Cell Calcium 39 (2006) 539–550

Presenilin mutations linked to familial Alzheimer’s disease reduceendoplasmic reticulum and Golgi apparatus calcium levels�

Giancarlo Zatti a, Andrea Burgo a, Marta Giacomello a, Laura Barbiero b, Roberta Ghidoni b,Giulietta Sinigaglia a, Cristina Florean a, Silvia Bagnoli c, Giuliano Binetti b,

Sandro Sorbi c, Paola Pizzo a,∗∗, Cristina Fasolato a,∗a Department of Biomedical Sciences, University of Padua, Italy

b Neurobiology Lab-Memory Clinic, IRCCS “Centro San Giovanni di Dio-FBF”-AFaR, Brescia, Italyc Department of Neurological and Psychiatric Sciences, University of Florence, Italy

Received 25 January 2006; received in revised form 25 February 2006; accepted 3 March 2006

Abstract

Presenilin-1 and -2 (PS1 and PS2) mutations, the major cause of familial Alzheimer’s disease (FAD), have been causally implicated inthe pathogenesis of neuronal cell death through a perturbation of cellular Ca2+ homeostasis. We have recently shown that, at variance withprevious suggestions obtained in cells expressing other FAD-linked PS mutations, PS2-M239I and PS2-T122R cause a reduction and not anincrease in cytosolic Ca2+ rises induced by Ca2+ release from stores. In this contribution we have used different cell models: human fibroblastsfrom controls and FAD patients, cell lines (SH-SY5Y, HeLa, HEK293, MEFs) and rat primary neurons expressing a number of PS mutations,e.g. P117L, M146L, L286V, and A246E in PS1 and M239I, T122R, and N141I in PS2. The effects of FAD-linked PS mutations on cytosolicCa2+ changes have been monitored either by using fura-2 or recombinant cytosolic aequorin as the probe. Independently of the cell modelor the employed probe, the cytosolic Ca2+ increases, caused by agonist stimulation or full store depletion by drug treatment, were reducedor unchanged in cells expressing the PS mutations. Using aequorins, targeted to the endoplasmic reticulum or the Golgi apparatus, we hereshow that FAD-linked PS mutants lower the Ca2+ content of intracellular stores. The phenomenon was most prominent in cells expressingPS2 mutants, and was observed also in cells expressing the non-pathogenic, “loss-of-function” PS2-D366A mutation. Taken as a whole, ourfindings, while confirming the capability of presenilins to modify Ca2+ homeostasis, suggest a re-evaluation of the “Ca2+ overload” hypothesisin AD and a new working hypothesis is presented.© 2006 Elsevier Ltd. All rights reserved.

Keywords: Presenilin; Calcium stores; Alzheimer’s disease; Aequorin; Fura-2; Capacitative calcium entry

Abbreviations: AD, Alzheimer’s disease; FAD, familial form of Alzheimer’s disease; A�, amyloid �-peptide; APP, amyloid precursor protein; PS,presenilin; ER, endoplasmic reticulum; IP3, inositol 1,4,5-trisphosphate; Bk, bradykinin; Hist, histamine; Cch, carbachol; Tg, thapsigargin; SERCA, sar-coplasmic/endoplasmic reticulum Ca2+ ATPase; CCE, capacitative Ca2+ entry; CPA, cyclopiazonic acid; [Ca2+]i, intracellular Ca2+ concentration; [Ca2+]ER,endoplasmic reticulum Ca2+ concentration; [Ca2+]Go, Golgi apparatus Ca2+ concentration; mKRB, modified Krebs–Ringer buffer; Aeq, aequorin; cyt-Aeq,cytosolic aequorin; ER-Aeq, endoplasmic reticulum-targeted Aeq; Go-Aeq, Golgi apparatus-targeted Aeq; GFP, green fluorescent protein� The work was supported by the 2002 University of Padua Project (to P.P.); by the University of Padua grants (fellowship to A.B. and ex 60% 2003-05

to C.F. and P.P.) and the Italian Ministry of University FIRB 2001 (RBNE01ERXR). This wok was also supported in part by grant 2003/agreement numberPS/03/10, Italian Ministry of Health and grant 2003.1714/10.2034, CARIPLO Foundation, Milan, Italy, to G.B.; by San Paolo Company (Grant no. 4164IT/FA) and by Cassa di Risparmio of Florence (Grant no. 2003.1764) to S.S.

∗ Corresponding author at: Department of Biomedical Sciences, University of Padua, Via G. Colombo 3, 35121 Padova, Italy. Tel.: +39 049 8276065;fax: +39 049 8276049.∗∗ Corresponding author at: Department of Biomedical Sciences, University of Padua, Via G. Colombo 3, 35121 Padova, Italy. Tel.: +39 049 8276067;

fax: +39 049 8276049.E-mail addresses: paola.pizzo@unipd.it (P. Pizzo), cristina.fasolato@unipd.it (C. Fasolato).

0143-4160/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.ceca.2006.03.002

540 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

1. Introduction

Alzheimer’s disease (AD) is a progressive neurodegen-erative disorder leading to dementia. Deposition of the �-amyloid (A�) peptide in senile plaques and neurofibrillarydegeneration are common features to all forms of the dis-ease [1]. Mutations in three different genes are responsi-ble for the familial forms of the disorder (FAD) [2]. Thefirst gene encodes the �-amyloid precursor protein (APP),a type I membrane protein from which A� is derived byproteolytic cleavage. The other genes encode two homol-ogous proteins, presenilin-1 (PS1) and presenilin-2 (PS2),which are essential components of the APP cleavage com-plex [3]. All the FAD-linked mutations were shown to favourthe generation of longer, more amyloidogenic A� species.As integral membrane proteins, presenilins (PSs) are abun-dantly expressed with markers of the endoplasmic reticulum(ER), the Golgi complex and endo-exocytic vesicles; onlya minor fraction is present at the plasma membrane level[4,5]. Several lines of evidence suggest that PSs are alsoinvolved in the regulation of cellular Ca2+ homeostasis. Exag-gerated Ca2+ responses to agonists that stimulate release ofCa2+ from the ER were reported in PC12 cells expressingthe FAD-linked PS1-L286V [6]. Similarly, potentiation ofinositol-1,4,5-trisphosphate (IP3) evoked Ca2+ release wasreported in Xenopus oocytes expressing FAD-linked PS1 orPMtmalnCCbCPitgo[

mreefiorC2iMm

2. Materials and methods

2.1. Human primary skin fibroblasts

Patients carrying either the PS1-P117L or the PS1-M146Lmutation and control subjects were recruited at the Mem-ory Clinic of IRCCS “Centro San Giovanni di Dio-FBF”,Brescia and at the Department of Neurological and Psychi-atric Sciences, University of Florence, Italy and evaluatedas previously described [17]. Fibroblasts from patients andcontrols were obtained and cultured as previously described[14].

