gamma-Secretase is differentially modulated by alterations of homocysteine cycle in neuroblastoma and glioblastoma cells.

Abstract

Multiple aspects of homocysteine metabolism were studied to understand the mechanism responsible for hyperhomocysteinemia toxicity in Alzheimer disease. Besides oxidative stress and vascular damage, homocysteine has also a great importance in regulating DNA methylation through S-adenosylmethionine, the main methyl donor in eukaryotes. Alterations of S-adenosylmethionine and methylation were evidenced in Alzheimer disease and in elderly. In order to clarify whether DNA methylation can provide the basis for amyloid-beta overproduction, we used human SK-N-BE neuroblastoma and A172 glioblastoma cell lines. We tested the effects of folate, B12 and B6 deprivation and S-adenosylmethionine addition on methylation metabolism. Our results indicate that homocysteine accumulation induced through vitamin B deprivation could impair the "Methylation Potential" with consequent presenilin 1, BACE and amyloid-beta upregulation. Moreover, we found that homocysteine alterations had an effect on neuroblastoma but not on glioblastoma cells; this suggests a possible differential role of the two cell types in Alzheimer disease.

Full text

Page 1

Journal of Alzheimer’s Disease 11 (2007) 275–290
IOS Press
275
γ−Secretase is Differentially Modulated by
Alterations of Homocysteine Cycle in
Neuroblastoma and Glioblastoma Cells
Andrea Fusoa, Rosaria A. Cavallaroa, Alessandro Zampellia, Fabrizio D’Anselmia, Paola Piscopob,
Annamaria Confaloniband Sigfrido Scarpaa,∗
aDepartment of Surgery “P. Valdoni”, University of Rome “La Sapienza”, Rome, Italy
bSection of Degenerative Inflammatory and Neurological Disease, Department of Cell Biology and Neurosciences,
Istituto Superiore di Sanit` a, Roma, Italy
Abstract. Multiple aspects of homocysteine metabolism were studied to understand the mechanism responsible for hyperho-
mocysteinemia toxicity in Alzheimer disease. Besides oxidative stress and vascular damage, homocysteine has also a great
importance in regulating DNA methylation through S-adenosylmethionine, the main methyl donor in eukaryotes. Alterations of
S-adenosylmethionine and methylation were evidenced in Alzheimer disease and in elderly. In order to clarify whether DNA
methylationcanprovidethebasisforamyloid-β overproduction, weused humanSK-N-BEneuroblastoma andA172 glioblastoma
cell lines. We tested the effects of folate, B12 and B6 deprivation and S-adenosylmethionine addition on methylation metabolism.
Our results indicate that homocysteine accumulation induced through vitamin B deprivation could impair the “Methylation
Potential” with consequent presenilin 1, BACE and amyloid-β upregulation. Moreover, we found that homocysteine alterations
had an effect on neuroblastoma but not on glioblastoma cells; this suggests a possible differential role of the two cell types in
Alzheimer disease.
Keywords: S-adenosylmethionine, homocysteine, folate, vitamin B, DNA methylation, amyloid processing, presenilin 1
INTRODUCTION
Several papers point the attention of Alzheimer dis-
ease (AD)researchersonthe importanceofhighserum
homocysteine (HCY) as a risk factor in the sporadic
form of disease [1–4]. Nevertheless, an explanation
of the mechanism of HCY toxicity is still missing.
Even the role of hyperhomocysteinemia, as a real risk
factor or a consequence of the disease, is still under
debate. At present, a few mechanisms are suggest-
ed as protagonists in the toxic pathway of HCY in
AD onset: oxidative stress and neurotoxicity [5,6],
vascular damage [7,8], alteration of cholesterol and
∗Corresponding author: SigfridoScarpa, ViaA.Scarpa, 14,00161
–Roma,Italy. Tel.: +390649766600; Fax: +390649766606; E-mail:
sigfrido.scarpa@uniroma1.it.
lipids [9] and deregulationof gene expression by DNA
methylation [10]. HCY cycle is a complex biochemi-
cal pathway regulated by the presence of folate, vita-
min B12 and B6 (among other metabolites) and lead-
ing to the production of methyl donor molecule S-
adenosylmethionine(SAM). Folate and B12 are cofac-
tors in trans-methylation reactions transforming HCY
to methionine, whereas B6 is the cofactor in trans-
sulfuration reactions, responsible for HCY removal
and glutathione synthesis [11,12]. SAM can donate
a methyl group to different substrates (lipids, proteins
andDNA)beingconvertedinS-adenosylhomocysteine
(SAH), a strong competitive inhibitor of methyltrans-
ferases. The SAM/SAH ratio is an important indica-
tor of cellular methylation status and is usually indi-
cated as Methylation Potential (MP) [13,14]. In addi-
tion, HCY accumulationcould lead to SAH accumula-
tion because of the reversibility of reaction converting
ISSN 1387-2877/07/$17.00  2007 – IOS Press and the authors. All rights reserved

