Metabolomics of cerebrospinal fluid reveals changes in the central
nervous system metabolism in a rat model of multiple sclerosis
Marek J. Noga•Adrie Dane•Shanna Shi•Amos Attali•Hans van Aken•
Ernst Suidgeest•Tinka Tuinstra•Bas Muilwijk•Leon Coulier•
Theo Luider•Theo H. Reijmers•Rob J. Vreeken•Thomas Hankemeier
Received: 5 January 2011/Accepted: 30 March 2011/Published online: 16 April 2011
? The Author(s) 2011. This article is published with open access at Springerlink.com
(EAE) is the most commonly used animal model for
Multiple Sclerosis (MScl). CSF metabolomics in an acute
EAE rat model was investigated using targetted LC–MS
and GC–MS. Acute EAE in Lewis rats was induced by co-
injection of Myelin Basic Protein with Complete Freund’s
Adjuvant. CSF samples were collected at two time points:
10 days after inoculation, which was during the onset of
the disease, and 14 days after inoculation, which was
during the peak of the disease. The obtained metabolite
profiles from the two time points of EAE development
show profound differences between onset and the peak of
Experimental Autoimmune Encephalomyelitis
the disease, suggesting significant changes in CNS
metabolism over the course of MBP-induced neuroin-
flammation. Around the onset of EAE the metabolome
profile shows significant decreases in arginine, alanine and
branched amino acid levels, relative to controls. At the
peak of the disease, significant increases in concentrations
of multiple metabolites are observed, including glutamine,
O-phosphoethanolamine, branched-chain amino acids and
putrescine. Observed changes in metabolite levels suggest
profound changes in CNS metabolism over the course of
EAE. Affected pathways include nitric oxide synthesis,
altered energy metabolism, polyamine synthesis and levels
of endogenous antioxidants.
EAE ? LC–MS ? GC–MS
Metabolomics ? CSF ? Multiple sclerosis ?
Multiple sclerosis (MScl) is a chronic demyelinating neu-
rodegenerative central nervous system (CNS) disorder of
autoimmune origin affecting over 1 million people world-
wide. Development of the disease includes emergence of
inflammatory lesions in white matter and dysfunction of
neural conductivity. This leads to impairment of sensory
and motor functions. Molecular processes associated with
the onset and progression of MScl and its etiology are still
Because of obvious ethical and safety constraints, access
to human brain and spinal cord system samples is very
limited. Therefore in-depth investigation of molecular
mechanisms of disorders within CNS is hampered. Bio-
chemistry of CNS activity of affected patients can be
monitored only indirectly, using cerebrospinal fluid (CSF)
Marek Noga and Adrie Dane contributed equally to this manuscript.
Electronic supplementary material
article (doi:10.1007/s11306-011-0306-3) contains supplementary
material, which is available to authorized users.
The online version of this
M. J. Noga (&) ? A. Dane ? S. Shi ? T. H. Reijmers ?
R. J. Vreeken ? T. Hankemeier
Leiden/Amsterdam Center for Drug Research, Leiden
University, P.O. Box 9502, 2300, RA, Leiden, The Netherlands
A. Attali ? H. van Aken ? E. Suidgeest ? T. Tuinstra
Abbott Healthcare Products B.V, Weesp, The Netherlands
B. Muilwijk ? L. Coulier
TNO Zeist, Zeist, The Netherlands
A. Dane ? S. Shi ? L. Coulier ? T. H. Reijmers ?
R. J. Vreeken ? T. Hankemeier
Netherlands Metabolomics Centre, Leiden/Amsterdam Center
for Drug Research, Leiden University, Leiden, The Netherlands
Erasmus Medical Center, Rotterdam, The Netherlands
Metabolomics (2012) 8:253–263
for diagnostic purposes. Also explorative studies of
molecular mechanisms involved in the onset and progres-
sion of neurological diseases except for brain tumors are
limited to CSF samples. The metabolites present in the
CSF represent actual metabolism of CNS and the balance
between blood and CSF. Its analysis can be helpful in the
determination of markers for neurological disorders.
However, sampling human CSF is an invasive procedure,
hence access to such samples is much more limited com-
pared to blood or urine. In order to partially avoid these
limitations, animal models of neurological diseases have
been introduced. An additional major advantage of animal
models is less variation between individual animals com-
pared to variation between individuals in clinical cohorts.
Therefore the discovery of dominating factors of the onset
and progression of the disease is more straightforward. On
the other hand, biomarkers discovered by means of animal
models may not be relevant for the human situation and
such results need to be validated and confirmed using
The most commonly used animal model of MScl is the
experimental autoimmune encephalomyelitis (EAE). The
course of EAE resembles the symptoms of MScl and can
be used to study some aspects of MScl interconnected with
neuroinflammation and neurodegeneration. In this model,
disease is induced by inoculating animal with myelin-
specific antigens. This paper describes a study investigating
changes in metabolites caused by acute EAE; an autoim-
mune reaction was provoked using MBP co-injected with
an adjuvant containing Mycobacterium tuberculosis pro-
teins. In this model, gradual and reversible impairment of
motor function is observed, however no clear myelin loss is
Metabolomics, the comprehensive analysis of a wide
range of metabolites, provides a novel perspective for the
search of new disease biomarkers and drug targets, being
an alternative and complementary approach to more
established omics techniques such as genomics, transcri-
ptomics or proteomics. Recent progress in metabolomics
resulted in an increasing number of metabolomics appli-
cations in neurological research. In a recent review,
Wishart et al. (2008) summarized the current status of CSF
metabolomics reporting 308 metabolites together with their
concentrations in CSF. Recent articles investigating CSF
metabolome focused on untargeted approaches (Crews
et al. 2009; Myint et al. 2009; Carrasco-Pancorbo et al.