2.2. Ca2+ measurements

Both the fura-2 and the aequorin approaches were rou-tinely employed. Whenever possible, i.e. with easy trans-fectable cells, the latter tool was the method of choice becauseit allows monitoring Ca2+ signals arising from a larger num-ber of cells (103–104 times). When the two techniques wereemployed in HeLa cells, the obtained results were in agree-ment. Ca2+ changes when measured as peaks or areas abovebasal level were also in agreement. For presentation peak val-ues are shown and expressed as mean ± S.E.M. (n = numberof independent experiments; *p < 0.05, **p < 0.001, unpairedStudent’s t test). Fura-2 approach: Fibroblasts on coverslipswfsa2aaG2s(miCf(eiFlactm2icCra

S2 mutants, in particular, PS1-M146V, PS2-N141I and PS2-239V [7,8]. In oocytes expressing PS1-M146V, a lower IP3

hreshold level was also found by confocal analysis of ele-entary Ca2+ release events, an effect that was attributed to

n abnormal elevation of the ER Ca2+ content (“Ca2+ over-oad”) [9]. Enhanced Ca2+ responses were also triggered ineurons derived from transgenic mice [10]. The increaseda2+ release was often coupled to a reduced capacitativea2+ entry (CCE), the Ca2+ influx pathway that is activatedy emptied stores [11]. The same authors also suggested thatCE reduction was causally linked to A� production since inS1 expressing cell lines, prolonged treatment with the CCE

nhibitor SKF96365 induced an increase in the synthesis ofhe amyloidogenic A�42 peptide [11, but see also 12]. Alto-ether these findings allowed the breakthrough of the “Ca2+

verload” hypothesis as a central, pathogenetic event in AD10,13].

We have recently demonstrated that the FAD-linked PS2utations M239I and T122R reduce rather than increase Ca2+

elease in human fibroblasts and cell lines stably or transientlyxpressing the PS2 mutants [14,15]. Given the anti-apoptoticffect of a low store Ca2+ level [16], we here investigaterst whether a reduced Ca2+ release could be associated tother PS mutants, and secondly, whether it corresponds to aeduced store Ca2+ level. Cytosolic, ER and Golgi apparatusa2+ concentrations ([Ca2+]) were thus measured by fura-and different recombinant, organelle-targeted aequorins

n cell models expressing either FAD-linked PS1 (A246E,146L, P117L, L286V) or PS2 (M239I, T122R, N141I)utations.

ere incubated with fura-2/AM (2 �M) at room temperatureor 60 min in MEM containing 10% FCS, 0.04% pluronic andulfinpyrazone (250 �M). The coverslips were washed withmodified Krebs–Ringer buffer (mKRB, in mM:140 NaCl,.8 KCl, 2 MgCl2, 1 CaCl2, 10 HEPES, 11 glucose, pH 7.4)nd processed for imaging analysis with a 20× UV perme-ble objective (CellR, Olympus Biosystems GmbH, Planegg,ermany). Neurons on coverslips were incubated with fura-/AM (5 �M) at 37 ◦C and processed for imaging analy-is as described for fibroblasts. For presentation, the ratiosF340/F380) were off-line averaged (30–40 cells) and nor-alised to the resting value. Traces are representative of 5–10

ndependent experiments. Where indicated, in the mKRBaCl2 was omitted and EGTA (150 �M) was added (Ca2+-

ree medium), or NaCl was substituted by KCl (140 mM)K+-based medium). Aequorin approach: The cells on cov-rslips were incubated with coelenterazine (5 �M) for 1–2 hn mKRB and then transferred to the perfusion chamber.or reconstitution of ER- and Golgi (Go)-aequorin (Aeq),

uminal [Ca2+] was reduced before coelenterazine (5 �M)ddition by exposing the cells to ionomycin (5 �M) in mKRBontaining EGTA (600 �M). Upon 1 h incubation at 4 ◦C inhe same medium, the cells were extensively washed with

KRB supplemented with bovine serum albumin (BSA,%). All the luminescence measurements were carried outn mKRB at 37 ◦C. The experiments were terminated byell permeabilization with digitonin (100 �M) in a hypotonica2+-rich solution (10 mM CaCl2 in H2O) to discharge the

emaining unused Aeq pool. The light signal was collecteds previously described [18].

G. Zatti et al. / Cell Calcium 39 (2006) 539–550 541

2.3. Plasmids construction

The pcDNA3 constructs, as well as the pEF6/V5-His-TOPO ones, containing the cDNA encoding for different PS2(M239I, N141I, T122R, D366A) or PS1 (A246E, M146L,P117L, L286V) mutants were created by site directed muta-genesis of pcDNA3/PS2-wt or pEF6-V5-His-TOPO/PS1-wt, respectively (QuikChange Site-directed mutagenesis Kit,Stratagene, La Jolla, CA). The constructs were checkedby sequence analysis (ABI Prism Genetic Analyzer 310,Applied Biosystems, Monza, Italy).

2.4. Cell lines and transfection

HeLa and SH-SY5Y cells were grown in DMEM sup-plemented with 10% FCS containing penicillin (100 U/ml)and streptomycin (100 �g/ml). Before transfection, cellswere seeded on coverslips (13 mm diameter) and allowedto grow to 50% confluence. At this stage, transfections ofHeLa cells were carried out using the Ca2+-phosphate tech-nique in the presence of 4 �g of DNA [3 �g PS-cDNA orvoid vector (pcDNA3 or pEF6/V5-His-TOPO) plus 1 �gAeq cDNA]. SH-SY5Y cells were transfected by meansof LipofectamineTM 2000 using 1.5 �g of DNA (1 �g PS-cDNA or void vector plus 0.5 �g Aeq cDNA). IntracellularCa2+ measurements were carried out 48 or 24 h after trans-ftmooaPlkGDc1A

2

tb(ccg(andAT

cules, CA, USA) using the cDNA for EGFP (0.5 mg) togetherwith the cDNA for the PS mutants or the void vectors(pcDNA3 or pEF6/V5-His-TOPO) (1.0 mg). After 24 h, cov-erslips were incubated in the mKRB with fura-2/AM andanalysed by Ca2+ imaging.

2.6. Protein extracts preparation and Western blotanalysis

The different cell types were harvested and treated as pre-viously reported [15]. PS immunodetection was carried outwith the polyclonal antibody anti-human PS1, loop domain(263–407) (Calbiochem, Merck, Darmstadt, Germany) orthe monoclonal anti-V5 antibody (Invitrogen, Carlsbad, CA,USA) and with the polyclonal antibody anti-PS2 (324–335),Ab-2 (Calbiochem, Merck, Darmstadt, Germany). Detectionwas carried out by incubation with alkaline phosphatase-conjugated anti-rabbit or anti-mouse IgG (Vector Laborato-ries, Burlingame, CA, USA) for 1 h. The proteins were visu-alised by the chemiluminescence reagent CDP-star (Perkin-Elmer Life Sciences, Boston, MA, USA).

