activation of β2-adrenergic receptor stimulates γ-secretase activity and accelerates amyloid...
TRANSCRIPT
Activation of b2-adrenergic receptor stimulatesg-secretase activity and accelerates amyloidplaque formationYanxiang Ni1,3, Xiaohui Zhao1,3, Guobin Bao1, Lin Zou1, Lin Teng1, Zhu Wang1, Min Song2, Jiaxiang Xiong2,Yun Bai2 & Gang Pei1
Amyloid plaque is the hallmark and primary cause of Alzheimer disease. Mutations of presenilin-1, the c-secretase catalytic
subunit, can affect amyloid-b (Ab) production and Alzheimer disease pathogenesis. However, it is largely unknown whether and
how c-secretase activity and amyloid plaque formation are regulated by environmental factors such as stress, which is mediated
by receptors including b2-adrenergic receptor (b2-AR). Here we report that activation of b2-AR enhanced c-secretase activity
and thus Ab production. This enhancement involved the association of b2-AR with presenilin-1 and required agonist-induced
endocytosis of b2-AR and subsequent trafficking of c-secretase to late endosomes and lysosomes, where Ab production was
elevated. Similar effects were observed after activation of d-opioid receptor. Furthermore, chronic treatment with b2-AR agonists
increased cerebral amyloid plaques in an Alzheimer disease mouse model. Thus, b2-AR activation can stimulate c-secretaseactivity and amyloid plaque formation, which suggests that abnormal activation of b2-AR might contribute to Ab accumulation
in Alzheimer disease pathogenesis.
Alzheimer disease is the most common neurodegenerative disorder ofthe central nervous system. Amyloid plaques, which are largely com-posed of amyloid-b (Ab), are a major characteristic of Alzheimerdisease neuropathology and considered to be the primary cause of thedisease1. Diffusible oligomeric Ab is neurotoxic and potentially relatedto the disease2. Ab is generated from Ab precursor protein (APP) viasequential cleavages by b- and g-secretases3. The g-secretase is pivotal,because it determines the ratio of two main Ab species (Ab40 andAb42)4, and mutations in its catalytic subunit presenilin-1 (PS1)account for most cases of familial Alzheimer disease (FAD)5. BecauseFAD accounts for fewer than 10% of cases of Alzheimer disease andsporadic Alzheimer disease accounts for most cases6, it is important toinvestigate how environmental influences have an impact on and con-tribute to the pathogenesis of the disease. Ab production in cell culturescan be reduced by activation of intracellular signaling pathways7,8 or byactivation of the muscarinic acetylcholine receptor or estrogen recep-tor9,10. Further, Ab levels and amyloid plaque formation can be reducedby environmental enrichment or somatostatin11,12. Therefore, it is ofinterest to know whether and how environmental factors can enhanceg-secretase activity, amyloid plaque formation or both.
Environmental factors such as stress activate receptors includingb-adrenergic receptors (b-ARs) and d-opioid receptor (DOR)13–15.Once activated, these receptors couple to heterotrimeric guaninenucleotide–binding proteins (G proteins) and modulate the levels of
intracellular second messengers, such as cyclic AMP (cAMP)16,17. Theactivated receptor also undergoes clathrin-mediated endocytosis, whichis crucial for receptor desensitization and signal transduction18,19.These receptors, especially b2-AR, are well studied, typical G protein–coupled receptors (GPCRs)19–21 and are expressed in hippocampus andcortex22, the main brain regions involved in Alzheimer disease2. So it isintriguing to examine whether some important GPCRs, like b2-AR, canenhance Ab production. In this study, we found that activation of b2-AR enhances g-secretase activity, resulting in increased Ab production.Activation of DOR induced similar effects. Furthermore, in vivoexperiments with an Alzheimer disease mouse model (APPswe/PS1DE9 double-transgenic mouse) showed that cerebral amyloidplaque formation was increased after chronic treatment with the b2-AR agonists isoproterenol or clenbuterol. In contrast, amyloid plaqueformation was reduced after chronic treatment with a b2-AR antago-nist, ICI 118,551. Thus, these findings provide direct evidence thatg-secretase activity and amyloid plaque formation can be enhanced byb2-AR–mediated extracellular signals, implying that b2-AR antagonistscould be potential drugs for the treatment of Alzheimer disease.
RESULTS
Activation of b2-AR increases Ab production
First, we investigated the effect of b2-AR activation on Ab productionin HEK293 cells that possessed functional GPCR signaling pathways
Received 2 March; accepted 22 September; published online 19 November 2006; doi:10.1038/nm1485
1Laboratory of Molecular Cell Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences; GraduateSchool of the Chinese Academy of Sciences, 320 Yue Yang Road, Shanghai 200031, China. 2Department of Medical Genetics, Third Military Medical University,30 Gao Tan Yan, Chongqing 400038, China. 3These authors contributed equally to this work. Correspondence should be addressed to G.P. ([email protected]).
1390 VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 NATURE MEDICINE
ART ICL ES©
2006
Nat
ure
Pub
lishi
ng G
roup
ht
tp://
ww
w.n
atur
e.co
m/n
atur
emed
icin
e
and exhibited normal Ab secretion. We cotransfected HEK293cells with b2-AR combined with an APP mutant (APPswe), whichharbors FAD-linked ‘Swedish’ mutations23. Stimulation of b2-ARwith its agonist isoproterenol increased the levels of Ab40 andAb42 (Fig. 1a), without reducing cell viability (data not shown).The b2-AR antagonist propranolol abolished the effect of isoproter-enol on the levels of secreted Ab, whereas propranolol alone had noeffect. Furthermore, pretreatment with a specific g-secretase inhi-bitor L685,458 abolished the increase of secreted Ab, indicatingthe involvement of g-secretase in isoproterenol-induced increase inAb production.
