Content uploaded by Raúl González-Domínguez
Author content
All content in this area was uploaded by Raúl González-Domínguez on Oct 02, 2016
Content may be subject to copyright.
Send Orders for Reprints to reprints@benthamscience.ae
Current Alzheimer Research, 2016, 13, 000-000 1
1567-2050/16 $58.00+.00 © 2016 Bentham Science Publishers
Metabolomic-Driven Elucidation of Serum Disturbances Associated with
Alzheimer’s Disease and Mild Cognitive Impairment
Raúl González-Domíngueza,b,c, Francisco Javier Rupérezd, Tamara García-Barreraa,b,c,
Coral Barbasd, and José Luis Gómez-Arizaa,b,c,*
aDepartment of Chemistry and CC.MM. Faculty of Experimental S cience. University of Huelva. Cam-
pus de El Carmen. 21007 Huelva. SPAIN; bCampus of Excellence International ceiA3. University of
Huelva. SPAIN; cResearch Center of Health and Environment (CYSMA). University of Huelva. Cam-
pus de El Carmen. 21007 Huelva. SPAIN; dCenter for Metabolom ics and Bioanalysis (CEMBIO),
Pharmacy Faculty, Campus Monteprincipe, Universidad San Pablo-CEU, 28668 Boadilla del Monte,
Madrid, SPAIN
Abstract: Numerous efforts have been made in the last years to discover potential biomarkers of Alz-
heimer’s disease and its progression from mild cognitive impairment, considered as an intermediate
phase in the development of Alzheimer’s disease from normal aging. However, there is still a consid-
erable lack of understanding about pathological mechanisms underlying to disease. In the present study, serum me-
tabolomics based on ultra-high-performance liquid chromatography-mass spectrometry was applied to investigate meta-
bolic differences between subjects with Alzheimer’s disease and mild cognitive impairment, as well as healthy controls.
The most important findings can be associated with impaired metabolism of phospholipids and sphingolipids leading to
membrane breakdown, wherein the nature of the fatty acids contained in the structure in terms of acyl chain length and
degree of unsaturation appears to play a crucial role. Furthermore, several discriminant metabolites were found for the
first time in relation to known pathological processes associated with Alzheimer’s disease, such as the accumulation of
acylcarnitines in relation to mitochondrial dysfunction, decreased levels of oleamide and monoglycerides as a result of de-
fects in endocannabinoid system, or increased serum phenylacetylglutamine, which could reveal alterations in glutamine
homeostasis. Therefore, these results represent a suitable approximation to understand the pathogenesis and progression of
the disease.
Keywords. Alzheimer’s disease, disease progression, membrane breakdown, metabolomics, mild cognitive impairment, patho-
logical mechanisms.
1. INTRODUCTION
Alzheimer’s disease (AD) is the most prevalent neurode-
generative disorder worldwide, principally among older peo-
ple, characterized by a complex etiology in which multiple
pathological processes are involved. A great challenge in AD
research is early diagnosis, given that currently it can be only
detected at advanced stages of disease by exclusion of other
pathologies based on clinical criteria defined by the
NINCDS-ADRDA [1]. Despite these criteria have been re-
cently revised, and the use of molecular biomarker is ex-
pected to enhance the pathophysiological specificity of diag-
nosis of AD, clinical criteria are still the main tools for diag-
nosis in clinical practice [2]. Therefore, new diagnostic tools
are required for predicting the development of dementia
from older people with very mild symptoms of cognitive
dysfunction. This pre-dementia phase, known as mild cogni-
tive impairment (MCI), is considered an intermediate stage
in the development of Alzheimer’s disease from normal ag-
*Address correspondence to this author at the Department of Chemistry and
CC.MM, Faculty of Experimental Science, University of Huelva, Campus
de El Carmen, 21007 Huelva, Spain; Tel: +34959219968;
Fax: +34 959 219942; E-mail: ariza@uhu.es
ing. However, MCI is a heterogeneous syndrome with sev-
eral possible outcomes, so that although up to 80% of pa-
tients develop AD, other subjects show a benign form of
MCI as part of the normal aging course [3]. Thus, numerous
efforts have been made to discover biomarkers of Alzheimer
and its progression, in order to differentiate between AD,
MCI and healthy control subjects. Conventional markers of
AD such as cerebrospinal fluid levels of Aβ peptides and tau
protein have proven useful in the study of MCI [4], and dif-
ferent neuroimaging techniques have been also extensively
applied for detection of brain abnormalities in AD and MCI
patients [5]. Alternatively, it is recognized that metabolomics
plays nowadays an important role in health survey and bio-
markers discovery, because changes in the metabolome level
may be representative of pathological situations. Further-
more, the non-targeted character of this approach allows a
broader investigation of disease and better understanding of
underlying pathological mechanisms. In this way, several
authors have previously described the use of metabolite pro-
filing for monitoring progression of AD from MCI. On the
one hand, it has been demonstrated the occurrence of early
metabolic changes associated with the onset of dementia in
MCI patients compared to age-matched controls. 1H NMR
Please provide
corresponding author(s)
photograph
size should be 4" x 4" inches
2 Current Alzheimer Research, 2016, Vol. 13, No. 4 González-Domínguez et al.
spectroscopy of serum samples underlined the association
between MCI and altered lipid metabolism, particularly with
low relative amount of ω3 fatty acids and metabolic syn-
drome [6]. Moreover, Zheng et al. developed a metabolomic
approach based on liquid chromatography-mass spectrome-
try combined with an isotope dansylation labeling method
for the analysis of human salivary metabolites, whose appli-
cation to MCI patients revealed the implication of taurine in
its pathogenesis [7]. On the other hand, other authors fo-
cused on determining altered profiles associated with the
progression of dementia by means of comparative studies
between MCI and AD patients. Greenberg et al. found a con-
siderable increase of three bile acids in serum from AD and
MCI patients [8]. In other prospective cohort study, AD pa-
tients presented diminished levels of different classes of lip-
ids, and 2,4-dihydroxybutanoic acid, dipalmitoyl-
phosphatidylcholine and an unidentified organic acid were
found as predictive markers of progression to AD in the fol-
low-up [9]. More recently, Armirotti et al. developed a po-
larity-driven sample preparation protocol coupled to or-
thogonal hydrophobic-hydrophilic liquid chromatography for
“shotgun” plasma metabolomics, which has been success-
fully applied to the discovery of metabolomic alterations that
accompany disorders of human cognition, such as MCI and
AD [10]. Similarly, Wang et al. described two biomarker
panels consisting in several plasma metabolites able to dif-
ferentiate between AD, MCI and normal controls using an
integrated analytical platform based on gas and liquid chro-
matography coupled to mass spectrometry [11]. Further-
more, we optimized in a previous work a serum metabolomic
approach based on ultrafiltration and analysis by capillary
electrophoresis for obtaining representative fingerprints of
the polar metabolome in order to distinguish between pa-
tients with Alzheimer’s disease, mild cognitive impairment
and healthy controls [12]. Alternatively, different analytical
platforms have been also applied to examine metabolic dif-
ferences in CSF from patients with different cognitive de-
cline related to AD progression, including capillary electro-
phoresis-mass spectrometry [13], ultra-high performance
liquid chromatography-mass spectrometry [14] and liquid
chromatography-electrochemical coulometric array detection
[15]. Finally, integrated analysis of plasma and CSF samples
has been also performed in order to evaluate in a more com-
prehensive manner the early disease mechanisms shared in
progression from normal aging to MCI and AD, which al-
lowed the discovery of numerous canonical pathways sig-
nificantly disturbed [16]. Nevertheless, despite the consider-
able amount of literature on this subject, there is still a con-
siderable lack of understanding about pathological mecha-
nisms underlying to this neurodegenerative disorder.
