Molecular Psychiatry (2004) 9, 946–952 & 2004 Nature Publishing Group All rights reserved 1359-4184/04 $30.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE GM1 regulates the proteolysis of amyloid precursor protein Q Zha1, Y Ruan1, T Hartmann2, K Beyreuther2 and D Zhang1 1Department of Biochemistry, Institute of Mental Health, Peking University, Beijing, China; 2ZMBH, Center for Molecular Biology, University of Heidelberg, Im Neuenheimer Feld 282, Heidelberg, Germany

Plaques containing amyloid b-peptides (Ab) are a major feature in Alzheimer’s disease (AD), and GM1 ganglioside is an important component of cellular plasma membranes and especially enriched in raft. GM1-bound Ab (GM1/Ab), found in exhibiting early pathological changes of AD including diffuse plaques, has been suggested to be involved in the initiation of amyloid fibril formation in vivo by acting as a seed. However, the role of GM1 in amyloid b- protein precursor (APP) processing is not yet defined. In this study, we report that exogenous GM1 ganglioside promotes Ab biogenesis and decreases sAPPa secretion in SH-SY5Y and COS7 cells stably transfected with human APP695 cDNA without affecting full-length APP and the sAPPb levels. We also observe that GM1 increases extracellular levels of Ab in primary cultures of mixed rat cortical neurons transiently transfected with human APP695 cDNA. These findings suggest a regulatory role for GM1 in APP processing pathways. Molecular Psychiatry (2004) 9, 946–952. doi:10.1038/sj.mp.4001509 Published online 30 March 2004 Keywords: amyloid beta-protein; Alzheimer’s disease; GM1 ganglioside; amyloid beta-protein precursor; proteolysis

Formation of senile plaques, composed of a 4 kDa signaling machinery reside within small peptide—the amyloid b-peptide (Ab)—is one of microdomains, or rafts, enriched in and the hallmarks of Alzheimer’s disease (AD). Ab is cholesterol. through the plasma generated by the consecutive cuts of two proteases, membrane is thought to proceed within raft and might known as b- and g- secretase,1 which liberate the be influenced by .8,9 The ganglioside GM1, amyloidogenic peptide from its precursor, the Alzhei- enriched in raft, has been well-studied because of its mer amyloid precursor protein (APP). An alternative role as the cell surface receptors for viral and bacterial cleavage before residue 17 of Ab by a-secretase toxins. It has been reported that an increased sialidase releases sAPPa and prevents the generation of Ab. activity in the AD would result in an increase in Ab aggregation is thought to result in the formation of gangliosides, preferentially GM1, and significant neurotoxic protofibrils,2 and the deposition of amyloid changes in the ganglioside patterns of AD brains have plaques. Genetic studies show that increased Ab42 been reported.10,11 GM1 has been found in amyloid production is associated with inherited familial AD.1,3 plaques and binds to Ab.12,13 They discovered GM1- Factors regulating Ab secretion could thus become bound Ab in brains exhibiting early pathological very important for the etiology of AD. The predomi- changes of AD and suggested that GM1-Ab complexes nant cleavage product is sAPPa and this may serve to may serve as a seed for toxic amyloid fibril formation. prevent the production of Ab.1,3 The proteolytic Recently, Matsuloka et al14 use GM1 as peripheral processing of APP to sAPP and Ab is regulated by agent with high affinity to Ab to reduce brain amyloid protein kinase C (PKC),4 protein tyrosine kinase, load in transgenic mice. Here we found that ganglio- muscarinic receptors, estrogens and cholesterol.5 sides do not only bind to Ab but also clearly stimulate Gangliosides are found in the Ab production on a cellular level in cultures of African outer leaflet of mammalian plasma membranes, espe- green monkey epithelial kidney cells (COS7) and cially abundant on neuronal cell surfaces. GM1, GD1a, human neuroblastoma cells (SH-SY5Y) and in mixed GD1b, GT1b are major types of gangliosides found in rat cortical neurons, which transfected with human brain.6 Gangliosides have been reported to participate APP695 cDNA. We examined the direct effect of GM1 in proliferation and differentiation of various types of on APP processing in detail. cell.7 Many components of the plasma membrane