C99 derives from b-secretase–mediated cleavage of APP and func-tions as a direct substrate for g-secretase as well as an immediateprecursor for Ab (ref. 24). We cotransfected C99 and b2-AR intoHEK293 cells and found that stimulation of b2-AR with isoproterenolcaused an increase in Ab production (Fig. 1b) that was similar to thatwhen APPswe was used for cotransfection. This increase was againabolished by the presence of propranolol, which had no effect per se.Thus, the increase in secreted Ab was probably due to enhancedg-secretase activity. We further tested another GPCR, DOR, andobserved a similar increase in secreted Ab levels in response toDADLE treatment in C99-transfected HEK293 cells (Fig. 1c). Experi-ments in primary hippocampal cultures transfected with C99 alsoshowed that stimulation of endogenous b2-AR or DOR elevated levelsof secreted Ab (Fig. 1d). Finally, the pulse-chase experiment showed
that C99 turnover was more rapid in DADLE-treated HEK293 cellscotransfected with DOR and C99 than in similarly labeled control cells(Fig. 1e); this implied that activation of the receptor facilitated C99cleavage. Thus, activation of b2-AR or DOR enhances Ab productionand secretion, probably due to promoted g-secretase–mediated clea-vage of C99.
Activation of b2-AR enhances c-secretase activity
The promotion of C99 cleavage could have resulted from enhancedexpression or enzymatic activity of g-secretase. We next analyzedthe effect of b2-AR activation on C99 cleavage with an expressedsubstrate assay25. In this assay, C60 is directly produced from theg-secretase–mediated cleavage of C99 after a 2-h incubation of cellmembrane fractions in vitro, and can be detected by western blotting.We found an elevation of C60 level after isoproterenol treatmentof the C99-transfected HEK293 cells (Fig. 2a) but did not observeany change of the expression level of PS1, which exists as theheterodimer of the amino- and carboxyl-terminal fragments (PS1-NTF and PS1-CTF).
To directly measure the enzymatic activity of g-secretase, we applieda fluorogenic substrate assay, which is based on the secretase-depen-dent cleavage of a g-secretase–specific substrate peptide conjugatedwith fluorescent reporter molecules26. We found that g-secretaseactivity was enhanced 30 min after stimulation of endogenousb2-AR in C6 glioma (Fig. 2b). We confirmed this effect by stimulating
endogenous b2-AR in acute hippocampalslices (Fig. 2c) and by stimulating endogen-ous DOR in SH-SY5Y neuroblastoma andacute hippocampal slices (SupplementaryFig. 1 online). The results from the twog-secretase assays showed that g-secretaseactivity peaked at 30 min and returned tothe basal level after about 60 min of b2-ARstimulation (Fig. 2a and SupplementaryFig. 2 online). Presenilin deficiency abolishedthe isoproterenol-induced enhancement ofg-secretase activity in Psen1�/�Psen2�/�
mouse embryonic fibroblasts (Fig. 2d), con-firming the specificity of the fluorogenicsubstrate assay. Taken together, our dataclearly demonstrate that activation of b2-ARstimulates g-secretase activity, leading to theincrease in Ab production.
C99
Chase time (h):
DADLE
0 3
– + – +
2.0
1.5
1.0
0.5
0.0
Sec
rete
d A
β le
vels
(fol
d of
bas
al)
IsoDADLE
––
++––
2.0
1.5
1.0
0.5
0.0
Sec
rete
d A
β le
vels
(fol
d of
bas
al)
DADLENALT
––
++––
++
2.0
1.5
1.0
0.5
0.0
Sec
rete
d A
β le
vels
(fol
d of
bas
al)
IsoPro
––
++––
++
2.0
1.5
1.0
0.5
0.0
Sec
rete
d A
β le
vels
(fol
d of
bas
al)
IsoPro
––
++––
+L685,458 – –– –
+–+
––+
+
Aβ40Aβ42
Aβ40
Aβ42
Aβ40
Aβ42
Aβ40Aβ42
a b c d
e
* ** *
* * * * * *
Figure 1 GPCR stimulation increases Ab production in cell lines and primary hippocampal cultures. (a–d) ELISA
determination of secreted Ab40 and Ab42 levels in the conditioned medium. Data represent mean ± s.e.m. of three
independent experiments, presented as ratio of the basal levels. *P o 0.01. (a) Effects of isoproterenol (Iso) on the
levels of secreted Ab in HEK293 cells coexpressing b2-AR and APPswe, with or without L685,458 pretreatment
(1 mM). (b,c) Effects of isoproterenol (b) and DADLE (c) on levels of secreted Ab in HEK293 cells coexpressingC99 and either b2-AR (b) or DOR (c). (d) Effects of isoproterenol or DADLE on levels of secreted Abin C99-transfected primary hippocampal cultures. (e) Pulse-chase analysis of C99 cleavage after DOR
stimulation. Pro, propranolol; DADLE, [D-Ala2, D-Leu5]-enkephalin; NALT, naltrindole.
+ – – – –30 0 10 30 60Iso (min)
Pro
C99
C60
PS1-NTF
PS1-CTF
** *
* *
2.5 C6
1.5
2.0
1.0
0.5
0.0
γ-se
cret
ase
activ
ity(f
old
of b
asal
)
IsoPro
– + – +– – + +
3.0
1.5
2.5
2.0
1.0
0.5
0.0
γ-se
cret
ase
activ
ity(f
old
of b
asal
)
IsoPro
– + – +– – + +
1.5
2.0
1.0
0.5
0.0
WT Psen–/–
γ-se
cret
ase
activ
ity(f
old
of b
asal
)
a b c dConIso
Figure 2 b2–AR stimulation enhances g-secretase activity. (a) Analysis of C99 cleavage and C60
production in HEK293 cells using an expressed substrate assay with different durations of isoproterenoltreatment. (b–d) Determination of isoproterenol effects on g-secretase activity in C6 glioma (b), rat
acute hippocampal slices (c), and b2-AR–transfected wild-type (WT) and Psen1�/�Psen2�/� (Psen–/–)
mouse embryonic fibroblasts (d), using a fluorogenic substrate assay. Data represent mean ± s.e.m. of
at least three independent experiments, presented as fold values of the basal activity. *P o 0.001.
Con, vehicle control.