In this work we present a metabolomic study of Alz-
heimer’s disease pathogenesis and its progression from mild
cognitive impairment in serum samples. For this, a large
population sample has been enrolled, comprising AD and
MCI patients, and aged-control subjects (n=137). Me-
tabolomic screening of serum samples was performed by
ultra-high performance liquid chromatography-MS, which
has become the main workhorse in this field due to its high
resolution and sensitivity, short analysis time and great po-
tential for biomarker identification [17]. Thus, numerous
compounds were identified as potential markers of AD and
MCI, which may contribute to deepen into underlying patho-
logical mechanisms related to neurodegenerative processes.
2. MATERIAL AND METHODS
2.1. Chemicals
Methanol, ethanol and acetonitrile (HPLC-grade), as well
as formic acid (MS-grade) were purchased from Sigma-
Aldrich (Steinheim, Germany), and water was purified with
a Milli-Q Plus 185 system (Millipore, Bedford, USA).
Purine and HP921 (standard reference solutions) were ob-
tained from Agilent Technologies (USA).
2.2. Blood Collection
Blood samples were collected from 137 subjects re-
cruited by the Neurological Service of Hospital Juan Ramón
Jiménez (Huelva, Spain), including healthy controls (HC),
mild cognitive impairment (MCI) and Alzheimer’s disease
(AD) patients. Blood was extracted by venipuncture of the
antecubital region after 8 hours of fasting, and collected in
BD Vacutainer SST II tubes with gel separator and Advance
vacuum system, previously cooled in a refrigerator. It should
be noted that all these samples were collected in the morning
in order to avoid the influence of circadian rhythm. Samples
were immediately cooled and protected from light for 30
minutes to allow clot retraction, and then centrifuged at 3500
rpm for 10 minutes. Serum was aliquoted in Eppendorf tubes
and frozen at -80°C until analysis. Alzheimer’s disease pa-
tients (n=75, 33 male and 42 female, medium age 79.9±5.0
y) were newly diagnosed of sporadic AD according to the
criteria of the NINCDS-ADRDA [1], and only subjects that
had not yet received any type of medication were included in
the study. Mild cognitive impairment patients (n=17, 10
male and 7 female, medium age 76.1±5.5 y) reported cogni-
tive decline and impairment on objective cognitive tasks, but
they were not demented and did not meet the NINCDS-
ADRDA requirements for a possible or probable diagnosis
of Alzheimer [3]. Finally, matched healthy controls in sex
and age (n=45, 18 male and 27 female, medium age 72.4±5.5
y) were enrolled after examination by neurologists to con-
firm the absence of neurological disorders, whom had not
more than two reported cases of Alzheimer’s disease in their
families. Demographic characteristics of subjects enrolled in
this study, including age, gender, comorbidities, medication
and family history of AD, are listed in Table S1, available as
Supplementary Material. All subjects gave informed consent
for the extraction of peripheral venous blood, and the study
was performed in accordance with the principles contained
in the Declaration of Helsinki and approved by the Ethical
Committee of University of Huelva.
2.3. Metabolomic Fingerprinting
Metabolomic analysis of serum samples was performed
following a procedure previously optimized [18]. For protein
precipitation and metabolite extraction, 50µL of serum were
mixed with 150 µL of a cold mixture of methanol:ethanol
(1:1). Then, samples were briefly vortexed and maintained
for 5 min in an ice bath. Finally, protein precipitate was re-
moved by centrifugation at 13000 rpm for 20 min at 4ºC, and
the supernatant was filtered through a 0.22 µm nylon filter.
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 3
Samples were subsequently fingerprinted by ultra-high per-
formance liquid chromatography (Agilent 1290) coupled to a
quadrupole-time-of-flight mass spectrometry system
equipped with electrospray source (Agilent 6550). Separa-
tion was performed in a reversed-phase column (Zorbax Ex-
tend C18, 2.1x50 mm, 1.8µm) thermostated at 60ºC, with an
injection volume of 0.5 µL. Solvents were delivered at a
flow rate 0.6 mL/min, using water with 0.1% formic acid
(solvent A) and acetonitrile with 0.1% formic acid (solvent
B). The gradient program was as follows: initial conditions
were 5% B for 1 min, followed by a gradual increase to 80%
B in 6 min and finally 100% B in other 4.5 min. Then, sys-
tem returns to initial conditions in 0.5 min, and column is
equilibrated for 1 min with 5% B. Therefore, total analysis
time was 15 min. MS operated in positive and negative po-
larities in separated runs, acquiring full scan spectra in the
m/z range 50-1000. The capillary voltage was set to 3000 V,
with 1000 V of nozzle voltage, 175 V of fragmentor voltage
and 65 V of skimmer voltage. Nitrogen was used as drying
and nebulizer gas, whose temperature was fixed at 250ºC.
Drying gas was supplied at 12 L/min, while nebulizer gas
pressure was 52 psi.
2.4. Data Processing
Raw data was preprocessed using the Molecular Feature
Extraction (MFE) tool in MassHunter Qualitative Analysis
Software (Agilent Technologies) in order to cleaning back-
ground noise and unrelated ions (elimination of spurious
signals from the mass spectra). The MFE algorithm uses the
accuracy of the measurements for grouping related ions by
charge state envelope, isotopic distribution, and/or the pres-
ence of adducts and dimmers, and then creates a list of all
possible components (or features) described by mass, reten-
tion time and abundance [19]. Thus, processing was per-
formed by applying an abundance cutoff of 200 counts and
enabling the search of different ion species (M+H+, M+Na+,
M+K+, M+NH4+, M-H2O in positive mode; M-H+,
M+HCOO-, M+Cl- in negative mode). In addition, for iso-
tope grouping, the peak spacing tolerance was set at 0.0025
m/z, and the charge states were limited to 2. Then, alignment
and filtering were conducted using the Mass Profiler Profes-
sional software (Agilent Technologies). For this, data was
filtered by selecting features into the range 0.05-11.5 min,
and then, peaks were aligned applying a retention time win-
dow of 0.15 minutes and a mass window of 20 ppm.
2.5. Data Analysis
First of all, data was filtered in Mass Profiler Profes-
sional to remove non reproducible signals before to perform
any statistical analysis. For this purpose, features were fil-
tered by choosing masses present in at least 75% of samples
in one of the compared groups, and then features were again
filtered on sample variability, selecting only variables with a
coefficient of variation less than 50% within each group.
Then, data were processed by partial least squares discrimi-
nant analysis (PLS-DA) in SIMCA-P™ software (version
11.5, Umetrics AB, Umeå, Sweden), in order to find differ-
ences between the groups of study. For this, data was sub-
mitted to Pareto scaling, for reducing the relative importance
of larger values, and logarithmic transformation, in order to
approximate a normal distribution [20]. In addition, quality
of the model was assessed by the R2 and Q2 values, supplied
by the software, which provide information about the class
separation and predictive power of the model, respectively.
These parameters are ranged between 0 and 1, and they indi-
cate the variance explained by the model for all the data ana-
lyzed (R2) and this variance in a test set by cross-validation
(Q2). Finally, potential biomarkers of disease and its progres-
sion were found by two-class comparisons, AD vs. HC, MCI
vs. HC (markers of advanced and early dementia, respec-
tively), and AD vs. MCI (markers of dementia progression).