Materials and methods Correspondence: D Zhang, No 51, Hua Yuan Bei Road, Haidian and treatment District, Beijing 100083, China. E-mail: [email protected] Received 17 February 2004; revised 17 February 2004; accepted Mixed cortical neurons were prepared from 14-day- 19 February 2004 old fetal rats as described.15 Recombinant Semliki GM1 regulates APP proteolysis Q Zha et al 947 Forest virus (SFV) encoding human APP695 was length APP (APPfl) was determined by lysing cells prepared as described previously.16 Neurons were with RIPA buffer immediately after labeling, followed infected for 60 or 90 min at 371C under a 5% CO2/95% by immunoprecipitation with 22C11 or W02 (anti- air atmosphere with recombinant SFV-APP695. The Ab4-10) and protein G agarose (Sigma). Secreted virus solution was replaced by maintenance medium. APPa/b was immunopreciptated from the medium COS7 and SH-SY5Y cell lines both transfected with with 22C11. Total Ab,Ab40, Ab42 were immunopre- APP695cDNA, SH-SY5Y-SPA4CT were cultivated in cipitated from the medium with W02, G2–10, G2–11 MEM/F-12 medium containing 10% fetal calf serum respectively. (FCS), 2 mML-glutamine, 1% nonessential amino acids and 300 mg/ml Hygromycin (Roche). GM1 Western blots purchased from Sigma were dissolved in PBS. Cell Medium and cell extracts were analyzed by Western culture medium contains no FCS when treated with blot as previously described.18 Detection was made by GM1 for the presence of FCS interfered with GM1 ECL method (Amersham). Band was acquired using a binding to cells.17 CCD camera (BioRad) and band intensity was per- formed by the applied software Quantity One. Monoclonal antibodies W02 (for total Ab and sAPPa), Statistical methods G2-10 (for Ab40) and G2-11 (for Ab42) have been The data were analyzed by ANOVA followed by previously described.18 22C11 (purchased from Sig- unpaired Student’s t-tests. Values of Po0.05 were ma) recognizes amino acids 66–81 of the N-terminus considered significant. on the APP molecule, specific for sAPPa/b and full- length APP. Polyclonal anti-Ab1-40 692 Results equally precipitates Ab and P3. Effect of GM1 on APP gene transcription Cell toxicity To test the role of GM1 on APP gene transcription, we To determine the effect of GM1 on cell toxicity, measured the ratio of APP/b- by RT-PCR method plasma membrane damage associated with acute cell in SY5Y695 cells. There was no significant difference death was analyzed by measuring the release of the in the ratio indicating that GM1 does not influence lactate dehydrogenase (LDH) into the medium with the transcription of APP gene (data not shown). GM1 the Cytotoxicity Detection Kit (Promega). treatment has no cell toxicity, as assessed by LDH released into the media (Table 1). RT-PCR Total RNA was isolated using the Trizol kit according Effect of GM1 on APP turnover to the manufacturers protocal (Gibco/BRL). Comple- To study the effect of GM1 on the half-life of full- mentary DNA (cDNA) was prepared by reverse length APP (APPfl) and the amount of secreted a b transcription (RT) of 1 mg of total RNA according to soluble APP (sAPP / ), we performed pulse-chase the manufacturer’s protocal (Gibco/BRL). The PCR experiment in SH-SY5Y cells transfected with m program was designed according to a sequence APP695. Cells were pretreated with 50 M GM1 1 or vehicle for 8 h, then metabolically labeled with including a denaturation step at 95 C for 3 min, 35 following by 25 cycles for APP, or 20 cycles for S-methionine for 1 h and chased for different a b b-actin, each consisting of denaturation at 941C for periods of time. APPfl and sAPP / from the condi- 30 s, annealing at 601C for APP or 521C for b-actin for tioned media were immunoprecipitated with anti- 60 s, and elongation at 701C for 60 s. The amplifica- body 22C11. No significant difference in APPfl levels tion was terminated by a delay time of 10 min at 721C. between treated and untreated cells was observed The PCR products were separated by electrophoresis either at timepoint 0 or throughout the pulse chase on 2% agarose gel that included ethidium bromide (Figure 1, upper half ). Since 22C11 crossreacts with a (0.1 mg/ml). The specific primers for APP gene were 50 related protein APLP2, further confirmation of pro- CCA AGC TTA TGA GGG GCC GCA AGC AG 30 as tein specificity was obtained using the APP-specific sense primer, and 50 GCG AAT TCC TCT TTG GCG antibody W02, which recognizes epitopes 4–10 amino ACG GTG TGC 30 as antisense primer (amplify the residues 100–156 of APP695). The specific primers Table 1 GM1 did not affect the release of the cytosolic enzyme LDH into the conditioned media for b-actin were: 50 GTG GAC ATC CGC AAA GAC 30 0 as sense primer, and 5 AAA GGG TGT AAC GCA ATC 0 mM1mM10mM50mM AA30 as antisense primer defining a 302 base pair target of the human cDNA. LDH (U/l) 28.671.5 29.372.0 28.972.3 32.172.6 Metabolic labeling and immunoprecipitation For pulse-chase experiments cells were pretreated SH-SY5Y cell lines were treated with different concentra- with or without 50 mM GM1 for 8 h, labeled with tion of GM1 for 8 h. LDH release in the conditioned media 35S-methionine 250 mCi/ml for 1 h and chased for was analyzed using the enzymatic assay from Promega, as different periods of time. Total labeled cellular full- described by the manufacturer.