2009; Wuolikainen et al. 2009) detecting high number of
features. In this article we used a related approach
emphasizing targeted metabolomics. We applied two
platforms, allowing the fully validated analysis of 39
identified metabolites by LC–MS and 64 identified
metabolites by GC–MS, with an overlap of 18 metabolites
between both methods.
2 Materials and methods
2.1 Induction of acute EAE in the Lewis rat
Male Lewis rats (Harlan Laboratories B.V., the Nether-
lands) kept under normal housing conditions with water
and food ad libitum, weighing between 175 and 225 g at
the start of the experiment, were inoculated on day 0 as
previously described (Hendriks et al. 2004). Briefly, a
100 ll saline based emulsion containing 50 ll Complete
Freund’s Adjuvant H37 RA (CFA, Difco Laboratories,
Detroit, MI), 500 ll Mycobacterium tuberculosis type
H37RA (Difco) and 20 lg guinea pig myelin basic protein
(MBP) was injected subcutaneously in the pad of left hind
paw of Isoflurane anaesthetized animals. Next to these
MBP challenged rats, referred to as the EAE group, two
control groups were included: a group of animals receiving
the same emulsion without MBP (CFA group) and a
healthy group undergoing anesthesia only (Healthy group,
H). Each group consisted of 30 animals. Of each group,
half of the animals were sacrificed to collect plasma and
CSF on day 10 (day of onset of disease in EAE group)
resulting in groups further referred as H10, CFA10 and
EAE10, and the other half on day 14 (peak of disease in
EAE group)—groups H14, CFA14and EAE14.
Animals were grouped and housed three per cage and
cages were randomized across treatments and disease
duration. Disease symptoms and weights of all animals
were recorded daily. The following scores for motor dys-
functions were used: 0, healthy animal with normal curling
reflex at the tail; 1, paralysis of the tip of the tail; 2, loss of
muscle tone at the base of the tail; 3, low posture of hind
limbs; 4, instability at hips; 5, partial hind limb paralysis; 6,
complete hind limb paralysis; 7, paralysis include midriff;
8, quadriplegia; 9, moribund; 10, death due to EAE. The
animal experiments described were approved by the local
Ethical Committee for Animal Experiments.
2.2 CSF sampling
On day 10 and 14, animals were euthanized with CO2/O2
and the head of the rat was fixed in a holder. Terminal CSF
samples were obtained by direct insertion of an insulin
syringe needle (Myjector, 29G 9 1/200) via the arachnoid
membrane into the Cisterna Magna. For this purpose a skin
incision was made followed by a horizontal incision in the
musculus trapezius pars descendens to reveal the Arach-
noid membrane. A maximal volume of 60 ll was collected
per animal. Each sample was centrifuged within 20 min
after sampling, for 10 min at 20009g at 4?C. After cen-
trifugation, the supernatants were stored at -80?C for
further analysis. Previous experiments have shown that
collecting up to 60 ll using this technique and conditions
254 M. J. Noga et al.
provided hemoglobin-free CSF samples measured by ESI-
Orbitrap (data not shown). As an additional check, fresh
samples, supernatant and pellet size were visually scored
for hemolysis and samples were discarded if positive (data
not shown). After inspection, samples were aliquoted in
volumes of 10 ll for analysis by different analytical
Due to the disturbed physiological state of animals of
full EAE group at day 14 (EAE14), the success rate for CSF
sampling for these animals was significantly lower, limit-
ing the number of available samples. From 90 animals used
in the experiment, 84 CSF samples were available for
the LC–MS platform (11 in group EAE14) and 84 for the
GC–MS platform (10 in group EAE14).
2.3 LC–MS/MS method
The LC–MS method employed is targeted to the analysis of
primary and secondary amines using AccQ-Tag derivati-
zation. 10 ll of CSF sample was spiked with 10 ll of an
internal standard solution containing13C15N-labeled amino
acids, followed by addition of 100 ll of MeOH for
deproteination. The mixture was vortexed for 10 s and
centrifuged at 94009g for 10 min at 10?C. The supernatant
was transferred to a deactivated autosampler vial (Waters)
and dried under N2. The residue was reconstituted in 80 ll
of borate buffer (pH 8.5) vortexed for 10 s and treated with
20 ll of AccQ-TagAQC derivatization reagent (Waters).
After keeping the mixture for 10 min at 55?C, the vials
were transferred to an autosampler tray and cooled to 10?C
until the injection. 1.0 ll of the reaction mixture was
injected into the UPLC–MS/MS system.
An ACQUITY UPLC system with autosampler (Waters,
Etten-Leur, The Netherlands) was coupled online with a
Quattro Premier XE tandem quadrupole mass spectrometer
(Waters) operated using Masslynx data acquisition software
(version 4.1; Waters). The samples were analyzed by
UPLC–MS/MS using an AccQ-Tag Ultra 100 mm 9
2.1 mm (1.7 lm particle size) column (Waters). A binary
gradient system of water–eluent A (10:1, v/v) (AccQ-Tag,
Waters) and 100% eluent B (AccQ-Tag, Waters), was used.
Elution of the analytes was achieved by ramping the per-
a combination of both linear and convex profiles. The flow-
rate was 0.7 ml/min and the column temperature was
maintained at 60?C. After each injection, the injection nee-
dle was washed with 200 ll 95:5 v/v ACN:water and sub-
sequently with 600 ll 5:95 v/v ACN:water.