For this purpose, univariate statistical analyses with Bonfer-
roni correction for multiple testing (t-test, p≤0.05) were per-
formed, and loadings plots from PLS-DA were inspected to
select altered metabolites according to the Variable Impor-
tance in the Projection, or VIP (a weighted sum of squares of
the PLS weight, which indicates the importance of the vari-
able in the model), considering only variables with VIP val-
ues higher than 2, indicative of significant differences among
groups.
2.6. Identification of Metabolites
Identification of significant compounds was made match-
ing the experimental accurate mass and tandem mass spectra
(MS/MS) with those available in metabolomic databases
(HMDB, METLIN, KEGG and LIPIDMAPS), using a mass
accuracy of 20 ppm. In addition, the identity of lipids was
confirmed based on characteristic fragmentation patterns
described in literature. Phospholipids presented characteristic
fragments in positive ionization mode of 184, 104, and 86
m/z for phosphatidylcholines (PCs) and lyso-phosphati-
dylcholines (lysoPCs), and [M+H-141]+ for phosphatidyl-
ethanolamines (PEs) and lyso-phosphatidylethanolamines
(lysoPEs), while in negative mode these distinctive signals
are found at 168 and 196, respectively. Furthermore, frag-
mentation in the glycerol backbone by losses of the fatty acyl
substituents enables the identification of individual species
of phospholipids [21]. For sphingolipids, typical product
ions appear at m/z 264 and 282 due to the fragmentation in
the sphingosine moiety, and in the particular case of sphin-
gomyelins, cleavage of phosphocholine headgroup generates
characteristic fragments of 184 and 168 m/z, in positive and
negative modes respectively [22]. Finally, acylcarnitines
were confirmed based on characteristic fragments of 60 and
85 m/z [23].
3. RESULTS AND DISCUSSION
3.1. Metabolite Fingerprinting for Samples Classification
Metabolomic fingerprints obtained from serum samples
in both positive and negative ionization modes (Fig. 1) were
submitted to multivariate statistical analysis to check the
potential of the methodology to discriminate between AD,
MCI and healthy control subjects. For data collected in posi-
tive mode, the PLS-DA model showed a clear classification
of the three groups under investigation (Fig. 2A), whose
quality was appropriated in terms of variance explained
(R2=0.84) and variance predicted (Q2=0.263). On the other
hand, negative data provided worse statistical models
(R2=0.844, Q2=0.032), which leads to less robust separation
between the three groups (Fig. 2B). Finally, to obtain more
4 Current Alzheimer Research, 2016, Vol. 13, No. 4 González-Domínguez et al.
Fig. (1). Metabolomic fingerprints from serum samples obtained by UHPLC-ESI-QTOFMS in positive (A) and negative (B) ion mode.
detailed information about the variables involved in dis-
crimination according the stage of dementia, groups were
compared by pairs: AD vs. HC, MCI vs. HC and AD vs.
MCI. Scores plots demonstrated a total separation between
the different groups (Fig. 2C-E), but quality parameters were
again lower for statistical models built with data collected in
negative mode.
3.2. Selection of Potential Biomarkers
Statistically significant compounds identified are summa-
rized in Tables 1-3, including their retention times and ex-
perimental accurate masses, the mass error, the ionization
mode used for detection, the percentage of change observed
in each comparison and the p-value. All these metabolites
were significantly altered in AD patients respect to healthy
controls (p≤0.05), but in addition, some of them were also
perturbed in MCI. Thus, changes in the comparison MCI/HC
can be associated with failures in the early development of
this disorder, while altered metabolites between AD and
MCI patients would be related to the advance of cognitive
dysfunction. Furthermore, column plots representing the
mean values with standard deviation bars for these selected
metabolites showed that inter-individual variability within
each group is less important than differences between groups
(Fig. 3), thus corroborating that metabolic alterations de-
tected can be attributed to disease state, as previously dem-
onstrated [24].
3.3. Biological Meaning
The use of blood samples in AD research has been tradi-
tionally relegated to the background due to the difficulty to
interpret the association between blood-based measures and
brain processes. However, recent works have demonstrated
close similarities in metabolomic abnormalities observed in
serum [25-26] and brain samples [27-28] from transgenic
models of AD. Furthermore, there is growing evidence that
Alzheimer’s disease might be a systemic disorder [29], thus
demonstrating the utility of peripheral samples in the inves-
tigation of pathological mechanisms associated with AD. In
this study, numerous metabolites were found significantly
altered in serum from AD and MCI patients, which show the
deep impact that this disorder causes on multiple essential
pathways in the organism. One of the most important
changes was observed in phospholipids and related com-
pounds, listed in Table 1. Alzheimer’s disease has been pre-
viously associated with increased degradation of brain phos-
pholipids due to the over-activation of phospholipases [30],
leading to the breakdown of cellular membranes, although it
could be also related to changes in the fatty acid composition
of these lipids. Thereby, the loss of unsaturated fatty acids
due to oxidative stress, such as docosahexaenoic and
araquidonic acid that are highly enriched in neurons, seems
to contribute to AD pathogenesis via membrane damage, in
line with previous studies reporting reduced levels of phos-
pholipids containing PUFAs in brain and blood of patients
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 5
Fig. (2). PLS-DA scores plots from UHPLC-MS data. (A) Scores plot for all samples analyzed in positive ion mode (3 components, R2=0.84,
Q2=0.267); (B) Scores plot for all samples analyzed in negative ion mode (3 components, R2=0.844, Q2=0.032); (C) Scores plot for two-class
comparison, AD vs. HC (positive mode: 4 components, R2=0.996, Q2=0.44; negative mode: 2 components, R2=0.915, Q2=0.125); (D) Scores
plot for two-class comparison, MCI vs. HC (positive mode: 4 components, R2=0.999, Q2=0.654; negative mode: 3 components, R2=0.998,
Q2=0.085); (E) Scores plot for two-class comparison, AD vs. MCI (positive mode: 4 components, R2=0.995, Q2=0.496; negative mode: it was
not possible to build any model). HC: open triangles (n=45); MCI: open diamonds (n=17); AD: black dots (n=75).
with Alzheimer’s disease [31-32]. On the other hand, satu-
rated fatty acids (SFA) have received much less attention,
and most authors found no differences in their distribution
among phospholipids. Nevertheless, Söderberg et al. showed
that the decrease in PUFA-containing phosphatidylethano-
lamines in AD brain is paralleled by an increase in the rela-
tive amounts of the saturated myristic, palmitic and stearic
acids [33], and more recently, it has been demonstrated that
these imbalances in the ratio saturated/polyunsaturated fatty
acids are also present in serum phosphatidylcholines [21,
34]. In the present work, this rationale was corroborated by
increased levels of SFA-containing phosphocholines and
phosphoethanolamines, together with a significant decrease
in several unsaturated phosphocholines, containing linoleic,
arachidonic and docosahexaenoic acid (Table 1). Moreover,
several plasmalogen species were also down-regulated dur-
ing the development of Alzheimer. Previous studies have
shown that ethanolamine plasmalogen deficiency is closely
related to AD [31], but also plasmenylcholines appear to be
involved in neurodegenerative processes [34]. However,
based on our experimental results, we can conclude that
these pathological processes occur at different stages of the
disease, which has not been previously described to our
knowledge. Choline plasmalogen levels were significantly
decreased in serum from AD and MCI patients compared to
healthy controls, which supports that its metabolism is trig-
gered in the onset of neurodegeneration. On the other hand,
MCI and healthy controls were indistinguishable in terms of
ethanolamine plasmalogens, indicating that this impairment
must occur at more advanced stages of disease. Besides the
already described changes in serum levels of phospholipids,
lyso-phospholipids were also involved in pathogenesis of
AD. In previous studies, total lyso-phosphocholines concen-
tration tended to be lower in blood of AD patients [21, 35],
reflecting alterations in the metabolism of choline-containing
phospholipids that may be attributed to impairments in the
6 Current Alzheimer Research, 2016, Vol. 13, No. 4 González-Domínguez et al.
Table 1. Discriminant lyso-phospholipids and phospholipids identified in serum from AD and MCI patients.