Molecular Psychiatry GM1 regulates APP proteolysis Q Zha et al 948

Figure 1 Temporal profile of APP expression and secretion by SH-SY5Y. (a) The autoradiograph represents results from one of three experiments conducted. (b) Quantitative analysis of APPfl in cell lysate and total secreted APP in the medium as a result of a- and b-secretase cleavage (sAPPa/b). Relative intensity of the bands was calculated as percentage of the intensity of APPfl protein band in cell lysate at the beginning of the chase (time 0). Mean7SE, n ¼ 3. *Po0.05, **Po0.005.

acids of Ab (neither APLP1 nor APLP2 contains the smaller sAPP fragments. An example of immunoblots Ab region). GM1 treatment did not affect the level of with these preparations is shown in Figure 2c. GM1 cellular APPfl or the ratio between mature and strongly decreased the sAPPa signal but had little immature APP forms, indicating normal protein effect on the remaining sAPP levels. Since the maturation. majority of this cleavage products are sAPPb,itis likely that the decrease in sAPP is not caused by an GM1 inhibits the secretion of sAPPa alteration in sAPPb but sAPPa levels. In the pulse-chase experiment, immunoprecipitation of sAPP from the conditioned medium, however, b showed that the secretion of soluble APP products GM1 increases A production was significantly decreased in GM1 treated cells as To examine the effect of GM1 on Ab secretion, COS7 compared to the untreated cells (Figure 1, bottom cells stably transfected with APP695 were treated half ). Here the sAPP includes both sAPPa and sAPPb. with different concentrations of GM1 for 8 h. Ab in We then evaluated the effect of GM1 on the sAPPa the conditioned media was detected by IP-Western secretion by immunoblotting in the media of COS7 blot using polyclonal antibody 692. Figure 3 shows cells stably transfected with APP695 cDNA. Figure 2a that Ab increases in a dose-dependent manner by illustrates the dose–response pattern of the effects of GM1 treatment. Importantly, a parallel increase in GM1 on sAPPa after 8 h of GM1 exposure, sAPPa GM1 short Ab fragments including a-secretase generated P3 concentration dependently decreased. Figure 2b was observed, indicating that this GM1 effect is not shows the time course of the effect of 50 mM GM1 on specific to b-secretase. As Ab and P3 increased sAPPa secretion by COS7 cells. In the control simultaneously in COS7 cells, the GM1 induced condition the amount of sAPPa in the medium increase in Ab could be caused by either a parallel increased progressively with time, with a plateau increase in a- and b-secretase activity or by a after approx. 8 h. When the cells were exposed to stimulation of g-secretase activity. SP-C99 is a GM1, sAPPa in the medium was decreased relative to truncated APP molecule that is a poor substrate of a- the controls, but reached its plateau at approximately secretase and is not cleaved by b-secretase. SY5Y-SP- the same time. Similar effects were observed with C99 has been previously shown to secrete Ab SY5Y695 (data not shown). peptides into the culture media.22 Significantly, These results were obtained with the antibody GM1 increased Ab generation in a dose-dependent W02, which recognizes the Ab epitopes 4–10. This manner (Figure 4). excludes b-ceavage but includes b0 (BACE1 cleavage It has been shown that most of the Ab released from at Glu11, a minor by-product in COS and SH-SY5Y cell lines terminates at residue 40. However, a small cells) and a-cleavages. To see whether b-secretase proportion (5–10%) extends to residue 42. It has been product sAPPb was influenced by GM1 treatment, the postulated that the major pathologic culprit in AD is media of COS7 cell treated with vehicle or GM1 this subpopulation of Ab, because this longer, more (50 mM) for 8 h were precleared with W02 to discard aggregation-prone species deposits preferentially in sAPPa and 22C11 was then used to detect sAPPb and both sporadic and familial AD brains.19,20 Then we

