol described above, we prepared
DRMs from human brain Six fractions were collected
from the top of the tube and subjected to western blot
analysis using antibodies directed to the c-secretase
components BACE, APP, APP C-terminal fragments (APP-CTFs) and raft and non-raft markers Fraction 4 (at the 35–45% interface) and fraction 5 (45% sucrose) were enriched in the non-lipid raft markers, calnexin (ER) and adaptin-c (trans-Golgi network) PS1-NTF, nicastrin, Aph-1aL and Pen-2 were found in the DRM fraction, while only around 10% of the total protein was found in this fraction (Fig 2A,C) Interestingly, the majority of BACE1, full-length APP and APP-CTFs were distributed to fractions 4 and 5 The procedure was repeated using rat brain, and the results were in line with those obtained for human brain (Fig 2B,D) However, in rat brain, the localization of flotillin-1 and caveolin-1 differed between preparations, and they were also found in fraction 5 to a varying extent This could possibly be due to the more heterogenous and more lipid-rich starting material as the whole rat brain was used
The mature form of nicastrin is found in DRMs
In the active c-secretase complex, nicastrin is glycosy-lated [32] To determine the glycosylation status of nicastrin in DRMs and fractions 4 and 5 from human and rat brain, endoglycosidase H (Endo H) or N-gly-cosidase F (peptide-N-glyN-gly-cosidase F, PNGase F) were applied to deglycosylate nicastrin Endo H works on a more limited range of substrates than PNGase F Untreated DRMs contained a nicastrin species of approximately 125 kDa (Fig 2E) Endo H decreased the apparent molecular weight of nicastrin from approximately 125 kDa to approximately 100 kDa, indicating the presence of high-mannose oligosaccha-rides Upon treatment with PNGase F, which also removes complex oligosaccharides, the deglycosylation was more pronounced, resulting in a diffuse band at approximately 80 kDa (Fig 2E) The deglycosylation pattern of nicastrin was the same in fractions 4 and 5
as in DRMs, and no differences between human and rat brain were observed The above results suggest that the nicastrin that is present in DRMs (fraction 2) as well as in fractions 4 and 5 is highly glycosylated, including high-mannose oligosaccharides and complex oligosaccharides The results were confirmed using another nicastrin antibody (BD Biosciences, San Jose,
CA, USA, data not shown)
DRMs containing c-secretase elute in a high-molecular-weight SEC fraction
To further purify and investigate the approximate molecular weight of DRMs containing the c-secretase complex, we injected the DRM fraction from human
Trang 4brain onto a Superose 6 SEC column, collected
frac-tions and analyzed them by western blotting using
antibodies directed to all the known c-secretase
com-plex components When using 0.25% CHAPSO as the
mobile phase, the c-secretase components, APP and
flotillin-1 eluted with the void volume (> 2000 kDa)
(Fig 3A,B) DRMs from rat brain gave similar results
(Fig 3C) The relatively narrow peak indicates that
the complex is stable during separation When using
2% CHAPSO as the mobile phase, most of the
nicas-trin and PS1-NTF eluted in the void volume, although
a second peak around 230 kDa could be observed
(corresponding to fractions 17–21, Fig 3F) Thus, the
c-secretase complexes are mainly present in large
DRMs (> 2000 kDa)
The c-secretase complex components can be co-immunoprecipitated from DRMs
The stability of the complex was further evaluated
by co-immunopreciptation The starting material for DRM preparation (P3), fraction 2 (DRMs) and frac-tion 5 (soluble fracfrac-tion) from the rat brain DRM preparation were immunoprecipitated using an anti-body against nicastrin (Fig 4) Western blotting showed that PS1-CTF, Aph-1aL and Pen-2 co-immu-noprecipitated with nicastrin in P3 and the DRM frac-tion, and, to a lower degree, in the soluble fraction Flotillin-1 did not co-immunoprecipitate with nicastrin, indicating that flotillin-1 and c-secretase are present in different lipid rafts
Fig 1 DRMs prepared in 2% CHAPSO are
enriched in c-secretase components DRMs
from (A) rat brain and (B) SH-SY5Y cells
were isolated by sucrose gradient
centrifugation after treatment with 2.0%
CHAPSO In rat brain, six fractions were
collected from the top of the tube:
frac-tion 1, fracfrac-tion 2 (DRM fracfrac-tion, interface
between 5% and 35% sucrose), fraction 3,
fraction 4 (interface between 35% and 45%
sucrose), fraction 5 and fraction 6 (pellet) In
SH-SY5Y cells, 12 fractions were collected
from the top of the tube: fraction 1–2,
frac-tion 3 (DRM fracfrac-tion, interface between 5%
and 35% sucrose), fraction 4–9, fraction 10
(interface between 35% and 45% sucrose),
fraction 11 and fraction 12 The fractions
were subjected to western blot analysis
using flotillin-1 and caveolin-1 (lipid raft
markers), calnexin (a non-raft marker), and
PS1-NTF In SH-SY5Y cells, the ganglioside
GM1 (a lipid raft marker) was detected by
binding of cholera toxin subunit B using a
dot-blot assay In (C) and (D), the DRMs
were isolated after treatment with 1%
Tri-ton X-100 of (C) rat brain and (D) SH-SY5Y
cells.