change (%)
metabolite
mass (D a)
mass error
(ppm)
RT (min)
ion mode
AD/HC
AD/MCI
MCI/HC
p value
Lyso-phospholipids
LPE (16:0)
453.2807
-10.6
5.51
P
+16.2
+39.9
NS
0.036
P
+21.2
+25.5
NS
0.001
LPE (18:2)
477.2858
0.6
5.41
N
+23.2
+29.9
NS
0.006
P
+21.8
+26.1
NS
0.040
LPC (16:1)
493.3111
-11.6
5.12
N
+21.3
+22.9
NS
0.047
LPC (16:0)
495.3310
-3.0
5.64
N
+11.8
+17.4
NS
0.028
LPC (O-18:0)
509.3783
-12.2
6.48
P
+10.7
NS
NS
0.042
LPC (20:5)
541.3115
-9.8
5.37
N
-16.6
NS
-13.2
0.009
LPC (22:6)
567.3281
-7.8
5.44
P
-20.4
NS
-18.9
0.007
LPC (22:5)
569.3415
-11.6
5.75
P
-14.6
NS
NS
0.012
Phospholipids
PE(16:0/18:0)
719.5469
0.6
9.99
P
+31.9
NS
+44.0
0.035
PC(16:1/16:0)
731.5395
-9.6
9.67
N
+25.9
+32.5
NS
0.012
PC(16:0/16:0)
733.5593
-3.8
10.5
N
+39.9
NS
+50.7
0.001
PPC(16:0/18:2)
741.5612
-8.2
10.2
N
-24.3
NS
-34.3
0.003
PC(15:0/18:2)
743.5428
-4.8
9.45
N
+20.4
NS
NS
0.025
PC(15:0/18:1)
745.5640
2.5
10.1
P
+19.5
NS
NS
0.042
P
-10.3
-10.2
NS
0.039
PPE(16:0/22:6)
747.5155
-6.4
9.74
N
-17.2
-20.6
NS
0.014
P
-28.4
-24.1
NS
0.002
PPE(18:1/20:4)
749.5370
1.5
10.1
N
-14.4
NS
NS
0.046
PC(16:0/18:2)
757.5558
-8.3
9.85
N
+47.3
NS
+53.7
0.026
PC(16:0/20:3)
783.5772
-0.8
10.3
P
-88.3
NS
-88.1
0.037
PC(16:1/22:6)
803.5478
1.6
9.60
N
-25.4
NS
-41.8
0
PC(18:0/20:4)
809.5803
-16.2
10.3
P
-42.4
NS
NS
0.019
PC(20:4/20:4)
829.5533
-10.6
10.3
P
-22.9
NS
NS
0.011
PC(18:1/22:6)
831.5739
-4.7
9.6
P
-9.7
NS
NS
0.047
PC(20:4/22:6)
853.5538
-9.7
8.84
P
-13.4
NS
NS
0.041
Abbreviations. LPE: lyso-phosphoethanolamine; LPC: lyso-phosphocholine; PE: phosphoethanolamine; PC: phosphocholine; PPC: choline-plasmalogen;
PPE: ethanolamine-plasmalogen; NS: non significant
deacylation-reacylation cycle of phospholipids. However, it
is noteworthy that lyso-phospholipids presented a similar
distribution to that of their precursors in our analysis, with
increased levels of saturated compounds respect to unsatu-
rated ones in both ethanolamine and choline species (Table
1). Therefore, it can be concluded that the imbalance ob-
served in phospholipids is finally reflected in their degrada-
tion products, which makes them useful markers of destabi-
lization of neuronal membranes. Furthermore, the compari-
son of the three study groups lead us to believe in a temporal
evolution of the processes leading to phospholipids degrada-
tion given that, while unsaturated lysophospholipids are
down-expressed in both AD and MCI patients, the increase
of saturated species is only observed in AD. Thus, it could
be hypothesized that, in the membrane destabilization proc-
ess, the first step supposes the degradation of PUFA-
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 7
Table 2. Discriminant sphingolipids identified in serum from AD and MCI patients.
change (%)
metabolite
mass (D a)
mass error
(ppm)
RT (min)
ion mode
AD/HC
AD/MCI
MCI/HC
p value
Sphingoi d bases
S1P
379.2433
-14.5
5.00
N
+7.9
NS
+16.8
0.016
Ceramides
CER(d18:1/16:0)
537.5049
-13.4
10.8
P
+38.9
NS
+34.4
0.003
Sphingomyelins
SM(d18:1/12:0)
646.4976
-11.4
7.44
P
+23.6
NS
NS
0.031
SM(d18:1/14:0)
674.5287
-11.1
8.11
P
+13.2
NS
+29.7
0.023
P
+14.9
NS
+25.7
0.005
SM(d18:1/16:0)
702.5606
-9.9
8.89
N
+31.05
NS
+46.1
0.001
SM(d18:1/18:2)
726.5598
-10.7
8.47
P
-16.8
NS
NS
0.022
P
+9.9
NS
+29.4
0.043
SM(d18:1/18:1)
728.5739
-12.8
9.12
N
+15.0
NS
+29.9
0.008
SM(d18:1/18:0)
730.5986
11.8
9.86
P
+11.1
NS
+47.6
0.005
Glycosphingolipids
P
+33.2
NS
+89.5
0.001
Hex-CER(d18:1/16:0)
699.5597
-7.4
9.50
N
+28.2
NS
+34.04
0.041
Hex-CER(d18:1/18:0)
741.5648
-14.4
10.3
P
+41.1
NS
+47.2
0.011
SULF(d18:1/18:0)
807.5563
4.1
9.73
P
-8.1
-10.6
NS
0.026
Lac-CER(d18:1/14:0)
833.5812
-6.2
8.17
P
+50.7
NS
+86.7
0.008
Lac-CER(d18:1/16:1)
859.5951
-2.2
8.32
P
+31.6
NS
+45.4
0.048
Lac-CER(d18:1/16:0)
861.6127
-5.8
8.95
P
+20.1
NS
+44.8
0.018
Abbreviations. S1P: sphingosine-1-phosphate; CER: ceramide; SM: sphingomyelin; Hex-CER: hexosyl-ceramide; SULF: sulfatide; Lac-CER: lactosyl-
ceramide; NS: non significant
Table 3. Other discriminant metabolites identified in serum from AD and MCI patients.