ART ICL ES
NATURE MEDICINE VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 1391
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
Activity enhancement is independent of cAMP signaling
Once stimulated, b2-AR induces Gs protein–dependent adenylyl cyclaseactivation and elevates the level of intracellular cAMP (ref. 17). Todefine the molecular mechanism by which b2-AR activation enhancedg-secretase activity, we first used a new b2-AR mutant (b2-ART68F, Y132G, Y219A or b2-AR TYY), which is incapable of Gs proteinactivation19. We found that stimulation of this b2-AR mutantdid not abolish the enhancement of g-secretase activity (Fig. 3a);this excluded the possibility that Gs protein signaling was involvedin the effect of b2-AR on g-secretase. Furthermore, we treatedcells with some reagents (including cholera toxin, forskolin anddybutyl-cAMP) that mimic Gs-protein activation and cAMP levelelevation, and found that none of them could enhance g-secretaseactivity (Fig. 3b). DOR activates pertussis toxin–sensitive Gi/o
protein27 and decreases cAMP levels by inhibiting adenylyl cyclase16,17.Pretreatment of SH-SY5Y neuroblastoma with pertussis toxin didnot alter the enhancement of g-secretase activity in response toDOR stimulation (Supplementary Fig. 3 online). These resultsfurther confirm that regulation of g-secretase by b2-AR or DORactivation is independent of G-protein signaling or the canonicalcAMP pathway.
Receptor endocytosis correlates with activity enhancement
As the endocytosis of b2-AR also initiates specific signaling18, we theninvestigated whether the mechanism for enhanced g-secretaseactivity involves agonist-induced b2-AR endocytosis. We found thatthe effect of isoproterenol on g-secretase activity was abolished bypretreatment with endocytosis inhibitors such as concanavalin A,hypertonic solution and potassium-depleted medium (Fig. 3c), orby the expression of a dominant-negative version of dynamin (DynK44A) (Fig. 3d) that inhibits clathrin- and caveolin-mediated endo-cytosis28. As b2-AR internalizes mainly in a clathrin-dependentmanner28, we used small interfering RNA (RNAi) against clathrinheavy chain to deplete the cellular expression of clathrin, and observedthat the induced g-secretase activity was abolished (Fig. 3e). Thus,b2-AR–induced enhancement of g-secretase activity is mediated bysignaling mechanisms dependent on agonist-induced and clathrin-mediated endocytosis.
To further confirm the requirement of agonist-induced b2-ARendocytosis for the enhancement of g-secretase activity, we appliedanother b2-AR mutant (b2-AR L339,340A, or b2-AR LL) and anotherintrinsic adrenergic receptor, b3-AR, both of which are deficient inagonist-induced endocytosis20,29. We found that there was no receptorendocytosis (Fig. 3f) or enhanced g-secretase activity (Fig. 3g) wheneither of the receptors was stimulated in HEK293 cells, thoughelevation of cellular cAMP level did occur (data not shown). Theseresults suggested that agonist-induced, clathrin-mediated endocytosisof b2-AR is mechanically involved in the regulation of g-secretase.Then we investigated whether clathrin-mediated endocytosis per se issufficient to induce the enhancement of g-secretase activity. Treatmentof HEK293 cells with transferrin, which results in constitutive clathrin-dependent endocytosis of the transferrin receptor, did not enhanceg-secretase activity (Supplementary Fig. 4 online). Taken together,our data indicate that clathrin-mediated endocytosis is necessary butnot sufficient for b2-AR to enhance g-secretase activity.
The endocytic pathway consists of intracellular compartmentsthat are regulated by Rab guanosine triphosphatases. It has beenestablished that endocytic transports from the plasma membrane toearly endosomes and then to late endosomes and lysosomes (LEL) canbe blocked by Rab5 S34N and Rab7 T22N, respectively, which are thedominant-negative mutants of the early endosome marker Rab5 andthe LEL marker Rab7 (ref. 30). Expression of Rab5 S34N or Rab7T22N in HEK293 cells abolished b2-AR–stimulated enhancement ofg-secretase activity (Fig. 4a) and Ab production (Fig. 4b), indicatingthat LEL is critically involved in the regulation of g-secretase by b2-AR.Then, we immunoisolated vesicles of LEL from Flag-Rab7–transfectedcells with Flag-specific M2 beads, and subsequently blotted the isolateswith another early endosome marker, early endosome antigen 1(EEA1), or with the LEL marker, lysosome-associated membraneprotein-1 (LAMP-1)31. We found that the amount of Ab, but notFlag-Rab7 or LAMP-1, in LEL increased after 1 h of stimulation withb2-AR (Fig. 4c), indicating that Ab production in LEL was enhancedby b2-AR stimulation. Fractionation of subcellular vesicles alsoshowed that g-secretase activity in LEL was enhanced after DORstimulation (Supplementary Fig. 5 online). Taken together, theseresults suggest that the LEL is critical in the regulation of g-secretase
* * * * **
*
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
2.5
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
ConIso
ConIso
ConIso
ConIso
ConIso
β 2AR
β2AR β2AR LL β3AR
β 2AR
β 2AR L
L
β 3AR
β 2AR T
YY Con Iso
Choler
a to
xin
Forsk
olin
Dybut
yl-cA
MP
Moc
k
Conca
nava
lin A
Sucro
se
K+ d
plβ-g
al
Dyn K
44A
NSRNAi
ClathrinRNAi
Clathrin
Actin
Con
Iso
a b c d e
f gFigure 3 Receptor endocytosis is associated with enhanced g-secretase
activity. (a–e) Determination of g-secretase activity using a fluorogenic
substrate assay. (a) g-secretase activity in HEK293 cells expressing b2-AR
or b2-AR T68F,Y132G,Y219A (b2-AR TYY) in response to isoproterenol
treatment. (b) g-secretase activity in C6 glioma treated with cholera toxin
(1 mg/ml), forskolin (10 mM) or dybutyl cAMP (1 mM). (c) Pretreatment
of C6 glioma with concanavalin A (0.25 mg/ml), sucrose (0.5 M) or
potassium-depleted medium (K+ dpl) abolished isoproterenol-enhanced
g-secretase activity. (d) Ectopic expression of Dyn K44A abolished
enhanced g-secretase activity in C6 glioma. (e) Depletion of clathrin expression in HEK293 cells with clathrin RNAi abolished enhanced g-secretase activity.
The cytoplasm fractions were blotted with antibodies to clathrin and actin. (f,g) b2-AR L339,340A (b2-AR LL) and b3-AR did not endocytose (f) or enhance
g-secretase activity (g) in response to isoproterenol treatment. Scale bar, 8 mm. Data represent mean ± s.e.m. of at least three independent experiments,
presented as fold values of the basal activity. *P o 0.001. b-gal, b-galactosidase. NS, nonspecific.