2.6.1. Chemicals and reagentsFura-2/AM, pluronic, antibiotics, sera, culture media,

plasmids and LipofectamineTM 2000 were purchased fromInvitrogen (Carlsbad, CA, USA). Fugene was from Roche,Mru

3

3c

msloeo[aCepatsNhWws

ection, for HeLa and SH-SY5Y, respectively, by means ofhe Aeq technique as previously described [18] and sum-

arised above. HEK293 and SH-SY5Y stable clones werebtained as previously described [14]. Selection was carriedut with G418 (1 mg/ml) or blasticidin (7 �g/ml) for HEK293nd SH-SY5Y clones, respectively. PS2-null (PS2−/−) orS1/PS2-null (PS1−/−, PS2−/−) mouse embryonic fibrob-

asts (MEFs), obtained as previously described [19,20], wereindly provided by Dr. Bart De Strooper (Center for Humanenetics, KUL, VIB, Leuven, Belgium). Cells, grown inMEM supplemented with 10% FCS and 100 U/ml peni-

illin/streptomycin, were transfected by Fugene employing.5 �g of DNA (1 �g PS-cDNA or void vector plus 0.5 �geq cDNA).

.5. Primary neuronal cultures and transfection

Cells from rat neonatal cortices were dissociated inrypsin (0.8 mg/ml) for 10 min at 37 ◦C and digestion waslocked by the trypsin inhibitor (6.3 �g/ml) plus DNAse I40 �g/ml). Dissociated cells were plated on poly-l-lysine-oated (100 �g/ml) glass cover-slips at a density of 5 × 105

ells. The growth medium consisted of MEM containing:lucose (20 mM), l-glutamine (0.5 mM), N2 supplement1%), B27 supplement (0.5%) biotin (0.875 mg/L), pyruviccid (1 mM), penicillin (25 �g/ml) streptomycin, (25 �g/ml),eomycin (50 �g/ml) and horse serum (10%). Cytosine-�--arabinofuranoside (3 �M) was added 24 h after plating.fter 4–5 DIV, cells were transfected by means of theransfectinTM Lipid Reagent (Bio-Rad Laboratories, Her-

olecular Biochemicals (Indianapolis, IN, USA). All othereagents were from Sigma Chemical Co. (St. Louis, Mo)nless otherwise stated.

. Results

.1. Effects of FAD-linked PS mutants on cytosolic Ca2+

hanges

We have previously shown that the FAD-linked PS2utants, M239I and T122R, expressed either stably or tran-

iently in different cell models, altered Ca2+ dynamics byowering Ca2+ release from intracellular stores [14,15]. Thebserved phenomenon was substantially different from whatxpected on the basis of the “Ca2+ overload” hypothesis,riginally suggested to explain Ca2+ dys-regulation in FAD13]. This discrepancy forced us to investigate other PS2nd PS1 mutants to further characterize their effects ona2+ signaling. The FAD-linked PS mutant set was thusxpanded to include mutations which were, or were not,reviously characterized in terms of Ca2+ homeostasis, i.e.n additional PS2 mutation (N141I) and four PS1 muta-ions (P117L, M146L, L286V, A246E) [21–24]. Fig. 1Ahows representative traces describing the effect of PS2-141I on two common parameters linked to cellular Ca2+

andling, i.e. Ca2+ release from intracellular stores and CCE.e used the same technical approach previously employedith the other PS2 mutants [14,15]. HeLa cells were tran-

iently co-transfected with the cDNAs encoding cyt-Aeq

542 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

Fig. 1. Effects of FAD-linked PS mutants on cytosolic Ca2+ changes in HeLa cells and human fibroblasts. (A–C) HeLa cells were transiently co-transfectedwith the cDNAs coding for cyt-Aeq and one mutant form of PS2 (M239I, T122R or N141I) (A–B) or PS1 (M146L, L286V, A246E or P117L) (C). Cells,transfected with the void vectors (pcDNA3 and pEF6/V5-His-TOPO, for PS2 and PS1 constructs, respectively) were used as controls. Forty-eight hoursupon transfection, Ca2+ dynamics were analyzed following cyt-Aeq reconstitution. (A) Representative traces of control (dashed) and PS2-N141I expressing(continuous) cells, bathed in Ca2+-free, EGTA-containing medium and challenged with CPA (20 �M) plus histamine (Hist, 100 �M). Upon store depletion,CaCl2 (1 mM) was added in the continuous presence of CPA. (B–C) Bars represent mean [Ca2+]i peak values above basal level, reached, respectively afterCPA/Hist and CaCl2 additions, in cells expressing PS2 (B) or PS1 (C) mutants. (D) Human FAD-fibroblasts carrying PS1-M146L (upper panel, continuoustrace) or PS1-P117L (lower panel, continuous trace) and age-matched control fibroblasts (dashed traces) were loaded with fura-2. [Ca2+]i changes are expressedas the ratio (F340/F380), normalized to the average value obtained within the first minute of the experiment. Traces are averages of 20–30 fura-2 loaded cells.n: number of experiments. See Table 1 for statistics.

and one mutant form of PS2 (M239I, T122R or N141I) orPS1 (M146L, P117L, L286V or A246E). Cells, transfectedwith the void vectors (pcDNA3 and pEF6/V5-His-TOPO, forPS2 and PS1 constructs, respectively) were used as suitablecontrols. Forty-eight hours upon transfection, Ca2+ dynam-ics were analyzed following cyt-Aeq reconstitution. Underthese experimental conditions, all PS mutants were efficientlyexpressed as verified by Western blot (data not shown). Com-plete discharge from intracellular Ca2+ stores was induced ina Ca2+-free EGTA-mKRB (see Section 2) by simultaneousperfusion with CPA (20 �M), a SERCA inhibitor, and his-tamine (Hist, 100 �M), an IP3-generating agonist. Upon storedepletion, the amplitude of CCE was estimated by perfusionwith mKRB containing CaCl2 (1 mM). A summary of theeffects on Ca2+ release and CCE is shown in Fig. 1B and C forPS2 and PS1, respectively. All PS2 expressing cells showed a

significant reduction of both intracellular Ca2+ release (rang-ing from 30 to 45%), and CCE (ranging from 20 to 35%). Asfar as PS1 expressing cells was concerned, a decrease, albeitof lower entity, was found in both Ca2+ release and CCE withall the tested PS1 mutants, except with PS1-L286V whoseeffect on Ca2+ release, but not CCE, was missing (Fig. 1C).