Page 2

276
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
SAH to HCY, altering MP. The role of DNA methyla-
tion in the regulation of gene expression is widely ac-
cepted in developmental biology and cancer [15–17],
but the involvement in cell senescence and in onset of
senescence-related disease seems still muddled. Re-
gardingAD,clinicaldataevidencedecreasedSAMlev-
els in senescence and AD [18], but the most promis-
ing perspective concerns MP alteration (i.e., SAM and
SAH relativechanges,ratherthanabsolutelevels)[19].
Involvement of MP is indicated by recent studies of
AD patients showing that the best correlation between
HCY and AD is found when also folate, vitamin B12
and vitamin B6 are measured [1,4], stressing the im-
portance of the other metabolites of HCY cycle, rather
than hyperhomocysteinemia itself. Moreover, studies
on HCY cycle in relation to dementia and methylation
open an interesting way for understanding the connec-
tion between nutritional factors and AD [20]. Since
AD is a multi-factorial pathology, this approach could
represent a promising starting point to understand the
“environmental” components of the disease.
For this reason, we decided to study the alteration
of HCY cycle by vitamin B deprivation and SAM ad-
ministration; in particular we were interested in eluci-
dating the impact of these alterations on specific and
global DNA methylation. In fact, it has been known
for a long time that modulation of DNA methylation
can have a sequence- or cell-specific pattern [21,22];
thisevidenceseemsconfirmedbothwhenHCYcycleis
altered genetically [23] and when the alteration is due
toexogenousadministrationoffolate[24]orSAM [10,
25]. In our previous works in human neuroblastoma
cells we demonstrated that folate and B12 deprivation
or SAM administration were able to deregulate prese-
nilin1(PS1)andBACEgenes[10],butawidescaleap-
proachusing a cDNA arrayevidencedjust 1% of genes
weremodulatedintheCNS[25]. Moreover,werecent-
ly confirmed in mice carrying transgenic amyloid-β
protein precursor (AβPP) that vitamin B deprivation is
able to induce hyperhomocysteinemia,PS1 and BACE
up-regulation and increased amyloid-β (Aβ) deposi-
tion (data submitted elsewhere). Here we compared
humanneuroblastomaandglioblastomacell lines (SK-
N-BE and A172, respectively) to study the effects of
folate, B12, B6 deprivationandSAM, alone or in com-
bination, on DNA methylation and gene expression.
METHODS
Media and cell cultures
Neuroblastoma SK-N-BE and glioblastoma A172
celllinesweremaintainedinF14mediumwith8%FCS
(growth medium) and shifted in complete differentia-
tion medium (DM; with 1% FCS plus 10 µM retinoic
acid) or in differentiation medium deprived of folate
and vitamin B12 (-FB), deprived of vitamin B6 (-B6)
or deprived of all the three vitamins (DDM). SAM,
100 µM, was added to media according to the experi-
mental design. Cultures were re-fed every second day
andstoppedafter72hours(DNAmethylationanalysis),
96 hours (gene expression and SAM and SAH analy-
ses) or 144 hours (protein); the times indicated refer to
medium shift as day 0. Experiments were repeated at
least three times, giving similar results. Cell growth
was evaluated by trypsinization and counting with a
Z2BeckmanCoulter(CoulterCorporation,Miami,FL,
USA).
Analysis of SAM, SAH and total DNA methylation
Cell lysis and macromoleculeprecipitationfor SAM
and SAH analysis were carried out as previously de-
scribed [26]. HPLC measurements were carried out
using a Varian HPLC system (Varian, Walnut Creek,
CA, USA) equipped with a Prostar 210 pump and UV
detector (Prostar 320) set to 254 nm. Remote con-
trol of HPLC system, data acquisition and calcula-
tion of peak areas were performed via computer based
data-system (Varian Star 5.3). Samples were inject-
ed through an injection valve with a 20 µl sample
loop. Mobile phase (NH+
10% MeOH) was pumped at a flow-rate of 0.8 ml/min.
Samples were injected onto the reverse-phase column
HIRPBC18(HIRPB-250A,length250mm,ID4.6mm,
5 µm, Hichrom, Reading, Berkshire, UK).
ForDNAmethylationanalysis,DNAextractionfrom
cell cultures was performed by standard procedures.
Then,100µgofDNAsampleswerehydrolyzedin90%
formicacid(2µl/µgDNA)at170◦Cfor30min. Formic
acid was evaporatedundervacuumand the residue dis-
solved in 0.01 N HCl before HPLC analysis. HPLC
measurements were carried out as for SAM and SAH
but with a mobile phase containing 5% MeOH and
detected at a λ of 280 nm.
StandardcurvesofSAM,SAH,cytosineandmethyl-
cytosine (all from Sigma; St. Louis, MO, USA) dis-
solved in distilled water were run before and after the
experimental samples.
4COO−10 mM, pH 3 and
RNA silencing
Transient transfection of SK-N-BE cells with PS1
silencer RNA (Santa Cruz Biotechnology, Santa Cruz,

Page 3

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
277
CA) was performed as suggested by the manufacturer
using a Random silencer RNA as a control. Trans-
fected and control cells were stopped after 48 hour of
treatment with DM or DDM and analyzed as below
described for gene and protein expression assays.
Quantitative Real-Time PCR
RNA was extracted from cell cultures with the
RNeasy Mini Kit (Qiagen, Milano, Italy); 1 µg of total
RNA was used for cDNA synthesis, with 50 pmol of
oligo-d(T)16and 50 U of M-MuLV reverse transcrip-
tase (Stratagene, La Jolla, CA) at 42◦C for 1h, as in-
dicated by the manufacturer. 1 µg of total cDNA was
used for each real-time reaction; analyses were per-
formedin triplicate for each sample. cDNA was mixed
with 0.6 µl of both forward and reverse primers and
10 µl of 2 x Master mix (DyNAmoTMHS SYBR
GreenqPCR Kit, Finnzymes,Espoo,Finland)toafinal
reaction volume of 20 µl. Reactions were performed
in an Opticon2 DNA Engine (MJ Research, Celbio,
Milano, Italy) with a hot start step of 95◦C for 15 min
followed by 30 to 40 amplification cycles (95◦C, 10 s;
62◦C, 30 s; 72◦C, 30 s), by a melting step from 72◦C
to 95◦C andby a final 72◦C extensionfor 10min. Am-
plification efficiency for each primer pair was deter-
minedbyamplificationofalinearstandardcurve(from
0.16 ng to 100 ng) of total cDNA assessed by ultravi-
olet spectrophotometry. The standard curves showed
good linearity and amplification efficiency (90–99%)
for each primer pair of experimental genes and refer-
ence (β-actin) gene. As negative controls, each sam-
ple was previously run with β-actin primers without
reverse transcription to detect genomic DNA contami-
nation; moreover, blank controls were assayed in each
reaction and for each primer pair to detect DNA con-
taminationofthereagents. Experimentalsampleswere
compared to a gene specific standard curve to deter-
mine the amount of specific cDNA present in the stan-
dard reaction. Standards were prepared from highly
purifiedPCRproductsamplifiedfrompositivecontrols;
concentration and purity of the standards was assessed
by A260/A280spectrophotometric measurement. The
standard curve ranged linearly from 5 attograms (5 ×
10−18) to 5 picograms (5 × 10−12). Total cDNA lev-
elswerestandardizedbynormalizingtoβ-actincontrol
and presented as total levels of cDNA detected (fem-
tograms of specific cDNA/µg of total cDNA). Addi-
tionally, data are presented as n◦of fold increase (ratio
of the experimental gene value/actin gene value) ver-
sus the control sample. GenBank accession numbers,
primer names, sequence and position, expected prod-
ucts and annealing temperatures of primers used for
quantitative real-time PCR assays are listed in Table 1.
Protein analysis
Western blot analysis of total protein extracts was
performed as previously described [10]. Western blot
signals were acquired and analyzed by a Fluor-S den-
sitometer (BIO-RAD, Hercules, CA); optical densities
(O.D.) from at least three different experiments were
calculatedforeachsampleandnormalizedwiththecor-
responding 14,3,3 β signal, then the O.D. ratios were
compared with the control (DM) value assumed as 1.
Total protein extracts were also used for ELISA
test with Aβ1−40 and Aβ1−42 Immunoassay kits
(BioSource International, Belgium); ELISA kits guar-
anteeagoodlinearsensitivityuntil5pg/mL(1–40)and
10 pg/mL (1–42). All measurements were performed
in triplicate.
For co-immunoprecipitation,cells were lysed in 1%
CHAPS-containing buffer (25 mM HEPES, pH 7.4,
150 mM NaCl, 2 mM EDTA) supplemented with pro-
teaseinhibitors. Aftercentrifugationat11,000rpmina
microfuge, supernatants were precleared, immunopre-
cipitatedwithanti-PS1andanalyzedbywesternblotas
described above, using specific antibodies.
Primary antibodies and band sizes are listed in Ta-
ble 2; secondaryantibodieswere from an ECL western
blotting reagents kit (GE Healthcare Europe GmbH).
Sequence specific DNA methylation analysis
HpaII/PCR analysis for PS1 promoter methylation
was performed with the procedure and primers previ-
ously described [10].
Bisulphite analysis of BACE promoter methylation
was performedas previouslydescribed[27]with direct
sequencingofPCRproductsandconsequentevaluation
of cytosine average methylation status. Primers used
for BACE promoter amplification are listed in Table 1.
Statistical analysis
One-way ANOVA was computed and Bonferroni
post test was used to evaluate any significant (P <
0.05) difference reported in this paper. All the assays
were repeated at least three times; all histograms show
the mean value ± s.e.m. When the differencesbetween
experimental conditions are shown as the “number of
fold”, s.e.m. was consequently transformed.