change (%)
metabolite
mass (D a)
mass error
(ppm)
RT (min)
ion mode
AD/HC
AD/MCI
MCI/HC
p value
Monoglycerides
MG(16:0)
330.2732
-11.5
7.02
P
-49.3
-52.3
NS
0.047
MG(18:0)
358.3034
-13.7
7.66
P
-34.6
-47.3
NS
0.007
Acyl-carnitines
CAR(16:0)
399.3352
0.8
5.25
P
+12.3
NS
NS
0.012
CAR(18:2)
423.3299
-11.8
5.08
P
+18.9
NS
+19.7
0.011
CAR(18:1)
425.3468
-8.7
5.37
P
+9.4
NS
+29.9
0.004
CAR(18:0)
427.3620
-9.6
5.71
P
+10.5
NS
+17.4
0.049
Others
8 Current Alzheimer Research, 2016, Vol. 13, No. 4 González-Domínguez et al.
(Table 3) contd….
change (%)
metabolite
mass (D a)
mass error
(ppm)
RT (min)
ion mode
AD/HC
AD/MCI
MCI/HC
p value
histidine
155.0684
-7.1
0.26
N
-13.7
-11.3
NS
0.039
PAG
264.1072
-14.4
0.28
N
+38.0
NS
+57.2
0.040
oleamide
281.2690
-10.3
6.89
P
-63.1
NS
-86.1
0.025
PREGS
396.1922
-12.1
4.89
N
-17.8
-24.1
NS
0.039
Unknowns
UK1
162.1025
-
3.67
P
-42.2
-46.1
NS
0.003
UK2
203.1281
-
5.51
P
+59.5
+42.8
NS
0
UK3
353.1752
-
6.52
P
+18.2
NS
NS
0.007
UK4
394.2793
-
7.64
N
-29.2
-30.1
NS
0.001
UK5
414.1996
-
4.86
P
+39.4
NS
+64.4
0.002
UK6
414.2259
-
7.02
P
-34.0
-32.8
NS
0.018
UK7
430.3029
-
5.46
P
+19.9
NS
+26.0
0.003
Abbreviations. MG(16:0): monopalmitin; MG(18:0): monostearin; CAR(16:0): palmitoyl-carnitine; CAR(18:2): linoleyl-carnitine; CAR(18:1): oleyl-
carnitine; CAR(18:0): stearoyl-carnitine; PAG: phenyl-acetyl-glutamine; PREGS: pregnenolone sulfate; NS: non significant.
containing phospholipids (probably due to oxidative stress),
and secondly, their replacement by saturated ones. Among
these lysophospholipids, one deserves special attention:
LPC(O-18:0), an alkyl ether lysophospholipid involved in
the synthesis of platelet activating factor (PAF) by the re-
modeling pathway. In this context, Ryan et al. found three
PAF species significantly elevated in AD cortex: C16:0
PAF, its precursor C16:0 lyso-PAF, and C18:1 lyso-PAF
[36]. Thus, the increase observed in serum C18:0 lyso-PAF
(Table 1) could reveal a selective disruption of stearoyl-PAF
biosynthesis, complementary to findings by Ryan et al.,
which evidences that these impairments in lipid metabolism
are finally reflected in peripheral fluids.
The involvement of perturbed sphingolipid metabolism
also emerges as a pivotal event in membrane degeneration in
AD, as can be observed in Table 2. Defects in sphingolipid
metabolism have been previously associated with Alz-
heimer’s disease, with up-regulated activities of different
enzymes such as ceramide synthases [37] and acid sphingo-
myelinase [38], which suggests a shift in metabolism to-
wards the accumulation of ceramides. In this sense, high
levels of total ceramides have been previously found in brain
[37-39] and blood [40-41] of AD patients, which is in
agreement with our findings in serum (Table 2). On the other
hand, studies of sphingomyelin (SM) levels in AD are less
clear. Results from post-mortem brain analyses are contra-
dictory, showing increased [42] or decreased [38] total con-
centrations of SMs, while plasma levels appear to be de-
creased [37]. The reasons for these discrepancies are not
understood to date, but they could be related to differences in
the acyl chain composition. It has been demonstrated that the
increase of ceramides in AD brain is more pronounced for
species containing very long chain fatty acids [43], which is
consequent with the elevated expression of long-chain cera-
mide synthase [37]. Similarly, when individual SM species
are considered, most authors noted a decrease in very long
chain sphingomyelins [40, 43], while other species are usu-
ally over-expressed [42]. In this work, this hypothesis was
corroborated by the increased serum levels of several me-
dium- and long-chain sphingomyelins (MCFA-SM and
LCFA-SM) in AD and MCI patients, from 12 to 18 carbon
atoms (Table 2), while very long chain species remained
unchanged. Alternatively, levels of SM(d18:1/18:2) were
decreased in diseased subjects, suggesting a possible impli-
cation of oxidative stress in its degradation, as occurred with
analogous phospholipids (Table 1). Furthermore, besides the
described differences according the fatty acid contained in
the structure, it is remarkable the trend observed in sphingo-
myelin levels between the three groups of study. For most of
these compounds, the percentage of change was slightly
more pronounced in the comparison MCI vs. HC than in AD
vs. HC, which suggests that alterations in metabolism of
sphingomyelins probably contribute to early pathological
mechanisms. In this sense, Kosicek et al. found significantly
increased SM levels in the CSF from patients with prodro-
mal AD compared to healthy controls, mild and moderate
AD [44], and Mielke et al. have demonstrated that blood
sphingomyelins and ceramides vary along the progression of
memory impairment [45]. Therefore, peripheral levels of
these sphingolipids could be good predictors of memory im-
pairment and markers of progression. Sphingosine-1-
phosphate (S1P) is a signaling molecule produced via degra-
dation of ceramides by the enzyme ceramidase and subse-
quent phosphorylation by sphingosine kinase. Previous stud-
ies reported that acid ceramidase is up-regulated in AD
brains versus normal controls [38], as well as the activity of
sphingosine kinase 2 [46], which supports the increased lev-
els of S1P found in serum samples (Table 2). Finally,
changes in hexosylceramides, lactosylceramides and sulfa-
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 9
tides (Table 2) also support the involvement of glycosphin-
golipids in pathogenesis of AD. On the one hand, a substan-
tial reduction of sulfatide levels was observed in serum sam-
ples from AD patients, as previously reported in brain [36].
On the other hand, the increase of different species of lacto-
sylceramides (Table 2) could be indicative of an impaired
metabolism of gangliosides. Like sulfatides, earlier studies in
AD showed alterations of ganglioside metabolism resulting
in reduced levels in the majority of brain regions [47], but
increased serum lactosylceramide has been associated with
risk of AD only once [48]. Complementarily, hexosylcera-
mides, which comprise isomeric glucosyl and galactosyl
species indistinguishable by MS, were also increased in se-
rum samples. Glucosyl- and galactosyl-ceramides are poten-
tial degradation products of glycosphingolipids, so their
over-expression support the depletion observed in gangli-
osides and sulfatides, respectively. Finally, it should be
pointed out that the increase of hexosyl- and lactosyl-
ceramides was sharper in MCI than in AD, which suggests
that depletion of glycosphingolipids occurs in the onset of
disease, as previously mentioned for sphingomyelins. There-
fore, it can be concluded that lipid homeostasis might be a
primary target in pathogenesis of Alzheimer’s disease, in-
volving both phospholipids and sphingolipids forming part
of cellular membranes and/or lipoproteins. This abnormal
metabolism of phospholipids and sphingolipids observed
during the progression of disease resulted in important bio-
chemical changes in serum samples, listed in Tables 1 and 2.
Thus, membrane breakdown could be considered as a source
of potential markers of Alzheimer, as summarized in Fig.
(4).
Fig. (3). Column plots with standard deviation bars for the discriminant metabolites identified in serum from AD and MCI patients. HC:
black columns (n=45); MCI: light gray columns (n=17); AD: dark gray columns (n=75).