Molecular Psychiatry GM1 regulates APP proteolysis Q Zha et al 949

Figure 2 GM1 inhibits the secretion of sAPPa, not sAPPb. (a) Dose–response pattern of the effect of GM1 on sAPPa secretion in COS7 cells. (b1) Time course of the effect of GM1 (50 mM) on sAPPa. (b2) Quantitative analysis of sAPPa in the medium, relative intensity of the bands was calculated as percentage of the intensity of sAPPa in 16 h time point. Means7SE, n ¼ 3. *Po0.05, **Po0.005. (c) COS7 cells exposed for 8 h to 50 mM GM1, immunoprecipitated with W02, which recognizes only sAPPa, a subsequent immunoprecipitation was performed with 22C11 to determine the sAPPb.

Figure 4 GM1 increases Ab production by promoting g-cleavage of APP. SY5Y-SPA4CT, a good substrate for g-secretase, treated with different concentrations of GM1 Figure 3 GM1 increases Ab and P3 secretion in COS7 for 8 h. Total Ab detected by IP-Western blot, Lanes 1–4, 0, cells. COS7 cells stably transfected with APP695 were 1, 10, 50 mM GM1; Lane 5, 5 ng standard Ab40. treated with different concentrations of GM1 for 8 h. IP-

Western blot shows total Ab(40) and P3(40) in the conditioned medium. encoding the human APP (SFV-APP).16,21–23 GM1 treatment also resulted in increased Ab secretion (Figure 6). These results showed that ganglioside performed the examinations using SH-SY5Y cells GM1 strongly increased both Ab40 and Ab42 secre- stably transfected with human APP695 in the same tion in neuronal and non-neuronal cells and cultured set, Ab40 and Ab42 were determined by immunopre- neurons. cipitation using the monoclonal antibodies G2-10 and In principle, increased Ab levels in the conditioned G2-11, respectively (Figure 5a). Both Ab40 and Ab42 media might be due to either increased Ab secretion increased dose-dependently by GM1 in the range of or decreased degradation. To examine GM1 effect on 10–50 mM. Ab increased slightly when cells were intracellular Ab, we use SH-SY5Y treated with treated with 1 mM GM1. In the presence of 10 mM various concentrations of GM1 for 8 h, both Ab40 GM1, Ab40 and Ab42 release was increased to and Ab42 levels increased in the detergent soluble 2.570.35 and 2.670.3 fold of that in the absence of pool of the cell lysates (Figure 7). To see whether this GM1. In the presence of 50 mM GM1, a 5.670.59 and effect was due to inhibition of Ab degradation, we 4.570.51 fold increased Ab40 and Ab42 levels was detect the Ab level after removal of GM1 treatment observed (Figure 5b). medium. Ab secretion continuously increased after To examine the effect of GM1 on cultured neurons, another 8 h incubation (Figure 8). Apparently, in- we expressed the human APP gene in primary creased Ab levels were not due to inhibition of Ab rat mixed cortical neurons by infection with SFV degradation.

Molecular Psychiatry GM1 regulates APP proteolysis Q Zha et al 950

Figure 8 Ab secretion still increases after another 8 h incubation after removing GM1 treatment. SH-SY5Y trans- fected with APP695 treated with different concentrations of GM1 for 8 h. Lanes 1–4, GM1 0, 1, 10, 50 mM. After removing the medium, cells were cultured for another 8 h with fresh medium. Total Ab in the culture medium was detected. A continuous increased Ab secretion was observed.