Page 4

278
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
Table 1
Characteristics of oligonucleotides used as PCR primers
Gene
name
β−actin
Accession
number
M10277
Primer namePRIMER SEQUENCE (5’–3’) Primer
position
1566–1584
1659–1678
533–553
626–645
854–873
944–963
1344–1363
1401–1420
1686–1705
1787–1806
653–672
761–780
237–256
326–345
774–793
872–891
Melt.
T◦
61
60
59
59
60
60
62
62
60
60
62
62
62
62
60
60
Ampl.
size
94bp
Anneal.
T◦
62◦C HSBACTL1
HSBACTR1
HSPS1L1
HSPS1R1
HSPS2L1
HSPS2R1
HSAPPL1
HSAPPR1
HSBACEL3
HSBACER3
HSAPH1AL1
HSAPH1AR1
HSPEN2L1
HSPEN2R1
HSNCTL3
HSNCTR3
CAACCGCGAGAAGATGACC
AGAGGCGTACAGGGATAGCA
GGTCGTGGCTACCATTAAGTC
GCCCACAGTCTCGGTATCTT
CGCTGCTACAAGTTCATCCA
TGAGCACTTCCCCAAGGTAG
GAACTACATCACCGCTCTGC
CGCGGACATACTTCTTTAGC
AACGAATTGGCTTTGCTGTC
AGCCACAGTCTTCCATGTCC
GAGACGGTACTGGGCTTTGG
GACTGCATAGATGGGCAGCA
ATTGAACCTGTGCCGGAAGT
GCCTCTCGGAAGAACCAGAA
CACTATGTGCCATGCAGCTC
GGTTGATGCTGAAGGTGCTT
PS1 L42110 94bp62◦C
PS2NM 00044791bp 62◦C
AβPPY0026477bp 62◦C
BACE AF201468102bp 62◦C
Aph1A NM 016022
109bp62◦C
Pen2NM 172341
90bp62◦C
Nicastrin NM 015331
99bp 62◦C
Primers for BACE bisulphite analysis:
BACE AY542689HSBACEbisL1
HSBACEbisR1
TGACAGACGGGAGGTGTG
CCCACGTCCGTCCCT
3555–3572
4172–4186
59
59
618bp60◦C
Table 2
Characteristics of antibodies
Manufacturer
monoclonal Chemicon International, Temecula, CA
monoclonalChemicon International, Temecula, CA
polyclonalSanta Cruz Biotechnology, Santa Cruz, CA
polyclonalSanta Cruz Biotechnology, Santa Cruz, CA
polyclonal Santa Cruz Biotechnology, Santa Cruz, CA
polyclonalCovance, Berkeley, CA
polyclonalSanta Cruz Biotechnology, Santa Cruz, CA
monoclonalChemicon International, Temecula, CA
polyclonalSanta Cruz Biotechnology, Santa Cruz, CA
polyclonal Calbiochem, Germany, EU
polyclonalCovance, Berkeley, CA
polyclonalCovance, Berkeley, CA
Protein
PS1N-t
PS1C-t
PS2
BACE
NCT
Aph1
14,3,3β
AβPP
Pen2
Pen2
Pen2
Pen2
AntibodyBand Size
32 kDa
18–20 kDa
50 kDa
70 kDa
110 kDa
30 kDa
30 kDa
110 kDa
12 kDa
12 kDa
12 kDa
12 kDa
MAB1563
MAB5232
sc-1456
sc-10055
sc-14369
PRB-550P
sc-629
MAB348
Sc-30232
NE 1008
PRB-560C
PRB-599C
RESULTS
Cell growth
Both neuroblastoma (Fig. 1a) and glioblastoma
(Fig. 1b) cells showed a slight, not significant, de-
crease in growth rate when fed with deficient media.
SAM added to culture media induced instead a signifi-
cant decrease after 144 hours of culture (medium with
SAM compared to respective medium without SAM)
confirming SAM cytostatic properties. General slow
growth rate was due to the use of differentiating me-
dia (1% FCS) in order to reproduce a neuron-like or
astrocyte-likemorphologyasconfirmedbyobservation
with optical microscopy.
Methylation potential
HPLC analysis revealed a marked decrease of intra-
cellular SAM levels in cells (both neuroblastoma and
glioblastoma) fed with deficient media; as expected,
exogenous SAM administration induced an approxi-
mately two-fold increase of intracellular SAM both in
controlanddeficientmedia(Fig.2aandbrespectively).
SAHlevelsweretoolowtobedetectedundermostcul-
ture conditions; it was only detectable in cells fed with
DDM (data not shown). This indicates, however, that
vitamin B deprivation can decrease SAM/SAH ratio,
i.e., decrease MP; exogenous SAM is able (at least in
DDM) to increase MP (since SAH was not detectable
in DDM+SAM). Variations of SAM levels in glioblas-
tomacellsculturedwithdeficientmedia(Fig.2b),were
lower than in neuroblastomacells, thereforenonstatis-