The finding that, in HeLa cells transiently expressingFAD-linked PS1 mutations, Ca2+ release from intracellu-lar stores was unchanged or even decreased, rather thanincreased, was surprising. To further investigate this cru-cial issue, we analysed fibroblasts from patients carryingthe FAD-linked PS1 mutation M146L or P117L [22–24].Ca2+ dynamics were studied by means of fura-2, i.e. thesame approach originally employed with human fibroblastscarrying the PS2 mutation M239I or T122R [14,15]. WhenCPA was used as a stimulus, a modest reduction, and not

G. Zatti et al. / Cell Calcium 39 (2006) 539–550 543

Table 1Clinical and demographic characteristics of the fibroblast donors

Subjects Age (years) Gender MMSE Number of exp. Ca2+ release (mean ± S.E.M.) CCE (mean ± S.E.M.)

PS1 M146LI 40 F 12 9 0.66 ± 0.05 (*) 0.15 ± 0.01 (*)II 53 M 16 12

Healthy controlsHC-1 38 F 28 12 0.82 ± 0.02 0.20 ± 0.01HC-2 43 F 30 14

PS1 P117LI 30 M 19 30 0.41 ± 0.02 (*) 0.14 ± 0.01

Healthy controlsHC-1 26 M 30 6 051 ± 0.01 0.15 ± 0.02HC-2 37 M 30 9HC-3 40 F 28 7HC-4 34 F 29 12HC-5 33 M 30 4

Ca2+ release and CCE mean values were measured as described in Fig. 1D.

an increase, in Ca2+ release was again found in fibroblastsexpressing either PS1-M146L or PS1-P117L, compared tocontrol fibroblasts obtained from healthy, age-matched, indi-viduals (see Fig. 1D and Table 1). On average, Ca2+ releasewas reduced by 20% whereas a significant reduction of CCEwas found only in fibroblasts expressing PS1-M146L. Sim-ilar results were obtained when the IP3-generating agonistbradykinin (Bk, 100 nM) was used as an appropriate stimu-lus (not shown).

Up until now our analysis was restricted to non-neuronalcell models where PS mutants were expressed either tran-siently (HeLa) or stably (human fibroblasts and HEK293clones, see also [14]). The observation that FAD-linked PSmutants reduce Ca2+ release from intracellular stores washere extended to the human derived neuroblastoma cell lineSH-SY5Y [25]. As shown in Fig. 2A, when PS2-T122R, themost effective PS2 mutant in non-excitable cells, was tran-siently expressed with cyt-Aeq in these cells, a reduction inCa2+ release induced by CPA (20 �M) plus Bk (100 nM)was obtained. Again, PS2-T122R was more potent thanPS1-A246E or PS1-M146L: the mean Ca2+ peak value wasreduced on average by 50% with the PS2 mutant, howeverby only 20% with PS1-A246E, and no effect was found withPS1-M146L (Fig. 2B). Similar data were obtained by induc-ing store depletion with CPA alone (not shown). Note thatttFmsec

3

f

Fig. 2. Effects of FAD-linked PS mutants on cytosolic Ca2+ changes inneuroblastoma cells. SH-SY5Y cells were transiently co-transfected with thecDNAs coding for cyt-Aeq and one PS mutant as described in Fig. 1. Twenty-four hours after transfection, Ca2+ dynamics were analyzed following cyt-Aeq reconstitution. (A) Representative traces of control (dashed) and PS2-T122R expressing (continuous) cells, bathed in Ca2+-free EGTA-containingmedium and challenged with CPA (20 �M) plus bradykinin (Bk, 100 nM).Upon store depletion, CaCl2 (2 mM) was added in the continuous presence ofCPA. B, Bars represent mean [Ca2+]i peak values above basal level, reached,respectively after CPA/Bk and CaCl2 additions.

hese two PS1 mutations were purposely chosen because ofheir significant effect in HeLa cells (A246E, M146L) andAD-fibroblasts (M146L). Interestingly, in SH-SY5Y, all PSutants exerted no major effect on CCE, both in a standard

olution (Fig. 2B) and in a K+-based medium (see Section 2),mployed to exclude alterations due to membrane potentialhanges (not shown).

.2. Effect of PS mutants in stable clones

A reduced Ca2+ release by PS2-M239I was previouslyound not only in over-expressing HeLa cells but also in

544 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

Fig. 3. Effect of PS mutants in stable clones. HEK293/PS1-M146L, SH-SY5Y/PS2-T122R and the corresponding void-vector clones were transiently transfectedwith cyt-Aeq as described in Section 2. (A) Clones were tested for PS expression with the monoclonal anti-V5 antibody for PS1 and the polyclonal antibodyanti-PS2 (324-335) for PS2. For PS1, the full length protein is shown because of the poor reactivity of the V5 antibody with the C-terminal fragment. (B–C) Barsrepresent mean [Ca2+]i peak values above basal level obtained for each clone (left) or their average (right) upon addition of CPA (20 �M) plus Cch (100 �M) inHEK293 clones or plus Bk (100 nM) in SH-SY5Y clones, following a protocol similar to that described in Fig. 1. (D, E) Representative traces (D) and statisticsof Ca2+ release (E) induced by ATP (100 �M) and CPA (20 �M) in MEFs PS1−/−PS2−/− and MEFs PS1+/+PS2−/− transfected with the cDNAs coding forcyt-Aeq. (F) Statistics of Ca2+ release induced by ATP (100 �M), CPA (20 �M) or both stimuli in MEFs PS1−/−PS2−/− transfected with the cDNAs codingfor cyt-Aeq and PS2-T122R or the void-vector (vv).

G. Zatti et al. / Cell Calcium 39 (2006) 539–550 545

FAD-fibroblasts and in stable HEK293 clones, where theexpression level of the mutant protein appeared to be similarto that of the endogenous protein [14]. We thus investi-gated other mutations in stable clones. Fig. 3 shows Ca2+

release measured by cyt-Aeq in HEK293 and SH-SY5Yclones expressing, respectively, PS1-M146L and PS2-T122R(see Fig. 3A for PS immunodetection). Five HEK293/PS1-M146L clones were tested and no significant effect wasfound in Ca2+ release, induced by CPA plus carbachol (Cch,100 �M) addition, if compared to the values obtained in threevoid-vector transfected clones (Fig. 3B). In contrast, uponstimulation with CPA and Bk (100 nM), a reduced Ca2+

release was found in three SH-SY5Y/PS2-T122R cloneswhen compared to three clones obtained following void-vector transfection (Fig. 3C). As previously found in neurob-lastoma cells transiently expressing PS2-T122R, the mutantprotein did not reduced CCE also in the stable clones (data notshown).