Trang 5DRMs contain active c-secretase complex
To investigate whether the c-secretase complex present
in DRMs is active, we incubated fractions 2 (DRMs), 4
and 5 from rat brain in the absence or presence of 1 lm
of the c-secretase inhibitor L-685,458 (Fig 5A), and found that AICD was produced in DRMs only in the absence of L-685,458 Although there were higher amounts of full-length APP and APP-CTFs in frac-tions 4 and 5, the highest amount of AICD was clearly
Fig 2 DRMs containing the c-secretase complex can be isolated from human and rat brain The protein concentration was analyzed by BCA assay in (A) human brain and (B) rat brain (C) Human brain membranes were treated with 2.0% CHAPSO, fractionated on a sucrose gradient, and subjected to western blot analysis using antibodies directed to the c-secretase complex components’ BACE1, APP, APP-CTFs, flotillin-1 and caveolin-1 (lipid raft markers) and calnexin and adaptin-c (non-raft markers) (D) The experiment was repeated using rat brain The higher-molecular-weight form of nicastrin (> 125 kDa, labeled with an asterisk) was only detected by one antibody, and this was due to non-specific binding (E) Fractions 2 (DRMs), 4 and 5 for human brain and rat brain were denatured and incubated overnight at 37 C with glycosidases (Endo H and PNGase F) The samples were analyzed by western blot using anti-nicastrin serum The control was incubated overnight at 4 C.
Trang 6generated in DRMs The immunoreactive band co-migrated with a 50-residue synthetic AICD peptide and was detected by several antibodies (data not shown) Using an exogenous substrate, C99-FLAG, and a sensi-tive sandwich ELISA method, we were also able to detect Ab production, which was inhibited by L-685,458
in the DRM fraction from rat brain (Fig 5B)
In the next step, we determined whether c-secretase activity could be detected in a human brain sample with a postmortem time of 22 h, and were able to detect AICD production that was inhibited by L-685,458 (Fig 5C) Thus, DRMs isolated from post-mortem human brain tissue contain active c-secretase that cleaves endogenous APP-CTFs
Fig 3 DRM-associated c-secretase is present in a high-molecular-weight complex, as shown by size-exclusion chromatography (SEC) The DRM fraction was injected onto a Superose 6HR column and fractions were collected from 10–50 min at a flow rate of 0.5 mLÆmin)1 Solubiliza-tion buffer with 0.25% CHAPSO was used as the mobile phase (A) The absorbance at 254 nm was monitored The DRM chromatogram was normalized to the standard chromatogram (B,C) Every second fraction was analyzed by western blot for (B) human brain and (C) rat brain The rat DRM fraction was further analyzed by SEC using solubilization buffer with 0.25% or 2.0% CHAPSO as the mobile phase and the absorbance
at 254 nm was monitored (D) (E,F) Every second fraction was analyzed by western blot for (E) 0.25% CHAPSO and (F) 2.0% CHAPSO.
Fig 4 The c-secretase complex immunoprecipitates in DRMs Rat
membranes (P3), the DRM fraction (fraction 2) and fraction 5 were
co-immunoprecipitated with anti-nicastrin serum or control rabbit
IgG PS1-CTF, Aph-1aL, Pen-2 and flotillin-1 were identified by
western blotting.