Page 5

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
279
Fig. 1. Effect of the HCY cycle alteration on cell growth. a) Growth
rate of SK-N-BE neuroblastoma cells in control (DM) or deficient
media(DDM,-FB,-B6), withorwithoutexogenous100µMSAM.b)
Growth rate of A172 glioblastoma cells in control (DM) or deficient
media (DDM, -FB, -B6), with or without exogenous 100 µM SAM.
Media with SAM vs. media without SAM, p < 0.05.
tically significant although relevant (except for DDM
vs. DM).
PS1 and BACE overexpression in neuroblastoma
Quantitative analysis of gene expression by Real-
Time PCR revealed that all the genes investigated, in-
volved in Aβ processing, were generally expressed
at lower levels in glioblastoma than in neuroblastoma
(Figs 3 and 6). AβPP expression showed some vari-
ability in both cell lines, but this modulation seemed
to be independent of culture conditions (Fig. 3a and
e). PS2 also was not modulated by vitamin B depri-
vation or by exogenous SAM administration, in either
cell lines (Fig.3d and h). Instead, PS1 and BACE were
up-regulatedin neuroblastomaSK-N-BEcells fedwith
vitamin B deprived media, whereas SAM was able to
revert the up-regulation; no effect was detectable in
A172 glioblastomacells (Fig. 3b–c and f–g). In partic-
ular, whereas PS1 up-regulation in SK-N-BE was evi-
dentbothintotallyandpartiallydeficientmediaandthe
down-regulationwhen SAM is added to complete DM
Fig. 2. SAM level modulation in deficient media. a) Relative SAM
concentration in SK-N-BEneuroblastoma cells treated with deficient
media with or without 100 µM SAM vs. control medium (DM)
assumed as 1. b) Relative SAM concentration in A172 glioblastoma
cells treated with deficient media with or without 100 µM SAM vs.
control medium (DM) assumed as 1. *: p < 0.05.
wasslightbutsignificant(Fig.3b),BACEup-regulation
was evident only in DDM and –FB media and no re-
pressionwas observedin DM+SAM (Fig.3c). This re-
sultsuggeststhattheexistenceofdifferentmechanisms
of regulation for the two genes.
Westernblotanalysisontotalcellextractsconfirmed
gene expression data (Fig. 4): no differences were ob-
served in protein levels in glioblastomacells (Fig. 4m–
r),butincreasesinPS1fulllength(FL),PS1C-terminal
fragment (C-term), PS1 N-terminal fragment (N-term)
and in BACE (Fig. 4a–c and g–l), were seen in neurob-
lastoma cells. Surprisingly, unlike in gene expression,
BACE resulted up-regulated also in –B6 medium.

Page 6

280
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
Fig. 3. PS1 and BACE mRNAs are modulated only in neuroblastoma. a-d) Quantitative Real-Time PCR expression analysis in SK-N-BE
neuroblastoma cells of AβPP, PS1, BACE and PS2 mRNAs respectively.
glioblastoma cells of AβPP, PS1, BACE and PS2 mRNAs respectively. Results are presented as femtograms of specific mRNA per µg of total
cDNA. Additionally, data are expressed as relative expression, with control value assumed as 1, in the legend box on the right of each panel.
Significant differences are indicated below legend boxes.
e-h) Quantitative Real-Time PCR expression analysis in A172

Page 7

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
281
Fig. 4. PS1 and BACEpeptides are modulated only in neuroblastoma. a-f) Representative western blot analysis of full length and C-terminal PS1,
N-terminal PS1, BACE, AβPP, PS2 and 14,3,3 β respectively, in SK-N-BE neuroblastoma cells; molecular weights are indicated on the right
of each panel. g-l) Histograms showing relative expression of Full Length, C-terminal, N-terminal PS1, and BACE peptides respectively, with
control value (DM) assumed as 1; data represent Optical Density (O.D.) values of several replicates for each sample normalised with O.D. value
of relative signals for 14,3,3 β peptide, **: p < 0.01. m-r) Representative western blot analysis of full length and C-terminal PS1, N-terminal
PS1, BACE, AβPP, PS2 and 14,3,3 β respectively, in A172 glioblastoma cells; molecular weights are indicated on the right of each panel.

Page 8

282
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
Fig. 5. Aβ1−40and1−42peptides are modulated by HCY cycle alteration. a) ELISA test for Aβ1−40in supernatants of neuroblastoma (left
side of panel) and glioblastoma (right side of panel) cell cultures after 96 hour. b) ELISA test for Aβ1−42in supernatants of neuroblastoma after
96 hour; concentration in glioblastoma cell cultures (right side of panel) was below test sensitivity. c) Aβ1−40vs. Aβ1−42ratio. Additionally,
data are presented as relative quantity, with control value assumed as 1, in the legend box on the right of each panel. Significant differences are
indicated below legend boxes.

Page 9

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
283
Fig. 6. Pen2, Aph1A and NCT expression in neuroblastoma and glioblastoma. a-c) Quantitative Real-Time PCR expression analysis in SK-N-BE
neuroblastoma cells of Pen2, Aph1A and NCT mRNAs respectively. d-f) Quantitative Real-Time PCR expression analysis in A172 glioblastoma
cells of Pen2, Aph1A and NCT mRNAsrespectively. Results are presented as femtograms of specific mRNA per µg of total cDNA. Additionally,
data are expressed as relative expression, with control value assumed as 1, in the legend box on the right of each panel.
Amyloid-β processing is regulated by methylation
Aβ1−40and Aβ1−42were both increased when neu-
roblastoma cells were treated with vitamin B defi-
cient media; the highest concentrations were reached
in DDM, whereas no differences were detectable be-
tween –FB and –B6 media. SAM was able not only
to revert Aβ overproduction, but also to downregulate
it when compared with DM (Fig. 5a and b, left side
of panels). Glioblastoma cells showed markedly low-
er levels of Aβ with respect to neuroblastoma cells, it
was even below test sensitivity threshold for Aβ1−42
peptide, without any evident modulation (Fig. 5a and
b, right side of panels). Aβ1−42 was less abundant
than Aβ1−40peptide, but 1-40/1-42 ratio appeared to
be similar in all tested conditions (Fig. 5c).
PS1 overexpression enhances γ-secretase assembly
Since it has been reported that overexpressed trans-
genic PS1 holoprotein is rapidly degraded without af-
fectingγ-secretaseactivity [28–30],we wonderedhow
thisfindingwascompatiblewiththeobservedAβ over-
productionin our experiments. First, we analysed mR-
NA levels of Aph1A, Nicastrin (NCT) and Pen2 (the
threecomponentsofγ-secretasecomplextogetherwith
PS1). Gene expression appeared un-modulated both
in SK-N-BE (Fig. 6a–c) and in A172 (Fig. 6d–f) cells
grown with complete or deficient media; also these
three genes, in glioblastoma cells showed lower levels
of mRNAs. In order to ascertain if these factors were
regulated at a post-transcriptional level, we analysed
the relative proteins by western blot in neuroblastoma
cells. Also in this case, we could not detect evidence
any modulation of Aph1A, NCT (Fig. 7a–c) and Pen2
(data not shown) peptides. Thus, the only possibili-
ty left was that the “excess” of PS1 might be able to
recruit more “partner” molecules from the basal pool
of Aph1A, NCT and Pen2. To test this possibility we
performed a co-immunoprecipitation assay using the
anti-PS1 (C-terminal) antibody, followed by western
blot with PS1, Aph1A, NCT (Fig. 7d–g) or Pen2 anti-