The Quattro Premier XE was used in the positive-ion
electrospray mode and all analytes were monitored in
Selective Reaction Monitoring (SRM) using nominal mass
resolution; the width of the isolation/fragmentation win-
dow was set to 0.7 Da. All target compounds were
monitored via the transition from the protonated molecule
of the AccQ-Tag derivative to the common fragment at
m/z 171. Collision energy and collision gas (Ar) pressure
were 22 eV and 2.5 mbar, respectively. The identities of all
compounds were confirmed by high resolution MS and
identical retention times as authentic standards (data not
shown). The acquisition run was divided into six time
segments in order to decrease the number of simultaneous
SRM transitions. All samples were analyzed in duplicate,
and measurement order was randomized.
Acquired data were evaluated using Quanlynx software
(Waters), by integration of assigned SRM peaks and nor-
malization using proper internal standards. For analysis of
amino acids, their13C15N-labeled analogs were used. For
other amines—the closest-eluting internal standard was
employed. Resulting relative responses from duplicate
injections were averaged after visual inspection.
2.4 GC–MS method
The GC–MS platform used in this study requires 30 ll of
rat CSF for sufficient sensitivity and coverage. Because
only 10 ll CSF per rat was available, samples were pooled
per three. This resulted in 30 ll pooled rat CSF samples
and group sizes of 5 (groups H10, H14, CFA10, CFA14,
EAE10) and 3 (group EAE14, limited number of samples
available, see Section 2.2). The samples (30 ll) were
deproteinized by adding 250 ll methanol and subsequently
centrifuged for 10 min at 11800 rpm. The supernatant was
dried under N2followed by derivatization with methyl-N-
(trimethylsilyl)-trifluoroacetamide (MSTFA) in pyridine
similar to Koek et al. (2006). During the different steps in
the sample work-up, i.e. prior to deproteinization, deriva-
tization and injection, different (deuterated) internal stan-
dards were added at a level of approx. 20 ng/ll. The end
volume was 45 and 1 ll aliquots of the derivatized samples
were injected in splitless mode on a HP5-MS 30 m 9
0.25 mm 9 0.25 lm capillary column (Agilent Technolo-
gies, USA) using a temperature gradient from 70 to 320?C
at a rate of 5?C/min. GC–MS analysis was performed using
an Agilent 6890 gas chromatograph coupled to an Agilent
5973 mass selective detector (Agilent Technologies, USA).
Detection was carried out using MS detection in electron
impact mode and full scan monitoring mode (m/z 15–800).
The electron impact for the generation of ions was 70 eV.
Samples were analyzed in duplicate.
Data-pre-processing was carried out by composing tar-
get lists of peaks detected in the samples based on retention
time and mass spectra and these peaks were integrated for
all samples. All peak areas were subsequently normalized
using internal standards. The resulting target lists were
used for further statistical analysis. Identities were assigned
based on the presence of identical mass spectra in an
CSF metabolomics of rat EAE model 255
in-house database based on authentic standards (data not
3 Data analysis
The significance of MBP treatment for all metabolites was
assessed by linear modeling (Draper and Smith 1998).
Three models were fitted for each metabolite. The first
global model (Eq. 1) includes all data for each metabolite:
y ¼ b0þ b1weightday0þ b2CFA þ b3MBP
þ b4time þ b5CFA ? time þ b6MBP ? time
in which y = metabolite response; CFA = CFA injected
(1 for CFA and EAE groups, otherwise 0); MBP = MBP
injected (1 for EAE groups, otherwise 0); b0…b6= model
coefficients; time = number of days since start of the
experiment (10 or 14); weightday0= weight at day 0 (for
LC–MS, individual rat LC–MS; for GC–MS, average of
rats in pooled sample) [g].
To obtain a higher confidence, two additional single-day
models were fitted for the samples collected on day 10 and
day 14, respectively (Eq. 2):
ydayX¼ a0þ a1weightday0þ a2textCFA þ a3MBP
in which (other symbols are explained above) ydayX=
metabolite response at day X (X is 10 or 14) and
a0…a3= model coefficients.
For all three models, stepwise linear regression (Massart
et al. 1997) was used to remove non-significant terms. The
significance of the regression coefficients in the final
models was obtained from a t test (H0: ai= 0, H1: ai= 0
or H0: bi= 0, H1: bi= 0).
A significant EAE effect in an individual single day
model is present when there is a significant a3. However,
here a distinction between a time dependent and a time
independent effect cannot be made. For the combined
model, a time-dependent effect is given by a significant b6
and a time-independent effect is given by a significant b3.
We distinguish between three effects: early stage EAE,
full stage EAE and EAE progression. For an effect to be
significant it must be present in both the combined model
and at least one single day model, otherwise results are
considered to be inconclusive.
A significant early stage EAE effect is reported when a
significant a3in the day 10 model is observed together with
a significant b3and/or b6in the combined model. A sig-
nificant a3in the day 14 model together with a significant b3
and/or b6in the combined model indicates full stage EAE.
An EAE progression effect is reported when a significant b6
in the combined model is observed in correlation with a
significant a3in at least one of the single day models.
Fold changes are obtained by taking ratios between
average model calculated metabolite responses per treat-
ment group. The fold changes for the onset (EAE10/CFA10)
and peak (EAE14/CFA14) of the disease are calculated from
the respective single day model results (Eq. 2). Fold
changes for EAE progression (EAE14/EAE10) are obtained
from the global model results (Eq. 1).