The role of PS was also analyzed in MEFs, knockedout (KO) for both PSs (PS1−/−PS2−/−), or PS2 alone(PS1+/+PS2−/−). Ca2+ release induced by ATP (100 �M),with or without CPA (20 �M), was effectively higher incells expressing only PS1 (MEFs PS1+/+PS2−/−) comparedto the double KO cells (Fig. 3D and E). However, it wasunchanged when store depletion was accomplished by CPA(20 �M) alone (Fig. 3E). Noteworthy, in the absence of aPtr(

3E

gobbicwEoEta

afoBtsii

PS2-T122R expressing HeLa cells and in control, void-vectortransfected cells. It can be noted that the steady-state Ca2+

level was markedly different in the two cases. In particu-lar, expression of PS2-T122R caused a 20% reduction in the[Ca2+]ER. Fig. 4B and C shows the statistics of similar exper-iments, carried out in HeLa and SH-SY5Y cells, with PS2and PS1 mutants, respectively. Noteworthy, whereas expres-sion of different PS2 forms was associated to a significantdecrease of [Ca2+]ER, expression of PS1 mutants causedonly a minor reduction which became significant only withPS1-A246E and -P117L in HeLa but not in SH-SY5Y cells.Similar results were also obtained when ER Ca2+ levels weremeasured in permeabilized HeLa and SH-SY5Y cells (datanot shown).

3.4. FAD-linked PS2 and PS1 mutants strongly affect theCa2+ level of the Golgi apparatus

PSs are integral membrane proteins that are mainly locatedin internal membranes, i.e. those of the ER, the Golgi appa-ratus, and the vesicles of the lysosomal and endo-exocyticnetwork. Although the ER is considered the major agonist-sensitive intracellular Ca2+ store, it recently became evidentthat also the Golgi apparatus is an important Ca2+ store[27,28]. We therefore extended our analysis to this organelleby employing a previously established recombinant form ofAtntflufA[cami[rncte

3o

aWPtW

S1 background (MEFs PS1−/−PS2−/−), PS2-T122R, uponransient expression, was still capable of reducing the Ca2+

elease induced by both ATP, ATP plus CPA or CPA aloneFig. 3F).

.3. Effects of FAD-linked PS2 and PS1 mutations onR Ca2+ handling

Cytosolic Ca2+ changes caused by CPA and IP3-enerating agonists do not simply reflect the Ca2+ contentf intracellular stores. In fact, they are strongly affected alsoy other factors, such as Ca2+ buffers and plasma mem-rane extrusion systems, which might also be, directly orndirectly, PS targets. Therefore, we investigated the mainellular organelle involved in Ca2+ release, the ER, to seehether PS mutants affect Ca2+ handling by this organelle.R Ca2+ concentration ([Ca2+]ER) was measured by meansf a recombinant Aeq (ER-Aeq) with lower Ca2+ affinity andR selective retention [26]. HeLa and SH-SY5Y cells were

hus transiently co-transfected with the cDNAs for ER-Aeqnd one of the different PS forms.

Because of the high Ca2+ concentration inside the ERnd the intrinsic properties of the probe, which is consumedollowing Ca2+ binding, ER-Aeq reconstitution was carriedut in store-depleted cells (see Section 2 and ref. [26]).riefly, the cells were incubated with the co-factor coelen-

erazine in the absence of external Ca2+, following expo-ure to the Ca2+ ionophore ionomycin. Upon washing theonophore, the steady-state [Ca2+]ER was monitored follow-ng CaCl2 addition. Fig. 4A shows the refilling process in

eq (Go-Aeq) that, by comprising the trans-membrane por-ion of sialyltransferase, is selectively retained in the Golgietwork [27]. HeLa and SH-SY5Y cells were transientlyransfected with the cDNAs coding for Go-Aeq and the dif-erent PS mutants. Since the Golgi apparatus has an elevateduminal Ca2+ concentration ([Ca2+]Go), due to active Ca2+

ptake mechanisms, the procedure for Go-Aeq reconstitutionollowed the same protocol described for cells expressing ER-eq (see Section 2). As shown in Fig. 5A, the steady-state

Ca2+]Go was obtained upon cell perfusion with a Ca2+-ontaining medium. Fig. 5B and C shows the statistics for theverage [Ca2+]Go, measured in cells expressing PS2 and PS1utants, respectively. Interestingly, the percentage reduction

n this parameter was even larger than that measured for theCa2+]ER. In particular, in HeLa cells the maximal reductioneached 50% with PS2-N141I and 30% with PS1-A246E. Ofote, PS1-A246E in SH-SY5Y cells and PS1-M146L in HeLaells significantly, albeit modestly, reduced the Golgi but nothe ER steady-state Ca2+ level. Both mutations were howeverffective on cytosolic Ca2+ release (see Figs. 1C and 2B).

.5. Ca2+ dysregulation by PS2 mutants is independentf γ-secretase activity

We previously showed that reduction in Ca2+ release waslso mimicked by over-expression of wild-type PS2 [14,15].

hen focussing on ER and Golgi Ca2+ levels, both wild-typeS2 and PS1 behaved as their corresponding mutant pro-

eins upon over-expression in HeLa cells (data not shown).hereas these findings are still consistent with a “gain-of-

546 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

Fig. 4. [Ca2+]ER measurements in HeLa and SH-SY5Y cells expressing FAD-linked PS2 or PS1 mutants. Cells were co-transfected with the cDNAs coding forER-Aeq and one PS mutant form or the void-vector. After Aeq reconstitution in a Ca2+-free, EGTA containing medium in the presence of ionomycin (5 �M),control (dashed trace) and PS2-T122R expressing (continuous trace) cells were washed and bathed in the same medium without ionomycin. CaCl2 (1 mM) wasadded to follow the ER refilling process. (B–C) Bars represent mean [Ca2+]ER steady-state values obtained upon expression of PS2 (B) or PS1 (C) mutants.

function”, they are not informative on their dependence on�-secretase activity. To test this issue, we analyzed HeLacells transiently co-transfected with the cDNAs encodingcyt-Aeq and PS2 carrying the “loss-of-function” mutationD366A [29]. The cells were first challenged with CPA/Histin a Ca2+-free, EGTA-containing medium, and subsequentlybathed in a Ca2+ containing medium, as described in Fig. 1A.The results were compared with those obtained in control,void-vector transfected cells. As shown in Fig. 6A, in PS2-D366A expressing cells, a marked reduction in both intracel-lular Ca2+ release and CCE was recorded. The entity of suchreduction was comparable with that measured in HeLa cellsexpressing PS2-T122R, i.e. 45% and 40% reduction in Ca2+

release and CCE, respectively.