10 Current Alzheimer Research, 2016, Vol. 13, No . 4 González-Domínguez et al.
Aside from changes related to membrane defects, the de-
crease observed in monoglycerides (monopalmitin and
monosterain) and oleamide (Table 3) could point to distur-
bances in endocannabinoid (EC) system. This system plays a
role in numerous physiological processes, principally in the
central nervous system where acts as neuromodulator. How-
ever, accumulating data show an imbalance in key elements
of the EC system associated with AD and other neurodegen-
erative disorders. In this sense, the expression levels of can-
nabinoid receptors (CB) have been found altered in AD
brains, including a decrease of CB1 receptors and a comple-
mentary up-regulation of CB2 receptors in the hippocampus
[49]. In addition, the activities of enzymes fatty acid amide
hydrolase (FAAH) and monoacylglycerol lipase (MAGL),
which are involved in the metabolism of the major endocan-
nabinoids, anandamide and 2-arachidonoyl-glycerol respec-
tively, are affected in AD. Benito et al. reported that the ac-
tivity of FAAH seems to be elevated in the plaques and sur-
rounding areas of AD brains [49], which could be responsi-
ble for the decrease of oleamide found in serum of AD and
MCI patients. Oleamide is a bioactive fatty acid amide
catabolically related to anandamide, which may be also de-
graded by FAAH thereby serving as indicator of its over-
expression. On the other hand, there is also evidence for the
involvement of monoglyceride lipase in AD, given that it has
been found that microglia expressing MAGL accumulates
around senile plaques [50]. This lipase is the major enzyme
inactivating 2-arachidonoyl glycerol, but other monoglyc-
erides are also potential substrates. In this way, reductions of
circulating levels of monopalmitin and monostearin could be
considered as indicators of MAGL over-expression.
Long chain acylcarnitines were also increased in serum
of AD and MCI patients, which may indicate an incomplete
fatty acid β-oxidation in the onset of disease. Elevation of
palmitoyl-, stearoyl-, oleyl- and linoleyl-carnitines is the
characteristic feature of carnitine palmitoyltransferase II de-
ficiency, an inherited metabolic impairment that prevents
mitochondrial oxidation of long chain fatty acids [51]. In this
sense, mitochondrial dysfunction appears to be one of the
primary events in the course of AD, which can perturb cellu-
lar bioenergetics [52]. Therefore, this impaired metabolism
of lipids could be involved in neurodegenerative energetic
failures as a supplementary pathway to those previously de-
scribed, such as deregulated tricarboxilic acid cycle or oxida-
tive phosphorylation system [53].
Histidine is an amino acid with antioxidant and anti-
inflammatory properties, like other imidazole-containing
compounds. Thus, oxidative stress may account for the de-
crease of this amino acid (Table 3), as previously reported
[31, 34]. Pregenenolone sulfate (PREGS) is a neurosteroid
precursor of estrogens, progestins and androgens, which
plays important roles in the aging nervous system. Weill-
Engerer et al. found a trend toward lower levels of PREGS
and other steroids in brain from AD patients, and a negative
correlation between PREGS and β amyloid levels [54]. In the
present study, pregnenolone sulfate was decreased in serum
from AD patients compared to MCI and controls, corroborat-
ing the potential of neurosteroids as diagnostic markers of
Alzheimer. Finally, increased serum levels of phenylacetyl-
glutamine (PAG) were found in AD and MCI patients, which
could be associated with elevated levels of glutamine, given
that PAG can be also synthesized in the liver by condensa-
Fig. (4). Metabolomic changes observed in phospholipids and sphingolipids due to membrane breakdown. (↑) increased compounds, (↓)
decreased compounds in AD and/or MCI respect to healthy controls (individual changes are shown in Tables 1-2). Abbreviations: PL: phos-
pholipid; PUFA: polyunsaturated fatty acid; SFA: saturated fatty acid; Pls: plasmalogen; lyso-PL: lyso-phospholipid; SM: sphingomyelin;
MCFA: medium chain fatty acid; LCFA: long chain fatty acid; Cer: ceramide; ; S1P: sphingosine-1-phosphate; Glc-Cer: glucosyl-ceramide;
Lac-Cer: lactosyl-ceramide; Gal-Cer: galactosyl-ceramide; Hex-Cer: hexosyl-ceramide.
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 11
tion of glutamine with phenylacetyl-CoA. Therefore, the
increase observed in PAG could be considered a novel
marker of impaired glutamate-glutamine homeostasis in rela-
tion to the regulation of ammonia levels in the organism
[55].
CONCLUSION
Metabolomic fingerprinting of serum samples by
UHPLC-ESI-QTOFMS has been demonstrated as a suitable
approach to distinguish between patients with Alzheimer’s
disease, mild cognitive impairment and healthy controls.
Multiple metabolites, principally lipids, were identified as
potential markers for diagnosis, which indicated the in-
volvement of different pathological processes in the devel-
opment of neurodegeneration. Thus, it was observed a deep
dysregulation in metabolism of phospholipids and sphin-
golipids in terms of altered levels of saturated/unsaturated
fatty acids contained in the structure, but also due to differ-
ences in the acyl chain length. In addition, these imbalances
were accompanied by significant changes in related com-
pounds (ceramides, lyso-phospholipids) that pointed to
membrane breakdown. On the other hand, novel potential
biomarkers not previously described were found for well-
known pathological situations associated with AD: defects in
energy metabolism (long chain acylcarnitines) and endocan-
nabinoid dysfunction (monoglycerides and oleamide). Fi-
nally, reductions in histidine and PREGS were related to
oxidative stress and altered neurosteroid metabolism, respec-
tively, while increase of PAG could reveal impaired glu-
tamine homeostasis in pathogenesis of AD. Furthermore, the
comparative study between AD and MCI patients allowed
exploring these altered pathways in more detail for better
understanding of pathogenesis and progression of disease
from pre-clinical stages of dementia. As future plan, it would
be interesting to extend this study to other types of dementia
in order to assess the specificity of these potential markers
against other neurodegenerative disorders.
ABBREVIATIONS
AD = Alzheimer’s disease
MCI = mild cognitive impairment
HC = healthy control
UHPLC-MS = ultra-high performance liquid chromatog-
raphy mass spectrometry
PLS-DA = partial least squares discriminant analysis
VIP = variable importance in the projection
LPE = lyso-phosphoethanolamine
LPC = lyso-phosphocholine
PE = phosphoethanolamine
PC = phosphocholine
PPC = choline-plasmalogen
PPE = ethanolamine-plasmalogen
S1P = sphingosine-1-phosphate
CER = ceramide
SM = sphingomyelin
Hex-CER = hexosyl-ceramide
SULF = sulfatide
Lac-CER = lactosyl-ceramide
MG = monoglyceride
CAR = carnitine
PAG = phenyl-acetyl-glutamine
PREGS = pregnenolone sulfate
PUFA = polyunsaturated fatty acid
SFA = saturated fatty acid
CONFLICT OF INTEREST
The author(s) confirm that this article content has no con-
flicts of interest.
ACKNOWLEDGEMENTS
This work was supported by the projects CTM2012-
38720-C03-01 from the Ministerio de Ciencia e Innovación
and P008 FQM-3554 and P009-FQM-4659 from the Conse-
jería de Innovación, Ciencia y Empresa (Junta de An-
dalucía). Raúl González Domínguez thanks the Ministerio de
Educación for a predoctoral scholarship. The authors also
thank to Dr. Alberto Blanco and Carlos Salgado from Hospi-
tal Juan Ramon Jimenez for providing serum samples.