activity of g-secretase, demonstrated by SY5Y- SPA4CT cell model. It has no influence on the expression of APP gene by RT-PCR analysis and has Figure 5 GM1 ganglioside exposure to SH-SY5Y cells little effect of rates of APP turnover by pulse-chase increases both Ab40 and Ab42 levels. (a) IP-Western blot of experiment. GM1 is known to have Ab binding conditioned media. Ab40 (antibody G2-10) and Ab42 affinity, to ensure that the binding of GM1 to Ab did (antibody G2-11) levels in the media increase after incuba- not reduce the sensitivity of the assay because of tion in the presence of 0, 1, 10, 50 mM GM1 (lanes 2–5, 7–10) epitope masking, IP-Western blot was performed on a for 8 h. Lane 1, standard Ab40 5 ng; lane 6, standard Ab42 known concentration of Ab with and without GM1, 5 ng. (b) Quantification of Ab40 and Ab42 were based on six the epitope of the detection antibody was not masked independent experiments. Student’s t-test was used for to any great degree, because IP-Western blot detected unpaired-wise comparisons, Values are means7SE, 96–105% of added human Ab (data not shown). *Po0.05, **Po0.005. Gangliosides are a containing complex glycosphingolipids found in most of the vertebrate cells, and are present in high concentrations in the neuron-rich regions of the nervous systems. They are amphophilic in nature because they have both and . The lipid portion of the molecule is embedded in the hydrophobic part of the mem- brane and its portion extends into the Figure 6 GM1 ganglioside exposure to mixed cortical extracellular environment. Topologically, ganglio- neurons increases total Ab levels. Immunoprecipitation sides are ideally situated to modulate biologic effects b and Western blot analysis of A (antibody W02) from on cells through their interactions with various untreated cells and GM1-treated cells. Lane 1, 0 mM GM1; 24,25 Lane 2, 10 mM GM1; Lane 3, 50 mM GM1. ligands. When administered exogenously, either in vivo or in vitro, they affect several different cellular events. How does GM1 alter APP processing and secretion? The APP processing is complex and includes APP post-translational modifications and the activity of at least three secretases in different subcellular compartments through the secretory as well as endocytic pathways.26 Fishman et al27,28 observed that neuroblastoma cells that contain GM1 Figure 7 GM1 promotes intracellular Ab production. Ab40 were able to metabolize exogenous GM1. Exogenous and Ab42 production was measured by IP-Western blot. gangliosides can be incorporated into the plasma Lanes 1–4, 5–8 show 0, 1, 10, 50 mM GM1 treatment for 8 h membrane in a manner analogous to endogenous in SH-SY5Y cells. gangliosides under appropriate conditions. Wiegandt and co-workers29,30 have reported that exogenous Taken together, the above data indicate that GM1 gangliosides are taken up by cells, but are incorpo- increases Ab generation by affecting g-secretase rated in the plasma membrane in a different way than activity directly or indirectly. endogenous gangliosides and are not degraded. They also observed that most of the gangliosides were removed by adding serum or serum albumin to the Discussion culture medium. Differences in types of cells used Our results demonstrate that exogenous addition of might account for some of these contrasting results. GM1 in cultured cells and neurons enhances Ab Over many years, extensive studies on the binding to production, which are paralleled by inhibition of and uptake of gangliosides by biological membranes sAPPa secretion. In addition, we found that GM1 were performed. In principle, the binding of ganglio- regulates the processing of APP by affecting the sides to plasma membranes occurs in three different