Page 10

284
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
Fig. 7. Aph1A and NCT peptides association with PS1 in neuroblastoma. a-c) Representative western blot analysis of Aph1A, NCT and
14,3,3 β respectively, in SK-N-BE neuroblastoma cells. d-f) Western blot analysis of PS1, Aph1A and NCT respectively on protein fraction
immunoprecipitated with anti-C terminal PS1 (left panels) and on related supernatant fractions (right panels) in SK-N-BE neuroblastoma cells
treated with control medium (DM) or medium deprived of folate, B12 and B6 (DDM); total lysates and fraction immunoprecipitated with
pre-immune serum were used as positive and negative control respectively. g) Western blot analysis of 14,3,3 β peptide was used to normalise
supernatant fraction. Molecular weights are indicated on the right of each panel.
bodies. Data clearly demonstrated an increment of co-
immunoprecipitatedpeptidesinDDMandaconcurrent
decrease of unbound peptides in supernatants. More-
over, western blots for NCT showed an upper (about
120 kDa) band relative to fully glycosylated protein
(Fig. 7b) that appeared to increase in media without
SAM. This finding shows that over-expressed PS1 can
promote recruitment of Aph1A, NCT and Pen2 (that
displayedunvariedconcentrationincells)toincreaseγ-
secretase complex assembly, maintaining the (hypoth-
esised) 1:1:1:1 stoichiometric ratio between the four
peptides.
Regarding Pen2 peptide analysis, we could only ob-
tain high molecular weight complex signals (30-200
kDa), instead ofthe expected12kDa band; this despite
the use of different lysis buffer, different denaturation
and electrophoresis conditions and the use of four dif-
ferent antibodies recognizingN-, C-terminal and inter-
nal epitopes (also in different cell lines and in mice;
data submittedelsewhere). For these reasons, although
wethinkthatthebandscorrespondedtohighmolecular
weightcomplexescontainingPen2,wecouldnotascer-
tain thecorrectidentityof detectedbands; we therefore
chose not to show results concerning Pen2 protein.
Methylation pattern of PS1 and BACE promoters
HpaII/PCR analysis of the PS1 5’-flanking region
confirmed the previously observed demethylation of a
specific CCGG sequence in neuroblastoma cells after
72 hours in medium deficient of folate and Vitamin
B12 [10] (Fig. 8a). Moreover, the new data indicates
thatdeprivationofvitaminB6hadasimilareffectonthe
methylation pattern whereas simultaneous deprivation
offolate,B12andB6inducedamoreevidentdemethy-
lation (Fig. 8a, lane “DDM”); as expected, SAM was
able to restore DNA methylation similar to control.
Glioblastoma cells, that showed no modulation of PS1
gene expression, were not affected by folate, vitamin
B12 and/or vitamin B6 deprivationwith respect to PS1
promoter methylation (Fig. 8b); this finding indirect-
ly confirms that overexpression of PS1 in neuroblas-
toma is a direct consequence of the observed DNA hy-
pomethylation in these cells.
The HpaII/PCR technique was not suitable for anal-
ysis ofBACE methylationbecauseof thehighCG con-
tent of gene promoter; so we decided to study BACE
DNA methylation through bisulfite modification tech-
nique. In contrast to what we expected, analysis of a

Page 11

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
285
Fig.8. PS1promoterandtotalDNAmethylation ismodulated onlyinneuroblastoma. a-b)HpaII/PCR analysisofDNAmethylation on5’-flanking
region of PS1 gene in SK-N-BE and A172 cells respectively; on the right of panels are indicated the molecular weights of analysed sequences in
5’-flanking region (containing a CCGG site) and of control region (without any CCGG site) used as Internal Standard; MM = molecular weight
marker. c) HPLC analysis of total genomic DNA methylation indicated as methyl-Cytosine vs. total Cytosine ratio in neuroblastoma (left) and
glioblastoma (right) cells; **: p < 0.05.
618 bp sequence in 5’-flanking region of the gene, re-
vealed a basal hypomethylated status without any fur-
ther demethylation in DDM medium (both in neurob-
lastomaandinglioblastoma)andwithoutanyhyperme-
thylation in SAM treated cells in neuroblastoma (data
not shown).
HPLC analysis of global DNA methylation demon-
strated that at genomic level, only neuroblastoma cells
showed modulation of methyl-Cytosine (mC) con-
tents (Fig. 8c); indeed, mC/totalC ratio increased in
DM+SAManddecreasedin DDMandin DDM+SAM
in comparison to controls.
BACE upregulation is independent of PS1
The observationthat BACE upregulationwas not di-
rectly induced by hypomethylation of its promoter led
us to hypothesise the involvement of some other regu-
lating factor. Among all the countless possible factors,
we wanted to verify whether PS1 upregulation itself
could be the inducer of BACE. In order to test this hy-
pothesis, we performed a “RNA silencing” assay us-
ing a PS1 silencer RNA (siRNA): neuroblastoma cells
transientlytransfectedwithPS1orrandomsiRNAwere
incubated for 48 hours with DM or DDM, then RNA
and proteins were extracted. PS1 siRNA was able to
repress PS1 expression in DM and to inhibit PS1 over-
expression in DDM, whereas control (random) siRNA
causedonlyaslightinhibitiononPS1overexpressionin
DDM (Fig. 9a and d). In contrast, neither basal BACE
levels(inDM),norBACEupregulation(inDDM)were
affected by PS1 silencing (Fig. 9b and e). These re-
sult (together with BACE methylation) indicated that
BACE was indirectly regulated by MP but that PS1
upregulation had no effect on BACE increase.
We also tested the PS2 expression pattern to deter-
mine whether it could act in the place of PS1; we ob-
servedthatevenwhenPS1 was silenced, PS2 RNA and
peptide levels remained unchanged (Fig. 9c and f).
DISCUSSION
The aim of this work was to explore the correla-
tion between alterations of HCY cycle and pathogenic
mechanisms of AD, with particular attention to poten-
tialdifferencesbetweenneuronsandglia. Althoughthe
useofneuroblastomaandglioblastomacellscouldlim-
it the applicationof ourconclusionto a normalsystem,
at the same time it offers an “easy-to-handle” and well
characterised model of neural cells. Growing cells in
lowseruminthepresenceofretinoicacid,allowedusto
obtaincellswithadifferentiatedphenotype,i.e.,similar
to neurons or astrocytes [31]. This approach provides
some advantages in consideration of the difficulty in
maintaining primary neuronal cultures for a long time