Figure 1 shows the progression of animal growth (weight,
left axis) for all experimental groups. For animals from
EAE14,progression of EAE is also shown (neurological
scores, right axis). The weight change during the course of
the experiment is dependent on multiple factors and
therefore difficult to use as quantitative measure in the data
analysis. Nevertheless, it accounts for the overall well-
being of the animal. Because of the fast growth of animals
(4–6 g per day), the complex treatment and use of two
sampling time-points, growth of the animals may affect the
metabolic profile. In addition, after the onset of the disease,
animals from EAE group started losing weight and this
decrease is correlated with the increase of neurological
scores (see Fig. 1). As a consequence, the weight of the
animals from EAE14group was significantly lower than
from all the other groups. The exact cause for this effect is
not certain. It is possible that progressing paralysis
impaired the animals ability to feed themselves, however
animals were able to eat and drink while lying. The degree
to which weight loss influences the volume of CSF and
concentration of metabolites is hard to access.
Principal component analysis (PCA) was applied for
preliminary visual inspection of the metabolite profiles of
Fig. 1 Weight trends (left axis) and neurological scores (right axis)
during the course of the animal experiment. Standard deviation of
weights within each group was in range of 8–12 g, bars not shown for
sake of clarity
256M. J. Noga et al.
the CSF samples belonging to the different groups (Figs. 2, 3).
Prior to PCA, each metabolite was scaled to zero mean and
unit variance. Clear separation between the full-stage EAE
group (EAE14) and the other groups is observed for both
platforms. The GC–MS data (Fig. 3) also show partial
separation of early stage EAE samples (group EAE10). For
LC–MS (Fig. 2) this separation is not observed although
some separation can be seen in higher principal compo-
nents (see Figure S1 in supplementary material).
Table 1 summarizes the LC–MS linear modeling results.
The table also indicates overlapping GC–MS targets and
where the GC–MS linear modeling results confirm effect
directionality. Although directionality is not always con-
firmed by linear modeling, visual inspection of the GC–MS
data indicates agreement. There are no cases where LC–
MS and GC–MS effects have opposite direction. The
smaller groups in GC–MS as a result of sample pooling
restrict the ability to draw unambiguously reliable biolog-
ical conclusions based on that platform alone. Therefore,
GC–MS linear modeling results are supplied as supple-
Significant changes of metabolite levels detected by
linear models were verified by pair wise comparison using
double-sided t tests and visually inspected by boxplot per
molecule. Figure 4 below shows a boxplot presenting
concentration changes of putrescine—the metabolite that
showed biggest fold-change between the experimental
The design of the biological experiments, with three
treatments groups—healthy controls, a CFA group with
animals subjected to peripheral inflammation only and a
full EAE treatment, allowed for separation of the concen-
tration changes caused by EAE from possible variations
due to adjuvant injection, including variations due to ani-
mal growth or peripheral inflammation.
Several significant changes in concentrations of metab-
olites both at the onset and on the peak of the disease were
detected. Dissimilar patterns of changes observed on both
time points may suggest activation of different metabolic
pathways in the early and late stage of the disease. The
early-stage EAE samples, taken on day 10 post inoculation
resemble the moment of maximal CNS infiltration by blood
monocytes and T-cells, however, at this moment no neg-
ative signs on health status are observed and animals do not
have any neurological scores. Metabolite level changes in
the early stage of the disease may be caused by processes
related to the development of EAE and disruption of the
The full-stage EAE samples are taken on day 14, at the
peak of the disease, when the paralysis of the animals is
most severe, which is expressed in maximum neurological
scores (see Fig. 1). The EAE model employed in this study
involves spontaneous recovery of animals. In fact, due to
the slightly heterogeneous course of the disease, 2 out of 15
animals from the EAE14group had neurological scores
which had already decreased relative to the previous day.
Therefore, one can assume that some metabolites observed
as significantly regulated in this sample series might be
connected to pathological processes involved in EAE
progression or recovery. The observed weight loss of
EAE14animals poses a difficult challenge in interpretation
Fig. 2 PCA scores plot of LC–MS data. Filled symbols represent the
day 14 samples while the open symbols represent the day 10 samples.
Clear separation between samples from group with full stage EAE
(group F, enclosed in ellipse) is observed
Fig. 3 PCA scores plot of GC–MS data. Filled symbols represent the
day 14 samples while the open symbols represent the day 10 samples.
Ellipses show grouping of samples from EAE-affected animals—
early stage EAE (group E, partially separated as indicated by dashed
line) and full stage of EAE (group F, solid line)
CSF metabolomics of rat EAE model 257
of observed changes. It was shown in the literature that
weight loss and dietary restriction may cause changes in
metabolite profiles in body fluids (Bollard et al. 2005).
However, it is unknown how these changes affect CSF
metabolic profile. For that reason the brief interpretation of
biological relevance of observed changes presented below
is mostly limited to compounds involved in processes
previously associated with MScl and EAE.
Table 1 Statistically significant concentration ratios as a result of MBP treatment for compounds detected by LC–MS
Name HMDB IDPubchem CID EAE onsetPeak of EAE EAE progression
Cyclic GMPHMDB01314 24316
Homo-L-arginine HMDB00670 50302.6??
Homocysteine HMDB00742 8144236
L-Alpha-aminobutyric acidHMDB00452 10217858
L-Arginine HMDB0051710318557 0.82-
L-Aspartic acidHMDB00191 3351
HMDB00687 6106 0.87-
Taurinel HMDB002511123 1.1?
Single and double symbols denote respectively 95 and 99% significance levels of the observed EAE effects (- denotes decrease, ? denotes
EAE onset = EAE10/CFA10, Peak of EAE = EAE14/CFA14, EAE progression = EAE14/EAE10. See Section 3 for details
* Effect directionality confirmed by GC–MS
aCompounds that were also measured by GC–MS
258M. J. Noga et al.