ART ICL ES
1392 VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 NATURE MEDICINE
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
by b2-AR, and are consistent with previous reports of endocyticcompartments being the optimal environment for g-secretase32–34.
Then we used immunofluorescence microscopy to examine whetherb2-AR stimulation promoted the localization of g-secretase in LEL.For experiments in HEK293 cells, we marked LEL by means ofGFP-Rab7 expression and observed colocalization of PS1 with GFP-Rab7 30 min after b2-AR stimulation (Fig. 4d). For acute hippo-campal slices, we marked LEL with LAMP-1–specific antibody andobserved increased colocalization of PS1 or nicastrin (anotherg-secretase component) with LAMP-1 in response to b2-AR stimula-tion (Supplementary Fig. 6 online). Thus we concluded that b2-ARstimulation promotes the localization of g-secretase in LEL, an effectleading to an enhancement of g-secretase activity and Ab production.
Association between PS1/c-secretase and b2-ARWe also found that coexpression of Dyn K44A or Rab5 S34Neffectively prevented the elevated localization of PS1 in LEL followingb2-AR stimulation (Fig. 5a). This suggested that PS1 may be trans-ported to LEL from the plasma membrane. We then used b-adaptin tospecifically mark clathrin-coated pits and vesicles, and observed
colocalization of PS1 with b-adaptin and the endocytosed receptorafter 3-min stimulation of DOR in HEK293 cells (Fig. 5b). Given thatb2-AR or DOR can mediate endocytosis of other transmembraneproteins by forming heterodimers with them17, this implies coendo-cytosis of PS1 with the activated receptor (b2-AR or DOR). As it isalready known that PS1 constitutively interacts with membraneproteins such as APP and Notch35,36, we further examined theassociation between PS1 and b2-AR or DOR using a coimmunopre-cipitation assay. The four indispensable g-secretase components—PS1,nicastrin, anterior pharynx defective 1 homolog A (APH1A) andpresenilin enhancer 2 (PEN2)3—coprecipitated with b2-AR or DORin 3-[(3-cholamidopropyl)dimethylammonio]-2-hydroxy-1-propane-sulfonate (CHAPSO)-containing buffer (Fig. 5c), in which g-secretasestays as a complex37,38. The detergent Triton X-100 is known todissociate g-secretase complex37,38. Replacement of CHAPSO withTriton X-100 disrupted the coprecipitation of receptors with nicastrin,APH1A and PEN2, but not with PS1; this demonstrated that b2-ARand DOR associated with g-secretase by means of direct binding toPS1. We also observed the association of endogenous DOR and PS1 inrat hippocampus (Supplementary Fig. 7 online). However, another
0′
0′
3′
30′
30′30′β-gal Dyn K44A
Rab5 S34Nβ-gal
PS1/Rab7 PS1/DOR/β-adaptinFlag-β2ARFlag-DORFlag-B2R
IB: PS1-NTF
IB: nicastrin
IB: APH1A
IB: PEN2
IB: Flag
5% lyaste IP: Flag
1% CHAPSO 0.1% Triton–––
+––
–+–
––+
–––
+––
–+–
––+
–––
+––
–+–
––+
Flag-B2RFlag-DORFlag-β2AR
β2AR B2R
2.0
1.5
1.0
0.5
0γ-se
cret
ase
activ
ity(f
old
of b
asal
)
*
0 030 3015 45
Time (min)Iso BK
a b c d
Figure 5 Enrichment of g-secretase requires endocytic transport. (a,b) Immunofluorescence assay analysis of PS1 localization in transfected HEK293 cells.
(a) The effect of Dyn K44A or Rab5 S34N on the colocalization of PS1-NTF (red) and Flag-Rab7 (green) was analyzed in cells transfected with b2-AR and
the indicated plasmid (b-gal, Dyn K44A or Rab5 S34N). Scale bar, 8 mm. (b) Colocalization of PS1-NTF (red), HA-DOR (green) and b-adaptin (blue) were
detected after 3 min of DADLE treatment. Arrows, punctal structures containing PS1-NTF, HA-DOR and b-adaptin (insets). Scale bar, 8 mm. (c) Flag-b2-AR
and Flag-DOR, but not Flag-B2R, coprecipitated with endogenous g-secretase components (PS1, nicastrin, APH1A and PEN2) in CHAPSO-containing buffer.
(d) g-secretase activity in B2R-transfected HEK293 cells was determined by fluorogenic substrate assay after treatment with 1 mM bradykinin (BK).
*P o 0.001. HA, hemagglutinin.
* *
2.5
2.0
1.5
1.0
0.5
0
γ-se
cret
ase
activ
ity(f
old
of b
asal
)
ConIso
Rab 5/7
GFP
GFP
GFP-Rab
5 S34
N
GFP-Rab
7 T22
N
GFP-Rab
7
IsoC99
Aβ
– + – + – + – +
GFP
GFP-Rab
5 S34
N
GFP-Rab
7 T22
N
GFP-Rab
7
IsoFlag-Rab7
LAMP-1
EEA1
C99Aβ40Aβ42
– + – +5% lysate IP: Flag
Con
PS1 Rab7 PS1/Rab7
Iso
a b
c
d
Figure 4 Increased g-secretase and Ab in endocytic compartments. (a,b) HEK293 cells cotransfected with b2-AR and the indicated plasmids were treated
with isoproterenol and subjected to either a fluorogenic substrate assay (a), or to Ab-specific immunoprecipitation and western blot (b). The cytoplasm
fractions were blotted with antibody to GFP. *P o 0.001. (c) Immunoisolation of LEL showing that isoproterenol treatment increased Ab production in LELfrom HEK293 cells cotransfected with b2-AR, C99 and Flag-Rab7. (d) Immunofluorescence assay analyzing the colocalization of PS1 (red) and GFP-Rab7
(green) after 30 min of isoproterenol treatment. Arrow, punctal structure containing PS1 and GFP-Rab7. Scale bar, 8 mm. IP, immunoprecipitation.