Molecular Psychiatry GM1 regulates APP proteolysis Q Zha et al 951 operationally determined manner: loosely associated N-terminus could result in conformational changes of micelles removable by serum proteins, a protein- APP N-terminus such that a-secretase cleavage is bound, serum-resistant but trypsin-labile ganglioside inhibited (in preparation). On the other hand, GM1 fraction, and trypsin-stable associated gangliosides has been reported to have an inhibitory action on PKC largely comprising inserted monomers.31,32 The inser- and sAPPa is positively regulated by PKC activa- tion into plasma membranes of gangliosides from tion.40–42 GM1 may inhibit sAPPa secretion by their micellar state is a slow process. The maximal inhibiting the PKC activity. Both hypotheses require incorporation into cells of GM1 was reached after and deserve further experimental work. 72 h. The exogenous GM1 that has been incorporated Many studies are exploring the roles of molecules into the outer surface of the plasma membrane such as cholesterol and ganglioside in the formation becomes internalized. Once inside the cell, some of of Ab plaques. Our study raises the possibility that the GM1 must enter the Golgi apparatus where it is GM1 increases the level of intracellular Ab and causes converted to GD1a and some of it must be taken up by a greater extent of Ab secretion into the extracellular where it is degraded to GM2. APP is a space, and thus may contribute to b-amyloid deposi- transmembrane glycoprotein33 and the APP proces- tion. The above hypothesis remains to be tested in sing enzymes, or secretases, appear to be membrane vivo. In conclusion, our results indicate that the associated.34 The interaction of APP with the secre- processing of APP and generation of Ab can be tases are believed to occur within several subcellular modulated depending on the cellular content of compartments, including the endoplasmic reticulum GM1 ganglioside. (ER), Golgi stacks, trans-Golgi networks, post-Golgi vesicles, endosomes and plasma membrane. All of the membranes within these compartments contain GM1. Acknowledgements Thus, exogenous GM1 may interfere with APP This work was supported by National Natural processing in many different subcelluar organelles. Sciences Foundation of China (39870276) and grant In our study, the GM1 promotes Ab secretion in from the Major State Basic Research Development SY5Y-A4CT cells indicated that GM1 could increase Program of China (G1999064007) and EU-QLK-2002- g-secretase activity directly or indirectly. The illusive 172. g-secretase seems to be a complex of multiple integral membrane proteins35,36 and it was shown to be raft associated. This finding fits well with recent reports References that amyloidogenic processing of the Alzheimer b- 37 1 Selkoe DJ. Alzheimer’s disease: genes, proteins, and therapy. amyloid precursor protein depends on lipid rafts. Physiol Rev 2001; 81: 741–766. Rafts are lateral assemblies of sphingolipids and 2 Nilsberth C, Westlind-Danielsson A, Eckman CB, Condron MM, cholesterol within the membrane. The involvement Axelman K, Forsell C et al. The ‘Arctic’ APP mutation (E693G) of rafts in Ab production is further supported by causes Alzheimer’s disease by enhanced Abeta protofibril forma- previous work showing that GM1 and cholesterol tion. Nat Neurosci 2001; 4: 887–893. 3 Beyreuther K, Pollwein P, Multhaup G, Monning U, Konig G, both bind to Ab and facilitate amyloid fibril forma- Dyrks T et al. Ann NY Acad Sci 1993; 695: 91–102. 12,13 tion Rafts are small and highly dispersed at the 4 Cedazo-Minguez A, Wiehager B, Winblad B, Huttinge M, Cowburn cell surface and are suggested to contain only a subset RF. Effects of apolipoprotein E (apoE) isoforms, beta-amyloid of B10–30 protein molecules.38 Therefore, the like- (Abeta) and apoE/Abeta complexes on protein kinase C-alpha g (PKC-alpha) translocation and amyloid precursor protein (APP) lihood is low that APP and -secretase are in the same processing in human SH-SY5Y neuroblastoma cells and fibro- individual raft. For g-cleavage to occur rafts would blasts. Neurochem Int 2001; 38: 615–625. have to be clustered to get APP and g-secretase into 5 Zheng H, Xu H, Uljon SN, Gross R, Hardy K, Gaynor J et al. the same raft platform. How and where clustering is Modulation of A(beta) peptides by estrogen in mouse models. accomplished during GM1 uptake from plasma J Neurochem 2002; 80: 191–196. 6 Molander-Melin M. Blennow K, Bogdanovic N, Mansson JE, membrane is not yet known. Other such examples of Fredman P. Lipid analyses of human brain rafts. J Neurochem raft-dependent processes are known. The conforma- 2003; 85(Suppl 2): 34. tional change of the cellular prion protein (PrPc) to its 7 Fusco M, Vantini G, Schiavo N, Zanotti A, Zanoni R, Facci L et al. pathogenic scrapie form critically depends on the Gangliosides and neurotrophic factors in neurodegenerative 39 diseases: from experimental findings to clinical perspectives. integrity of rafts. However, more work will be Ann N Y Acad Sci 1993; 695: 314–317. necessary to demonstrate whether g-cleavage of APP 8 Marmor MD, Julius M. Role for lipid rafts in regulating interleukin- is indeed directly dependent on the raft lipid 2 receptor signaling. Blood 2001; 98: 1489–1497. environment. 9 Gniadecki R, Christoffersen N, Wulf HC. Cholesterol-rich plasma As for the inhibition of a-secretase cleavage by GM1 membrane domains (lipid rafts) in keratinocytes: importance in the baseline and UVA-induced generation of reactive oxygen there are two possible mechanisms. On the one hand, species. J Invest Dermatol 2002; 118: 582–588. the stiffening of the membrane due to GM1 loading 10 Blennow K, Davidsson P, Wallin A, Fredman P, Gottfries CG, may decrease sAPPa by inhibiting lateral movement Karlsson I et al. Gangliosides in cerebrospinal fluid in ‘probable and the required contact between enzyme and Alzheimer’s disease’. Arch Neurol 1991; 48: 1032–1035. 11 Svennerholm L, Bostrom K, Jungbjer B, Olsson L. Membrane lipids substrate. This result is supported by our previous of adult human brain: lipid composition of frontal and temporal study that there is a GM1 binding site located in the lobe in subjects of age 20 to 100 years. J Neurochem 1994; 63: N-terminus of APP, and binding of GM1 to APP 1802–1811.