3.6. Effects of PS2-T122R and PS1-A246E mutants inrat primary neuronal cultures

PS mutants were also transiently co-expressed with greenfluorescent protein (GFP) in rat cortical neurons at 4–5 DIV.Ca2+ release was studied at 24 h upon transfection by meansof the fura-2 approach. It is well known that neuronal Ca2+

stores have a low Ca2+ content under resting conditions [30];in fact, as shown in Fig. 6B, the response to CCh (50 �M)was practically undetectable in standard mKRB, whereas itwas substantially increased following a three minutes expo-sure to a depolarizing medium, containing KCl (30 mM).Among GFP-positive cells, those expressing PS2-T122R, butnot PS1-A246E, showed a potent reduction of Ca2+ release

G. Zatti et al. / Cell Calcium 39 (2006) 539–550 547

Fig. 5. [Ca2+]Go measurements in HeLa and SH-SY5Y cells expressing FAD-linked PS2 or PS1 mutants. Cells were co-transfected with the cDNAs codingfor Go-Aeq and one PS mutant form or the void-vector. (A) After Aeq reconstitution as described in Fig. 4, control (dashed trace) and PS2-T122R expressing(continuous trace) HeLa cells were bathed in a Ca2+-free, EGTA containing medium. CaCl2 (1 mM) was added to follow the Golgi apparatus refilling process.(B–C) Bars represent the mean [Ca2+]Go steady-state values obtained upon expression of PS2 (B) or PS1 (C) mutants.

from intracellular stores, when compared to GFP-positive,void-vector transfected cells (Fig. 6C and D).

4. Discussion

It is commonly accepted that AD is linked to Ca2+ dys-regulation. Both a Ca2+ overload of intracellular stores anda reduced CCE were suggested to play a relevant role in thepathogenesis of the familial form of the disease [13]. Themajority of the data deals with various mutations in PS1(M146V, M146L, L286V, A246E, L250S, �E9) but onlyvery few in PS2 (N141I, M239V). The employed approacheswere heterogeneous, ranging from transient expression of themutant protein in Xenopus oocytes to stable expression inclones from neuronal (SH-SY5Y, PC12) and non-neuronalcell lines (CHO, HEK293), or in neurons and fibroblasts fromtransgenic knock-in mice. Ca2+-sensitive dyes have beenemployed to detect cytosolic Ca2+ changes, mainly induced

by IP3-generating agonists, IP3 itself (i.e. caged-IP3) or, lessfrequently, by SERCA inhibitors.

We have recently shown that two previously uncharacter-ized FAD-linked PS2 mutations (M239I and T122R) reduceCa2+ release from intracellular stores when present both athigh levels, such as in over-expressing HeLa or HEK293cells, and at endogenous levels, such as in human FAD fibrob-lasts or stable HEK293 clones [14,15]. The oddness of thisfinding, when compared to the “Ca2+ overload” hypothesisfor AD, convinced us firstly to extend our investigation toadditional PS mutants, including other FAD variants (PS2-N141I, PS1-A246E, PS1-L286V, PS1-M146L, PS1-P117L),and secondly, to proof the “calcium overload” hypothesis bymeasuring calcium levels directly inside the stores in differ-ent model cells expressing pathogenic presenilin mutants.

Concerning the first objective, two PS1 mutations, P117Land M146L, were for the first time investigated also infibroblasts from FAD patients. We have here demonstratedthat, in these cells, Ca2+ release was not augmented com-

548 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

Fig. 6. The effect of PS2 mutant is independent of �-secretase activity and is observable also in rat primary neurons. (A) HeLa cells were co-transfected withthe cDNAs coding for cyt-Aeq and the “loss-of-function” PS2-D366A or the void-vector. Ca2+ release and CCE were estimated with the protocol shown inFig. 1A. (B–D) Neurons loaded with fura-2 were bathed in mKRB. (B) Cytosolic Ca2+ changes from control, non-transfected, neurons challenged with Cch(50 �M) before and after exposure to KCl (30 mM). The trace represents the normalized ratio (F340/F380) averaged from six neurons of a typical experiment.(C) Cytosolic Ca2+ changes from GFP positive neurons, transfected with the void-vector (continuous trace, n = 3) or the cDNA coding for PS2-T122R (dashedtrace, n = 3) or PS1-A246E (not shown) as described in Section 2. The cells were challenged with Cch (50 �M) after 6 min exposure to KCl (30 mM). (D) Barsrepresent mean [Ca2+]i changes above basal level measured upon Cch addition and expressed as � ratio (F340/F380), normalized to the average value obtainedwithin the first minute of the experiment (n = number of GFP positive neurons).

pared to control, age-matched fibroblasts. In particular, bothtypes of FAD-fibroblasts suffered a significant, albeit modest,decrease of Ca2+ release. Noteworthy, in FAD fibroblasts asimilar reduction in Ca2+ release could be observed also byother investigators only when controls were carefully selectedfrom age-matched subjects [31].

An overview of the effects of the various PS mutants oncytosolic Ca2+ changes was obtained by transient expressionof the proteins in HeLa cells in association with cyt-Aeq.Whereas the experiments with human fibroblasts were carriedout by means of the fura-2 approach, the standard techniquegenerally employed by other groups, the Aeq approach washere chosen to compare Ca2+ responses arising from both thecytosol and the organelles, by taking advantage of the selec-tive targeting of recombinant aequorins. The effect on Ca2+

release was studied by measuring cytosolic Ca2+ changesupon addition of CPA and an IP3-coupled agonist in a Ca2+-free medium, a protocol that induces a complete and quickdischarge of Ca2+ pools. Under these conditions, PS2 mutants(M239I, T122R, N141I) significantly reduced Ca2+ release.A modest reduction was also observed with PS1 mutants, witha lack of significance only for PS1-L286V. Notably, a reducedCa2+ release was observed also in human SH-SY5Y neu-roblastoma cells when co-transfected with cyt-Aeq and PSmutants. In these cells, PS2-T122R was again more effectivethan PS1-A246E, whereas PS1-M146L was practically idle.

As far as CCE was concerned, we noticed that PS mutantsdifferently affected the store-operated pathway in the twomodel cells. In fact, whereas in HeLa cells a reduction ofvariable entity was always observed, especially with PS2

G. Zatti et al. / Cell Calcium 39 (2006) 539–550 549

mutants, in SH-SY5Y cells no effect was reported with eitherPS2 or PS1 mutants. The reason for that discrepancy is notclear and it might be due to different store-operated pathways.While this issue was not further investigated, these findingsstrengthen the idea that the primary target of PS mutants is theCa2+ handling capability of intracellular stores. At this stage,we cannot anticipate how the variable effects of PS mutants atthe CCE level might influence the onset and the progressionof the disease. Notwithstanding, we have recently shown thatno causal relationship could be found between CCE reductionand A�42 production, thus further supporting a modulatoryrather than a causative role for Ca2+ in AD [12].

By employing recombinant aequorins targeted to both ERand Golgi networks, we have here proved, for the first time,that PS mutants exert their effect directly on intracellular Ca2+

stores. In fact, they reduced the ER and, more pronouncedly,also the Golgi Ca2+ concentration, with a rank order potencysimilar to that observed in cytosolic Ca2+ changes uponinduction of store depletion. The reduction was detectableboth in intact and permeabilized HeLa and SH-SY5Y cells,further indicating its independence on CCE.