Several metabolites targeted in this study and found
significantly up/down-regulated were previously associated
with molecular processes interconnected with development
of MScl and EAE. Glutamic acid (glutamate), glutamine,
gamma-aminobutyric acid (GABA), asparagine and taurine
are involved in excitotoxicity and energy metabolism.
Arginine and citrulline play a key role in NO synthesis.
Arginine, ornithine and putrescine are involved in poly-
amine synthesis. Alanine and branched-chain amino acids
(valine, leucine, isoleucine) are involved in energy
metabolism. See Table 2 for overview.
L-glutamic acid (glutamate) is a ubiquitous neurotransmit-
ter with several identified receptors and transport proteins
both in neural and glial cells (Danbolt 2001). Over-stim-
ulation of glutamate receptors may cause cell death due to
excitotocity. Indeed, elevated levels of CSF glutamate were
reported in clinical cases of multiple sclerosis (Sarchielli
et al. 2003). Contrary to previous reports, as shown in
Table 1, the level of glutamic acid in this study shows a
statistically significant decrease in concentration at the
peak of the disease, while no significant change is observed
at the early stage. However, the glutamine, which is closely
interconnected with glutamate through glutamine-gluta-
mate cycle in neurons and astrocytes, shows down-regu-
lation at the onset and up-regulation with full stage EAE.
Conversion of glutamate into glutamine by glutamine
synthetase (GS) enzyme localized in astrocytes is a key
(Norenberg and Martinez-Hernandez 1979) and the only
pathway of ammonia disposal due to the lack of complete
urea cycle in CNS (Wiesinger 2007). Available evidence
shows up-regulation of GS in astrocytes present in demy-
elinating lesions as assessed in a study of post mortem
brain tissue of MS patients (Newcombe et al. 2008).
However, a similar study performed in MBP-induced
mouse EAE shows down-regulation of this enzyme
(Hardin-Pouzet et al. 1997).
At the peak of the disease we observe a significant
gamma-aminobutyric acid (GABA) increase. GABA is
synthesized from glutamate by glutamate decarboxylase. It
was shown that GABA, an inhibitory neurotransmitter,
may attenuate glutamate-induced excitotoxicity (Ohkuma
et al. 1994). In our study we observe significant increase
in concentration of another inhibitory neurotransmitter,
taurine, which was also shown to be protective against
excitotoxicity (El Idrissi and Trenkner 1999).
It should also be emphasized that glutamate, besides
having a profound role as a neurotransmitter in CNS, is one
of the key metabolites in energy metabolism of neurons,
glia and immune cells (Newsholme et al. 1999). It is pos-
sible that the global CSF levels of glutamine and glutamate
involved in energy metabolism dominate over local high
and excitotoxic concentrations in the most affected regions
of brain and spinal cord. The observed discrepancy
between available results on glutamate up- and down-reg-
ulation in EAE/MS suggests it may be involved in multiple
different molecular processes associated with neuroin-
flammation and neurodegeneration.
5.2 Cytotoxicity interconnected with peroxynitrylation
Peroxynitrile anion (ONOO-) is a very reactive oxidizing
agent, capable of inducing cell death through multiple
pathways (Szabo ´ et al. 2007). It emerges in the reaction
between nitric oxide, produced mainly by inductive nitric
oxide synthetase (iNOS) in activated immune cells and
microglia, and the free radical superoxide O2-, coming
from oxidative phosphorylation cycle in mitochondria.
Fig. 4 Box plot showing concentration of putrescine in all experi-
mental groups. Note significant increase in concentration in EAE14
group. The outlying rat 76, with the highest level of putrescine,
started to recover during the last day prior to sampling. See text for
Table 2 Overview of most important biological processes intercon-
nected with significantly regulated changes in metabolome detected in
Molecular process Related significantly regulated compounds
Excitotoxicity Glutamic acid, glutamine, asparagine,
Peroxynitrilation Arginine, citrulline
Polyamine synthesisArginine, ornithine, putrescine
Oxidatative stress Taurine
Energy metabolism Glutamine, alanine, BCAA
(leucine, isoleucine, valine)
CSF metabolomics of rat EAE model 259
Currently available evidence suggests, besides promotion of
peroxynitrile synthesis, that NO plays a complex regulatory
role in mediating the immune response (van der Veen et al.
1997). Numerous reports suggest activity of peroxynitrile in
in EAE (van der Veen et al. 1997).
In our study we observe significant down-regulation of
arginine, the main substrate for NO synthesis during the
early stage of EAE. This effect may be interconnected with
intense production of NO by cytotoxic T-cells, macro-
phages and activated microglia during development of
neuroinflammation. However, the exact role of NO in EAE
and MScl development remains unclear. Available reports
show that inhibition of NO synthesis may attenuate (Ding
et al. 1998) or enhance (Zielasek et al. 1995) EAE.
Interestingly, a second product of NO synthesis, citrul-
line, is also significantly down-regulated in the early stage
of EAE, while one could expect increased production of
citrulline linked to increased NO synthesis. However,
immune and CNS cells are able to re-synthesize arginine in
the so-called arginine-NO cycle (see Fig. 5, solid line),
This cycle can be considered as a shortcut of the urea cycle
(Fig. 5, dashed line), which due to the lack of ornithine
carbamoyltransferase (OCT) is not complete in the CNS
(Wiesinger 2001). It was demonstrated that arginine defi-
ciency does not inhibit NO synthesis by T-cells, which
under such circumstances are able to increase uptake of
citrulline and up-regulate enzymes responsible for arginine
synthesis (Bansal et al. 2004). Due to the involvement of
arginine in other reactions, the citrulline–NO cycle does
not operate in a stoichiometric manner, resulting in
simultaneous decrease of arginine and citrulline levels.