Molecular Psychiatry GM1 regulates APP proteolysis Q Zha et al 952 12 Yanagisawa K, Odaka A, Suzuki N, Ihara Y. GM1 ganglioside- 27 Fishman PH, Bradley RM, Hom BE, Moss J. Uptake and bound amyloid beta-protein (A beta): a possible form of pre- of exogenous gangliosides by cultured cells: effect amyloid in Alzheimer’s disease. Nat Med 1995; 1: 1062–1066. of choleragen on the turnover of GM1. J Lipid Res 1983; 24: 13 Yanagisawa K, Ihara Y. GM1 ganglioside-bound amyloid beta- 1002–1011. protein in Alzheimer’s disease brain. Neurobiol Aging 1998; 19: 28 Fishman PH, Pacszka T, Hom B, Moss J. Modification of ganglio- S65–S67. side GM1. Effect of lipid moiety on choleragen action. J Biol Chem 14 Matsuoka Y, Saito M, LaFrancois J, Saito M, Gaynor K, Olm V et al. 1980; 255: 7657–7664. Novel therapeutic approach for the treatment of Alzheimer’s 29 Callies R, Schwarzmann G, Radsak K, Siegert R, Wiegandt H. disease by peripheral administration of agents with an affinity to Characterization of the cellular binding of exogenous gangliosides. beta-amyloid. J Neurosci 2003; 23: 29–33. Eur J Biochem 1977; 80: 425–432. 15 Tienari PJ, Ida N, Ikonen E, Simons M, Weidemann A, Multhaup G 30 Radsak K, Schwarzmann G, Wiegandt H. Studies on the cell et al. Intracellular and secreted Alzheimer b-amyloid species are association of exogenously added sialo-. Hoppe Seylers generated by distinct mechanisms in cultured hippocampal Z Physiol Chem 1982; 363: 263–272. neurons. Proc Natl Acad Sci USA 1997; 94: 4125–4130. 31 Mobius W, Herzog V, Sandhoff K, Schwarzmann G. Intracellular 16 de Hoop MJ, Olkkonen VM, Ikonen E, Williamson E, von Poser C, distribution of a biotin-labeled ganglioside, GM1, by immunoelec- Meyn L et al. Semliki Forest virus as a tool for protein expression tron microscopy after endocytosis in fibroblasts. J Histochem in cultured rat hippocampal neurons. Gene Ther 1994; 1: S28–S31. Cytochem 1999; 47: 1005–1014. 17 Facci L, Leon A, Toffano G, Sonnino S, Ghidoni R, Tettamanti G. 32 Albrecht B, Pohlentz G, Sandhoff K, Schwarzmann G. Synthesis Promotion of neuritogenesis in mouse neuroblastoma cells by and mass spectrometric characterization of digoxigenin and biotin exogenous gangliosides: relationship between the effect and labeled ganglioside GM1 and their uptake by and metabolism in the cell association of ganglioside GM1. J Neurochem 1984; 42: cultured cells. Chem Phys Lipids 1997; 86: 37–50. 299–305. 33 Gandy S, Petanceska S. Regulation of Alzheimer beta-amyloid 18 Ida N, Hartmann T, Pantel J, Schroder J, Zerfass R, Forstl H et al. precursor trafficking and metabolism. Biochim Biophys Acta 2000; Analysis of heterogenous A4 peptides in human cerebrospinal 1502: 44–52. fluid and blood by a newly developed sensitive western blot assay. 34 Vassar R. The beta-secretase, BACE: a prime drug target for J Biol Chem 1996; 271: 22908–22914. Alzheimer’s disease. J Mol Neurosci 2001; 17: 157–170. 19 Yang AJ, Knauer M, Burdick DA, Glabe C. Intracellular A beta 1-42 35 De Strooper B. Aph-1, Pen-2, and nicastrin with presenilin aggregates stimulate the accumulation of stable, insoluble amyloi- generate an active gamma-secretase complex. Neuron 2003; 38: dogenic fragments of the amyloid precursor protein in transfected 9–12. cells. J Biol Chem 1995; 270: 14786–14792. 