Trang 7To investigate the effect of postmortem time on
c-secretase activity, we collected rat brains after 0, 6,
12 or 24 h postmortem time (see Experimental
proce-dures), prepared P3 and analyzed this fraction for
c-secretase activity Compared to freshly prepared
membranes (postmortem time 0 h), the c-secretase activity, measured as AICD production, was decreased
by more than 80% at a postmortem time of 6 h After this time, the activity continued to decrease but was still detectable at 24 h postmortem (Fig 5D) and remained even after 48 h (data not shown) Thus, c-secretase activity decreases most rapidly after short postmortem times, but can be observed in brain tissue
at all time points studied
c-Secretase co-localizes with lipid rafts in human brain sections
The presence of a protein in DRMs suggests that it is associated with lipid rafts To further investigate whether the c-secretase components are associated with lipid rafts, we performed immunofluorescence labeling
on human brain sections Multiple fluorescent staining was used to study co-localization of PS1, nicastrin and APP with the lipid raft marker GM1 Confocal microscopy revealed that GM1 immunoreactivity was most pronounced in the plasma membrane of the cells PS1 and nicastrin immunoreactivity overlapped exten-sively with the lipid raft marker (Fig 6A,B), but the overlap of APP and GM1 was limited (Fig 6C) Thus, confocal microscopy supports the view that c-secretase
is localized to lipid rafts in human brain
Discussion
Previous studies have shown that APP, BACE and c-secretase partially localize to lipid rafts, and it has been suggested that the clustering of these proteins in lipid rafts increase Ab production [15,21] These stud-ies were performed in cell lines, which in many cases overexpressed APP or c-secretase proteins Recently, c-secretase was also found to be associated with DRMs in adult mouse brain [23] However, the associ-ation of c-secretase with lipid rafts in human brain has not been investigated, and biochemical evidence for c-secretase activity in DRMs is limited
Due to their lipid composition, lipid rafts are resis-tant to certain detergents Therefore, isolation of DRMs by treatment with detergents such as Triton X-100 followed by flotation in a discontinuous sucrose gradient is frequently used for studying lipid raft com-ponents It should be noted that these preparations are dependent on the nature and concentration of the detergent used [20] Previously, 2% CHAPSO has been used to isolate DRMs containing an active c-secretase from SH-SY5Y neuroblastoma cells [22], and 0.5% Lubrol WX has been used to isolate c-secretase-rich DRMs from N2a neuroblastoma cells and mouse brain
Fig 5 c-Secretase activity was observed in DRMs by monitoring
AICD and Ab production (A) The production of AICD was assayed in
fractions 2 (DRMs), 4 and 5 by incubation of 100 lg of protein for
16 h at 37 C in the absence or presence of the c-secretase inhibitor
L-685,458 The supernatant was subjected to western blot using the
antibody C1 ⁄ 6.1 (B) The DRM fraction (approximately 12 lg protein)
was incubated for 16 h at 37 C in the absence or presence of the
c-secretase inhibitor L-685,458 Twenty nanograms of C99-FLAG
were added to the samples Ab40 levels were analyzed by sandwich
ELISA (C) The production of AICD from human brain was measured
and detected as in (A) (D) Solubilized membranes from rat brain
obtained at various postmortem times were incubated as in (A) and
AICD production was measured as in (A).
Trang 8[23,33] We and others have previously studied the
effect of various detergents on the activity of
c-secre-tase prepared from rat brain, and found that 0.4%
CHAPSO resulted in the highest activity [31] but Triton
X-100 abolished the activity [34] Hence, we used
CHAPSO to prepare DRMs from human and rat
brain To preserve c-secretase as an active complex, we
started with CHAPSO concentrations in the range
0.25–1.0% However, it was necessary to increase the
CHAPSO concentration to 2.0% to obtain a good separation between raft and non-raft markers We also noted that the separation was better in SH-SY5Y cells than in brain samples Difficulties in obtaining pure DRM fractions from brain tissue are probably due to the heterogeneity of the sample and high levels of myelin
In the case of human brain, the DRM fraction resulting from treatment with 2.0% CHAPSO was
Fig 6 Confocal microscopy shows partial
co-localization of lipid rafts and c-secretase
in human brain tissue Immunofluorescence
labeling was performed on human brain
sections The nucleus was stained with
4’,6-diamidino-2-phenylindole (DAPI) The
cholera toxin subunit B (CT-B) that labels
the lipid rafts is shown by the green
fluores-cence of the Alexa Fluor 488-coupled goat
anti-rabbit serum Expression of PS1-CTF,
nicastrin and APP is shown by the red
fluorescence using secondary anti-mouse
Alexa Fluor 594 conjugates (A) PS1-CTF.
(B) Nicastrin (C) APP Scale bar = 1 lm.