It should be also noted that the most significant alter-
native reaction of arginine utilization is the synthesis of
ornithine by the enzyme arginase (ARG). This reaction is
known to take part in regulating NO synthesis by reducing
the amount of available arginine (Bansal and Ochoa 2003).
In the early stage of EAE we observe a decrease in the
concentration of ornithine, supporting the hypothesis that
arginine is probably being utilized by NO synthesis.
5.3 Polyamine synthesis
During the peak of EAE, arginine levels do not differ from
controls, while citrulline levels continue to decrease. This
may be explained by the reduction of NO synthesis inter-
connected with continued activity of other citrulline-NO
cycle enzymes. At the same time we observe up-regulation
of products of an alternative pathway of arginine utiliza-
tion. The ornithine levels return to their normal states after
significant decreases during the EAE onset, and this cor-
relates with the significant 20-fold increase of putrescine
levels, synthesised from ornithine by ornithine decarbox-
ylase (see Fig. 4). These results may suggest a switch in
arginine metabolism from nitric oxide production towards
the synthesis of ornithine and further to polyamine syn-
thesis, which could be associated with compensatory anti-
inflammatory response and recovery. Interestingly, the rat
with the highest observed level of putrescine is a clear
outlier from the rest of the group (animal 76 on Fig. 4) and
is one of two animals in the group that showed a neuro-
logical score decrease on the last day of experiment.
Putrescine was reported to be up-regulated in EAE
(Bolton and Paul 2006), however this change was associ-
ated with disease progression rather than recovery mech-
anisms. On the other hand, the beneficial effect of
increased polyamine synthesis on post-injury recovery of
CNS was demonstrated in literature (Gilad et al. 1996; Cai
et al. 2002). Available reports show the potential of
putrescine to block NMDA glutamate receptors (Williams
1997) and therefore could possibly attenuate excitotoxic
effects of glutamate.
It was postulated previously that over the course of
inflammation, metabolism of arginine switches from NO
synthesis in the early stage to arginine utilization in poly-
amine synthesis during tissue repair and wound healing
(Nelin et al. 2007).
Fig. 5 Overview of pathways of arginine metabolism in CNS.
Hexagons depict metabolites and squares—reactions. Compounds
measured in our study are marked in grey. Orinithine transcarbamo-
ylase (OTC), argininosuccinate synthetase (ASS), argininosuccinate
lyase (ASL) and arginase (ARG) are enzymes in the urea cycle
(dashed circle), however OTC (dashed) is not expressed in CNS.
Nitric oxide synthetase (NOS) is used for synthesis of nitric oxide and
part of arginine-NO cycle (solid ellipsis). Ornithine decarboxylase
(ODC) produces putrescine that is later used in synthesis of other
polyamines. This figure was generated using Cytoscape (www.
260M. J. Noga et al.
5.4 Protection from oxidative stress
It was shown that antioxidants, such as uric acid, inhibit
inflammation in EAE (Hooper et al. 2000). The inhibitory
neurotransmitter taurine, observed as up-regulated in our
study, is also known for its antioxidant and anti-inflam-
matory activity (Li et al. 2007). Levels of taurine increase
over the course of EAE, suggesting that they may play a
role in protection and recovery mechanisms. As the
neurotoxicity of peroxynitrile might be dependent on the
availability of superoxide rather than nitric oxide alone, it
was suggested that MScl may be in fact connected with a
malfunction of mitochondria (Mao and Reddy 2010)
or imbalance of ATP synthesis and energy metabolism
(Cadoux-Hudson et al. 1991; Amorini et al. 2009).
5.5 Energy metabolism
Besides changes discussed above, we also observe other
changes in the metabolome that cannot be clearly con-
nected to already established hypotheses of MScl/EAE
mechanisms. In the early stage of the EAE we observe
decreased levels of alanine and branched-chain amino-
acids (BCAA: leucine, isoleucine and valine). Alanine and
BCAAs are known to be utilized as a source for pyruvate
for energy metabolism or de novo synthesis of macro-
molecules within neural (Hutson et al. 2007) and immune
cells (Li et al. 2007). As the onset of EAE is associated
with maximum infiltration of the CNS by blood monocytes
and T-cells, the observed decrease of BCAAs and alanine
may suggest these metabolites are utilized for energy
metabolism by invading cells. Alternatively, pro-inflam-
matory cytokines released during EAE development may
cause changes in energy metabolism of native CNS cells.
5.6 Phospholipid metabolism
In the full stage EAE, among other changes, we observe a
significant increase of O-phosphoethanolamine concentra-
tion, which is an important intermediate in phospholipid
metabolism. Wheeler et al. (2008) reported changes in lipid
contents in normal appearing white and grey matter of
MScl patients, suggesting changes in lipid metabolism
including increased sphingolipid turnover and phospholipid
synthesis. The detected increase of O-phosphoethanol-
amine might be interconnected with these processes.
6 Conclusions and future prospects
A targeted metabolomic approach of metabolites contain-
ing amino groups was applied to study changes in CNS
metabolism in a MScl rat model. The obtained results
showed significant changes in metabolic processes during
the onset and progression of EAE. Excessive nitric oxide
production along with changes in energy metabolism
appear to play a key role in the development of the EAE,
while a switch from arginine metabolism to polyamine
synthesis may be associated with processes involved in
spontaneous recovery of affected animals. Further research
is required to confirm these hypotheses.