Considering the mutations in PS1, we concluded that Ca2+

release (and accordingly the Ca2+ level within the store),instead of being increased, showed a tendency to a decrease(A246E, P117L, M146L) or was unchanged (L286V). Weargued that this discrepancy could be ascribed to neither thetffcaciaecwpnps

rmgePrtataiPPa

In summary, we found that four mutations in PS2 (M239I,T122R, N141I, D366A) significantly reduce the Ca2+ contentof intracellular stores, whereas four in PS1 (P117L, M146L,L286V, A246E) exert modest or null effects. When consider-ing the anti-apoptotic effect of a low ER Ca2+ level, and thenotion that FAD-linked PS2 mutations are associated to lateronsets and milder phenotypes, we suggest that mutations inPS2 might be endowed of a lower toxicity, if compared tothose in PS1. In fact, the differential effects on Ca2+ home-ostasis can influence the neuronal (and glial) response todeath stimuli, such as exposure to A� and reactive-oxygenspecies, and thus favouring or delaying disease onset andprogression.

Acknowledgments

We thank Dr. B. Nacmias for providing FAD/PS1-M146Lfibroblasts, Dr. B. De Strooper for the MEFs cells and G.Ronconi and M. Santato for technical assistance. We aregrateful to Drs. T. Pozzan and P. Magalhaes for usefulsuggestions.

References

[

echniques themselves nor the model systems employed. Inact, reductions in Ca2+ release were detected by using eitherura-2 or recombinant aequorin, not only in over-expressingell lines (HeLa, SH-SY5Y, MEFs) and neuronal cultures, butlso in human FAD-fibroblasts and in stable clones (HEK293)hallenged with either the agonist alone or plus the SERCAnhibitor. Noteworthy, in the presence of an IP3-generatinggonist, an increased Ca2+ release was observed in MEF cellsxpressing the only PS1 when compared to the double KOells. Such a difference was however abolished when CPAas used as discharging agent. This finding is consistent withrevious data suggesting that an exaggerated Ca2+ release isot a property of the store but might be a consequence of theositive modulation exerted by PS1 on the IP3-generatingystem [8,32].

Our data suggest that PS mutants act mainly at the Golgiather than at the ER level. This finding could explain theodest effect of PS1 mutants on cytosolic Ca2+ changes,

iven that this organelle only partially contributes to rapidly-xchanging Ca2+ pools. These data also support the idea thatS1 and PS2 act on different targets. Indeed, PSs are nowegarded as multi-faceted proteins whose roles can be func-ionally distinguished and independent on their enzymaticctivity [33,34]. In this regard, we have here demonstratedhat the “loss-of-function” PS2-D366A, which lacks a keyspartate residue, was still capable of reducing Ca2+ releasen over-expressing HeLa cells. In addition, we showed thatS2-T122R exerts its effect independently of endogenousS1, in fact, a similar reduction in Ca2+ release was observedlso in double KO MEF cells.

[1] D.W. Dickson, Neuropathological diagnosis of Alzheimer’s disease:a perspective from longitudinal clinicopathological studies, Neuro-biol. Aging 18 (1997) S21–S26.

[2] D.J. Selkoe, The genetics and molecular pathology of Alzheimer’sdisease: roles of amyloid and the presenilins, Neurologic Clinics 18(2000) 903–922.

[3] B. De Strooper, P. Saftig, K. Craessaerts, H. Vanderstichele, G.Guhde, W. Annaert, K. Von Figura, F. Van Leuven, Deficiency ofpresenilin-1 inhibits the normal cleavage of amyloid precursor pro-tein, Nature 391 (1998) 387–390.

[4] W.G. Annaert, L. Levesque, K. Craessaerts, I. Dierinck, G. Snellings,D. Westaway, P.S. George-Hyslop, B. Cordell, P. Fraser, B. DeStrooper, Presenilin 1 controls �-secretase processing of amyloidprecursor protein in pre-Golgi compartments of hippocampal neu-rons, J. Cell Biol. 147 (1999) 277–294.

[5] K.S. Vetrivel, H. Cheng, W. Lin, T. Sakurai, T. Li, N. Nukina, P.C.Wong, H. Xu, G. Thinakaran, Association of �-secretase with lipidrafts in post-Golgi and endosome membranes, J. Biol. Chem. 279(2004) 44945–44954.

[6] Q. Guo, K. Furukawa, B.L. Sopher, D.G. Pham, J. Xie, N. Robin-son, G.M. Martin, M.P. Mattson, Alzheimer’s PS-1 mutation perturbscalcium homeostasis and sensitizes PC12 cells to death induced byamyloid �-peptide, Neuroreport 8 (1996) 379–383.

[7] M.A. Leissring, I. Parker, F.M. LaFerla, Presenilin-2 mutations mod-ulate amplitude and kinetics of inositol 1,4,5-trisphosphate-mediatedcalcium signals, J. Biol. Chem. 274 (1999) 32535–32538.

[8] M.A. Leissring, B.A. Paul, I. Parker, C.W. Cotman, F.M.LaFerla, Alzheimer’s presenilin-1 mutation potentiates inositol 1,4,5-trisphosphate-mediated calcium signaling in Xenopus oocytes, J.Neurochem. 72 (1999) 1061–1068.

[9] M.A. Leissring, F.M. LaFerla, N. Callamaras, I. Parker, Subcellularmechanisms of presenilin-mediated enhancement of calcium signal-ing, Neurobiol. Dis. 8 (2001) 469–478.

10] I.F. Smith, K.N. Green, F.M. LaFerla, Calcium dysregulation inAlzheimer’s disease: recent advances gained from genetically mod-ified animals, Cell Calcium 38 (2005) 427–437.

550 G. Zatti et al. / Cell Calcium 39 (2006) 539–550

[11] A.S. Yoo, I. Cheng, S. Chung, T.Z. Grenfell, H. Lee, E. Pack-Chung,M. Handler, J. Shen, W. Xia, G. Tesco, A.J. Saunders, K. Ding,M.P. Frosch, R.E. Tanzi, T.W. Kim, Presenilin-mediated modulationof capacitative calcium entry, Neuron 27 (2000) 561–572.

[12] F. Facchinetti, C. Fasolato, E. Del Giudice, A. Burgo, S. Furegato,M. Fusco, E. Basso, R. Seraglia, A. D’Arrigo, A. Leon, Nimodip-ine selectively stimulates �-amyloid 1–42 secretion by a mechanismindependent of calcium influx blockage, Neurobiol. Aging 27 (2006)218–227.

[13] F.M. La Ferla, Calcium dyshomeostasis and intracellular signallingin Alzheimer’s disease, Nature Rev. Neurosci. 3 (2002) 862–872.

[14] G. Zatti, R. Ghidoni, L. Barbiero, G. Binetti, T. Pozzan, C. Fasolato,P. Pizzo, The presenilin 2 M239I mutation associated with familialAlzheimer’s disease reduces Ca2+ release from intracellular stores,Neurobiol. Dis. 15 (2004) 269–278.