Trang 9enriched in lipid raft marker proteins and all four
known c-secretase complex components, while APP,
APP-CTFs and BACE were mainly found in other
fractions The association of c-secretase with DRMs is
in line with previous results from studies in cells or
mouse brain [22,23,33], and suggests that the majority
of c-secretase in human brain is localized to lipid rafts
Importantly, the localization of PS1 and nicastrin to
lipid rafts was confirmed by immunofluoresence
con-focal microscopy on human brain sections
In accordance with our data, previous studies show
that < 25% of BACE is associated with lipid rafts
[21,23,24,26] Interestingly, it has been suggested that
raft association is necessary for BACE activity [21],
and thus decreasing the amount of raft-associated
BACE could result in lower levels of Ab Our
observa-tion that most of the APP occurs outside the DRM
fraction is in line with previous studies, where the
reported association of APP with DRMs varied from
zero up to 20% [21,23,24,26] The low levels of BACE,
its substrate and the product (see below) in the DRM
fraction could indicate that the initial step in the
amy-loidogenic pathway occurs outside lipid rafts
Alterna-tively, the processing may be initiated by a transient
localization of APP and BACE to lipid rafts
With regard to APP-CTFs, the published
informa-tion is limited due to the difficulties in detecting
endo-genous APP-CTFs in cell lines Using a c-secretase
inhibitor, APP-CTFs accumulated and were detectable
in both DRMs and the soluble fraction from
SH-SY5Y neuroblastoma cells or Chinese hamster ovary
(CHO) cells [22,23,25] In brain tissue, the situation is
different, and endogenous APP-CTFs are readily
detected In a previous study using adult mouse brain,
the majority of APP-CTFs were found in DRMs,
while full-length APP was found in soluble fractions
[23] In contrast, we detected APP-CTFs mainly in the
soluble fraction, and it is possible that the choice of
detergent (2% CHAPSO versus 0.5% Lubrol WX)
could explain this discrepancy
Despite the low substrate levels, AICD production
was easily detected in DRMs from human and rat
brain, while only minor amounts of AICD were
gener-ated in the other fractions To our knowledge, this is
the first study to show c-secretase processing of an
endogenous substrate in DRMs, and the first to show
c-secretase activity in mammalian brain DRMs
Inter-estingly, we could not detect APP and APP-CTFs in
DRMs after incubating the sample at 37C for 16 h
As shown in Fig 2C,D, APP and APP-CTFs are
pres-ent in the DRM fraction before the start of the activity
assay However, the levels of those fragments were
clearly lower in DRMs than in fraction 4 and 5 We
speculate that the low levels of APP-CTFs might be degraded by non-specific protease activity during incu-bation at 37C, which also explains why blocking c-secretase activity using an inhibitor did not lead to the accumulation of APP-CTFs in DRMs APP and APP-CTFs are generally more difficult to detect in human brain material than in rat brain, probably due
to proteolysis during the long postmortem time We were not able to detect production of endogenous Ab, but, after addition of an exogenous substrate, Ab could be detected in the DRM fraction from rat brain using sandwich ELISA Both AICD and Ab produc-tion were inhibited by the c-secretase inhibitor L-685,458
In cell studies, only mature nicastrin, which is the form that associates with the active c-secretase com-plex [35], is localized to DRMs, while immature nicas-trin is detected in other fractions [22,33] However, in accordance with our previous results [36], nicastrin was highly glycosylated in all fractions in the brain study Thus, there is a clear difference in the maturation pro-cess of nicastrin between cell lines and mammalian brain
The predicted molecular weight of the c-secretase complex at a stoichiometry of 1 : 1 : 1 : 1 (PS : nicas-trin : Aph-1 : Pen-2) is approximately 220 kDa [37,38] Previous studies on soluble c-secretase have estimated the size of the complex to vary between 200 and