Results obtained in this study are partially in agreement
with established knowledge on EAE and multiple sclerosis.
However, in some cases we observe changes conflicting
with previously published results. It should be noted that the
vast majority of existing results were obtained in different
experiments, executed in different paradigms, investigating
similar, however still distinct biological conditions. Estab-
lished EAE models differ from each other and still they do
not cover all aspects of multiple sclerosis (Lassmann 2007).
The ability to simultaneously measure a wide range of
compounds is advantageous in systems biology approaches,
like the one presented here. The use of a common experi-
ment to probe multiple pathways in a biological system
yields a chance for obtaining more comprehensive and
coherent picture of molecular mechanisms involved.
Our results prove that metabolomics provides useful
insights into the mechanisms underlying biological pro-
cesses in the onset and development of multiple sclerosis
using a rat model. High sensitivity and a wide range of
metabolites were covered by the present analytical plat-
forms, which allowed for a reliable screening of potential
changes in CNS metabolism. In contrast to untargeted
approaches, our LC/GC–MS analysis focuses on the mea-
surement of well defined and confidently identified com-
ponents. This allowed a more accurate description of the
potential molecular mechanisms involved.
der Kloet (LACDR) and Siem Heisterkamp (MSD) for batch design
for LC–MS and GC–MS experiments, respectively. We also thank
Amy Harms and Vanessa Gonzalez (both LACDR) for reading and
commenting on this manuscript. This study was financially supported
by Top Institute Pharma project D4-102. This study was supported by
the research program of the Netherlands Metabolomics Centre
(NMC), which is a part of The Netherlands Genomics Initiative/
Netherlands Organization for Scientific Research.
Authors would like to acknowledge Frans van
Creative Commons Attribution Noncommercial License which per-
mits any noncommercial use, distribution, and reproduction in any
medium, provided the original author(s) and source are credited.
This article is distributed under the terms of the
Amorini, A. M., Petzold, A., Tavazzi, B., et al. (2009). Increase of
uric acid and purine compounds in biological fluids of multiple
sclerosis patients. Clinical Biochemistry, 42, 1001–1006.
CSF metabolomics of rat EAE model261
Bansal, V., & Ochoa, J. B. (2003). Arginine availability, arginase, and
the immune response. Current Opinion in Clinical Nutrition and
Metabolic Care, 6, 223–228.
Bansal, V., Rodriguez, P., Wu, G., et al. (2004). Citrulline can
preserve proliferation and prevent the loss of CD3 zeta chain
under conditions of low arginine. Journal of Parenteral and
Enteral Nutrition, 28, 423–430.
Bollard, M. E., Stanley, E. G., Lindon, J. C., Nicholson, J. K., &
Holmes, E. (2005). NMR-based metabonomic approaches for
evaluating physiological influences on biofluid composition.
NMR in Biomedicine, 18, 143–162.
Bolton, C., & Paul, C. (2006). Glutamate receptors in neuroinflam-
matory demyelinating disease. Mediators of Inflammation, 2006,
Cadoux-Hudson, T. A., Kermode, A., Rajagopalan, B., et al. (1991).
Biochemical changes within a multiple sclerosis plaque in vivo.
Cai, D., Deng, K., Mellado, W., Lee, J., Ratan, R. R., & Filbin, M. T.
(2002). Arginase I and polyamines act downstream from cyclic
AMP in overcoming inhibition of axonal growth MAG and
myelin in vitro. Neuron, 35, 711–719.
Carrasco-Pancorbo, A., Nevedomskaya, E., Arthen-Engeland, T.,
et al. (2009). Gas chromatography/atmospheric pressure chem-
ical ionization-time of flight mass spectrometry: analytical
validation and applicability to metabolic profiling. Analytical
Chemistry, 81, 10071–10079.
Crews, B., Wikoff, W. R., Patti, G. J., et al. (2009). Variability
analysis of human plasma and cerebral spinal fluid reveals
statistical significance of changes in mass spectrometry-based
metabolomics data. Analytical Chemistry, 81, 8538–8544.
Danbolt, N. C. (2001). Glutamate uptake. Progress in Neurobiology,
Ding, M., Zhang, M., Wong, J. L., Rogers, N. E., Ignarro, L. J., &
Voskuhl, R. R. (1998). Antisense knockdown of inducible nitric
oxide synthase inhibits induction of experimental autoimmune
encephalomyelitis in SJL/J mice. Journal of Immunology, 160,
Draper, N. R., & Smith, H. (1998). Applied regression analysis. New
York: John Wiley & Sons.
against excitotoxicity by stabilizing calcium homeostasis and
energy metabolism. Journal of Neuroscience, 19, 9459–9468.
Gilad, V. H., Tetzlaff, W. G., Rabey, J. M., & Gilad, G. M. (1996).
Accelerated recovery following polyamines and aminoguanidine
treatment after facial nerve injury in rats. Brain Research, 724,
Hardin-Pouzet, H., Krakowski, M., Bourbonnie `re, L., Didier-Bazes,
M., Tran, E., & Owens, T. (1997). Glutamate metabolism is
down-regulated in astrocytes during experimental allergic
encephalomyelitis. Glia, 20, 79–85.
Hendriks, J. J., Alblas, J., van der Pol, S. M., van Tol, E. A., Dijkstra,
C. D., & de Vries, H. E. (2004). Flavonoids influence monocytic
GTPase activity and are protective in experimental allergic
encephalitis. Journal of Experimental Medicine, 200, 1667–1672.