Page 12

286
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
Fig. 9. BACE upregulation is independent of PS1. a-c) Quantitative Real-Time PCR expression analysis in SK-N-BE neuroblastoma cells of
PS1, BACE and PS2 mRNAsrespectively, in control (DM) and deficient (DDM) medium in cells transiently transfected with PS1 silencer mRNA
(PS1 siRNA) or with control random silencer mRNA (random siRNA). Results are presented as femtograms of specific mRNA per µg of total
cDNA. Additionally, data are expressed as relative expression, with control value assumed as 1, in the legend box on the right of each panel. d-g)
Representative western blot analysis of Full Length and C-terminal PS1, BACE, PS2 and 14,3,3 β respectively, in SK-N-BE neuroblastoma cells
transiently transfected with PS1 siRNA or with control random siRNA; molecular weights are indicated on the right of each panel.
and in obtaining pure neuronal or glial cells cultures.
Our data regarding PS1 and BACE up-regulation were
also confirmed in an AD transgenic mouse model with
vitamin B deficient diet (data submitted elsewhere); in
those experiments, we demonstrated that diet-induced
hyperhomocysteinemiawasresponsibleforanincrease
in intra- and extra-cellular Aβ deposition with conse-
quent cognitive impairment.
In our opinion the use of vitamin B deficient media
could represent a useful model to study the molecular
effectsofMPimpairment,sinceit is mediatedbyphys-
iological HCY metabolism, unlike the models based
on gene knockout or exogenous HCY administration.
Moreover, exogenous HCY could be rapidly metab-
olized and therefore be ineffective. Inhibition of the
HCY cycle through vitamin deprivation did not cause
a marked decrease of intracellular SAM concentration,
which increasedupon additionof exogenousSAM. On
the contrary, MP impairment is due to the increase of
SAH levels and the consequentdecrease of SAM/SAH
ratio. In this set of experiments, using cell cultures, we
were able to detect SAH only in cells grown in DDM
mediumbecauseoftheverylowlevelsinculturedcells;
this finding per se is sufficient to demonstrate that vi-

Page 13

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
287
tamin B deprivation induces MP impairment. Howev-
er, our experiments in mice, where SAH levels were
detectable both in serum and in brain lysates also in
controls, completely confirmedthe indications showed
here (data submitted elsewhere).
Another interesting observation on the HCY cycle
andmethylationreactionsconcernsthedifferenteffects
of vitamin B deprivationon PS1 promoter hypomethy-
lation. HpaII/PCR analysis showed that “partial” de-
privations of folate and B12 (FB medium) or B6 (B6
medium) induced hypomethylation to a lower extent
than “complete” deprivation (folate, B12 and B6 to-
gether; DDM medium). This data agrees with the role
of these cofactors in the HCY cycle: Folate and B12
deprivation inhibit SAM synthesis but still allow HCY
scavenging through the trans-sulfuration pathway, pre-
venting excessive accumulation of HCY and, conse-
quently, SAH; B6 deprivation inhibits efficient HCY
removal inducing its accumulation, but MP homeosta-
sis is partially maintained by HCY remethylation to
SAM; in contrast, deprivation of all three co-factors
inhibits both HCY remethylation and removal, with a
consequent higher impact on MP.
Whereas our data clearly demonstrate that PS1 gene
expression is directly regulated by DNA methylation,
BACE appears to be regulatedby indirect factors. This
hypothesis is supported by indirect and direct observa-
tions: BACE expression is up-regulated in DDM but
seems unaffected by SAM in DM, leading to the hy-
pothesis of a basal gene expression level that can be
incrementally increased by vitamin B deprivation but
cannot be reduced by methylation (i.e., adding SAM).
DNA methylation data confirm this idea, since the
BACE promoter was largely hypomethylatedunder all
tested conditions, with no change in the DNA methy-
lation pattern when DDM or SAM were used. The
contrasting observation of gene up-regulation in hy-
pomethylatingconditions(DDM) but in presenceof an
unchanged promoter methylation pattern, opens a new
scenario for BACE regulation. Thus, the effector of
BACE over-expression might be a transcription factor
in trans, i.e., a differentmethylation-dependentregula-
torygene. Theidentificationofthis master-genewould
require a wide-spectrum analysis technique; neverthe-
less, we wanted to test the hypothesis whether PS1
could be involved in this process, in consideration of
the fact that PS1 and BACE co-ordinately participate
in Aβ processing.Moreover, PS1 is considered a
regulator of signal transduction, due to its role in the
generation of intracellular signalling peptides like in-
tracellular sAβPP (after α-secretase cleavage), NICD
(Notch intracellular domain) and E-cadherin (through
the ε-secretase pathway) [32–36]. Recent papers re-
port the absence of BACE up-regulation under PS1
over-expression conditions, but using a substantially
different model based on transgenic overexpression of
mutated PS1 [9]. Our project was to test BACE ex-
pression in neuroblastoma cells under hypomethylat-
ing conditions (DDM medium; where both PS1 and
BACE were overexpressed) but in the presence of PS1
silencermRNA;inthatwaywecouldascertainifBACE
overexpressionwas dependent on PS1 or on some oth-
er methylation-dependent process. Our results indi-
cate that in the presence of PS1 siRNA, BACE con-
tinued to be up-regulated, indicating that hypomethy-
lation condition was sufficient to over-express BACE,
independent from promoter methylation or PS1 over-
expression. Thus, the BACE regulation factor(s) still
remain to be identified.
Interestingly, experiments with siRNA also indicat-
ed that, in our cellular model, PS2 is not a substitute
molecule for PS1; in fact, PS2 gene and proteins levels
remained unchanged in both control and deficient me-
diawhenPS1wasinhibitedbysiRNA;moreover,itwas
absent from Pen2 co-immunoprecipitation complexes
(data not shown).
Almost all the experiments reported here display a
different response in neuroblastoma and glioblastoma
cells to alteration of the HCY cycle. Although there
was a similar initial response to vitamin B deprivation
(impairment of MP in deprived media) and SAM ad-
ministration (increased SAM uptake and consequent
cytostatic effect), glioblastoma cells did not show any
change in DNA methylation, expression of AD genes
and Aβ processing. Indeed, both total DNA and PS1
promoter methylation were modulated only in neurob-
lastoma cells. Therefore,PS1, BACE and Aβ overpro-
duction were not found in glioblastoma cells. More-
over,Aβ levels werelower inglioblastomaalso in con-
trol condition. This result demonstrates that Aβ pro-
cessing is not a major pathway in glia and couldjustify
the lack of a methylation-basedmechanism to regulate
Aβ processing, similar to those observed in neuroblas-
toma. On this topic, DNA methylation analysis allows
us to discuss the biological significance of total and
sequence-specific methylation. Differences observed
in total methyl-Cytosine content in neuroblastoma do
not necessarily indicate the presence of a sequence-
specific methylation, since it has been demonstrated
(particularlyin a model of folate deficiency)that varia-
tioninthemethylationpatternofspecificgenescanalso
occurintheabsenceofgeneralisedchangesingenomic