2000 kDa, and the stoichiometry of the c-secretase complex is not clear [10,36,39–41] The diverse results might be due to differences in starting material, prepa-ration procedures and the techniques used (e.g SEC, blue native PAGE or gradient centrifugation) By SEC, the molecular weight of the DRM fraction was esti-mated to be > 2000 kDa This high-molecular-weight fraction contained the raft marker flotillin-1, the c-sec-retase complex and low amounts of APP and APP-CTFs We suggest that the estimated molecular weight reflects the size of the DRMs (including other proteins, lipids and CHAPSO) rather than the size of the c-secre-tase complex Elution of the soluble c-secrec-secre-tase com-plex has been shown to shift from the void volume to a lower-molecular-weight fraction when the CHAPSO concentration in the mobile phase is increased [42] We detected the majority of the c-secretase components in the high-molecular-weight fraction from DRMs even when 2% CHAPSO was used as the mobile phase These data show that the c-secretase complex is stably associated with DRMs In line with these SEC results,
it was also possible to co-immunoprecipitate PS1, Aph-1aL and Pen-2 using an anti-nicastrin serum in 2% CHAPSO Another indication of the stability of the c-secretase complex is that activity can be observed
Trang 10between lipid rafts and disordered domains of the
membrane seems to regulate processing We speculate
that lowering the levels of c-secretase or APP⁄
APP-CTFs in lipid rafts may be one way to decrease Ab
production Possibly, c-secretase inhibitors that
prefer-entially distribute to lipid rafts might show increased
selectivity for inhibition of Ab production, and thus be
useful for pharmacological treatment of AD
In conclusion, c-secretase is present in DRMs
pre-pared from human and rat brain, and confocal
micro-scopy on sections from human brain confirms that
c-secretase is indeed localized to lipid rafts The
DRM fraction shows high c-secretase activity although
the substrate levels are low, and DRMs prepared
from brain tissue are suitable for studies on active
c-secretase
Experimental procedures
Human brain material
The cortex from a postmortem human brain (postmortem
time 22 h) of a non-Alzheimer case was obtained from
Huddinge Brain Bank (Huddinge, Sweden) and stored at
)70 C before use
Animals
Male Sprague–Dawley rats (200–250 g) were obtained from
B&K Universal (Sollentuna, Sweden) The ethical permit
was granted by the Animal Trial Committee of Southern
Stockholm (no S60-05) The rats were killed by carbon
dioxide treatment The brains were dissected to remove
blood vessels and white matter
Cell culture
The human neuroblastoma cell line, SH-SY5Y, was
cul-tured in Dulbecco’s modified Eagle’s medium supplemented
with 10% fetal bovine serum and 1%
penicillin–strepto-mycin solution (GIBCO⁄ Invitrogen, Carlsbad, CA, USA)
Cells were grown in 5% CO2⁄ 95% air at 37 C Nearly
confluent SH-SY5Y cells in three 150 mm dishes were
heim, Germany) in lysis buffer (1 mL buffer⁄ 0.2 g tissue) containing 20 mm Hepes (pH 7.5), 50 mm KCl, 2 mm EGTA and Complete protease inhibitor mixture (Roche Applied Science, Indianapolis, IN, USA) All procedures were carried out on ice The samples were centrifuged at
1000 g for 10 min to remove nuclei and poorly homo-genized material The pellet was homohomo-genized and then centrifuged at 1000 g for 10 min, and the post-nuclear supernatants were pooled and centrifuged once more at
10 000 g for 30 min in order to remove mitochondria The supernatant was centrifuged once more, and the final supernatant was then centrifuged at 100 000 g for 1 h to yield the final pellet (P3)
Preparation of detergent-resistant membranes
DRMs were prepared as described previously [22] with some modifications To isolate DRMs from brain material
or cells, P3 or the cell pellet, respectively, were resus-pended in 600 lL of buffer containing 20 mm Tris⁄ HCl (pH 7.4), 150 mm NaCl, 1 mm EDTA, 2.0% CHAPSO or 1% Triton X-100, and Complete protease inhibitor mix-ture (Roche Applied Science) The samples were incubated with end-over-end rotation for 20 min at 4C The sample was adjusted to 45% sucrose and placed at the bottom of
a 14 mL Beckman Ultra-Clear centrifuge tube Then, 6.9 mL of 35% sucrose followed by 2.3 mL of 5% sucrose was overlaid The sample was centrifuged at 100 000 g for
16 h at 4C in a SW40Ti rotor (Beckman Coulter, Fuller-ton, CA, USA) Six fractions were collected from the top
of the tube using a 5 mL syringe (CODAN, Hørsholm, Denmark) In order to remove sucrose from the six fractions, PD-10 desalting columns (GE Healthcare, Piscat-away, NJ, USA) were used according to the manufac-turer’s instructions A buffer containing 20 mm Hepes (pH 7.4), 150 mm NaCl, 5 mm EDTA and Complete protease inhibitor mixture (Roche Applied Science) was diluted sevenfold and used to equilibrate the columns The samples were applied, eluted and concentrated to 1· buffer (seven times) using a vacuum centrifuge (Maxi Dry Lyo, Heto-Holten AIS, Allerød, Denmark) The protein concen-tration was determined by BCA protein assay according to the manufacturer’s instructions (Pierce, Rockford, IL, USA)