ART ICL ES
NATURE MEDICINE VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 1393
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
member of GPCRs, B2 bradykinin receptor (B2R), did not associatewith PS1 or g-secretase (Fig. 5c), or induce g-secretase activity(Fig. 5d). These results thus suggest that the b2-AR-PS1 or DOR-PS1 association was specific and provided a mechanistic base for theregulation of g-secretase.
In vivo c-secretase activity, Ab production and plaques
Finally, we investigated the effects of b2-AR activation in rats andmice. First, acute injection of norepinephrine, an endogenous ligandfor adrenergic receptors, enhanced g-secretase activity (Fig. 6a) andAb levels (Fig. 6b) in rat hippocampus. Then we used the b2-AR–selective agonist clenbuterol, which can readily cross the blood-brainbarrier39, to confirm the norepinephrine effect and show the specificinvolvement of b2-AR in this in vivo effect. Also, it is to be expectedthat chronic agonist treatments could deteriorate Alzheimer disease–related pathology in animal models. Our experiments in APPswe/PS1DE9 double-transgenic mice supported this idea: these miceexhibited increased cerebral amyloid plaques after chronic adminis-tration of isoproterenol or clenbuterol for 30 d (Fig. 6c–e). Incontrast, APPswe/PS1DE9 mice exhibited reduced cerebral amyloidplaques after chronic treatment with the b2-AR–selective antagonistICI 118,551 (Fig. 6f,g).
DISCUSSION
In this study, we demonstrated that activation of either b2-AR or DORincreases Ab production in vitro and amyloid plaque formation in vivoby enhancing g-secretase activity. In other investigations of themolecular mechanism, it has been shown that agonist-induced endo-cytosis of b2-AR or DOR is mechanically indispensable for g-secretaseenhancement. Given that some proteins, such as GRK2 and b-arrestin,are recruited by b2-AR and DOR and are critically involved in agonist-induced endocytosis of these receptors18,21, we expected that thedepletion of GRK2 or b-arrestins would abolish this enhancement.We tested this experimentally and found that this was indeed the case(data not shown). However, g-secretase activity is not enhanced byconstitutive endocytosis of the transferrin receptor or by activation ofother types of GPCRs, such as B2R, V2 vasopressin receptor and AT1Aangiotensin II receptor (ref. 40 and data not shown). These resultssuggest that agonist-induced endocytosis of b2-AR or DOR per se isnot sufficient and that enhancement of g-secretase activity does notresult from a generic effect of accelerating cellular endocytic processes.Furthermore, we propose that association between PS1 and b2-AR or
DOR is another indispensable mechanistic requirement. This associa-tion is constitutive and independent of regulatory proteins such asb-arrestin, as it cannot be abolished by depletion of cellular b-arrestinexpression (data not shown). This association is also specific to b2-ARand DOR; therefore, some receptor specificity exists in the regulationof g-secretase. Taken together, we propose the following model for themechanism: b2-AR constitutively associates with g-secretase bydirectly binding to PS1 at the plasma membrane; once agonisttreatment induces endocytosis of the activated b2-AR, endocytosisof the bound PS1 will also be triggered and subsequently PS1 willtraffic to LEL, which provide an optimal environment for g-secretaseactivity32–34, resulting in the enhancement of g-secretase activity.
As is well known, Ab production requires b-secretase– andg-secretase–mediated cleavages of APP (refs. 3,24,41). g-secretaseattracts great interest, partially because about 150 mutations in PS1have been found to be related to Alzheimer disease pathogenesis. Thecurrent study reveals that g-secretase contributes, at least partially, tothe increase in Ab production induced by the activation of b2-AR orDOR through a mechanism of promoted localization of g-secretase inthe endosomal and lysosomal systems. These findings are consistentwith the existence of Ab, APP and g-secretase in the endosomal andlysosomal systems32,42, as well as with a previous report that aberrantactivation of the endocytic pathway precedes Ab accumulation insporadic Alzheimer disease43.
We found that there is an apparent sex difference in cerebralamyloid plaque formation of APPswe/PS1DE9 mice. In controlmice, females showed more amyloid plaques than did age-matchedmales, which has been previously reported in Tg2576 APPswe trans-genic mice44. Estrogen reduces Ab generation10, which suggests thatestrogen deficiency may induce the sex difference. Unexpectedly,female mice seemed to be more sensitive to chronic treatment withb2-AR agonist than male mice: the increase in the formation ofcerebral amyloid plaques in females was greater than that in males.As estrogen signaling can be inhibited by b-adrenergic stimulation45,this difference in sensitivity may be interpreted as an inhibition ofestrogen signaling by the b2-AR agonist, which then further contri-butes to Ab production and amyloid plaque formation in female mice.
In the past decade, genes related to the production of Ab (APP,PSEN1, PSEN2 and APOE) have been reported to associate with theneuropathology of Alzheimer disease5. Here we showed that activationof b2-AR, which can mediate the effect of environmental influencessuch as stress, enhances g-secretase activity and accelerates amyloid
2.0
1.5
1.0
0.5
0.0γ-se
cret
ase
activ
ity(f
old
of b
asal
) * * * ***
Sal
Sal
Iso
Sal
Cle
NE Sal Cle
2.0
1.5
1.0
0.5
0.0A
β le
vels
(fol
d of
bas
al)
Sal NE Sal Cle
Aβ40Aβ42
0
14,00012,00010,0008,0006,0004,000
Are
a of
am
yloi
dim
mun
orea
ctiv
ityA
rea
of a
myl
oid
imm
unor
eact
ivity
2,000
n =
4
n =
6n
= 6
n =
6n
= 6
n =
7
Sal Iso
Sal
Sal
ICIICI
Sal Cle
0
400
800
1,200
1,600
2,000
a c d e
f g
b*
*
*Figure 6 Enhanced g-secretase activity and Ab production, and accelerated amyloid
plaque formation in vivo. (a,b) Acute treatment of rats with norepinephrine (NE) or
clenbuterol (Cle) enhanced g-secretase activity (a) and increased Ab40 and Ab42 levels (b)
in hippocampus. *P o 0.01. (c–g) Cerebral amyloid plaque formation in APPswe/PS1DE9
mice chronically administered isoproterenol (c), clenbuterol (d) or ICI 118,551 (ICI) (f).