SUPPLEMENTARY MATERIAL
Table S1. Demographic characteristics of subjects en-
rolled in the study.
Supplementary material is available on the publisher’s
web site along with the published article.
REFERENCES
[1] McKahnn G, Drachman D, Folstein M, Katzman R, Price D, Stad-
lan EM. Clinical diagnosis of Alzheimer´s disease: report of the
NINCDS-ADRDA Work Group under the auspices of Department
of Health and Human Services Task Force on Alzheimer´s disease.
Neurology 34: 939-44 (1984).
[2] McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR
Jr, Kawas CH, Klunk WE, Koroshetz WJ, Manly JJ, Mayeux R,
Mohs RC, Morris JC, Rossor MN, Scheltens P, Carrillo MC, Thies
B, Weintraub S, Phelps CH. The diagnosis of dementia due to Alz-
heimer's disease: recommendations from the National Institute on
Aging-Alzheimer's Association workgroups on diagnostic guide-
lines for Alzheimer's disease. Alzheimers Dement 7: 263-9 (2011).
[3] Petersen RC. Mild cognitive impairment as a diagnostic entity. J
Intern Med 256: 183-94 (2004).
[4] Dean RA, Shaw LM. Use of cerebrospinal fluid biomarkers for
diagnosis of incipient Alzheimer disease in patients with mild cog-
nitive impairment. Clin Chem 56: 17-9 (2000).
[5] Reiman ER, Jagust WJ. Brain imaging in the study of Alzheimer's
disease. NeuroImage 61: 505-16 (2012).
[6] Tukiainen T, Tynkkynen T, Mäkinen VP, Jylänki P, Kangas A,
Hokkanen J, et al. A multi-metabolite analysis of serum by 1H
NMR spectroscopy: Early systemic signs of Alzheimer’s disease.
Biochem Biophys Res Commun 375: 356-61 (2008).
[7] Zheng J, Dixon RA, Li L. Development of isotope labeling
LC−MS for human salivary metabolomics and application to profil-
ing metabolome changes associated with mild cognitive impair-
ment. Anal Chem 84: 10802-11 (2012).
12 Current Alzheimer Research, 2016, Vol. 13, No . 4 González-Domínguez et al.
[8] Greenberg N, Grassano A, Thambisetty M, Lovestone S, Legido-
Quigley C. A proposed metabolic strategy for monitoring disease
progression in Alzheimer’s disease. Electrophoresis 30: 1235-9
(2009).
[9] Oresic M, Hyötyläinen T, Herukka SK, et al. Metabolome in pro-
gression to Alzheimer’s disease. Transl Psychiatry 1: e57 (2011).
[10] Armirotti A, Basit A, Realini N, Caltagirone C, Bossù P, Spalletta
G, Piomelli D. Sample preparation and orthogonal chromatography
for broad polarity range plasma metabolomics: application to hu-
man subjects with neurodegenerative dementia. Anal Biochem 455:
48-54 (2014).
[11] Wang G, Zhou Y, Huang FJ, Tang HD, Xu XH, Liu JJ, et al.
Plasma metabolite profiles of Alzheimer’s disease and mild cogni-
tive impairment. J Proteome Res 13: 2649-58 (2014).
[12] González-Domínguez R, García A, García-Barrera T, Barbas C,
Gómez-Ariza JL. Metabolomic profiling of serum in the progres-
sion of Alzheimer’s disease by capillary electrophoresis-mass spec-
trometry. Electrophoresis 35: 3321-30 (2014).
[13] Ibáñez C, Simó C, Martín-Álvarez PJ, Kivipelto M, Winblad B,
Cedazo-Mínguez A, et al. Toward a predictive model of Alz-
heimer’s disease progression using capillary electrophoresis-mass
spectrometry metabolomics. Anal Chem 84: 8532-40 (2012).
[14] Ibañez C, Simo C, Barupal DK, Fiehn O, Kivipelto M, Cedazo-
Mínguez A, et al. A new metabolomic workflow for early detection
of Alzheimer’s disease J Chromatogr A 1302: 65-71 (2013).
[15] Kaddurah-Daouk R, Zhu H, Sharma S, et al. Alterations in meta-
bolic pathways and networks in Alzheimer’s disease. Transl. Psy-
chiatry 3: e244 (2013).
[16] Trushina E, Dutta T, Persson XMT, Mielke MM, Petersen RC.
Identification of altered metabolic pathways in plasma and CSF in
mild cognitive impairment and Alzheimer’s disease using me-
tabolomics. Plos ONE 8: e63644 (2013).
[17] Wilson ID, Nicholson JK, Castro-Perez J, Granger JH, Johnson
KA, Smith BW, et al. High resolution “ultra performance” liquid
chromatography coupled to oa-TOF mass spectrometry as a tool for
differential metabolic pathway profiling in functional genomic
studies. J Proteome Res 4: 591-8 (2005).
[18] Bujak R, García-Álvarez A, Rupérez FJ, Nuño-Ayala M, García A,
Ruiz-Cabello J, et al. Metabolomics reveals metabolite changes in
acute pulmonary embolism. J Proteome Res 13: 805-16 (2014).
[19] Godzien J, García-Martínez D, Martinez-Alcazar P, Ruperez FJ,
Barbas C. Effect of a nutraceutical treatment on diabetic rats with
targeted and CE-MS non-targeted approaches. Metabolomics 9:
188-202 (2013).
[20] van den Berg RA, Hoefsloot HCJ, Westerhuis JA, Smilde AK, van
der Werf MJ. Centering, scaling, and transformations: improving
the biological information content of metabolomics data. BMC Ge-
nomics 7: 142 (2006).
[21] Gonzalez-Dominguez R, Garcia-Barrera T, Gomez-Ariza JL.
Combination of metabolomic and phospholipid-profiling ap-
proaches for the study of Alzheimer's disease, J Proteomics 104:
37-47 (2014).
[22] Haynes CA, Allegood JC, Park H, Sullards MC. Sphingolipidom-
ics: Methods for the comprehensive analysis of sphingolipids. J
Chromatogr B 877: 2696-708 (2009).
[23] Vernez L, Hopfgartner G, Wenk M, Krahenbuhl S. Determination
of carnitine and acylcarnitines in urine by high-performance liquid
chromatography-electrospray ionization ion trap tandem mass spec-
trometry. J Chromatogr A 984: 203-13 (2003).
[24] Masson P, Spagou K, Nicholson JK, Want EJ. Technical and bio-
logical variation in UPLC-MS-based untargeted metabolic profiling
of liver extracts: application in an experimental toxicity study on
galactosamine. Anal Chem 83: 1116-23 (2011).
[25] González-Domínguez R, García-Barrera T, Vitorica J, Gómez-
Ariza JL. Application of metabolomics based on direct mass spec-
trometry analysis for the elucidation of altered metabolic pathways
in serum from the APP/PS1 transgenic model of Alzheimer’s dis-
ease. J Pharm Biomed Anal 107: 378-85 (2015).
[26] González-Domínguez R, García-Barrera T, Vitorica J, Gómez-
Ariza JL. Deciphering metabolic abnormalities associated with
Alzheimer’s disease in serum from the APP/PS1 mouse model us-
ing integrated metabolomics approaches. Biochimie 110: 119-28
(2015).