Page 14

288
A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
DNA methylation [24]. Our data on neuroblastoma,
showingbothalterationofPS1 (sequence-specific)and
genomic (non-specific) methylation could be charac-
teristic of this cancer cell line; however, we already
showed, by a cDNA array approach, that modulation
of genomic methylation is inconsistent with wide gene
expression modulation. Indeed, we observed that on-
ly few genes (about 1%) were modulated in a CNS
array in neuroblastoma cells treated with exogenous
SAM [25].This is probably due to modulation of
methylationin “non-functional”sequencesof genomic
DNA, which can be responsible for alterations in total
methylation without functional changes in regulatory
or coding regions. We can reasonably exclude that al-
teration of the HCY cycle featured in our experiments
could cause modulation of more than the few observed
genes. In fact, besides the cited experiments with cD-
NA arrays regarding CNS genes, we only found PS1
and BACE to be modulated among all the analysed
genesinvolvedinAβ processing(i.e.,AβPP,PS1,PS2,
BACE, Notch1, Adam10, TACE, Aph1A, NCT, Pen2,
Tau, Neprilisin) [10,37].
One of the main findings emergingfrom this work is
that PS1 upregulation influences assembly and/or sta-
bility of γ-secretase complex, in presence of constant
mRNA and protein levels of Aph1A, NCT and Pen2.
Several papers described how these four peptides asso-
ciate to form the complex; it is still to be ascertained if
active complex is composed by one or multiple (prob-
ably four) tetrameric subunits, but a hypothetical sto-
ichiometric 1:1:1:1 (PS1:Aph1A:NCT:Pen2) ratio ap-
pears possible [38–43]. Some points in the elucidation
of γ-secretase assembly remain to be clarified, such as
theeffectofPS1excessonγ-secretaseactivity; indeed,
data proving its ability to stabilise the complex [44]
contrast with its rapid degradation. Our experiments
showed an apparently paradoxical contrast, since up-
regulation of PS1 per se led, to Aβ overproduction.
The question was how PS1 overexpression could pro-
mote amyloid processing if Aph1A, NCT and Pen2
peptide remained constant. Our hypothesis, verified
by co-immunoprecipitationassays, was that the excess
of PS1 is able to recruit, from the basal pool, more
Aph1A, NCT and Pen2 peptides in γ-secretase com-
plex; in fact, these proteins appear increased in DDM
in the co-immunoprecipitated fraction, but decreased
in the supernatantfraction. An alternative explanation,
compatiblewithourresults,wouldbethattheexcessof
PS1 stabilises the complex by decreasing its turnover.
However, stabilisation of the complex is confirmed by
theobservedincreasein fullyglycosylatedNCTin me-
dia without SAM. Moreover, BACE over-expression
and consequent β-secretase cleavage could have some
regulatory function or could be a synergistic factor for
increased γ-secretase activity. Of course, other mod-
els of PS1 over-expressionmight not show this kind of
mechanism, since the HCY cycle alteration could also
affectthestatusofproteinandlipidmethylation. Inpar-
ticular, we are interested in studying cholesterol, lipid
methylationand membranefluidity in our model in the
near future. This interest arises from the observations
that cholesterol may play a major role in AD [45,46]
andthat γ-secretaseproteolyticactivity seems to be in-
fluencedbymembranefluidity[47]. Rapiddegradation
of the PS1 holoprotein observed in a PS1 overexpres-
sion model [28], is probably inhibited by a favourable
membrane conformation in our model. Since it is well
known that both cholesterol and lipid methylation are
modulating factors of membrane fluidity, this area of
study could give a unifying view of the role of HCY
and cholesterol in AD.
The etiological basis of sporadic AD appears to be
very complex and multi-factorial; this work lays the
foundation for a causal theory about the molecular ba-
sis of HCY toxicity in AD, although more studies fo-
cusing on the differential role of neurons and glia and
other possible effects of HCY are needed. Moreover,
our observations on γ-secretase regulation and the dif-
ferential role of neuron and glia provide new bricks
forbuildinga morecomprehensivetheoryonAD onset
and progression.
ACKNOWLEDGEMENTS
WewouldliketothankDr.VincenzinaNicoliaforthe
skilful technical assistance and Prof. Rinaldo Marini
Bettolo for his kind support. This work was supported
by MIUR grants (Ateneo 2004 and 2005 and FIRB
2003).
DISCLOSURE STATEMENT
All authors disclose any actual or potential conflicts
of interest including any financial, personal or other
relationships with other people or organizations that
could inappropriately influence their work.