Images in c,d and f: representative plaques in female (left) and male (right) mice. Scale
bar, 600 mm. In e: quantitative analysis of amyloid plaque areas in mice from c and d
(*P o 0.05). In g: same for mice in f (*P o 0.05). Sal, saline.
ART ICL ES
1394 VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 NATURE MEDICINE
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
plaque formation. Stress has been recently appreciated to be a riskfactor for Alzheimer disease46,47 and can elevate the concentration ofthe endogenous b2-AR ligands epinephrine and norepinephrine15.Our data suggest that abnormal b2-AR activation, which could resultfrom the response to stress, might contribute to Ab accumulation inthe pathology of Alzheimer disease. In this regard, our findings mightprovide an underlying molecular mechanism for the pathological roleof stress in this disease.
Our study showed that the b2-AR–selective antagonist ICI 118,551can ameliorate amyloid plaque pathology of APPswe/PS1DE9 mice. Asb-AR antagonists are widely used in the treatment of cardiovasculardisease, our study proposes another potential therapeutic use for b-ARantagonists. In this respect, it is also interesting to note that a recentepidemiological study showed that the use of b-AR antagonists seemsto correlate with decreased incidence of Alzheimer disease pathogen-esis48, which is in full agreement with our results.
METHODSReagents. All reagents were from Sigma, unless otherwise indicated. Agonists
and antagonists of b2-AR and DOR used for in vitro treatments included
isoproterenol (10 mM), propranolol (10 mM), DADLE (1 mM) and naltrindole
(1 mM). The RNAi plasmid for human clathrin heavy chain targeted the
sequence of 5¢-GCTGGGAAAACTCTTCAGATT-3¢. The nonspecific (NS)
RNAi was 5¢-GGCCGCAAAGACCTTGTCCTTA-3¢.
Animals and drug treatments. All animal experiments were in accordance with
the National Institutes of Heath Guide for the Care and Use of Laboratory
Animals and were approved by the Biological Research Ethics Committee,
Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences.
In acute experiments, we injected Sprague-Dawle rats (from Shanghai
SLAC Laboratory Animal Company) intracerebroventricularly with 2 mg
norepinephrine. The stereotaxic coordinates were as follows: anterior-posterior,
�0.9 mm; left-right, �1.5 mm; dorsal-ventral, �3.8 mm). Rats were given
acute intraperitoneal injections of 0.5 mg clenbuterol per kg (body weight). In
30-d chronic experiments, APPswe/PS1DE9 mice (Jackson Laboratory) up
to 5 months of age were implanted with cannulae (anterior-posterior,
�0.6 mm; left-right, �1.2 mm; dorsal-ventral, �1.8 mm) for daily injections
of saline (n ¼ 4: two females and two males) and 3 nM isoproterenol (n ¼ 6:
four females and two males). APPswe/PS1DE9 mice were subjected to daily oral
administration of saline (n ¼ 6: three females and three males), 2 mg/kg per d
clenbuterol (n ¼ 7: four females and three males), and 1 mg/kg per d ICI
118,551 (n ¼ 6: three females and three males).
Hippocampal cultures and brain slice/section preparation. We prepared
primary hippocampal cultures from 1-d-postnatal Sprague-Dawley rats. Cells
were electroporated using Amaxa Nucleofector system and maintained in B27/
neurobasal medium (Invitrogen) for 14 d. We prepared acute hippocampal
slices from Sprague-Dawley rats using a vibratome. For cryosections, we
anesthetized and killed mice by transcardiac saline perfusion. The forebrains
were bisected midsagittally, fixed with 4% paraformaldehyde for 5 h, and then
cryosectioned coronally into 10-mm sections.
ELISA for Ab. We collected conditioned medium 6 h after 1-h treatment of
cells and examined Ab40 and Ab42 with sandwich ELISA kits (Biosource). Rat
hippocampus was homogenized and centrifuged at 100,000g for 1 h. We
detected rat Ab40 and Ab42 in supernatants with BNT77/BA27 and BNT77/
BC05 sandwich ELISA kits (Wako) according to previous reports49. All
measurements were performed in duplicate.
Immunoprecipitation. Cells were lysed and immuoprecipitated with Flag-
specific M2 beads. The immunoprecipitated complexes and cell lysates were
separated by SDS-PAGE and blotted with the indicated antibodies.
Expressed substrate assay. We performed the experiments as described
previously25. We stimulated HEK293 cotransfected with C99 and b2-AR. We
lysed cells and centrifuged aliquots (containing 50 mg of protein) at 13,000g for
15 min. We then resuspended and incubated pellets at 37 1C for 2 h in 50 ml of
assay buffer (pH 6.5) containing protease inhibitors. We detected C60 gener-
ated from the incubated membrane fractions by western blot using the
hemagglutinin (HA)-specific antibody 12CA5.
Fluorogenic substrate assay. We performed the assay as reported26. After
centrifugation of cell lysate or tissue homogenate aliquots at 13,000g for
15 min, pellets were resuspended and incubated at 37 1C for 2 h in 50 ml of
assay buffer (pH 6.5) containing 12 mM fluorogenic substrates (Calbiochem).
We measured fluorescence using SpectraMax M5 spectrometer (Molecular
Devices) with the excitation wavelength set at 355 nm and the emission
wavelength set at 440 nm.
Pulse-chase assay. We cotransfected HEK293 cells with DOR and C99. Cells
were starved for 2 h in methionine and serum-free medium (Invitrogen), pulse-
labeled with 500 mCi [35S]methionine (Amersham Pharmacia) in the absence
or presence of DADLE for 1 h, and then chased for 3 h in medium containing
excess amounts of unlabeled methionine. The C99 in cell lysates was immu-
noprecipitated with 12CA5 and detected by autoradiography.
Immunoisolation of LEL. We modified the experiments from the vesicle
isolation protocol50. After centrifugation of cell homogenates at 500g for
10 min, the resulting supernatants were incubated with Flag-specific M2 beads
for 8 h. We then subjected 5% of the lysates and isolates to western blot with
antibodies to LAMP-1, EEA1 (BD Biosciences) and Ab (Chemicon).
Immunofluorescence microscopy and quantification of amyloid plaques.