[15] M. Giacomello, L. Barbiero, G. Zatti, R. Squitti, G. Binetti, T. Poz-zan, C. Fasolato, R. Ghidoni, P. Pizzo, Reduction of Ca2+ stores andcapacitative Ca2+ entry is associated with the familial Alzheimer’sdisease presenilin-2 T122R mutation and anticipates the onset ofdementia, Neurobiol. Dis. 18 (2005) 638–648.

[16] L. Scorrano, S.A. Oakes, J.T. Opferman, E.H. Cheng, M.D. Sor-cinelli, T. Pozzan, S.J. Korsmeyer, BAX and BAK regulation ofendoplasmic reticulum Ca2+: a control point for apoptosis, Science300 (2003) 135–139.

[17] M.F. Folstein, S.E. Folstein, P.R. McHugh, Mini-mental state”. Apractical method for grading the cognitive state of patients for theclinician, J. Psychiatr. Res. 12 (1975) 189–198.

[18] M. Brini, R. Marsault, C. Bastianutto, J. Alvarez, T. Pozzan, R.Rizzuto, Transfected aequorin in the measurement of cytosolic Ca2+

concentration ([Ca2+]c): a critical evaluation, J. Biol. Chem. 270(1995) 9896–9903.

[

[

[

[

gene bearing missense mutations in early-onset familial Alzheimer’sdisease, Nature 375 (1995) 754–760.

[23] T. Wisniewski, W.K. Dowjat, J.D. Buxbaum, O. Khorkova, S.Efthimiopoulos, J. Kulczycki, W. Lojkowska, J. Wegiel, H.M. Wis-niewski, B. Frangione, A novel polish presenilin-1 mutation (P117L)is associated with familial Alzheimer’s disease and leads to death asearly as the age of 28 years, Neuroreport 9 (1998) 217–221.

[24] S. Signorini, R. Ghidoni, L. Barbiero, L. Benussi, G. Binetti, Preva-lence of pathogenic mutations in an Italian clinical series of patientswith familial dementia, Curr. Alzheimer Res. 1 (2004) 215–218.

[25] R.A. Ross, B.A. Spengler, J.L. Biedler, Coordinate morphologicaland biochemical interconversion of human neuroblastoma cells, J.National Cancer Inst. 71 (1983) 741–747.

[26] M. Montero, M. Brini, R. Marsault, J. Alvarez, R. Sitia, T. Pozzan,R. Rizzuto, Monitoring dynamic changes in free Ca2+ concentrationin the endoplasmic reticulum of intact cells, EMBO J. 14 (1995)5467–5475.

[27] P. Pinton, T. Pozzan, R. Rizzuto, The Golgi apparatus is an inositol1,4,5-trisphosphate-sensitive Ca2+ store, with functional propertiesdistinct from those of the endoplasmic reticulum, EMBO J. 17 (1998)5298–5308.

[28] L. Missiaen, K. Van Acker, K. Van Baelen, L. Raeymaekers, F.Wuytack, J.B. Parys, H. De Smedt, J. Vanoevelen, L. Dode, R. Riz-zuto, G. Callewaert, Calcium release from the Golgi apparatus andthe endoplasmic reticulum in HeLa cells stably expressing targetedaequorin to these compartments, Cell Calcium 36 (2004) 479–487.

[29] G. Yu, F. Chen, M. Nishimura, H. Steiner, A. Tandon, T. Kawarai,S. Arawaka, A. Supala, Y.Q. Song, E. Rogaeva, E. Holmes, D.M.Zhang, P. Milman, P.E. Fraser, C. Haass, P.S. George-Hyslop, Muta-tion of conserved aspartates affects maturation of both aspartatemutant and endogenous presenilin 1 and presenilin 2 complexes,

[

[

[

[

[

19] A. Herreman, D. Hartmann, W. Annaert, P. Saftig, K. Craessaerts, L.Serneels, L. Umans, V. Schrijvers, F. Checler, H. Vanderstichele, V.Baekelandt, R. Dressel, P. Cupers, D. Huylebroeck, A. Zwijsen, F.Van Leuven, B. De Strooper, Presenilin 2 deficiency causes a mildpulmonary phenotype and no changes in amyloid precursor proteinprocessing but enhances the embryonic lethal phenotype of presenilin1 deficiency, Proceedings of the National Academy of Sciences ofthe United States of America 96 (1999) 11872–11877.

20] O. Nyabi, S. Pype, M. Mercken, A. Herreman, P. Saftig, K. Craes-saerts, L. Serneels, W. Annaert, B. De Strooper, No endogenousA� production in presenilin-deficient fibroblasts, Nature Cell Biol.4 (2002) E164 (author reply E165–6).

21] E. Levy-Lahad, W. Wasco, P. Poorkaj, D.M. Romano, J. Oshima,W.H. Pettingell, C.E. Yu, P.D. Jondro, S.D. Schmidt, K. Wang, A.C.Crowley, Y.-H. Fu, S.Y. Guenette, D. Galas, E. Nemes, E.M. Wijs-man, T.D. Bird, G.D. Schellenberg, R.E. Tanzi, Candidate gene forthe chromosome 1 familial Alzheimer’s disease locus, Science 269(1995) 973–977.

22] R. Sherrington, E.I. Rogaev, Y. Liang, E.A. Rogaeva, G. Levesque,M. Ikeda, H. Chi, C. Lin, G. Li, K. Holman, et al., Cloning of a

J. Biol. Chem. 275 (2000) 27348–27353.30] A.J. Irving, G.L. Collingridge, A characterization of muscarinic

receptor-mediated intracellular Ca2+ mobilization in cultured rat hip-pocampal neurones, J. Physiol. 511 (Pt 3) (1998) 747–759.

31] K.R. McCoy, R.D. Mullins, T.G. Newcomb, G.M. Ng, G. Pavlinkova,R.J. Polinsky, L.E. Nee, J.E. Sisken, Serum- and bradykinin-inducedcalcium transients in familial Alzheimer’s fibroblasts, Neurobiol.Aging 14 (1993) 447–455.

32] B.O. Popescu, A. Cedazo-Minguez, E. Benedikz, T. Nishimura, B.Winblad, M. Ankarcrona, R.F. Cowburn, �-secretase activity of pre-senilin 1 regulates acetylcholine muscarinic receptor-mediated signaltransduction, J. Biol. Chem. 279 (2004) 6455–6464.

33] J. Herms, I. Schneider, I. Dewachter, N. Caluwaerts, H. Kretzschmar,F. Van Leuven, Capacitive calcium entry is directly attenuated bymutant presenilin-1, independent of the expression of the amyloidprecursor protein, J. Biol. Chem. 278 (2003) 2484–2489.

34] Y. Akbari, B.D. Hitt, M.P. Murphy, N.N. Dagher, B.P. Tseng, K.N.Green, T.E. Golde, F.M. LaFerla, Presenilin regulates capacitativecalcium entry dependently and independently of �-secretase activity,Biochem. Biophys. Res. Commun. 322 (2004) 1145–1152.