Page 15

A. Fuso et al. / γ−Secretase is Differentially Modulated by Alterations of Homocysteine Cycle
289
References
[1]D.M. Kado, A.S. Karlamangla, M.H. Huang, A. Troen, J.V.
Rowe, J. Selhub and T.E. Seeman, Homocysteine versus the
vitamins folate, B6, and B12 as predictors of cognitive func-
tion and decline in older high-functioning adults: MacArthur
Studies of Successful Aging, Am J Med 118 (2005), 161–167.
J.A. Luchsinger and R. Mayeux,
Alzheimer’s disease, Lancet Neurology 3 (2004), 579–587.
M. Malaguarnera, R. Ferri, R. Bella, G. Alagona, A.
Carnemolla and G. Pennisi, Homocysteine, vitamin B12 and
folate in vascular dementia and in Alzheimer disease, Clin
Chem Lab Med 42 (2004), 1032–1035.
P.Quadri, C.Fragiacomo, R.Pezzati, E.Zanda, G.Forloni, M.
Tettamanti and U. Lucca, Homocysteine, folate, and vitamin
B-12 in mild cognitive impairment, Alzheimer disease, and
vascular dementia, Am J Clin Nutr 80 (2004), 114–122.
P.I.Ho, D.Ortiz, E.Rogers and T.B.Shea, Multiple aspects of
homocysteine neurotoxicity: glutamate excitotoxicity, kinase
hyperactivation and DNA damage, J Neurosci Res 70 (2002),
694–702.
I.I. Kruman, T.S.Kumaravel, A. Lohani, W.A.Pedersen, R.G.
Cutler, Y. Kruman, N. Haughey, J. Lee, M. Evans and M.P.
Mattson, Folic acid deficiency and homocysteine impair DNA
repair in hippocampal neurons and sensitize them to amy-
loid toxicity in experimental models of Alzheimer’s disease, J
Neurosci 22 (2002), 1752–1762.
K.S. McCully, Homocysteine and vascular disease, Nature
Med 2 (1996), 386–389.
A.M. Troen, The central nervous system in animal models
of hyperhomocysteinemia, Prog Neuropsychopharmacol Biol
Psychiatry 29 (2005), 1140–1151.
J. Pacheco-Quinto, E.B. Rodriguez de Turco, S. Derosa,
A. Howard, F. Cruz-Sanchez, K. Sambamurti, L. Refolo,
S. Petanceska and M.A. Pappolla, Hyperhomocysteinemic
Alzheimer’s mouse model of amyloidosis shows increased
brain amyloid beta peptide levels, Neurobiol Dis 22 (2006),
651–656.
A. Fuso, L. Seminara, R.A. Cavallaro, F. D’anselmi and S.
Scarpa, S-adenosylmethionine/homocysteine cycle alterations
modify DNAmethylation status with consequent deregulation
of PS1 and BACE and beta-amyloid production, Mol Cell
Neurosci 28 (2005), 195–204.
P.K. Chiang, Biological effects
adenosylhomocysteine hydrolase, PharmacolTher77(1998),
115–134.
M.A. Medina, J.L. Urdiales and M.I. Amores-Sanchez, Roles
of homocysteine in cell metabolism: old and new function,
Eur J Biochem 268 (2001), 3871–3882.
P.K. Chiang, R.K. Gordon, J. Tal, G.C.Zeng, B.P. Doctor, K.
Pardhasaradhi and P.P. McCann, S-Adenosylmethionine and
methylation, FASEB J 10 (1996), 471–480.
M.E. Lee and H. Wang, Homocysteine and hypomethylation.
A novel link to vascular disease, Trends Cardiovasc Med 9
(1999), 49–54.
P.A. Jones, DNA methylation and cancer, Oncogene
(2002), 5358–5360.
M. Ronemus and R. Mertienssen, RNA interference: methy-
lation mystery, Nature 433 (2005), 472–473.
A.P. Wolffe, P.L. Jones and P.A. Wade, DNA demethylation,
Proc Natl Acad Sci USA 96 (1999), 5894–5896.
T. Bottiglieri and K. Hyland, S-adenosylmethionine levels in
psychiatric and neurological disorders: a review, Acta Neurol
Scand Suppl 154 (1994), 19–26.
[2] Dietary factors and
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]of inhibitorsof S-
[12]
[13]
[14]
[15]
21
[16]
[17]
[18]
[19] B.P. Kennedy, T. Bottiglieri, E. Arning, M.G. Ziegler, L.A.
Hansen and E. Masliah, Elevated S-adenosylhomocysteine in
Alzheimer brain: influence on methyltransferases and cogni-
tive function, J Neural Transm 111 (2004), 547–567.
R.E. Tanzi, The synaptic Abeta hypothesis of Alzheimer dis-
ease, Nature Neurosci 8 (2005), 977–979.
F.Antequera and A.P.Bird, Number of CpGislands and genes
in human and mouse, Proc Natl Acad Sci USA 90 (1993),
11995–11999.
A.P. Bird, CpG-rich islands and the function of DNA methy-
lation, Nature 321 (1986), 209–213.
S.F. Chaumenkowitz, J. Selhub, P. Bagley, N. Maeda, M.R.
Nadeau, D.E. Smith and S.W. Choi, In the cystathionine beta-
synthase knockout mouse, elevations in total plasma homo-
cysteineincreasetissueS-adenosylhomocysteine, butrespons-
es of S-adenosylmethionine and DNA methylation are tissue
specific, J Nutr 132 (2002), 2157–2160.
M.S. Jhaveri, C. Wagner and J.B. Trepel, Impact of extracel-
lular folate levels on global gene expression, Mol Pharmacol
60 (2001), 1288–1295.
R.A. Cavallaro, A. Fuso, F. D’Anselmi, L. Seminara and S.
Scarpa, The effect of S-adenosylmethionine on CNS gene
expression studied by cDNA microarray analysis, J Alz Dis 9
(2006), 415–419.
A.Fuso,R.A.Cavallaro, L.Orr` u,F.R.Buttarelli andS.Scarpa,
Gene silencing by S-adenosylmethionine in muscle differen-
tiation, FEBS Lett 508 (2001), 337–340.
A. Fuso, S. Scarpa, F. Grandoni, R. Strom and M. Lucarel-
li, A reassessment of semiquantitative analytical procedures
for DNA methylation: comparison of bisulfite- and HpaII
polymerase-chain-reaction-based methods, Analyt Biochem
350 (2006), 24–31.
W.T. Kimberly and M.S. Wolfe, Identity and function of
gamma-secretase, J Neurosci Res 74 (2003), 353–360.
T.Ratovitski, H.H.Slunt, G.Thinakaram, D.L.Price S.SSiso-
dia and D.R. Borchelt, Endoproteolytic processing and stabi-
lization of wild-type and mutant presenilin, J Biol Chem 272
(1997), 24536–24541.
G. Thinakaram, D.R. Borchelt, M.K. Lee, H.H. Slunt, L.
Spitzer, G. Kim, T. Ratovitsky, F. Davenport, C Nordstedt,
M. Seeger, J. Hardy, A.I. Levey, S.E. Gandy, N.A. Jenkins,
N.G. Copeland, D.L. Price and S.S. Sisodia, Endoproteolysis
of presenilin 1 and accumulation of processed derivatives in
vivo, Neuron 17 (1996), 181–190.
V. Pijuan-Thompson, J.R. Grammer, J. Stewart, R.L. Silver-
stein, S.F. Pearce, G.P. Tuszynski, J.E. Murphy-Ullrich and
C.L. Gladson, Retinoic acid alters the mechanism of attach-
ment of malignant astrocytoma and neuroblastoma cells to
thrombospondin-1, Exp Cell Res 249 (1999), 86–101.
W.AnnaertandB.DeStrooper, Presenilins: molecularswitch-
es between proteolysis and signal transduction, Trends Neu-
rosci 22 (1999), 439–443.
P. Cupers, I. Orlans, K. Craessaerts, W. Annaert and B. De
Strooper, The amyloid precursor protein (APP)-cytoplasmic
fragment generated by gamma-secretase is rapidly degraded
but distributes partially in a nuclear fraction of neurones in
culture, J Neurochem 78 (2001), 1168–1178.
B. De Strooper, W. Annaert, P. Cupers, P. Saftig, K. Craes-
saerts, J.S. Mumm, E.H.Schroeter, V. Schrijvers, M.S .Wolfe,
W.J. Ray, A. Goate and R. Kopan, A presenilin-1-dependent
gamma-secretase-like protease mediates release of Notch in-
tracellular domain, Nature 398 (1999), 512–522.
G. Serban, Z. Kouchi, L. Baki, A. Georgakopoulos, C.M.
Litterst, J. Shioi and N.K. Robakis, Cadherins mediate both
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]