Hooper, D. C., Scott, G. S., Zborek, A., et al. (2000). Uric acid, a
peroxynitrite scavenger, inhibits CNS inflammation, blood-CNS
barrier permeability changes, and tissue damage in a mouse
model of multiple sclerosis. FASEB J, 14, 691–698.
Hutson, S. M., Sweatt, A. J., & LaNoue, K. F. (2007). Branched Chain
Amino Acids (BCAAs) in Brain. In S. S. Oja, P. Saransaari, & A.
Schousboe (Eds.), Handbook of neurochemistry and molecular
neurobiology: amino acids and peptides in the nervous system
(pp. 117–131). Springer-Verlag: Berlin Heidelberg.
Koek, M. M., Muilwijk, B., van der Werf, M. J., & Hankemeier, T.
(2006). Microbial metabolomics with gas chromatography/mass
spectrometry. Analytical Chemistry, 78, 1272–1281.
Lassmann, H. (2007). Multiple Sclerosis and Autoimmune Enceph-
alomyelitis. In A. Lajtha, M. B. H. Youdim, P. Riederer, S.
A. Mandel, L. Battistin, & H. Lassmann (Eds.), Handbook of
neurochemistry and molecular neurobiology (pp. 373–404). US:
Li, P., Yin, Y. L., Li, D., Kim, S. W., & Wu, G. (2007). Amino acids
and immune function. British Journal of Nutrition, 98, 237–252.
Mao, P., & Reddy, P. H. (2010). Is multiple sclerosis a mitochondrial
disease? Biochimica et Biophysica Acta, 1802, 66–79.
Massart, D. L., Vandeginste, B. M. G., Buydens, L. M. C., de Jong,
S., Lewi, P. J., & Smeyers-Verbeke, J. (1997). Multiple and
Polynomial Regression. Handbook of chemometrics and quali-
metrics Part A (pp. 263–303). Amsterdam: Elsevier.
Quantitative profiling of polar cationic metabolites in human
mass spectrometry. Analytical Chemistry, 81, 1121–1129.
Nelin, L. D., Wang, X., Zhao, Q., et al. (2007). MKP-1 switches
arginine metabolism from nitric oxide synthase to arginase
following endotoxin challenge. American Journal of Physiology
Cell Physiology, 293, C632–C640.
Newcombe, J., Uddin, A., Dove, R., et al. (2008). Glutamate receptor
expression in multiple sclerosis lesions. Brain Pathology, 18,
Newsholme, P., Curi, R., Pithon Curi, T. C., Murphy, C. J., Garcia,
C., & Pires de Melo, M. (1999). Glutamine metabolism by
lymphocytes, macrophages, and neutrophils: its importance in
health and disease. Journal of Nutritional Biochemistry, 10,
Norenberg, M. D., & Martinez-Hernandez, A. (1979). Fine structural
localization of glutamine synthetase in astrocytes of rat brain.
Brain Research, 161, 303–310.
Ohkuma, S., Chen, S. H., Katsura, M., Chen, D. Z., & Kuriyama, K.
(1994). Muscimol prevents neuronal injury induced by NMDA.
Japanese Journal of Pharmacology, 64, 125–128.
Sarchielli, P., Greco, L., Floridi, A., & Gallai, V. (2003). Excitatory
amino acids and multiple sclerosis: Evidence from cerebrospinal
fluid. Archives of Neurology, 60, 1082–1088.
Smith, K. J., & Lassmann, H. (2002). The role of nitric oxide in
multiple sclerosis. Lancet Neurology, 1, 232–241.
Szabo ´, C., Ischiropoulos, H., & Radi, R. (2007). Peroxynitrite:
Biochemistry, pathophysiology and development of therapeutics.
Nature Reviews Drug Discovery, 6, 662–680.
van der Veen, R. C., Hinton, D. R., Incardonna, F., & Hofman, F. M.
(1997). Extensive peroxynitrite activity during progressive
stages of central nervous system inflammation. Journal of
Neuroimmunology, 77, 1–7.
Wheeler, D., Bandaru, V. V. R., Calabresi, P. A., Nath, A., &
Haughey, N. J. (2008). A defect of sphingolipid metabolism
modifies the properties of normal appearing white matter in
multiple sclerosis. Brain, 131, 3092–3102.
Wiesinger, H. (2001). Arginine metabolism and the synthesis of nitric
oxide in the nervous system. Progress in Neurobiology, 64,
Wiesinger, H. (2007). Arginine, Citrulline, and Ornithine. In
S. S. Oja, P. Saransaari, & A. Schousboe (Eds.), Handbook of
neurochemistry and molecular neurobiology: Amino Acids and
peptides in the nervous system (pp. 99–115). Berlin Heiderberg:
Williams, K. (1997). Modulation and block of ion channels: a new
biology of polyamines. Cellular Signalling, 9, 1–13.
Wishart, D. S., Lewis, M. J., Morrissey, J. A., et al. (2008). The
human cerebrospinal fluid metabolome. Journal of Chromatog-
raphy B, 871, 164–173.
Wuolikainen, A., Hedenstro ¨m, M., Moritz, T., Marklund, S. L., Antti, H.,
& Andersen, P. M. (2009). Optimization of procedures for
262M. J. Noga et al.
collecting and storing of CSF for studying the metabolome in Download full-text
ALS. Amyotrophic Lateral Sclerosis, 10, 229–236.
Zielasek, J., Jung, S., Gold, R., Liew, F. Y., Toyka, K. V., & Hartung,
H. P. (1995). Administration of nitric oxide synthase inhibitors
in experimental autoimmune neuritis and experimental autoim-
mune encephalomyelitis. Journal of Neuroimmunology, 58,
CSF metabolomics of rat EAE model263