Cells were fed with the HA-specific antibody 12CA5 for 30 min, treated with
agonists, fixed, and then incubated with indicated antibodies. Mice brain
sections were incubated with 6E10 antibody and then with tetramethyl-
rhodamine isothiocyanate (TRITC)-conjugated secondary antibody. We
acquired images by using a confocal fluorescence microscope (Leica TCS
SP2). Area of amyloid plaques was quantified with Image-Pro Plus 5.1 software
(Media Cybernetic).
Statistical analysis. Data from in vitro experiments were analyzed using a
Student’s t-test for comparison of independent means, with pooled estimates of
common variances. We determined the statistical significance of in vivo data by
an analysis of variance (ANOVA) followed by Student’s t-test.
Note: Supplementary information is available on the Nature Medicine website.
ACKNOWLEDGMENTSWe thank M.M. Poo, D.S. Li, H. Zheng, Z. Zhang and L. Pu for comments onthe manuscript; Y.X. Zeng, G. Niu, P. Xia, Y.Y. Wang, W.B. Zhang and Y. Sun fortechnical support; R.J. Lefkowitz (Duke University Medical Center) for b2-ARTYY plasmid; S.L. Schmid (The Scripps Research Institute) for Dyn K44Aplasmid; S. Marullo (The Cochin Institute) for b3-AR plasmid; P. Wang(University of Minnesota School of Medicine) for various Rab5 and Rab7plasmids; and B. De Strooper (Katholieke Universiteit Leuven) and H.X. Xu(Burnham Institute for Medical Research) for wild-type and Psen1�/�Psen2�/�
mouse embryonic fibroblasts. This research was supported by grants from theMinistry of Science and Technology (2003CB515405 and 2005CB522406) and theNational Natural Science Foundation of China (30021003 and 30228018).
AUTHOR CONTRIBUTIONSThis study was designed by Y.N., X.Z. and G.P. The experiments were performedby Y.N. and X.Z. G.B. and L.T. contributed to the in vivo experiments. L.Z. andZ.W. contributed to the in vitro experiments. M.S., J.X. and Y.B. provided theAPPswe/PS1DE9 mice. G.P. supervised the project. Y.N., X.Z. and G.P.contributed to the writing of the paper.
COMPETING INTERESTS STATEMENTThe authors declare that they have no competing financial interests.
Published online at http://www.nature.com/naturemedicine
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Tanzi, R.E. The synaptic Aa hypothesis of Alzheimer disease. Nat. Neurosci. 8,977–979 (2005).
ART ICL ES
NATURE MEDICINE VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 1395
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine
2. Cleary, J.P. et al. Natural oligomers of the amyloid-b protein specifically disruptcognitive function. Nat. Neurosci. 8, 79–84 (2005).
3. Takasugi, N. et al. The role of presenilin cofactors in the g-secretase complex. Nature422, 438–441 (2003).
4. Sisodia, S.S. & St. George-Hyslop, P.H. g-secretase, Notch, Ab and Alzheimer’sdisease: where do the presenilins fit in? Nat. Rev. Neurosci. 3, 281–290 (2002).
5. Tanzi, R.E. & Bertram, L. Twenty years of the Alzheimer’s disease amyloid hypothesis: agenetic perspective. Cell 120, 545–555 (2005).
6. Koo, E.H. & Kopan, R. Potential role of presenilin-regulated signaling pathways insporadic neurodegeneration. Nat. Med. 10 (suppl.), S26–S33 (2004).
7. Cai, D. et al. Presenilin-1 uses phospholipase D1 as a negative regulator of b-amyloidformation. Proc. Natl. Acad. Sci. USA 103, 1941–1946 (2006).
8. Weggen, S. et al. A subset of NSAIDs lower amyloidogenic Ab42 independently ofcyclooxygenase activity. Nature 414, 212–216 (2001).
9. Nitsch, R.M., Slack, B.E., Wurtman, R.J. & Growdon, J.H. Release of Alzheimeramyloid precursor derivatives stimulated by activation of muscarinic acetylcholinereceptors. Science 258, 304–307 (1992).
10. Xu, H. et al. Estrogen reduces neuronal generation of Alzheimer b-amyloid peptides.Nat. Med. 4, 447–451 (1998).
11. Lazarov, O. et al. Environmental enrichment reduces Ab levels and amyloid depositionin transgenic mice. Cell 120, 701–713 (2005).
12. Saito, T. et al. Somatostatin regulates brain amyloid b peptide Ab42 through modula-tion of proteolytic degradation. Nat. Med. 11, 434–439 (2005).
13. Bierhaus, A. et al. A mechanism converting psychosocial stress into mononuclear cellactivation. Proc. Natl. Acad. Sci. USA 100, 1920–1925 (2003).
14. Yin, D., Tuthill, D., Mufson, R.A. & Shi, Y. Chronic restraint stress promoteslymphocyte apoptosis by modulating CD95 expression. J. Exp. Med. 191, 1423–1428 (2000).
15. Ferry, B., Roozendaal, B. & McGaugh, J.L. Role of norepinephrine in mediating stresshormone regulation of long-term memory storage: a critical involvement of theamygdala. Biol. Psychiatry 46, 1140–1152 (1999).
16. Hall, R.A. & Lefkowitz, R.J. Regulation of G protein-coupled receptor signaling byscaffold proteins. Circ. Res. 91, 672–680 (2002).
17. Jordan, B.A., Trapaidze, N., Gomes, I., Nivarthi, R. & Devi, L.A. Oligomerization ofopioid receptors with b2-adrenergic receptors: a role in trafficking and mitogen-activated protein kinase activation. Proc. Natl. Acad. Sci. USA 98, 343–348(2001).
18. Gonzalez-Gaitan, M. & Stenmark, H. Endocytosis and signaling: a relationship underdevelopment. Cell 115, 513–521 (2003).
19. Shenoy, S.K. et al. b-arrestin-dependent, G protein-independent ERK1/2 activation bythe b2 adrenergic receptor. J. Biol. Chem. 281, 1261–1273 (2006).
20. Gabilondo, A.M. et al. A dileucine motif in the C terminus of the b2-adrenergic receptoris involved in receptor internalization. Proc. Natl. Acad. Sci. USA 94, 12285–12290(1997).