36 Herreman A, Van Gassen G, Bentahir M, Nyabi O, Craessaerts K, 20 Yang AJ, Chandswangbhuvana D, Shu T, Henschen A, Glabe CG. Mueller U et al. Gamma-secretase activity requires the presenilin- Intracellular accumulation of insoluble, newly synthesized abe- dependent trafficking of nicastrin through the Golgi apparatus but tan-42 in amyloid precursor protein-transfected cells that have not its complex glycosylation. J Cell Sci 2003; 116: 1127–1136. been treated with Abeta1-42. J Biol Chem 1999; 274: 20650–20656. 37 Ehehalt R, Keller P, Haass C, Thiele C, Simons K. Amyloidogenic 21 Turner RS, Suzuki N, Chyung AS, Younkin SG, Lee VM. Amyloids processing of the Alzheimer beta-amyloid precursor protein beta40 and beta42 are generated intracellularly in cultured human depends on lipid rafts. J Cell Biol 2003; 160: 113–123. neurons and their secretion increases with maturation. J Bio Chem 38 Pralle A, Keller P, Florin EL, Simons K, Horber JK. - 1996; 271: 8966–8970. cholesterol rafts diffuse as small entities in the plasma membrane 22 De Strooper B, Simons M, Multhaup G, Van Leuven F, Beyreuther of mammalian cells. J Cell Biol 2000; 148: 997–1008. K, Dotti CG. Production of intracellular amyloid-containing 39 Baron GS, Wehrly K, Dorward DW, Chesebro B, Caughey B. fragments in hippocampal neurons expressing human amyloid Conversion of raft associated prion protein to the protease- precursor protein and protection against amyloidogenesis by resistant state requires insertion of PrP-res (PrP(Sc)) into contig- subtle amino acid substitutions in the rodent sequence. EMBO J uous membranes. EMBO J 2002; 21: 1031–1040. 1995; 14: 4832–4938. 40 Levites Y, Amit T, Mandel S, Youdim MB. Neuroprotection and 23 Tienari PJ, De Strooper B, Ikonen E, Simons M, Weidemann A, neurorescue against abeta toxicity and PKC-dependent release Czech C et al. The beta-amyloid domain is essential for axonal of nonamyloidogenic soluble precursor protein by green tea sorting of amyloid precursor protein. EMBO J 1996; 15: 5218–5229. polyphenol (À)-epigallocatechin-3-gallate. FASEB J 2003; 17: 24 Mutoh T, Tokuda A, Miyadai T, Hamaguch M, Fujiki N. Ganglio- 952–954. side GM1 binds to the Trk protein and regulates receptor function. 41 Solomonia RO, Morgan K, Kotorashvili A, McCabe BJ, Jackson AP, Proc Natl Acad Sci USA 1995; 92: 5087–5091. Horn G. Analysis of differential gene expression supports a role for 25 Mutoh T, Tokuda A, Guroff G, Fujiki N. The effect of the B subunit amyloid precursor protein and a protein kinase C substrate of toxin on the action of nerve growth factor on PC12 cells. (MARCKS) in long-term memory. Eur J Neurosci 2003; 17: J Neurochem 1993; 60: 1540–1547. 1073–1081. 26 Selkoe DJ, Schenk D. ALZHEIMER’S DISEASE: molecular under- 42 Racchi M, Mazzucchelli M, Pascale A, Sironi M, Govoni S. Role of standing predicts amyloid-based therapeutics. Annu Rev Pharma- protein kinase Calpha in the regulated secretion of the amyloid col Toxicol 2003; 43: 545–584. precursor protein. Mol Psychiatry 2003; 8: 209–216.

Molecular Psychiatry