[27] González-Domínguez R, García-Barrera T, Vitorica J, Gómez-
Ariza JL. Metabolomic screening of regional brain alterations in
the APP/PS1 transgenic model of Alzheimer’s disease by direct in-
fusion mass spectrometry. J Pharm Biomed Anal 102: 425-35
(2015).
[28] González-Domínguez R, García-Barrera T, Vitorica J, Gómez-
Ariza JL. Region-specific metabolic alterations in the brain of the
APP/PS1 transgenic mice of Alzheimer's disease. Biochim Biophys
Acta 1842: 2395-402 (2014).
[29] Morris JK, Honea RA, Vidoni ED, Swerdlow RH, Burns JM. Is
Alzheimer's disease a systemic disease? Biochim Biophys Acta
1842: 1340-9 (2014).
[30] Farooqui AA, Ong WY, Horrocks LA. Biochemical aspects of
neurodegeneration in human brain: involvement of neural mem-
brane phospholipids and phospholipases A2. Neurochem Res 29:
1961-77 (2004).
[31] González-Domínguez R, García-Barrera T, Gómez-Ariza JL. Me-
tabolomic study of lipids in serum for biomarker discovery in Alz-
heimer’s disease using direct infusion mass spectrometry. J Pharm
Biomed Anal 98: 321-6 (2014).
[32] Cunnane SC, Schneider JA, Tangney C, Tremblay-Mercier J, For-
tier M, Bennett DA, et al. Plasma and brain fatty acid profiles in
mild cognitive impairment and Alzheimer's disease. J Alzheimers
Dis 29: 691-7 (2012).
[33] Söderberg M, Edlund C, Kristensson K, Dallner G. Fatty acid com-
position of brain phospholipids in aging and in Alzheimer's disease.
Lipids 26: 421-5 (1991).
[34] González-Domínguez R, García-Barrera T, Gómez-Ariza JL. Using
direct infusión mass spectrometry forserum metabolomics in Alz-
heimer’s disease. Anal Bioanal Chem 406: 7137-48 (2014).
[35] Li N, Liu W, Li W, Li S, Chen X, Bi K, et al. Plasma metabolic
profiling of Alzheimer's disease by liquid chromatography/mass
spectrometry. Clin Biochem 43: 992-7 (2010).
[36] Ryan SD, Whitehead SN, Swayne LA, Moffat TC, Hou W, Ethier
M, et al. Amyloid-beta42 signals tau hyperphosphorylation and
compromises neuronal viability by disrupting alkylacylglycero-
phosphocholine metabolism. Proc Natl Acad Sci USA 106: 20936-
41 (2009).
[37] Katsel P, Li C, Haroutunian V. Gene expression alterations in the
sphingolipid metabolism pathways during progression of dementia
and Alzheimer’s disease: A shift toward ceramide accumulation at
the earliest recognizable stages of Alzheimer’s disease? Neuro-
chem Res 32: 845-56 (2007).
[38] He X, Huang Y, Li B, Gong CX, Schuchman EH. Deregulation of
sphingolipid metabolism in Alzheimer’s disease. Neurobiol Aging
31: 398-408 (2010).
[39] Han X, Holtzman D, McKeel DW, Kelley J, Morris JC. Substantial
sulfatide deficiency and ceramide elevation in very early Alz-
heimer’s disease: Potential role in disease pathogenesis. J Neuro-
chem 82: 809-18 (2002).
[40] Han X, Rozen S, Boyle SH, Hellegers C, Cheng H, Burke JR, et al.
Metabolomics in early Alzheimer’s disease: identification of al-
tered plasma sphingolipidome using shotgun lipidomics. PLoS
ONE 6: e21643 (2011).
[41] González-Domínguez R, Garcia-Barrera T, Gomez-Ariza JL. Ap-
plication of a novel metabolomic approach based on atmospheric
pressure photoionization mass spectrometry using flow injection
analysis for the study of Alzheimer’s disease. Talanta 131: 480-9
(2015).
[42] Chan RB, Oliveira TG, Cortes EP, Honig LS, Duff KE, Small SA,
et al. Comparative lipidomic analysis of mouse and human brain
with Alzheimer disease. J Biol Chem 287: 2678-88 (2012).
[43] Cutler RG, Kelly J, Storie K, Pedersen WA, Tammara A, Hatanpaa
K, et al. Involvement of oxidative stress-induced abnormalities in
ceramide and cholesterol metabolism in brain aging and Alz-
heimer’s disease. Proc Natl Acad Sci 101: 2070-5 (2004).
[44] Kosicek M, Zetterberg H, Andreasen N, Peter-Katalinic J, Heci-
movic S. Elevated cerebrospinal fluid sphingomyelin levels in pro-
dromal Alzheimer’s disease. Neurosci Lett 516: 302-5 (2012).
[45] Mielke MM, Bandaru VVR, Haughey NJ, Rabins PV, Lyketsos
CG, Carlson MC. Serum sphingomyelins and ceramides are early
predictors of memory impairment. Neurobiol Aging 31: 17-24
(2010).
[46] Takasugi N, Sasaki T, Suzuki K, Osawa S, Isshiki H, Hori Y, et al.
BACE1 activity is modulated by cell-associated sphingosine-1-
phosphate. J Neurosci 31: 6850-7 (2011).
[47] Ariga T, McDonald MP, Yu RK. Role of ganglioside metabolism
in the pathogenesis of Alzheimer’s disease - a review. J Lipid Res
49: 1157-75 (2008).
Metabolomic Study of Alzheimer’s Disease Progression Current Alzheimer Research, 2016, Vol. 13, No. 4 13
[48] Mielke MM, Bandaru VV, Haughey NJ, Xia J, Fried LP, Yasar S,
et al. Serum ceramides increase the risk of Alzheimer disease: the
Women's Health and Aging Study II. Neurology 79: 633-41 (2012).
[49] Benito C, Nuñez E, Tolon RM, Carrier EJ, Rabano A, Hillard CJ,
et al. Cannabinoid CB2 receptors and fatty acid amide hydrolase
are selectively overexpressed in neuritic plaque-associated glia in
Alzheimer’s disease brains. J Neurosci 23: 11136-41 (2003).
[50] Mulder J, Zilberter M, Pasquaré SJ, Alpár A, Schulte G, Ferreira
SG, et al. Molecular reorganization of endocannabinoid signalling
in Alzheimer’s disease. Brain 134: 1041-60 (2011).
[51] Sigauke E, Rakheja D, Kitson K, Bennett MJ. Carnitine palmitoyl-
transferase II deficiency: A clinical, biochemical, and molecular
review. Lab Invest 83: 1543-54 (2003).
[52] Maruszak A, Żekanowski C. Mitochondrial dysfunction and Alz-
heimer's disease. Prog Neuropsychopharmacol Biol Psychiatry 35:
320-30 (2011).
[53] Ferreira IL, Resende R, Ferreiro E, Rego AC, Pereira CF. Multiple
defects in energy metabolism in Alzheimer’s disease. Curr Drug
Targets 11: 1193-206 (2010).
[54] Weill-Engerer S, David JP, Sazdovitch V, Liere P, Eychenne B,
Pianos A, et al. Neurosteroid quantification in human brain re-
gions: comparison between Alzheimer’s and non-demented pa-
tients. J Clin Endocrinol Metab 87: 5138-43 (2002).
[55] Felipo V, Butterworth RF. Neurobiology of ammonia. Prog Neuro-
biol 67: 259-79 (2002).
Received: ???????????????? Revised: ???????????????? Accepted: ????????????????