21. Seachrist, J.L. & Ferguson, S.S.G. Regulation of G protein-coupled receptor endocy-tosis and trafficking by Rab GTPases. Life Sci. 74, 225–235 (2003).
22. Russo-Neustadt, A. & Cotman, C.W. Adrenergic receptors in Alzheimer’s diseasebrain: selective increases in the cerebella of aggressive patients. J. Neurosci. 17,5573–5580 (1997).
23. Wang, Y.P., Wang, Z.F., Zhang, Y.C., Tian, Q. & Wang, J.Z. Effect of amyloid peptides onserum withdrawal-induced cell differentiation and cell viability. Cell Res. 14, 467–472(2004).
24. Walter, J., Kaether, C., Steiner, H. & Haass, C. The cell biology of Alzheimer’s disease:uncovering the secrets of secretases. Curr. Opin. Neurobiol. 11, 585–590 (2001).
25. Pinnix, I. et al. A novel g-secretase assay based on detection of the putative C-terminalfragment-g of amyloid b protein precursor. J. Biol. Chem. 276, 481–487 (2001).
26. Farmery, M.R. et al. Partial purification and characterization of g-secretase from post-mortem human brain. J. Biol. Chem. 278, 24277–24284 (2003).
27. Law, P.-Y., Wong, Y.H. & Loh, H.H. Molecular mechanisms and regulation of opioidreceptor signaling. Annu. Rev. Pharmacol. Toxicol. 40, 389–430 (2000).
28. Gagnon, A.W., Kallal, L. & Benovic, J.L. Role of clathrin-mediated endocytosis inagonist-induced down-regulation of the b2-adrenergic receptor. J. Biol. Chem. 273,6976–6981 (1998).
29. Soeder, K.J. et al. The b3-adrenergic receptor activates mitogen-activated proteinkinase in adipocytes through a Gi-dependent mechanism. J. Biol. Chem. 274, 12017–12022 (1999).
30. Seachrist, J.L. & Ferguson, S.S. Regulation of G protein-coupled receptor endocytosisand trafficking by Rab GTPases. Life Sci. 74, 225–235 (2003).
31. Gruenberg, J. The endocytic pathway: a mosaic of domains. Nat. Rev. Mol. Cell Biol. 2,721–730 (2001).
32. Langui, D. et al. Subcellular topography of neuronal Ab peptide in APPxPS1 transgenicmice. Am. J. Pathol. 165, 1465–1477 (2004).
33. Takahashi, R.H. et al. Oligomerization of Alzheimer’s b-amyloid within processes andsynapses of cultured neurons and brain. J. Neurosci. 24, 3592–3599 (2004).
34. Vetrivel, K.S. et al. Association of g-secretase with lipid rafts in post-Golgi andendosome membranes. J. Biol. Chem. 279, 44945–44954 (2004).
35. Xia, W. et al. Presenilin complexes with the C-terminal fragments of amyloid precursorprotein at the sites of amyloid b-protein generation. Proc. Natl. Acad. Sci. USA 97,9299–9304 (2000).
36. Ray, W.J. et al. Evidence for a physical interaction between presenilin and Notch. Proc.Natl. Acad. Sci. USA 96, 3263–3268 (1999).
37. Li, Y.M. et al. Presenilin 1 is linked with g-secretase activity in the detergentsolubilized state. Proc. Natl. Acad. Sci. USA 97, 6138–6143 (2000).
38. Esler, W.P. et al. Activity-dependent isolation of the presenilin-g-secretase complexreveals nicastrin and a gamma substrate. Proc. Natl. Acad. Sci. USA 99, 2720–2725(2002).
39. Follesa, P. & Mocchetti, I. Regulation of basic fibroblast growth factor and nerve growthfactor mRNA by b-adrenergic receptor activation and adrenal steroids in rat centralnervous system. Mol. Pharmacol. 43, 132–138 (1993).
40. Guo, D.F., Sun, Y.L., Hamet, P. & Inagami, T. The angiotensin II type 1 receptor andreceptor-associated proteins. Cell Res. 11, 165–180 (2001).
41. Vassar, R. b-secretase cleavage of Alzheimer’s amyloid precursor protein by thetransmembrane aspartic protease BACE. Science 286, 735–741 (1999).
42. Pasternak, S.H. et al. Presenilin-1, nicastrin, amyloid precursor protein, andg-secretase activity are co-localized in the lysosomal membrane. J. Biol. Chem. 278,26687–26694 (2003).
43. Cataldo, A.M. et al. Endocytic pathway abnormalities precede amyloid b deposition insporadic Alzheimer’s disease and Down syndrome: differential effects of APOEgenotype and presenilin mutations. Am. J. Pathol. 157, 277–286 (2000).
44. Callahan, M.J. et al. Augmented senile plaque load in aged female b-amyloid precursorprotein-transgenic mice. Am. J. Pathol. 158, 1173–1177 (2001).
45. Walters, M.R. & Sharma, R. Cross-talk between b-adrenergic stimulation and estrogenreceptors: isoproterenol inhibits 17 b-estradiol-induced gene transcription in A7r5cells. J. Cardiovasc. Pharmacol. 42, 266–274 (2003).
46. Fratiglioni, L., Paillard-Borg, S. & Winblad, B. An active and socially integratedlifestyle in late life might protect against dementia. Lancet Neurol. 3, 343–353(2004).
47. Wilson, R.S. et al. Proneness to psychological distress is associated with risk ofAlzheimer’s disease. Neurology 61, 1479–1485 (2003).
48. Khachaturian, A.S. et al. Antihypertensive medication use and incident Alzheimerdisease: the Cache County study. Arch. Neurol. 63, 686–692 (2006).
49. Ramsden, M. et al. Androgens modulate b-amyloid levels in male rat brain.J. Neurochem. 87, 1052–1055 (2003).
50. Bahadoran, P. et al. Characterization of the molecular defects in Rab27a, caused byRAB27A missense mutations found in patients with Griscelli syndrome. J. Biol. Chem.278, 11386–11392 (2003).
ART ICL ES
1396 VOLUME 12 [ NUMBER 12 [ DECEMBER 2006 NATURE MEDICINE
©20
06 N
atur
e P
ublis
hing
Gro
up
http
://w
ww
.nat
ure.
com
/nat
urem
edic
ine